Intrapulmonary Delivery of CpG Microparticles Eliminates Lung Tumors

Lung cancer is the leading cause of cancer-related death in the United States, with non–small cell lung cancer (NSCLC) accounting for 85% of these tumors (1). Despite recent advances in multimodal targeted and tailored therapies, the 5-year survival rate for NSCLC remains poor (∼15%). Accumulating evidence suggests that controlling the primary tumor can enhance survival even in patients with advanced/metastatic disease (refs. 2, 3; NCT01242800, http://clinicaltrials.gov/show/NCT01242800). As most lung tumors are poorly immunogenic and resistant to immune surveillance, novel approaches to activating the pulmonary immune system may be key to promoting tumor regression (4, 5).

One strategy to trigger the innate immune system involves activating cells that express TLRs. Our group and others find that stimulating cells bearing TLR9 using synthetic oligonucleotides expressing unmethylated CpG motifs (CpG ODN) generates substantial antitumor activity (reviewed in ref. 6, 7). For example, treatment with CpG ODN in combination with chemotherapy significantly prolonged survival in a murine Lewis lung cancer model (8). On the basis of considerable preclinical data, clinical trials assessing the safety and efficacy of systemically administered CpG ODN were conducted in patients with advanced NSCLC (9, 10). While favorable results were observed in phase II studies, no survival benefit was observed in a phase III trial of CpG ODN (11, 12). Yet the objective response rate of patients who received CpG ODN exceeded that of patients receiving chemotherapy alone in those studies (28%–30% vs. 19%–23%). A possible explanation for these inconsistent clinical findings could be that systemically administered CpG ODN either did not reach or did not persist in many pulmonary tumors. In this context, recent studies demonstrate that intratumoral delivery optimizes the therapeutic utility of CpG ODN. While systemic administration promotes the development of cancer-specific CTLs, the lytic activity of such cells is inhibited by immunosuppressive cells in the tumor microenvironment. Local (but not systemic) delivery of CpG ODN has been shown to reverse this local immunosuppression (13).

Given the limited success of systemic CpG therapy for the treatment of lung cancer, we examined whether intrapulmonary delivery might be more efficacious. Previous studies demonstrated that immunomodulatory ODN could be safely delivered to mice via intratracheal instillation (mimicking inhalational therapy in humans). To improve tumor targeting and pulmonary retention, we adsorbed the CpG ODN onto polyketal microparticles previously shown to be safe and efficient for inhalational administration of therapeutic agents (14–17). Polyketal nanoparticles are biodegradable and can be formulated into microparticles optimized for delivery to distal bronchi (18, 19). When CpG-MP were delivered intratracheally to mice with Lewis lung cancer (a model of human NSCLC), they accumulated and persisted in the tumor microenvironment. This treatment promoted intratumoral immunity while reducing immunosuppression, culminating in a significant reduction in tumor burden and significantly improved survival.

A poly-(1,4-phenyleneacetone dimethylene ketal) matrix of nanoparticles was synthesized by Celagix Res Ltd. Polyvinylalcohol (PVA, Sigma) was used as the dispersing agent. Polyketal nanoparticles were produced via an acetal exchange reaction as previously described (18, 20). Briefly, 2.5 gm of polyketal nanoparticles were synthesized by dissolving 10 gm of 1,4-benzenedimethanol (36 mmol) and 8.6 gm of 1,5-pentanediol (8.67 mmol) in ethyl acetate and adding this to 100 mL of methylbenzene. Recrystallized p-toluenesulfonic acid (520 mg, 2.93 mmol) in 1.0 mL of ethyl acetate was then added followed by 2,2-dimethylpropane (9 mL, 37 mmol) to initiate the reaction. Additional 2,2-dimethylpropane (9 mL, 37 mmol) and methylbenzene was added to the reaction every 30 minutes for 2 hours. After 12 hours, the polymer was isolated by precipitation in ice-cold hexane (−20°C) followed by vacuum filtration.

