A novel immunochemotherapy based on targeting of cyclooxygenase and induction of immunogenic cell death

Abstract

Cyclooxygenase (COX) plays a crucial role in the “inflammogenesis of cancer”, which leads to tumor progression, metastasis, and immunotherapy resistance. Therefore, reducing “inflammogenesis” by COX inhibition may be a key perspective for cancer therapy. However, the role of tumor-derived COX in the actions of COX inhibitors remains incompletely understood. In this study, applying “old drug new tricks” to repurpose 5-aminosalicylic acid (5-ASA), a COX inhibitor, we examined the effect of 5-ASA, alone or in combination with doxorubicin (DOX), in several cancer cell lines with different levels of COX expression. To facilitate the evaluation of the combination effect on tumors in vivo, a new micellar carrier based on PEG-b-PNHS polymer-conjugated 5-ASA (PASA) was developed to enhance codelivery of 5-ASA and DOX. Folate was also introduced to the polymer (folate-PEG-NH₂-conjugated PASA (FASA)) to further improve delivery to tumors via targeting both tumor cells and tumor macrophages. An unprecedented high DOX loading capacity of 42.28% was achieved through various mechanisms of carrier/drug interactions. FASA was highly effective in targeting to and in inhibiting the growth of both 4T1.2 and CT26 tumors in BALB/c mice. However, FASA was more effective in CT26 tumor that has a high level of COX expression. Codelivery of DOX via PASA and FASA led to a further improvement in antitumor activity. Mechanistic studies suggest that inhibition of COX in vivo led to a more active tumor immune microenvironment. Interestingly, treatment with FASA led to upregulation of PD-1 on T cells, likely due to repressing the inhibitory effect of prostaglandin E2 (PGE₂) on PD-1 expression on T cells. Combination of FASA/DOX with anti-PD-1 antibody led to a drastic improvement in the overall antitumor activity including regression of some established tumors at a suboptimal dose of FASA/DOX. Our data suggest that FASA/DOX may represent a new and effective immunochemotherapy for various types of cancers, particularly those cancers with high levels of COX expression.

Introduction

Accumulating evidence indicates that chronic inflammation is a risk factor for various type of cancers[1]. Prostaglandin E2 (PGE₂), a prostanoid lipid derived from the action of cyclooxygenases, plays a predominant role in promoting inflammation and tumor progression by regulating downstream targets which control cell proliferation, angiogenesis and immunosuppression[2]. Cyclooxygenase (COX)-1 and 2, critical for the production of PGE₂, are upregulated in various malignant tumors, including colorectal, breast, stomach, lung, and pancreatic cancers[3, 4]. Moreover, COX-2 overexpression is indicative of a poor outcome and recurrence[5], low survival rate[6, 7], immune escape and resistance to cancer immunotherapy[8].

In contrast, absence of COX alters epidermal differentiation and attenuates the growth rate and incidence of papilloma formation[9]. Moreover, genetic ablation of COX-2 in mouse melanoma, colorectal, breast and pancreatic cancer cells renders them susceptible to immune-dependent tumor growth control[3, 10]. Furthermore, conventional type 1 dendritic cells (cDC1) and natural killer (NK) cells, which are essential for antitumor immunity, have been found to assemble in the tumor microenvironment of COX-deficient tumors[11].

Various COX inhibitors have been developed and examined for their antitumor activity and the underlying mechanisms. For example, pharmacological inhibition of COX by etodolac, a COX-2 selective inhibitor, has been shown to induce a dose dependent inhibition of endometrial cancer cells through G1 phase cell cycle arrest and inhibition of telomerase[12]. Another COX inhibitor, celecoxib significantly suppresses angiogenesis and tumor growth in CT26 tumor by extenuating PGE₂-mediated refractoriness to VEGF/VEGFR2 inhibition[13]. In addition, celecoxib enhances 5-Fluorouracil (5-FU) antitumor effects for esophageal squamous cell carcinoma by downregulating dihydropyrimidine dehydrogenase expression[14].

Despite the direct and indirect antitumor activity of various COX inhibitors, the underlying mechanisms remain incompletely understood, particularly the role of tumor-derived COX in the actions of these inhibitors. Among those COX inhibitors, 5-ASA is an FDA approved anti-inflammatory drug to treat inflammatory bowel disease, including ulcerative colitis and Crohn's disease[15]. Many retrospective correlative studies have showed that the long-term use of 5-ASA can prevent the tumorigenesis[16, 17]. However, the direct anti-tumor activity and underlying mechanisms of 5-ASA have seldom been reported. In the few reported studies, 5-ASA was used at concentrations significantly higher than the effective concentration required for inhibiting the COX activity. Few comparative studies were reported examining the antitumor activity of COX inhibitors on tumors with different expression levels of COX. More studies on the role of tumor-derived COX are needed to help develop a more rational treatment of cancer.

In this study, we compared the cytotoxicity of 5-aminosalicylic acid (5-ASA), alone or in combination with doxorubicin (DOX) in several human and murine cancer cell lines. We further compared the antitumor activity of 5-ASA in vivo using two murine tumor models (4T1.2 and CT26) that have different expression levels of COX. To facilitate the in vivo evaluation of the combination therapy, a new micellar carrier based on PEG-b-PNHS polymer-conjugated 5-ASA (PASA) was developed to facilitate selective codelivery of 5-ASA and DOX. In addition to examination of antitumor activity, the impact on tumor immune microenvironment was investigated (Scheme 1).

