Metronomic delivery of orally available pemetrexed-incorporated colloidal dispersions for boosting tumor-specific immunity

Abstract In this study, we developed oral pemetrexed (PMX) for metronomic dosing to enhance antitumor immunity. PMX was electrostatically complexed with positively charged lysine-linked deoxycholic acid (DL) as an intestinal permeation enhancer, forming PMX/DL, to enhance its intestinal permeability. PMX/DL was also incorporated into a colloidal dispersion (CD) comprised of the block copolymer of poly(ethylene oxide) and poly(propylene oxide), and caprylocaproyl macrogol-8 glycerides (PMX/DL-CD). CD-containing PMX/DL complex in a 1:1 molar ratio [PMX/DL(1:1)-CD] showed 4.66- and 7.19-fold greater permeability than free PMX through the Caco-2 cell monolayer and rat intestine, respectively. This resulted in a 282% improvement in oral bioavailability in rats. In addition, low-dose metronomic PMX led to more immunogenic cell death in CT26.CL25 cells compared to high PMX concentrations at the maximum tolerated dose. In CT26.CL25 tumor-bearing mice, oral metronomic PMX/DL-CD elicited greater antitumor immunity not only by enhancing the number of tumor-infiltrating lymphocytes but also by suppressing T cell functions. Oral PMX/DL-CD substantially increased programmed cell death protein ligand-1 (PD-L1) expression on tumor cells compared to the control and PMX-IV groups. This increased antitumor efficacy in combination with anti-programmed cell death protein-1 (aPD-1) antibody in terms of tumor rejection and immunological memory compared to the combination of PMX-IV and aPD-1. These results suggest that oral metronomic scheduling of PMX/DL-CD in combination with immunotherapy has synergistic antitumor effects.


Introduction
Pemetrexed (PMX) is a prototype anticancer chemotherapeutic that inhibits folate pathways and is a component of the standard treatment regimen for non-small cell lung cancer (NSCLC). A combination of PMX and cisplatin has been approved as first-line therapy for NSCLC. However, despite high therapeutic efficacy, acquired resistance and high recurrence rates lead to low overall survival rates in NSCLC patients (Darvin et al., 2018;Ng et al., 2019). A combination of conventional therapy and immune checkpoint inhibitors (ICIs), such as pembrolizumab and nivolumab, has recently been recommended as an alternative first-line therapy for NSCLS, and numerous clinical trials investigating combination therapies are in progress (Garon et al., 2019;Reck et al., 2019;Wu et al., 2020). Some of these clinical trials have demonstrated synergistic effects of combination therapies.
Conventional chemotherapy is administered at the maximum tolerated dose (MTD), leading to several adverse effects, including tumor resistance, neutropenia, myelosuppression, etc. To overcome these adverse effects, a drug-free period is scheduled, during which blood vessel regrowth and tumor recurrence can occur. As an alternative, low-dose metronomic chemotherapy (MCT) without drug-free periods was introduced, to maintain constant plasma drug levels and prevent blood vessel regrowth, and to achieve total remission with less toxicity (Simsek et al., 2019). MCT targets circulating endothelial and endothelial progenitor cells to prevent tumor angiogenesis (Andr e et al., 2014). Furthermore, MCT has evolved into a multi-targeted therapy with immunomodulatory and tumor dormancy-inducing effects (Gnoni et al., 2015;Chen & Mellman, 2017). It boosts immunity via induction of immunogenic cell death (ICD), promotion of antigen presentation by dendritic cells, and downregulation of suppressive cells to increase the cytotoxic effects of effector immune cells (Ghiringhelli et al., 2007;Banissi et al., 2009;Hao et al., 2014). MCT is also associated with reduced immunotoxicity, where severe lymphocyte toxicity is seen with the MTD (Wu & Waxman, 2018). The effects of immune-supportive drugs administered as low-dose MCT are further enhanced due to lower immunotoxicity. For frequent metronomic dosing, oral administration is required. However, most chemotherapeutics, including PMX, are not orally available, which is one of the limitations of applying the MCT. In addition, there are further limitations, including low volume of distribution and rapid clearance (Rinaldi et al., 1999;Hanauske et al., 2001;Soni et al., 2017).