The phosphorothioate ODN used in this study were CpG 1555 (GCTAGACGTTAGCGT) and control ODN 1612 (GCTAGAGCTTAGCGT). These were synthesized by IDT Technologies. ODNs were adsorbed onto polyketal nanoparticles using the water-in-oil-in-water double emulsion solvent evaporation method (21). ODN dissolved in water was added to DOTAP solution (Avanti) at a 1:1 (w/w) ratio with stirring. The ODN/DOTAP complexes were added to a 0.5% w/v PVA solution in 10 mmol/L sodium bicarbonate and then mixed with a solution of polyketal (500 mg) dissolved in 5 mL of dichloromethane. The mixture was emulsified by sonication (Sonifier 250, Branson, duty cycle 80%, output 5) for 3 minutes. The resulting emulsion was added to 125 mL of a 1% PVA solution and agitated with stirring at 500 rpm for 6 hours to remove dichloromethane. The ODN-loaded polyketal nanoparticles were collected by centrifugation at 4,500 rpm for 10 minutes, washed with 10 mmol/L sodium bicarbonate, and then freeze-dried to form microparticles. All materials in this study were endotoxin free as determined by Limulus amebocyte cell lysate assay (Cambrex Bioscience; sensitivity 0.1 U endotoxin/mg).

The release of CpG ODN from CpG-MP dissolved in physiologic saline (pH 7.4, 37°C) was examined. Supernatants were collected over time, centrifuged at 4,500 rpm for 10 minutes, and loaded onto a 3% agarose gel. Gel electrophoresis was performed at 100 V for 0.5 hours and the ODN visualized by SYBR staining (Invitrogen) in comparison to serial dilutions of free ODN. Loading efficiency (%) was calculated by the formula: (maximal release of ODN from CpG-MP)/(weight of ODN used in synthetic process) × 100%.

CpG-MP samples were suspended in physiologic saline, transferred onto carbon tape, and dried. They were then coated with a 30 nm layer of osmium using an osmium plasma coater (NL-OPC80NS, Nippon Laser & Electronics Laboratory), and visualized by scanning electron microscopy (JSM-6340F) at an acceleration voltage of 5.0 kV.

CpG-MP samples were also suspended in 70% ethanol, absorbed onto a 400 mesh formvar film-coated grid, and then visualized by transmission electron microscopy (JEM-1200EX; JEOL Ltd.) at an acceleration voltage of 80 kV. Digital images were taken with a CCD camera (VELETA, Olympus Soft Imaging Solutions GmbH).

C57BL/6, TLR9 KO, and Rag1 KO mice were obtained from the NCI (Frederick, MD) and studied at 5 to 6 weeks of age. Lewis lung carcinoma (LLC1) cells were obtained from ATCC in 2012. All experiments with this cell line were performed within 2 months of thawing the cryopreserved cells that were expanded in RPMI-1640 medium supplemented with 2% FBS, penicillin (100 U/mL), streptomycin (0.1 mg/mL), 2 mmol/L glutamate, and 1% NEAA. On the day of inoculation, cultured cells were trypsinized, washed, and suspended in 0.9% saline. Their viability exceeded 95%.

Tumor challenge studies were performed by instilling 10⁶ LLC cells in 50 μL of saline via orotracheal intubation to anesthetized mice as previously described (22, 23). One week later, free ODN or ODN-MP were administered either systemically (intraperitoneal) or locally (intratracheal). Instillation was achieved via orotracheal intubation using a 20 gauge 1″ catheter (TERUMO) under anesthesia. Four to 6 mice per group were used in each experiment and all results derived by combining data from 2 to 4 independent experiments. All animals were monitored 3 times per week and moribund mice euthanized as per Institutional Animal Care and Use Committee protocol.

Bronchoalveolar lavage (BAL) fluid was obtained by tracheal cannulation of anesthetized mice. Cell differentials were performed on cytocentrifuged BAL preparations after fixation and staining with Diff-Quick (Dade Behring).

IL12 levels in BAL and serum were determined by ELISA. Lungs were inflated and fixed by instilling 1 mL of 10% neutral-buffered formalin or periodate lysine paraformaldehyde (PLP) fixative (Wako Chemicals USA, Inc.) at 20 cm H₂O. Fixed tissue was embedded in paraffin, sectioned, and stained with hematoxylin and eosin for histopathologic assessment. Tumor area was evaluated in midline sections and quantified using ImageJ software ver.1.48 (NIH, Bethesda, MD).