Scheme 1. Differential effects of FASA/DOX in modulating the immune microenvironment of COX-low and COX-high tumors.

Self-assembled FASA/DOX nanoparticles were administered intravenously (i.v.) and selectively accumulated to the tumor by EPR effect and folate-mediated active targeting. FASA/DOX was more effective in improving the immune microenvironment of COX-high tumors, leading to better control of tumor growth.

Material and methods

Reagents

5-ASA was purchased from Frontier Scientific (UT, U.S.A). DOX·HCl was purchased from LC Laboratories (MA, U.S.A). 4-Cyano-4 [(dodecylsulfanylthiocarbonyl)sulfanyl] pentanoic acid, 2-Azobis-(isobutyronitrile) (AIBN), poly(ethylene glycol) methacrylate (average Mn = 950, PEG₉₅₀), 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT), Dulbecco's Modified Eagle's Medium (DMEM) and folic acid were purchased from Sigma-Aldrich (MO, U.S.A). AIBN was purified by recrystallization in anhydrous ethanol. N-Succinimidyl Methacrylate was purchased from TCI (U.S.A). RPMI-1640 medium and folate-deficient RPMI-1640 medium, fetal bovine serum (FBS) and penicillin-streptomycin solution were purchased from Invitrogen (NY, U.S.A). All solvents used in this study were of HPLC grade.

Cell culture

All cell lines used in this work were obtained from ATCC (Manassas, VA). 4T1.2 murine triple negative breast cancer cells and MDA-MB-468 human triple negative breast cancer cells were cultured in DMEM. CT26 murine colon cancer cells and HCT116 human colon cancer cells were maintained in RPMI-1640 medium. KB human epidermoid carcinoma cells were culture in FA-deficient RPMI-1640 medium. All cells were cultured at 37 °C in a humidified atmosphere with 5% CO₂. All medium was supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin.

Animals

Female BALB/c mice (4-6 weeks) were purchased from The Jackson Laboratory (ME, U.S.A). All animals were housed under pathogen free conditions according to AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care) guidelines. All animal-related experiments were performed in full compliance with institutional guidelines and approved by the Animal Use and Care Administrative Advisory Committee at the University of Pittsburgh.

Real-Time PCR

cDNA was generated from the purified RNA extracted from cultured cells, isolated tumor and stromal cells or tumor tissues using QuantiTect Reverse Transcription Kit (Qiagen, MD, U.S.A) according to the manufacturer’s instructions. Quantitative real-time PCR was performed using SYBR Green Mix on a 7900HT Fast Realtime PCR System. Relative target mRNA levels were analyzed using delta-delta-Ct calculations and normalized to GAPDH. The primers are shown in Table S2.

Western blot assay

Cultured cells, isolated tumor cells and stromal cells were lysed with RIPA lysis buffer (Thermo Fisher Scientific, MA, U.S.A) by gently shaking on ice for 30 min. Tumor tissues were homogenized with RIPA lysis buffer. After centrifugation at 12,500 g for 10 min, the supernatants were collected and the concentrations of proteins were measured using Pierce BCA Protein Assay Kit (ThermoFisher Scientific, MA, U.S.A). An equivalent amount of protein was resolved by 10% SDS-PAGE and transferred to PVDF membranes (Bio-Rad, CA, U.S.A). The membranes were then blocked in 5% non-fat powdered milk dissolved in phosphate buffer containing 0.05% Tween-20 (PBST) for 1h. Afterwards, the membranes were incubated with anti-COX-1 (Cell Signaling Technology, MA, U.S.A), anti-COX-2 (Cell Signaling Technology, MA, U.S.A) or anti-PD-L1 antibody (Abcam) in antibody dilution buffer (5% BSA in PBST) with gentle agitation overnight at 4 °C. After washing with PBST for three times, the membranes were subsequently incubated with the secondary antibody (Cell Signaling Technology, MA, U.S.A) for 1 h at room temperature. The membranes were then washed three times with PBST before being exposed to the SuperSignal West Dura Extended Duration substrate (Thermo Fisher Scientific, MA, U.S.A). Protein expression was normalized against GAPDH expression.

Synthesis of PEG-b-PNHS polymer

4-Cyano-4-(thiobenzoylthio) pentanoic acid (10.2 mg, 0.0366 mmol), AIBN (2 mg, 0.0124 mmol), N-Succinimidyl methacrylate (430 mg, 2.42 mmol), PEG₉₅₀ (356 mg, 0.375 mmol), and 2 mL of dried tetrahydrofuran were added into a Schlenk tube. Then the mixture was filled with N₂ and stirred at 80 °C for overnight. The reaction was quenched, and the mixture was precipitated in ethanol once and diethyl ether twice, separately. PEG-b-PNHS was collected as precipitate and dried. Conversion of PEG₉₅₀ polymer was 76% and the conversion of N-Succinimidyl methacrylate was 97%.

Synthesis of PASA polymer

PEG-b-PNHS (303 mg, 1 mmol NHS), 5-ASA (459 mg, 3 mmol) and TEA (416.2 μL, 3 mmol) were dissolved in DMSO (10mL) and stirred at 37 °C. After 48h reaction, the mixture was dialyzed against DMSO for two days, followed by dialysis against water for three days. The PASA polymer was obtained after lyophilization.