Previously, to overcome these problems, an electrostatic complex of PMX was prepared using positively charged lysine-linked deoxycholic acid (DA) (i.e. DL) as an intestinal permeation enhancer, forming PMX/DL, followed by incorporation into water-in-oil-in-water (w/o/w) multiple nanoemulsions, which provided significant increases in the intestinal membrane permeability as well as oral bioavailability of PMX with equivalent antitumor activities compared to intravenous (IV) injection of PMX (Mahmud et al., 2018;Pangeni et al., 2018aPangeni et al., , 2018b. However, to further improve the stability of PMX in the nanoemulsions, a solid oral powder formulation was designed by incorporation of PMX/DL with dispersing agents, such as Kolliphor P188 and Labrasol, to produce a colloidal dispersion (CD) of PMX/DL (PMX/DL-CD) Pangeni et al., 2019). In addition, in vitro permeability and oral absorption of PMX/DL-CD in rats were significantly increased by apical sodium-dependent bile acid transporter (ASBT)-facilitated endocytosis, caveola/lipid raftdependent endocytosis, macropinocytosis, and paracellular transport (Park et al., 2017;Pangeni et al., 2019). Furthermore, metronomic treatment of A549 xenograft mice using oral PMX/DL-CD exhibited enhanced antiangiogenic activities, resulting in greater tumor growth suppression effects compared to MTD-treated mice .
The present study was mainly performed to determine the optimum complexation ratio of PMX and DL, to enhance oral bioavailability, and identify immune-modulating activities of metronomic dosing of oral PMX/DL-CD, followed by synergistic immunogenic antitumor effects of the metronomic dosing schedule of oral PMX/DL-CD in combination with ICIs. In this study, the optimum complexation molar ratio of PMX and DL was determined based on the permeability through a Caco-2 cell monolayer and rat intestinal perfusion study. After confirming the dose-dependent effects of PMX/DL-CD on oral absorption in rats, its antitumor activity in a Lewis lung carcinoma (LLC) mouse model was assessed. Furthermore, the immune-stimulating effects of metronomic treatment with oral PMX/DL-CD were compared to those of IV PMX at the MTD via ICD induction in CT26.CL25 cells. Finally, the synergistic anticancer effects of daily oral PMX/ DL-CD and anti-programmed cell death protein-1 (aPD-1) antibody were evaluated in a CT26.CL25 colon cancer model in terms of complete response (CR) and immunological memory.
Ethical approval for this study was obtained from the Institutional Animal Care and Use Committee (IACUC) of Mokpo National University (Jeonnam, South Korea; approval nos. MNU-IACUC-2020-013, MNU-IACUC-2020-018, and MNU-IACUC-2021-001). All animal experiments were performed in accordance with the National Institute of Health Guidelines for the Care and Use of Laboratory Animals and IACUC guidelines.
Moreover, the re-dispersed forms of PMX/DLs and CDs comprising the complexes were further characterized based on particle sizes, polydispersity indices (PDIs), and zeta potentials at 25 C using a dynamic laser light scattering analyzer (Malvern Zetasizer Nano ZS90; Malvern Panalytical, Malvern, UK). Morphological evaluations were performed using high-resolution transmission electron microscopy (TEM) (JEM-200; JEOL Ltd., Tokyo, Japan) after negative staining with a 2% aqueous solution of phosphotungstic acid.

In vitro Caco-2 cell monolayer permeability of PMX/DL-CD
To compare the in vitro apparent permeability (P app ) of PMX, PMX/DL(1:1), PMX/DL(1:2), PMX/DL(1:1)-CD, and PMX/DL(1:2)-CD across a Caco-2 (ATCC V R HTB-37 TM ; American Type Culture Collection, Manassas, VA, USA) cell monolayer, Caco-2 cells were seeded onto 24-well Transwell V R filter inserts in Dulbecco's modified Eagle's medium (DMEM) containing 10% (v/v) fetal bovine serum and 1% penicillin/streptomycin, at a density of 1 Â 10 5 cells/well, cultured for 14-16 days to form a confluent monolayer with the transepithelial electrical resistance (TEER) >350 XÁcm 2 . After the media in the apical and basolateral compartments were replaced with Hank's balanced salt solution (HBSS) and stabilized for 20 min at 37 C, 100 mL of the drug solution diluted with HBSS (equivalent to 50 mg/mL PMX) and 600 mL of HBSS were loaded into the apical and basolateral compartments in each well, respectively. During incubation at 37 C, 100 mL of sample solution was withdrawn from each basolateral compartment, and the volume was replaced using fresh HBSS at 0.5, 1, 2, 3, 4, and 5 h. After passing through a polyvinylidene fluoride (PVDF) membrane filter (pore size: 0.45 mm), the concentration of PMX in each sample solution was measured using the HPLC system at 254 nm, with a Luna C18 column (4.6 Â 250 mm, 5 mm, 100 Å) at 25 C. The mobile phase consisted of water (pH 3.5 adjusted with phosphoric acid) and acetonitrile (80:20, v/v), at a flow rate of 1 mL/min. The P app of PMX from each formulation was estimated using the following formula: P app ¼ dM/dt Â 1/(S Â C i ), where dM/dt indicates the linear-appearance permeation rate of PMX across a monolayer (lg/s), C i is the initial concentration of PMX in the apical compartment (lg/mL), and S is the permeation area of the monolayer (cm 2 ).