Single spleen cell suspensions were prepared. A total of 5 × 10⁴ cells per well were cultured in 96-well flat bottomed microtiter plates in complete RPMI1640 medium supplemented with 10% fetal calf serum, 100 U/mL penicillin, 100 mg/mL streptomycin, 25 nmol/L HEPES, 1 mmol/L sodium pyruvate, NEAA, and 0.0035% 2-ME. The cells were stimulated with free ODN, ODN-MP, or R837 (TLR7 agonist) for 72 hours. Cell proliferation was assessed using the CCK-8 assay as per the manufacturer's recommendation (Dojindo Molecular Technologies, Inc.).

Tissue sections were deparaffinized with xylene and rehydrated with graded ethanol. Endogenous peroxidase activity was blocked by incubation with 0.3% H₂O₂ in methanol for 30 minutes followed by Protein Block (Dako). They were then stained with the following primary antibodies: rat anti-Foxp3 (eBioscience; clone FKJ-16s, dilution 1:200), rat anti-F4/80 (AbD Serotec; clone Cl:A3-1, dilution 1:300), rabbit anti-CD3 (LSBio; clone EPR4517, dilution 1:100), rat anti-CD8 (LSBio; clone CT-CD8a, dilution 1:100), hamster anti-CD11c (Abcam; clone N418, dilution 1:100), rat anti-CD45R/B220 (BD Pharmingen; clone RA3-6B2, dilution 1:20), rabbit anti-CD205 (Abcam; clone EPR5233, dilution 1:500), or rabbit anti-CD206 (Abcam; dilution 1:2,000). Isotype matched negative controls were included to insure specificity.

For immunohistochemistry, stained sections were incubated with HRP-conjugated anti-rat or rabbit IgG (simple stain mouse Max Po; Nichirei) plus 3-amino-9-ethyl carbazole (AEC) substrate for color development (Nichirei). The sections were counterstained with Mayer hematoxylin (Dako) and images were obtained using an IX50 inverted microscope equipped with a digital imaging system (Olympus). For immunofluorescent analysis, stained sections were incubated with secondary goat anti-rat antibody coupled to Texas Red (Vector Laboratories Inc.), rabbit anti-hamster coupled to DyLight 647 (Abcam), or goat anti- rabbit coupled to DyLight 549 (Vector). Cell nuclei were visualized with DAPI (Vector). Slides were mounted with VECTASHIELD (Vector). Fluorescent images were acquired with a laser-scanning confocal microscope (LSM 510; Zeiss), equipped with a 63×/1.4NA oil objective, or by using fluorescence microscope BZ-9000 and BZ-X700 (Keyence). Data were derived by counting the number of cells in 20 independent images per tumor group and expressed as cells/mm². M1 and M2 macrophages were identified on the basis of their expression of F4/80⁺CD206⁻ and F4/80⁺CD206⁺, respectively.

Formalin-fixed, paraffin-embedded sections were deparaffinized and processed using the ApopTag Peroxidase or the Fluorescence In Situ Apoptosis Detection Kit (EMD Millipore Corporation). DNA fragments were labeled with digoxigenin nucleotide followed by incubation with a peroxidase-conjugated antidigoxigenin antibody. Positive signals of peroxidase assay were visualized by AEC (Nichirei).

Cytokine levels in BAL, serum, and culture supernatants were measured by ELISA as previously described (24). Briefly, paired IL6 and IL12-specific monoclonal antibodies were purchased from BD Pharmingen. Ninety-six–well Immulon H2B plates (Thermo LabSystems) were coated with capture cytokine-specific antibodies and then blocked with PBS/1% BSA. Samples were added and bound cytokine detected by the addition of biotin-labeled secondary antibody, followed by phosphatase-conjugated avidin and a phosphatase-specific colorimetric substrate. Standard curves were generated using recombinant cytokines purchased from R&D Systems.

Statistical analyses were performed using MedCalc, version 13.0 (MedCalc Software). Differences in survival were determined using the log-rank test of Kaplan–Meier. Differences between groups were assessed using a one-way ANOVA followed by the Student–Newman–Keuls post hoc test. All tests were two-sided; P values < 0.05 were considered significant. All values are expressed as mean ± SE.

Breeding and experiments were reviewed and approved by the Institutional Animal Care and Use Committee of the NCI (Frederick, MD).