Synthesis of folate-PEG-NH₂-conjugated PASA (FASA) polymer

The PEG_3.5K-FA was first synthesized according to a previously published method[18]. PEG_3.5K-FA (78 mg, 0.02 mmol), PEG-b-PNHS (303 mg, 1 mmol NHS), and TEA (27.5 μL, 0.1 mmol) were then dissolved in DMSO (10 mL) and stirred at 37 °C for 48h. Then 5-ASA (459 mg, 3 mmol) and TEA (416.2 μL, 3 mmol) were added and the reaction mixture was stirred for another 48h. The mixture was dialyzed against DMSO for two days, followed by dialysis against water for three days. The FASA polymer was obtained after lyophilization.

Chemical characterization of synthesized polymer

¹H-NMR spectrum of synthesized polymer was examined on a Varian-400 FT-NMR spectrometer at 400.0 MHz with DMSO-d₆ as the solvent.

Preparation and physicochemical characterization of blank or drug-loaded micelles

Blank and DOX-loaded micelles were prepared by film hydration method. DOX solution (5 mg/mL) was first prepared by dissolving DOX·HCl in a mixture of dichloromethane/methanol (1:1, v/v) containing triethylamine (5 equiv.). Then DOX solution was mixed with PASA or FASA (5 mg/mL in dichloromethane) at different polymer/drug weight ratios. The solvent was removed by nitrogen flow, followed by 2 h in vacuum to further remove remaining solvent. The thin film formed was hydrated in 0.1 M PBS to give DOX-loaded micelles. The size distribution of prepared micelles was measured via dynamic light scattering (DLS) method. DOX concentrations in micelles were determined by Waters Alliance 2695 Separations Module combined with Waters 2475 Fluorescence Detector (excitation, 480 nm; emission, 510 to 620 nm; gain, 3; sensitivity (FUFS), 10,000). Drug loading capacity (DLC) and drug loading efficiency (DLE) of DOX were calculated according to the following equations: DLC (%) = [weight of loaded drug/(weight of polymer + input drug)] ×100%, DLE (%) = (weight of loaded drug/weight of input drug) × 100%. The colloidal stability of micelles was monitored at room temperature by following the changes in sizes of the particles or visible precipitates every hour in the first 12 h and daily after 12 h following sample preparation. The absorbance spectra of DOX, PASA and PASA/DOX were collected using a Varian 50 Bio UV-Vis spectrophotometer.

Critical micelle concentration (CMC) of PASA and FASA micelles

The CMC of PASA and FASA polymer was determined by fluorescence measurement using nile red as a fluorescence probe as described previously[19]. Briefly, nile red dichloromethane solution (0.05 mg/mL) was added to the test tubes and then the solvent was removed by evaporation at room temperature. Then, 2 mL of PASA or FASA micelles ranging from 1×10⁻⁴ to 5×10⁻¹ mg/mL was added to each tube with nile red respectively. The micelles were kept overnight to allow the solubilization equilibrium of nile red. Excitation was carried out at 550 nm with emission recorded from 570 to 720 nm wavelength.

Gel retardation assay

PASA/DOX micelles of different weight ratios (ranging from 1:1 to 20:1; DOX concentration was fixed at 0.5 mg/mL) were prepared at different pH (7.4 or 5) by film hydration method as mentioned above. These micelles were electrophoresed on agarose gel in Tris-acetate-EDTA (TAE) buffer of corresponding pH at 120 V for 20 min and the gel was subsequently visualized using a UV illuminator. Free DOX was used as a control.

In vitro drug release

The release of DOX from DOX-loaded PASA and FASA micelles at different pH was studied using a dialysis method. Briefly, 2 mL of PASA/DOX and FASA/DOX micelles containing 1 mg of DOX and 10 mg of polymer were placed in a dialysis bag (MWCO 3.5 kDa) and immersed into 40 mL of 0.1 M PBS solution containing 0.5% (w/v) Tween 80 at pH 5 and pH 7.4. The experiment was performed in an incubation shaker at 37 °C at 100 rpm. At selected time intervals, 10 μL solution in the dialysis bag and 1 mL medium outside the dialysis bag were withdrawn while same amount of fresh dialysis solution was added for replenishment. The concentration of DOX was examined by fluorescence spectrometry (excitation, 480 nm; emission, 510 to 620 nm). Free DOX was included as control.

Cellular uptake study

KB, 4T1.2, CT26, MDA-MB-468, HCT116 tumor cells and M2 macrophages were seeded to 6-well plates (3×10⁵/well), respectively. After overnight incubation, the culture medium was replaced by fresh medium containing free DOX, PASA/DOX and FASA/DOX micelles with or without 100 μM free folate, respectively, at an equivalent DOX concentration of 6 μg/mL (carrier/DOX ratio: 10/1 (w/w)). After incubation for 30 min at 37 °C, cells were washed with cold PBS and fixed with PBS containing 4% (w/v) formaldehyde. Nuclei were then stained by DAPI for 5 min. Cells were washed with cold PBS and observed under fluorescence microscope (BZ-X710, Japan).