In situ single-pass rat intestinal perfusion study
The in situ single-pass intestinal effective permeabilities (P eff ) of PMX, PMX/DL(1:1), PMX/DL(1:2), PMX/DL(1:1)-CD, and PMX/DL(1:2)-CD were investigated in rats. The rats were fasted overnight for 12-16 h with free access to water before the experiment and then anesthetized by intramuscular injection of tiletamineÁHCl-zolazepam mixture (1:1 [w/w]; 40 mg/kg). With the animal in the supine position, the abdominal cavity was opened using a midline longitudinal incision, and the intestine (duodenum, jejunum, and ileum) was exposed and subsequently intubated at the two ends with a polypropylene perfusion tube connected to a peristaltic pump (Harvard Apparatus, Holliston, MA, USA). After covering the entire excised area with a cotton pad soaked with warm normal saline, the cannulated intestinal segments were flushed with blank perfusion buffer at 37 C for 30 min, at a constant flow rate of 0.5 mL/min, to remove residual intestinal contents. Then, drug solutions of PMX, PMX/DL(1:1), PMX/DL(1:2), PMX/DL(1:1)-CD, or PMX/DL(1:2)-CD (equivalent to 100 lg/mL PMX) were perfused through the intestinal lumen at 37 C and a constant flow rate of 0.2 mL/min. After allowing 30 min to achieve a steady-state, outlet perfusate was collected every 15 min for 120 min. In addition, to assess the reliability of the perfusion study, 100 lg/mL fluorescein diluted in perfusion buffer was perfused separately under the same conditions and the collected samples were analyzed at excitation/emission wavelengths of 494/512 nm using a microplate reader (PerkinElmer multimode plate reader; PerkinElmer, Waltham, MA, USA) as a low-permeability marker. During the experiment, all animals were kept on heated pads under a heat lamp to maintain a normal body temperature. At the end of the experiment, the lengths and radii of the cannulated intestinal segments were measured, taking care to avoid stretching them. The collected samples were then filtered using 0.45-lm PVDF filters and stored at À20 C before analyses. The outlet concentration of PMX in the perfusate was analyzed by HPLC as described above. In addition, the errors caused by rat intestinal water absorption and secretion during perfusion were corrected by the net water flux (ratio of the inlet to outlet perfusate flows, Q out / Q in ) and the corrected drug concentration in perfusate for each formulation was calculated based on the following equation (i.e. gravimetric method): C out,corr ¼ C out Â (Q out / Q in ), where C out,corr is the corrected outlet PMX concentration (mg/mL), C out is the PMX outlet concentration (mg/mL), Q in (mL/min) is the flow rate entering the intestine, and Q out (mL/min) is the measured perfusate exit flow (net weight/ 15 min, assumed density of 1.0 g/mL) for the specified time interval. After correction, the P eff was calculated using the equation: where A is the intestinal area available for absorption (2prl) based on its length (l) and radius (r).