A well-established murine model of NSCLC was used to evaluate the antitumor activity of locally versus systemically administered CpG ODN. A total of 10⁶ LLC cells were instilled into the lungs of syngeneic C57BL/6 mice. Consistent with previous reports, these cells proliferated and formed peribronchial tumors resembling those present in patients with NSCLC (Supplementary Fig. S1; refs. 22, 25).

Mice challenged with LLC had a median survival time of 22 days (Fig. 1). Systemic (intraperitoneal) delivery of CpG ODN conferred a modest but statistically insignificant survival benefit [HR: 0.55; 95% confidence interval (CI), 0.24–1.26; P = 0.11]. Median survival time improved to 38 days (HR: 0.28; 95% CI, 0.13–0.59; P = 0.01) when 50 μg of CpG ODN was delivered via the intratracheal (i.t.) route directly into the lungs (Fig. 1). This effect was sequence specific as control ODN had no effect.

While local delivery of CpG ODN was superior to systemic administration, most mice in both treatment groups still succumbed to lung cancer. As free ODN rapidly diffuse from the lungs into the blood stream, we sought to prolong their pulmonary retention by adsorbing the ODN onto polyketal nanoparticles. Such particles have an excellent in vivo safety profile and form microparticles 1 to 5 μm in diameter (appropriate for delivery to the distal alveoli by inhalation) when freeze dried (Supplementary Fig. S2A; refs. 19, 26). Preliminary studies showed that 30 μg of ODN was adsorbed per mg of polyketal microparticle and that 80% of the ODN was released from the microparticles over 48 hours under physiologic conditions (Supplementary Fig. S2B and S2C).

To determine whether CpG ODN retained their ability to stimulate cells after polyketal adsorption, the response of splenocytes from WT and TLR9 KO mice was compared. WT cells responded to both free and adsorbed CpG ODN by proliferating and secreting IL6 and IL12 during culture (P < 0.05; Fig. 2). Free CpG was more effective immediately after delivery while CpG adsorbed onto microparticles (CpG-MP) was more effective after 3 days in culture, consistent with the release characteristics of ODN from such particles. As seen in Supplementary Fig. S2, less than half of the adsorbed ODN was available at the start of culture, with more being released over time. Two findings established the specificity of these responses: (i) cells from TLR9 KO mice failed to respond to CpG ODN (free or polyketal adsorbed) and (ii) control (non-CpG) ODN adsorbed onto microparticles had no effect on WT or TLR9 KO cells (Fig. 2).

The uptake and longevity of CpG instilled into the lungs of mice with LLC tumors was then examined using fluorescein-conjugated ODN. Six hours after intratracheal delivery, free ODN was present primarily on the mucosal surface of the bronchial tree (Fig. 3A). By 48 hours, little of this material remained in the bronchi or could be found in association with LLC tumors (Fig. 3B). CpG-MP was distributed in bronchial and alveolar spaces at 6 hours (Fig. 3C). Of interest, this material persisted through 48 hours, at which time it had accumulated in tumor nests (Fig. 3D). To identify the cells interacting with the CpG-MP, sections were counterstained with phenotype-specific antibodies. Results showed that fluorescein-labeled CpG-MP associated primarily with cells expressing F4/80 (macrophages) and CD205 (DC) within the tumors (Fig. 3E and F).

CpG-MP or saline was instilled into the lungs of normal mice. BAL fluid collected 2 days later from animals treated with saline contained on average 4 × 10⁴ cells/mL. Cellularity rose by ∼25% in mice treated with free CpG ODN (Fig. 4A). By comparison, the BAL of mice treated with CpG-MP contained ∼15 × 10⁴ cells/mL (P < 0.01, Fig. 4A). This increased cellular infiltrate consisted primarily of macrophages and lymphocytes. Consistent with previous studies, free CpG ODN induced a significant increase in pulmonary IL12 levels, an effect magnified >10-fold by delivery of CpG-MP (Fig. 4B). No such changes occurred in mice treated with control ODN-MP.