Cellular uptake of different DOX formulations was also quantified by flow cytometry. KB tumor cells seeded in 6-well plates (3×10⁵/well) were treated with various DOX formulations as described above at a DOX concentration of 6 μg/mL. Following incubation at 37 °C for 30 min, cells were washed with cold PBS, fixed in PBS containing 4% (w/v) formaldehyde, and resuspended in 500 μL PBS for flow cytometry analysis with CyAn ADP Analyzer (Beckman Coulter, Inc.). Fluorescence was examined at an excitation wavelength of 480 nm and an emission wavelength of 570 nm. 2×10⁴ events were collected for each sample.

In vitro cytotoxicity

Cytotoxicity assay was performed on different cancer cell lines (KB, 4T1.2, CT26, MDA-MB-468 and HCT116). Cells were seeded in 96-well plates at a density of 5×10³ cells/well with 100 μL of complete culture medium (DMEM or RPMI-1640 with 10% FBS and 1% streptomycin/penicillin).

To evaluate the combination effect of 5-ASA and DOX, cells were treated with various concentrations of free 5-ASA, DOX and the combinations, and MTT assay was performed 48 h later. The absorbances of each well were measured at 590 nm and the cell viability was determined via the following formula: (OD_treated-OD_blank)/(OD_control-OD_blank) × 100%

The cytotoxicity of PASA/DOX and FASA/DOX at a carrier/DOX ratio of 10/1 (w/w) were compared to free DOX and Doxil at various DOX concentrations. FASA control was added to cells at concentrations equivalent to the amounts of carrier in the corresponding DOX formulations. In order to confirm folate-mediated active targeting, free folate (100 μM) was added along with the FASA/DOX micelles. Cells were incubated for 30 min in drug-containing medium and then cultured for another 48 h in fresh medium prior to MTT assay.

Tissue biodistribution

For in vivo tissue biodistribution study, 4T1.2 and CT26 tumor bearing mice (~300 mm³) were intravenously (i.v.) injected with free DOX, DOX-loaded PASA and FASA micelles (carrier/DOX weight ratio: 10/1), respectively. The mice were sacrificed and perfused at 24 h post injection. Tumors and major organs including heart, liver, spleen, lung, and kidney were sectioned and observed under the fluorescence microscope (BZ-X710, Japan).

In vivo therapeutic study

In vivo antitumor efficacy of DOX-loaded PASA and FASA micelles was tested in syngeneic mouse breast (4T1.2) and colon (CT26) cancer models, respectively. Female BALB/c mice (4-6 weeks) were subcutaneously (s.c.) inoculated with 4T1.2 or CT26 cells (5×10⁵ cells per mouse). When the tumor volume reached ~ 50 mm³, mice were randomly divided into eight groups (n = 5), and treated via tail vein injection with PBS (control), DOX, 5-ASA+DOX, Doxil, blank PASA micelles, blank FASA micelles, DOX-loaded PASA micelles or DOX-loaded FASA micelles, respectively once every three days for three times (polymer: 50 mg/kg, DOX: 5 mg/kg, 5-ASA: 20 mg/kg). Tumor sizes were monitored every three days following the initiation of the treatment and calculated by the formula: (Length × Width²)/2. The relative tumor volume was calculated by the tumor size measured each time normalized by the tumor size prior to the 1^st treatment. Body weights were also followed as an indication of toxicity. After completion of the experiment, tumor and major organs were collected for hematoxylin and eosin (H&E) staining. Blood samples were collected for biochemical analysis of alanine transaminase (ALT) and aspartate aminotransferase (AST).

To evaluate the synergistic effects of anti-PD-1 and FASA/DOX, a syngeneic CT26 colon tumor model was established by inoculating 5×10⁵ CT26 cells into the flank of BALB/c mice. When the tumor volume reached ~ 100 mm³, mice were randomly grouped (n = 5), and treated with PBS (control), PD-1 antibody (BioCell), FASA/DOX, and anti-PD-1+FASA/DOX, respectively, every three days for a total of three times (polymer: 25 mg/kg, DOX: 2.5 mg/kg, anti-PD-1: 5 mg/kg). FASA/DOX and anti-PD-1 were given i.v. and intraperitoneally (i.p.), respectively. Tumor volumes were monitored every three days and calculated as described above. Body weights were also followed as an indication of systemic toxicity. After completing the in vivo experiment, tumor tissues and major organs were collected for histochemical staining. Blood sample were collected for ALT and AST analysis.

Quantification of tumor-infiltrating immune cells

CT26 tumor bearing BALB/c mice received various treatments with PBS as control via tail vein injection once every three days for three times. Tumors and spleens were harvested at 24 h after the last treatment. Single cell suspensions were prepared and stained for CD4, CD8, IFN-γ, Granzyme B, FoxP3 and macrophage (F4/80 and CD206) for flow cytometry analysis[20]. The abundance of immune cells in each group normalized by the number of control group was presented as relative abundance of immune cells.

RNA-Seq analysis

CT26 tumor bearing BALB/c mice received FASA or FASA/DOX treatment with PBS as control via tail vein injection once every three days for three times. Tumors were harvested at 24 h after the last treatment. Samples were sent to Health Sciences Sequencing Core, University of Pittsburgh for RNA extraction, library construction and sequencing. RNA-seq data were aligned to mouse reference genome GRCm38 using STAR. Gene expression levels were quantified, and count expression matrices were generated using RSEM from aligned reads. Count per million was used for further analysis.