In vivo pharmacokinetic study in rats
Oral absorption of PMX/DL(1:1)-CD was assessed to investigate the effects of ionic complex formation with DL, as well as CD formulations with Kolliphor P188 and Labrasol. Rats were orally administered 400 mL of PMX-Oral (20) (20 mg/kg PMX dissolved in water), PMX/DL(1:1)-CD (10) (equivalent to 10 mg/kg PMX), PMX/DL(1:1)-CD (20) (equivalent to 20 mg/kg PMX), and PMX/DL(1:1)-CD (40) (equivalent to 40 mg/kg PMX). Furthermore, to evaluate oral bioavailability, 150 mL of PMX-IV (10) (10 mg/kg PMX dissolved in water) was injected via the tail vein. Blood samples (150 mL) were drawn from the retroorbital plexus of rats at 0.5, 1, 1.5, 2, 4, 6, 8, and 10 h after oral administration or 0.25, 0.5, 1, 1.5, 2, 3, 4, and 8 h after IV injection under mild anesthesia. After mixing with 50 lL of 3.8% sodium citrate solution, the collected samples were immediately centrifuged at 2500 Â g and 4 C for 15 min. The separated plasma samples were kept at À80 C before analysis and the drug concentration in each plasma sample was determined by LC-MS. Then, 10 mL of DAMPA (5 mg/mL) was added to each sample as an internal standard (IS) and subjected to solid-phase extraction using Plexa Bond Elut PAX cartridge (30 mg, 1 mL; Agilent Technologies, Santa Clara, CA, USA), conditioned with 500 mL of methanol, followed by 500 mL of deionized water. After loading the standards or samples into the cartridges, the unadsorbed particles were washed out with 500 mL of deionized water and methanol. Finally, the adsorbents were eluted with 250 mL of 5% formic acid in methanol twice and the eluent was then dried in a centrifugal evaporator (Genevac Ltd., Ipswich, UK). The dried eluate was then reconstituted with 100 mL of 5% formic acid in methanol and the concentration of PMX in each sample was determined using an Agilent 6120 Quadruple LC-MS system equipped with a Luna C18 column (100 Â 2 mm, 3 mm). A solvent mixture consisting of acetonitrile and 0.34% formic acid solution (15:85, v/v) was used as the mobile phase, at a flow rate of 0.2 mL/min. PMX and IS were ionized using electrospray ionization (ESI) source in positive ion mode under a capillary voltage of 3.5 kV, drying gas flow rate of 3.1 L/min, and drying gas temperature of 300 C. Quantification of the protonated molecular ions was and DAMPA, respectively. In addition, the lowest limit of quantification (LOQ) and the limit of detection (LOD) of PMX were 5 ng/mL and 2.5 ng/mL, respectively.
Tumor volume was then calculated as a 2 Â b Â 0.52, where a and b were the width and length of the tumor, respectively. The mice were sacrificed and the tumor masses were measured 21 days after drug administration.

Immunofluorescence analysis of ICD induction by treatment with PMX/DL
CT26.CL25 cells, cultured in coverslips at 1 Â 10 4 cell density, were treated with 300 mM DL, 300 mM PMX, or PMX/DL (equivalent to 300 mM PMX) for 48 h. The PMX dose was converted from the C max of the pharmacokinetic profile in patients treated with 500 mg/m 2 PMX-IV (Li et al., 2007). The cells were then washed with PBS, fixed in 10% formalin solution, and incubated with a cell membrane marker, Texas Red-X conjugated wheat germ agglutinin solution (5 mg/mL), for 10 min at room temperature. The cells were washed and permeabilized with 1% Triton-X solution for 40 min at room temperature. The cells were then incubated overnight with primary anti-CRT (1:200) or anti-HMGB-1 (1:200) antibodies at 4 C after washing with PBS. On the following day, the cells were washed and incubated with secondary anti-rabbit Alexa

In vivo immunostimulatory effects of metronomic treatment with oral PMX/DL-CD
To investigate in vivo immunostimulatory effects of oral metronomic PMX, BALB/c mice were inoculated with 1 Â 10 6 of CT26.CL25 cells per mouse. The mice were randomly divided into three groups (10 mice per group) once the tumor volume had reached 50-70 mm 3 : control (untreated), PMX-IV (150 mg/kg PMX administered IV every 3 weeks), and PMX/ DL-CD (oral PMX/DL(1:1)-CD administered daily, equivalent to 20 mg/kg PMX). The mice were sacrificed 3 weeks after treatment and the tumors were isolated. The tumor tissues were converted into single cells by passing them through a 40-mm filter after digestion with collagenase A (0.2% w/v), DNAse (30 U/mL), and dispase (10 U/mL) in MACS (gentleMACS Octo Dissociator with Heaters; Miltenyi Biotec, Bergisch Gladbach, Germany) at 37 C for 45 min. Some cells were collected for tumor-infiltrating lymphocyte analyses and centrifuged using histopaque-1077 for 20 min at 450 g to isolate the lymphocytes. The isolated cells were then stained with fluorescenceconjugated antibodies and incubated. The cells were washed and suspended in FACS buffer and then collected into FACS tubes. The cells were then analyzed using flow cytometry (FACS Aria II) and the data were analyzed using FlowJo software (FlowJo LLC, Ashland, OR, USA). Cell populations were acquired by gating, as follows: tumor-infiltrating lymphocytes (DAPI À /CD45 þ ), helper T cells (DAPI À /CD45 þ /CD3 þ /CD4 þ ), PD-L1-expressing tumor cells (DAPI À /CD45 À /PD-L1 þ ), proliferating cytotoxic T cells (DAPI À /CD45 þ /CD3 þ /CD8 þ /ki67 þ ), and IFN-c-secreting T cells (DAPI À /CD45 þ /CD3 þ /CD8 þ /IFN-c þ ). Lymphocytes were isolated from the tumors using the procedure described previously; 2 Â 10 5 tumor lymphocytes per well, along with b-gal (5 mg/mL), were incubated on precoated 96-well plates in CTL test medium for 24 h at 37 C with 9% CO 2 . The standard protocol for the Immunospot assay kit (Cellular Technology Ltd., Shaker Heights, OH, USA) was followed. After the procedure, the plates were scanned and the spots were counted in the ELISpot counter (Immunospot V R S6 ULTIMATE analyzer; Cellular Technology Ltd.).