Mice were challenged with 10⁶ LLC as described above and then treated weekly for 1 month with CpG-MP starting on day 7. When delivered systemically (the route by which free CpG ODN was ineffective in phase III human trials; refs. 11, 12), CpG-MP nearly doubled median survival time (from 22 to 40 days, HR: 0.37; 95% CI, 0.14–0.95; P = 0.007, Fig. 5). Far better results were achieved when CpG-MP was instilled directly into the lungs: 82% of mice survived indefinitely (animals were followed for up to 1 year; HR: 0.054; 95% CI, 0.024–0.12; P < 0.0001 vs. untreated controls). Moreover, these surviving mice remained tumor free when rechallenged with LLC months after cessation of CpG-MP therapy (Fig. 6 legend). Control ODN-MP delivered intratracheally had no significant effect on survival (Fig. 5). The survival benefit conferred by CpG-MP was not observed in TLR9 KO or Rag1 KO mice (Supplementary Fig. S3), consistent with CpG-MP acting through host immune cells that express TLR9.

To clarify the basis of the antitumoral immunity elicited by local delivery of CpG-MP, serial sections were taken through the tumor beds and examined using immunohistology. Initial experiments replicated the survival studies described above: mice were treated with CpG-MP 7 and 14 days after LLC instillation and their tumors analyzed on day 20. Consistent with the survival data shown in Fig. 5, recipients of CpG-MP had significantly fewer and smaller tumor nodules than control animals with overall tumor burden being reduced by >90% (Fig. 6A). The frequency of apoptotic cells in these tumors rose by ∼3-fold while the number of CD8⁺ T cells infiltrating the tumor site increased nearly 8-fold (P < 0.01 for both parameters, Fig. 6B and C and Supplementary Figs. S4A, S4B, and S5A).

Several additional changes in immune cell frequency were observed after CpG-MP treatment. Immunohistologic analysis of serial lung sections showed that the frequency of immunosuppressive cells in the tumor microenvironment (Tregs plus M2 macrophages) declined by over half while the number of M1 macrophages doubled (P < 0.01, Fig. 6D and Supplementary Figs. S4C, S4D, and S5B). Thus, the relative frequency of M1:M2 macrophages rose nearly 4-fold in CpG-MP–treated animals (P < 0.01, Fig. 6E).

To gain insight into the relationships between these changes, the kinetics with which they occurred was evaluated by treating mice only once (18 days after LLC instillation) with CpG-MP. Tumors were studied 2 days later. At that early time point after treatment, increased tumor cell apoptosis was already evident (2-fold, P < 0.01, Fig. 6B) as was the increase in infiltrating CD8 T cells (P < 0.01, Fig. 6C and Supplementary Figs. S4 and S5). Concomitantly, the frequency of immunosuppressive Tregs and M2 macrophages decreased significantly (P = 0.05, Fig. 6D and Supplementary Figs. S4 and S5). However, no effect on M1 macrophages or M1:M2 ratio was detected 2 days after treatment (Fig. 6E). Together, these findings suggest that CpG-MP rapidly reduces local immune suppression, enabling the infiltration of tumoricidal CD8 T cells that support tumor cell apoptosis. The magnitude of these effects increases over time and with repeated therapy and is accompanied by a slower rise in the frequency of M1 macrophages (P < 0.01, Fig. 6E and Supplementary Fig. S4).

Cells that express TLR9 contribute to tumor-specific host immune responses (27–30). Yet systemic treatment with CpG ODN has been of limited benefit to patients or animals with lung cancer (11, 12, 31–33). Animal models of melanoma and kidney/colon carcinoma suggest that optimal CpG ODN therapy requires local (rather than systemic) delivery (6, 7). Building on that finding we explored the utility of instilling CpG ODN directly into the lungs of mice with LLC tumors (Fig. 1). The LLC model shares important features of primary human lung cancer, including the peribronchial localization of tumors nodules and the presence of tumor-induced bronchial-associated lymphoid tissue (Supplementary Fig. S1; refs. 22, 23, 34, 35).

Consistent with previous findings, systemic delivery of CpG ODN had little effect on survival (31). Outcomes improved when the ODN was delivered intratracheally but survival remained modest (Fig. 1). Biodistribution studies showed that free ODN (i) localized to the mucosal and submucosal regions of the bronchus rather than reaching tumor-associated immune cells and (ii) was rapidly cleared from the lungs (Fig. 3A and B).