Isolation and PGE₂ treatment of CD4⁺/CD8⁺ T cells

Spleen was harvested from BALB/c mice. Single cell suspensions were prepared. CD4⁺ and CD8⁺ T cells were isolated by CD4/CD8 MicroBeads (Miltenyi Biotec). Cells were cultured in RPMI-1640 medium and stimulated by anti-CD3 (Invitrogen) and anti-CD28 (Invitrogen) in the presence or absence of various concentrations of PGE₂ (TCI) (10nM, 100nM, 1000nM). Cells were stained for PD-1 for flow cytometry analysis after 24 h treatment.

Histopathological Analysis

Tumors and major organs including heart, liver, spleen, lung and kidney were excised and fixed in PBS containing 10% formaldehyde after completion of the in vivo therapy study, followed by embedment in paraffin. The paraffin embedded samples were sectioned into slices at 4 μm using an HM 325 Rotary Microtome. The tissue slices were then subjected to H&E staining for histopathological examination under a Zeiss Axiostar plus Microscope (PA, USA).

ALT and AST assessment

Mouse serum was obtained for blood biochemical assessment. ALT and AST were measured by ALT/SGPT or AST/SGPT liqui-UV assay kit following manufacturer’s protocols.

Prostaglandin E2 level analysis

To analyze the levels of prostaglandin E2 (PGE₂) production in different cell lines, the cell culture mediums were collected after overnight incubation. To test PGE₂ in the tumor tissues, the harvested tumors were homogenized, and the supernatants were obtained after centrifugation. PGE₂ in the supernatants was detected using Abcam Prostaglandin E2 ELISA Kit. The relative PGE₂ level was presented as the levels in each group normalized by the average value of the group with a lowest average level.

Statistical analysis

All values were presented as mean ± standard error of mean (SEM). Statistical analysis was performed with two-tailed Student’s t-test for comparison between two groups and one-way analysis of variance (ANOVA) for comparison between multiple groups. Results were considered statistically significant if p < 0.05.

Results

1. Expression levels of Ptgs1/2 in various cancer cell lines:

To elucidate a potential role of tumor cells-derived COX-1/2 (encoded by Ptgs1/2) in 5-ASA-mediated antitumor activity, we first examined the mRNA expression levels of Ptgs1 and Ptgs2 in four cancer cell lines including murine breast cancer cell line 4T1.2, murine colon cancer cell line CT26, human breast cancer cell line MDA-MB-468 and human colon cancer cell line HCT116. As shown in Fig. 1A-B, all four cancer cell lines examined had higher levels of Ptgs2 mRNA compared to their Ptgs1 mRNA counterparts. For either colon or breast cancer type, the murine cell line examined expressed higher levels of both Ptgs1 and Ptgs2 compared to the human cell line of the same cancer type. Finally, the murine colon cancer cell line CT26 showed higher mRNA level of COX-2 than the murine breast cancer cell line 4T1.2. The protein expression levels of COX-1 and COX-2 in the four cell lines were consistent with their mRNA expression levels (Fig. 1C). Fig. 1D show that the supernatants from CT26 cells had the highest level of PGE₂. The amounts of PGE₂ in the culture medium from the different cell lines follow the order of CT26 > 4T1.2 > HCT116 > MDA-MB-468. These data suggest that the COX enzymatic activities in the four cell lines examined were consistent with their expression at mRNA and protein levels.

Fig. 1. COX-1/2 levels and in vitro cytotoxicity of 5-ASA and DOX in tumor cell lines.

(A-B) Ptgs1 (A) and Ptgs2 (B) mRNA expression of different tumor cells (4T1.2, CT26, MDA-MB-468 and HCT116). (C) COX-1 and COX-2 protein expression of different tumor cells (4T1.2, CT26, MDA-MB-468 and HCT116). (D) Relative PGE2 levels of 4T1.2, CT26, MDA-MB-468 and HCT116 in the cell culture supernatant. (E-H) Proliferation inhibition of 4T1.2 (E), CT26 (F), MDA-MB-468 (G) and HCT116 (H). Tumor cell lines were treated with various concentrations of free 5-ASA (blue line), free DOX (red line) or the combination of 5-ASA and DOX (green line). After 48h, the cytotoxicity was determined by MTT assay. All data represent the means ± SEM (n=3). p values were determined by two-tailed Student's t-test. *p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001, ns, no significance.

2. Cytotoxicity of 5-ASA, alone or in combination with DOX in various cancer cell lines

Fig. 1E-H show the in vitro cytotoxicity of 5-ASA, alone or in combination with DOX in the four cancer cell lines. A concentration range of 0 ~ 1mM was chosen as the production of PGE₂ can be inhibited by 80% when 5-ASA concentration reached 1 mM (Fig. S1). 5-ASA did not show any significant anti-proliferation effect on all four cell lines examined (Fig. 1E-H) at the concentrations used. In addition, no improvement in cytotoxicity was noticed upon combination of 5-ASA with DOX. Several studies have demonstrated that the anti-inflammatory activity from COX inhibition can suppress tumor growth in vivo, either alone or in combination with other treatments[13, 21]. Therefore, we went to further examine the in vivo antitumor activity of 5-ASA, alone or in combination with DOX, in two murine cancer models (4T1.2 and CT26) that have different levels of COX activities. To better elucidate a role of tumor-derived PGE₂ in the antitumor effect, a PASA-based micellar carrier was designed to achieve enhanced selective delivery of 5-ASA or codelivery of 5-ASA and DOX to tumor tissues. In addition, folate was introduced into the carrier to further improve the selective delivery to tumors.