Immunofluorescence was performed to investigate the expression of various immune cells in the tumor tissues of mice, inoculated and treated as previously described. The isolated tumor tissues were fixed with zinc fixation solution (BD Biosciences), embedded in paraffin, cut into 4-mm sections, and loaded into slides. The slides were deparaffinized using xylene, rehydrated in a series of alcohol solutions, and incubated in citrate buffer (GenDEPOT) at 100 C for 60 min to retrieve the antigens. Slides were incubated with dye-conjugated CD45, CD8, F4/80, and PD-L1 antibodies overnight at 4 C, and the nuclei were stained with DAPI. An automated, multimodal tissue analysis system (Vectra; PerkinElmer) was used to obtain images.

Pharmacokinetics and statistical analyses
Pharmacokinetic parameters were estimated using a noncompartmental model in WinNonlin V R software (ver. 5.3; Pharsight Corp., Mountain View, CA, USA). All data were expressed as means ± standard deviations (SDs), or standard errors of the mean (SEMs). Statistical analysis was performed by one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test for unpaired data with more than three means. In all analyses, p < .05 was taken to indicate statistical significance.

In vitro permeability of PMX/DL-CD through a Caco-2 cell monolayer
The absorption potentials of PMX/DLs and PMX/DL-CDs across a Caco-2 cell monolayer were assessed after preparation of ion-pairing complexes between PMX and DL at molar ratios of 1:1 and 1:2. The P app of PMX/DL increased with an increasing complexation molar ratio of DL to PMX. PMX/DL(1:2) showed a 1.47-fold higher permeability than PMX/DL(1:1), resulting in a 297% increase in the P app of free PMX (Table 1). This increase in P app was attributed to the DL in the complex, which renders the PMX molecule amphiphilic and improves its partition coefficient (Table S1) (Moghimipour et al., 2015). After oral formulation of PMX/DL with Kolliphor P188 and Labrasol, the permeability of PMX from PMX/DL(1:1)-CD or PMX/DL(1:2)-CD increased 1.73-and 1.28-fold compared to PMX/DL(1:1) and PMX/DL(1:2), and resulted in increases of 366% and 408% in the P app of free PMX, respectively. The P app of PMX formulated only with Kolliphor P188 and Labrasol (i.e. PMX-CD) was 2.07 ± 0.482 (Â10 À6 cm/s), which was 1.73-fold higher than that of free PMX but 1.57-and 2.70-fold lower than those of PMX/DL(1:1) and PMX/DL(1:1)-CD, respectively. These results may have been due to the synergistic effects of DL and surfactants in PMX/DL-CD comprising the nano-micelles on their transor para-cellular permeation (Miyake et al., 2006;Moghimipour et al., 2015;Tian et al., 2017). The ionically conjugated DL in PMX/DL has been shown to facilitate penetration of a cargo molecule (i.e. PMX) through the cell membrane by ASBT Deng & Bae, 2020). In addition, DL, as well as Kolliphor P188 and Labrasol, can increase membrane flexibility to increase drug partitioning across the membrane (Gupta et al., 2013;DiMarco et al., 2017;Pavlovi c et al., 2018;Jha et al., 2020) DA in DL, as well as surfactants in CD, can reversibly open the tight junctions by autophosphorylation of epidermal growth factor receptor (EGFR) or dephosphorylation and rearrangement of zonula occludens-1 (ZO-1), which may facilitate the permeation of PMX or PMX/ DL released from PMX/DL-CD through the cell membrane via the paracellular pathway (Raimondi et al., 2008;Stojan cevi c et al., 2013;Zhou et al., 2013;Pavlovi c et al., 2018). Moreover, PMX/DL-CD is known to be transported by caveola/lipid raft-mediated endocytosis and micropinocytosis (Kou et al., 2018;Pavlovi c et al., 2018;Pangeni et al., 2019). Overall, the combined activities of DL complex and CD formation with Kolliphor P188 and Labrasol may markedly improve the P app of PMX/DL-CDs. However, the P app of PMX/ DL(1:1)-CD was not significantly enhanced by increasing the DL complexation ratio.