To improve the uptake and persistence of ODN in the tumor microenvironment, CpG ODN were adsorbed onto biodegradable polyketal nanoparticles and then freeze dried to form aggregates of optimize size for delivery throughout the bronchial tree (Supplementary Fig. S2; refs. 36, 37). This differed from earlier efforts in which ODN was simply mixed with 25 to 30 nm particles (38, 39). CpG ODN adsorbed onto MP retained their sequence specific ability to activate TLR9-expressing cells and were well tolerated (Fig. 2). Previous studies established that the small size and positive charge of free ODN resulted in their rapid redistribution through total body fluid. In contrast, CpG-MP remained in the lungs for up to 6 days where they induced an inflammatory response and accumulated within tumor nests after intrapulmonary instillation (Figs. 3 and 4). Systemic administration of CpG-MP did not achieve this goal (Fig. 5), consistent with earlier reports that such particles are primarily trapped and removed from the circulation by the liver (40). Preliminary studies found that 50 μg of CpG ODN delivered weekly for one month provided optimal protection in the LLC challenge model, an effect dependent on the presence of TLR9-expressing immune cells (Fig. 5 and Supplementary Fig. S3). Lowering the dose of ODN to 10 μg reduced long term survival from >80% to <40% (HR: 0.33; 95% CI, 0.11–1.03; P = 0.05 vs. no treatment). The immunity induced by CpG ODN therapy was persistent in that no tumors arose when survivors were rechallenged with LLC.

Increased numbers of Foxp3⁺ regulatory T cells are a negative prognostic indicator for patients with lung cancer (41–43). Consistent with that observation, we observed an inverse relationship between the frequency of Tregs and apoptotic cells in LLC tumors (Fig. 6, r = −0.47, P = 0.008). Clinical studies also indicate that patient survival correlates positively with M1- but inversely with M2-like macrophage frequency (44–46). This is not unexpected, as M1 macrophages support the inflammation that aids tumor rejection while M2 macrophages facilitate tumor growth by suppressing tumoricidal CTL and NK cells (47). Treatment with CpG-MP altered macrophage abundance, resulting in a 4-fold increase in the frequency of M1:M2 macrophages in the tumor bed (Fig. 6E).

Previous reports documented the ability of TLR9 agonists to promote the development of tumoricidal T and NK cells (28, 30, 48). Current findings establish that these CD8 T cells are better able to reach and lyse pulmonary tumors when the CpG ODN is absorbed onto microparticles and delivered to the lungs (Fig. 6C and Supplementary Fig. S4A and S4B). This therapeutic approach achieved 80% long-term survival of animals that would otherwise succumb to cancer in 22 days. Mechanistically, local administration of CpG-MP triggered a rapid decline in the number of Tregs and M2-like macrophages within the tumor (Fig. 6D and Supplementary Fig. S5B; refs. 13, 49). This was associated with an influx of CD8 T cells that correlated with increased tumor cell apoptosis (Fig. 6 and Supplementary Fig. S5A). The magnitude of these effects increased over time and with repeated therapy. As intratracheal instillation of LLC generated peribronchial tumors whether equivalent benefit will be observed in the treatment of primary or metastatic tumors in the lung periphery is uncertain. Biodistribution studies show that intratracheally administered CpG-MP do reach peripheral tumors but their effectiveness needs to be established in studies of additional lung cancer models. Available data show that adsorption onto polyketal microparticles enhances the delivery and persistence of immunostimulatory CpG ODN in the lungs, offering hope for substantially improved therapy of pulmonary tumors.

D.M. Klinman reports receiving royalties from the NIH concerning CpG ODN. No potential conflicts of interests were disclosed by the other authors.

Members of Dr. Klinman's lab have patents related to the use of CpG oligonucleotides. All rights to such patents have been assigned to the Federal Government. The assertions herein are the private ones of the authors and are not to be construed as official or as reflecting the views of the NCI at large.

Conception and design: T. Sato, D.M. Klinman

Development of methodology: T. Sato

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Sato

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Sato, D.M. Klinman

Writing, review, and/or revision of the manuscript: T. Sato, A. Ueda, Y. Ishigatsubo, D.M. Klinman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Shimosato, Y. Ishigatsubo, D.M. Klinman

Study supervision: Y. Ishigatsubo, D.M. Klinman

D.M. Klinman's lab is internally funded by the NCI. T. Sato received grants (no. 21790778, 23790917, and 15K09224) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

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.