3. Synthesis and characterization of PASA and FASA polymers

The synthesis routes of PASA and FASA were shown in Fig. S2. First, PEG-b-PNHS polymer was synthesized through reversible addition-fragmentation chain-transfer (RAFT) polymerization with PEG₉₅₀ and N-Succinimidyl Methacrylate. 5-ASA was subsequently conjugated to PEG-b-PNHS to yield PASA. FASA polymer was also obtained through reaction of PEG-b-PNHS with folate-PEG-NH₂ followed by reaction with 5-ASA. A relatively long PEG spacer (3.5 K) was introduced between folate and PASA polymer to overcome any potential steric hindrance for interaction with folate receptor (FR) on FR-overexpressing tumor cells. All the peaks of the polymers, PEG-b-PNHS, PASA, FA-PEG-b-PNHS and FASA (Fig. S3-S6), were well assigned in ¹H-NMR. For PEG-b-PNHS polymer, the average degree of polymerization of the PEG₉₅₀ monomer was calculated to be 8 according to the conversion of PEG₉₅₀ monomer. The average units of the NHS monomer were determined to be 64 (Fig. S3). After conjugation of 5-ASA to PEG-b-PNHS polymer, the characteristic peaks of benzene ring were observed in the ¹H-NMR at 6.56, 7.39 and 7.85 ppm (Fig. S4). By comparing these three peak intensities at 6.0-8.0 ppm with the proton intensities of methoxy in PEG at 3.24 ppm, the average ratio of 5-ASA to PEG₉₅₀ was about 4:1, both in PASA and FASA (Fig. S4&S6). The drug loading of 5-ASA is 39.2% and 31.4% in PASA and FASA polymer, respectively. The molar substitution of folate in FASA was ~2%, which was identified by methylene of PEG (3.51 ppm) linked to folate (Fig. S6).

4. Physicochemical characterization of blank and DOX-loaded micelles

Both PASA and FASA are amphiphilic molecules and can potentially self-assemble to form micellar carriers that can load other hydrophobic drugs, suggesting a 5-ASA prodrug-based new carrier platform for codelivery of 5-ASA and other drugs such as DOX (Fig. 2A). Fig. 2B shows that PASA and FASA had a critical micelle concentration (CMC) of 0.0033 mg/mL and 0.0038 mg/mL, respectively. The relatively low CMCs suggest a likely excellent stability of PASA and FASA micelles after dilution in blood following i.v. administration. Fig. 2C shows that PASA and FASA polymers formed micellar particles of around 197.6nm and 197.5nm, respectively. Interestingly, incorporation of DOX into micelles resulted in significant decreases in nanoparticle sizes and the sizes of the particles decreased gradually with an increase in the DOX/polymer ratio (Table S1). At a polymer/DOX ratio of 1/1 (w/w), the sizes of DOX-loaded PASA and FASA were 67.9 and 73.9 nm, respectively. We hypothesized this is likely due to several interactions between the polymer and DOX, including strong ionic interaction, ∏-∏ stacking and hydrophobic interaction, leading to the formation of a more compact structure. Indeed, the UV-Vis spectrum (Fig. 2D) showed a characteristic absorbance of free DOX at approximately 482 nm, whereas a 13 nm red-shift was observed in the PASA/DOX, likely due to Van der Waal’s interaction between the polymer and DOX[22]. The DOX electrophoresis (Fig. 2E) results showed that, under physiological pH (7.4), DOX stayed associated with PASA carrier when the PASA/DOX ratio reached 5:1 or higher. In contrast, an obvious release of DOX was observed at pH 5 even at a PASA/DOX ratio as high as 20/1, which might be ascribed to a significant reduction in the ionization of 5-ASA at an acidic condition. Table S1 shows the biophysical properties of DOX-loaded micelles at various carrier/drug ratios. At a PASA/DOX weight ratio of 1/1, an unprecedently high DLC of 42.28% was achieved with a DLE of 82.18%. The DOX-loaded micelles were stable at room temperature for over seven days (Table S1). Similar results were obtained for FASA/DOX (Table S1). Similar results were also obtained when PASA was used to carry imatinib that has several amines in the structure (data not shown). However, PASA was not effective in carrying several other hydrophobic agents tested such as paclitaxel and curcumin, highlighting the importance of ionic interaction for our PASA-based nanocarrier.

Fig. 2. Biophysical characterizations of 5-ASA based micelles.

(A) Schematic illustration of self-assembled FASA/DOX micelles. DOX and 5-ASA are located in the core of the nanoparticles with PEG and PEG-FA shielding outside. (B) Critical micelle concentration (CMC) of PASA and FASA polymer. (C) Size distribution of blank PASA, FASA and PASA/DOX, FASA/DOX micelles at a carrier/drug ratio 10/1 (mg/mg). (D) UV/Vis absorbance spectra of DOX, PASA/DOX and PASA in aqueous solution. Carrier/drug ratio was at 10/1 (mg/mg). (E) Gel electrophoresis of PASA/DOX at various ratios. (F) Cumulative DOX release profile of PASA/DOX and FASA/DOX micelles under different pH with free DOX as control. (upper panel) 0-72h and (bottom panel) 0 to 12h. DOX concentration was fixed at 0.5 mg/mL. Values reported are the means ± SEM for triplicate samples.