In situ single-pass intestinal perfusion of PMX/DL-CD in rats
The intestinal permeability of PMX was also investigated in rat intestinal segments using the in situ single-pass perfusion technique. After confirming the reliability of this study by measuring the P eff of fluorescein (0.001 ± 0.002, Â10 À4 cm/s), the P eff of PMX was 0.830 ± 0.044 (Â10 À4 cm/s), which increased 3.27-and 3.94-fold after ion-pairing with DL at molar ratios of 1:1 and 1:2, respectively. In addition, oral formulation of PMX/DL(1:1)-CD exhibited significantly higher intestinal permeability (5.97 ± 1.26, Â10 À4 cm/s) compared to PMX/DL(1:1) and free PMX. The P eff of PMX/DL(1:2)-CD was also 2.12-and 8.34-fold higher than those of PMX/DL(1:2) and free PMX, respectively (Table 1, Table S2). However, there were no significant differences in permeability between PMX/DL(1:1)-CD and PMX/DL(1:2)-CD. These results were consistent with the in vitro Caco-2 cell permeability data, which may have been induced by ASBT saturation (Kanda et al., 1998). Furthermore, the permeabilities of all formulations increased significantly in the perfusion study, possibly due to the greater available absorption area of the villi and microvilli (Dezani et al., 2016). Based on the overall in vitro as well as in situ permeability data, PMX/DL(1:1)-CD was selected as the optimum oral PMX formulation. Further studies, including analyses of in vivo oral absorption and in vivo synergistic antitumor effects of metronomic dosing of oral PMX in combination with aPD-1, were carried out using PMX/DL(1:1)-CD.
However, as the ion-pairing complex is less stable than the prodrug form, the possibility of premature complex dissociation in biological fluids before absorption is a major concern when utilizing the ion-pairing approach. On the other hand, the overall in vitro permeability and in vivo oral absorption data suggest that ion-pairing complexation with DL followed by formulation with surfactants, forming CDs, can deliver specific target molecules through the intestinal membrane more effectively via ASBT-mediated endocytosis and/or caveola/lipid raft-mediated endocytosis or micropinocytosis (Park et al., 2017;Pangeni et al., 2019), implying the formation of stable ion-pairing complexes were under physiological conditions after incorporation into micelles comprised of Kolliphor P188 and Labrasol. However, further studies to determine the stability and association constant of PMX/DL and its CD formulation in the gastric and intestinal fluids are required.

In vivo tumor growth inhibitory effect of metronomic oral PMX/DL-CD in LLC cellbearing mice
The dose-dependent tumor-growth inhibitory effects of oral PMX/DL-CD were investigated in the LLC cells-bearing mice model to assess the optimum oral dose for PMX/DL-CD. After once-a-day oral administration of PMX/DL-CD (10), the tumor growth was delayed by 50.3% compared to the control group (Figure 2(A)). In addition, PMX/DL-CD (10) exhibited a 1.34-fold higher tumor growth retardation rate compared to PMX-Oral (20), which indicates that enhanced permeability, followed by improved oral bioavailability may surpass the antitumor effects of oral PMX. As a daily oral dose of PMX/ DL-CD was increased to 20 mg/kg based on PMX, the tumor volume was suppressed by 67.4%, compared to the control group. Moreover, PMX/DL-CD (20) delayed tumor growth by 50.9% compared to PMX-Oral (20), equivalent to the PMX-IV (150) group, representing a 60.9% antitumor efficacy compared to the control group. Thus, after 21-day treatment, the PMX/DL-CD (20) and PMX-IV (150) groups resulted in a 334% and 213% greater reduction in the isolated tumor weights, respectively, compared to the control group (Figure 2(B,C)). The greater tumor-suppressing ability of PMX/DL-CD (20), despite lower plasma levels compared to IV administration, might be due to the enhanced oral absorption of PMX from PMX/DL-CD (20), achieving the plasma and tumor-tissue levels necessary to elicit an anticancer response. Oral PMX/DL-CD (40) also resulted in a greater tumor volume suppression, by 69.6 and 54.3%, compared to the control and PMX-Oral (20) groups, respectively. However, there was no further increase in its anticancer effects, compared to the PMX/DL-CD (20) or PMX-IV (150)-treated mice. Furthermore, the Table 2. Pharmacokinetic parameters of PMX in rats after intravenous or oral administration of PMX or PMX/DL(1:1)-CD.