5. In vitro drug release profile

The release kinetics of DOX from DOX-loaded PASA and FASA micelles was evaluated by dialysis method under physiological (pH 7.4) or acidic (pH 5) condition at 37°C. As depicted in Fig. 2F, free DOX was rapidly diffused across the dialysis membrane. Under physiological pH (7.4), less than 10% of DOX was released from PASA or FASA micelles in 2 h and a slow release of DOX was extended over 72 h. Instead, DOX was released much more rapidly under pH 5: around 20% of DOX was released in 2 h, and over 40% was released in 12 h. These data are consistent with the results of electrophoresis, likely due to a disruption of the interaction of the carrier with DOX by hydrogen ion under acidic condition. This pH-sensitive DOX release profile of our system suits well its application for drug delivery to tumors via i.v. route due to its excellent stability in blood and accelerated drug release upon reaching the acidic tumor environment, particularly the more acidic endosomal/lysosomal compartment after intracellular delivery.

6. In vitro cellular uptake

KB cells were chosen to investigate the cellular internalization of various DOX formulations, particularly the folate-mediated active targeting as these cells are known to overexpress folate receptor α (FRα)[23]. KB cells were treated with different DOX formulations with or without free folate at 37 °C and then observed by fluorescence microscope. After 30 min treatment free DOX was efficiently taken up by KB cells and the fluorescence signal was largely found in the nucleus (Fig. 3A). Incorporation of DOX into PASA micelles led to a decrease in the cellular uptake. However, the cellular uptake was significantly improved following conjugation with folate. The level of cellular uptake of FASA/DOX was even higher than that of free DOX. The improvement in cellular uptake was substantially abolished in the presence of excess amount of free folate (100 μM) and the cellular uptake decreased to a level comparable to that of free DOX or PASA/DOX. Free folate had no effect on the uptake of free DOX or PASA/DOX, suggesting that the enhanced cellular uptake of FASA/DOX was largely mediated by the FR. These data demonstrated that the conjugated folate ligand increased the endocytosis of FASA/DOX through folate receptor pathway in FR overexpressing cell line.

Fig. 3. In vitro cellular uptake and cytotoxicity of DOX-loaded PASA micelles in various cancer cell lines.

(A-C) Fluorescence microscope images of KB (A), 4T1.2 (B) and CT26 (C) cell lines after incubation with different DOX formulations for 30 min. DOX concentration was at 6 μg/mL. Free folate was at 100 μM. The scale bar is 50 μm. (D-F) Cytotoxicity of DOX-loaded 5-ASA based micelles and cell viability at highest concentration (histogram) in KB (D), 4T1.2 (E) and CT26 (F) cell lines after 48h treatment. The experiments were performed in triplicate and repeated three times. Data are presented as means ± SEM. p values were determined by two-tailed Student's t-test. *p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001.

The cellular uptake of various DOX formulations was further investigated by flow cytometry (Fig. S7). The data of the quantitative flow cytometric assay were consistent with the results of fluorescence microscopic examination. Similar results were observed in other four tumor cell lines, 4T1.2, CT26, MDA-MB-468, and HCT116 (Fig. 3B-C, Fig. S8A-B). However, a much lower magnitude of improvement in cellular uptake was seen for the folate-decorated FASA/DOX in all the four cell lines examined, likely due to much lower expression levels of the FR in these cell lines compared to KB cells.

7. In vitro cytotoxicity of DOX-loaded PASA micelles

The in vitro cytotoxicity of various DOX formulations was evaluated in several cell lines by MTT assay. To mimic the in vivo setting where non-targeted agents are likely to interact with tumor cells for a relatively short period of time, we treated cells for 30 min with drug-containing medium, followed by continuous culture for another 48h in drug-free fresh medium. Fig. 3D shows the cytotoxicity in KB cells. FASA was not active in inhibiting the tumor cell proliferation, even at the highest concentration of 200 μg/mL. DOX inhibited the cell proliferation in a dose-dependent manner. PASA/DOX was comparable to free DOX in potency. Coupling of folate to PASA/DOX led to a significant improvement in cytotoxicity and FASA/DOX was more effective than free DOX at several DOX concentrations tested. However, the improvement in cytotoxicity of FASA/DOX was essentially abolished in the presence of excess amount of free folate, which was consistent with the data of cellular uptake. Similar results were observed in 4T1.2, CT26, MDA-MB-468 and HCT116 tumor cells (Fig. 3E-F, Fig. S8C-D). Yet, the level of improvement in cytotoxicity for FASA/DOX varied among the four tumor cell lines tested, likely due to the different expression levels of FR in these cell lines. It was also noticed that PASA/DOX was less active than free DOX in the four tumor cell lines, which was consistent with a relatively lower level of cell uptake of PASA/DOX as shown in the cellular uptake study. Similar to the data in KB cells, FASA alone was essentially not active in all the four tumor cell lines.