Test material PMX-IV (10) PMX-Oral (  isolated tumor weights were similar in mice treated with PMX/DL-CD (20). These findings were in agreement with the dose-dependent plasma levels of PMX. The C max and AUC values for PMX/DL-CD (40) did not increase significantly and were two times greater than PMX/DL-CD (20). The body weights of mice treated with PMX or PMX/DL-CDs did not change significantly compared to the control group. This indicated that there were no additional toxic effects induced by the drug, or by oral formulations including DL, after repeated administration for 3 weeks (Figure 2(D)).
Collectively, these findings indicate that MCT of PMX/DL-CD (20) can deliver PMX effectively via the oral route, and can maintain sufficient drug levels in the tumor microenvironments to have significant antitumor efficacy compared to PMX-IV. Further in vivo synergistic anticancer efficacy studies of combined treatment with aPD-1 and oral PMX/DL-CD (20) were performed.

In vitro evaluation of ICD induction by treatment with PMX/DL
ICD elicits an immune response characterized by translocation of CRT to the cell surface and subsequent HMGB-1-release from the nucleus. The release of CRT and HMGB-1 following PMX treatment was revealed in CT26.CL25 cells by confocal microscopy, flow cytometry, ELISA, and immunoblotting. Confocal microscopy showed that CRT-fluorescence was more intense in PMX-treated cells compared to controls and DL-treated cells, indicating greater CRT release with PMX treatment (Figure 3(A)). Increased CRT release in PMX-treated cells was also observed using flow cytometry. The surface CRT-release was 3.73, 2.36, and 1.11 times higher in PMX/ DL-treated cells compared to untreated, DL-treated, and PMX-treated cells, respectively (Figure 3(B,C)). Similarly, HMGB-1-release was higher in PMX-treated cells compared to All values are means ± SEMs (n ¼ 12 for each group). ÃÃ p < .01, ÃÃÃ p < .001, and ÃÃÃÃ p < .0001 compared to untreated controls; # p < .05, and ### p < .001 compared to the PMX-IV (150) group; $$ p < .01, and $$$ p < .001 compared to the PMX-Oral (20) group; &&& p < .001 compared to the PMX/DL-CD (10) group.
controls and DL-treated cells according to confocal microscopy (Figure 3(D)). This was verified by HMGB-1 ELISA, which demonstrated 7.04, 1.87, and 1.07 times higher HMGB-1 release in PMX/DL-treated cells compared to untreated, DLtreated, and PMX-treated cells, respectively (Figure 3(E)). Furthermore, western blot analysis of CT26.CL25 cells demonstrated higher CRT-and HMGB-1-protein content in PMXtreated compared to untreated and DL-treated cells (Figure 3(F)). Taken together, the results suggested that PMX successfully induces ICD, which is key to the immune response. PMX/DL conjugate was a more potent ICD inducer than PMX alone, possibly due to increased cell permeability, as described previously.

Metronomic oral PMX/DL-CD treatment enhances both T cell population and function
As PMX provoked ICD in CT26.CL25 cells, we hypothesized that in vivo PMX treatment may promote antitumor immunity via tumor-infiltrating lymphocyte enhancement. To test this, we distinguished three groups: a control group, a group in which the MTD of PMX was injected into the tail vein (PMX-IV), and a metronomic oral PMX-treated (PMX/DL-CD) group. Tumor immune-modulatory effects were compared among the groups. The PMX-IV group had slightly increased CD45 þ -cell infiltration, while the PMX/DL-CD group had a significantly larger cell population compared to the control group (Figure 4(A,B)). Although the PMX-IV group had a substantially increased CD4 þ T-cell population compared to the control group, the increase was even greater in the PMX/DL-CD group (Figure 4(C,D)). However, only the PMX/DL-CD group demonstrated an increase in CD8 þ T-cell population compared to the control group (Figure 4(E,F)). These data suggest that although PMX treatment using MTD resulted in slightly elevated tumor-infiltrating lymphocytes, such as CD4 þ T-cells, the overall antitumor immunity remained inadequate; metronomic oral PMX treatment could elicit much stronger antitumor adaptive immunity. We also evaluated whether PMX treatment-induced PD-L1 expression in tumor cells modulates tumor immunity. We found that PD-L1 expression in PMX-IV and PMX/DL-CD groups was two and three times greater compared to the control group, respectively ( Figure 5(A,B)). This suggests that PMX treatment directly or indirectly increases PD-L1 expression. We further investigated the suppression of T-cell functions, such as proliferative capability and cytokine release, by PMX treatment. Oral PMX MCT significantly enhanced CD8 þ T-cell proliferation compared to the PMX MTD and control groups ( Figure 5(C,D)). Similarly, PMX MCT treatment substantially increased the IFN-c-releasing CD8 þ T-cell population compared to the control group ( Figure 5(E,F)). These data demonstrated that MCT treatment using PMX not only enhanced lymphocyte infiltration but also suppressed Tcell function.
macrophages, were substantially elevated in the PMX/DL-CD group compared to the control and PMX-IV groups ( Figure  6(A)). IFN-c-specific ELISPOT also showed that PMX/DL-CD was most effective in eliciting tumor-specific immunity among the three groups. These data demonstrated that PMX/DL-CD effectively boosts the tumor-specific immunity of tumor-infiltrating lymphocytes (Figure 6(B,C)).
Collectively, these data suggest that metronomic oral PMX treatment is more effective for tumor-specific immunity enhancement because of increased expression of lymphocytes in tumor tissue and suppression of tumorspecific immunity. These data also indicate that metronomic oral administration of PMX has a synergistic effect with ICIs. (E) Characterization of CD45 þ CD8 þ IFN-c þ T cells using flow cytometry and (F) quantification thereof. All values are means ± SEMs (n ¼ 5). ÃÃ p < .01, and ÃÃÃ p < .001 compared to untreated controls; # p < .05, and ## p < .01 compared to the PMX-IV group.
3.8. Synergistic antitumor effects of combined treatment with metronomic oral PMX/DL-CD and aPD-1 in CT26.CL25 cell-bearing mice To investigate a synergistic antitumor efficacy of oral PMX/ DL-CD in combination with aPD-1 using immunomodulatory effects of metronomic scheduling of PMX, we performed an in vivo combination study with aPD-1 in the CT26.CL25 tumor model (Figure 7(A)). Tumors in mice treated with PMX-IV (150) started to grow significantly 15 days after the injections (Figure 7(B)). Tumor growth was effectively suppressed with oral PMX/DL-CD (20) for 21 days, resulting in 74.8 and 61.4% maximal tumor growth inhibitions up to day 18, compared to the control and PMX-Oral (20) groups, respectively (Figure 7(B,C)). However, the remaining tumor tissues grew again quickly, soon after PMX/DL-CD (20) was stopped on day 21. Tumor growth in mice treated with aPD-1 alone was also significantly delayed up to day 18; however, the remaining tumor volume increased continuously, and only 50% of mice showed a CR (Figure 7(C,D)). On the other hand, a combination of PMX-IV (150) and aPD-1 (10) exhibited substantial anticancer effects, leading to 73.9 and 48.7% greater tumor-growth delays, respectively, compared to the mice treated with PMX-IV (150) or aPD-1 (10) alone ( Figure  7(B,C)). However, continuous tumor growth was observed in 50% of mice, with no significant improvement of incomplete rejection (Figure 7(D)). Co-administration of oral PMX/DL-CD drug administration, indicating no memory immune response (Figure 7(G,H)). Meanwhile, all rechallenged tumors in the mice treated with aPD-1 (10) alone, or in combination with PMX-IV (150) or PMX/DL-CD (20), were completely rejected up to day 21 (Figure 7(G,H)). These results suggest that the combined treatment with PMX/DL-CD (20) and aPD-1 (10) can provide durable antitumor effects by generating immunological memory.

Conclusion
Metronomic PMX treatment is an effective alternative for achieving the synergistic effects of combination treatment, compared to MTD treatment. However, there has been controversy regarding oral delivery of PMX for frequent metronomic dosing. In this study, we developed colloidal formulations of oral PMX (PMX/DL-CD) and demonstrated adequate absorption. We found that treatment with PMX/DL-CD successfully induced ICD, which resulted in the release of danger signals, such as HMGB-1 and CRT. We also demonstrated that low-dose metronomic oral PMX treatment was associated with increased numbers of tumor-infiltrating lymphocytes and enhanced T cell function, compared to control and PMX-IV groups. Furthermore, oral PMX/DL-CD elicited stronger antitumor effects in combination with aPD-1 antibodies compared to the PMX-IV group, indicating that metronomic oral PMX has stronger immune-supportive effects than IV MTD treatment. Our findings suggest that metronomic oral PMX has the potential for use in combination with immunotherapy to elicit synergistic antitumor effects with less toxicity.