8. Biodistribution study

The biodistribution of DOX in tumors and other major organs was examined following i.v. administration of free DOX, PASA/DOX and FASA/DOX, respectively. Both 4T1.2 and CT26 tumor models (s.c.) were investigated. Tumors and major organs were harvested for fluorescence microscopic examination at 24 h post-injection (Fig. 4A-B, Fig. S10-S14). Fig. 4A shows the data from 4T1.2 tumor model. Low level and scattered DOX fluorescence signals were observed in tumor tissues 24 h following a single injection of free DOX. The DOX signals were significantly stronger in PASA/DOX-treated tumors compared to those in free DOX-treated tumors. Incorporation of folate led to a further improvement in DOX accumulation at tumor tissues. In addition to an overall higher level of DOX signals, a more widespread DOX distribution was observed in FASA/DOX-treated tumors. Similar results were observed in CT26 tumor model (Fig. 4B). In addition, the targeting efficiency of PASA or FASA was similar in the two tumor models judging from the comparable levels of DOX fluorescence intensity in the corresponding treated groups.

Fig. 4. Biodistribution of DOX and in vivo therapeutic efficacy in syngeneic murine breast cancer model (4T1.2) and colon cancer model (CT26).

(A-B) Fluorescence microscopic examination of DOX distribution in 4T1.2 (A) and CT26 (B) tumor sections at 24 h after treatment with free DOX, DOX-loaded PASA and FASA micelles, respectively. The scale bar is 100 μm. (C-D) Relative 4T1.2 (C) and CT26 (D) tumor volume (n=5) changes in mice treated with various formulations. (E-G) Ptgs1 (E), Ptgs2 (F) mRNA expression and relative PGE₂ levels (G) in 4T1.2 and CT26 untreated tumor tissue. Values reported are the means ± SEM, n = 3. * p < 0.05; ** p < 0.01, *** p < 0.001.

9. In vivo antitumor efficacy and safety profile

For in vivo efficacy study, murine breast cancer 4T1.2 and colon cancer CT26 models were used. When the tumors reached about 50 mm³, mice were treated with various formulations every three days in the timeframe of a week (day 0, 3, 6). The effect of the treatments was followed up every three days by tumor volume measurement. Fig. 4C shows the results for 4T1.2 tumor model. PASA alone slightly inhibited the tumor growth and its antitumor activity was slightly improved following conjugation with folate. Free DOX exhibited a modest antitumor activity and its combination with free 5-ASA led to a slight improvement in efficacy. Doxil, a clinical liposomal DOX formulation, was more effective than the free 5-ASA+DOX combination but less effective compared to DOX formulated in PASA. Decoration of PASA/DOX with folate led to a further improvement in antitumor activity, the most effective one among all treatment groups. In CT26 tumor model (Fig. 4D), a more dramatic antitumor activity was observed for PASA or FASA alone, particularly FASA, its efficacy being similar to that of Doxil. Delivery of DOX via PASA, particularly FASA also led to more effective tumor growth control in CT26 model compared to 4T1.2 tumor model. The data of tumor weights and histopathological analysis of tumor tissues (Fig. S15) were consistent with the tumor growth curves.

Fig. 4E-G show the mRNA expression levels of Ptgs1/2 genes and the levels of PGE₂ in 4T1.2 and CT26 tumor tissues. Consistent with data from cultured tumor cells (Fig. 1B-D), CT26 tumor tissues showed higher mRNA expression levels of Ptgs2 gene and produced greater amounts of PGE₂ compared to 4T1.2 tumor tissues. However, CT26 tumor tissues showed higher mRNA levels of Ptgs1 compared to 4T1.2 tumor tissues (Fig. 4E-G), which is different from the data with cultured tumor cells (Fig. 1A-C). To elucidate the possible reason for this inconsistency, we isolated the tumor cells and stromal cells from the tumor tissues and examined the expression levels of COX-1/2 in the two cell subpopulations at both mRNA and protein levels. As shown in Fig. S16, the expression of COX-1 at both mRNA and protein levels in isolated 4T1.2 tumor cells was comparable to those in isolated CT26 tumor cells, which is consistent with in vitro data (Fig. 1). On the other hand, the mRNA expression levels of Ptgs1 in CT26 stromal cells were significantly higher than those in 4T1.2 stromal cells (Fig. S16). These data suggest that the inconsistency in the expression levels of COX-1 of the two tumor models in vitro and in vivo are likely attributed to the differences in the COX-1 expression levels in their stromal cells. CT26 stromal cells also expressed higher mRNA levels of Ptgs2 compared to 4T1.2 stromal cells (Fig. S16). These data, together with the data of in vivo therapy, suggest that high level of Ptgs1/2 may play a more oncogenic role in CT26 model, rendering it more responsive to FASA- or FASA/DOX-based therapy.

It has been reported that some chemotherapeutic agents like DOX can induce immunogenic cell death (ICD) [24]. To examine whether ICD was similarly induced by our DOX formulation, we examined the expression of calreticulin (CRT) and high mobility group box 1 (HMGB1), two makers of ICD[25], after FASA/DOX treatment, both in vitro and in vivo. As shown in Fig. S17A-B, the expression of CRT and HMGB1 was significantly induced following different DOX treatments including free DOX, 5-ASA+DOX and FASA/DOX in two cell lines (4T1.2 and CT26), compared to control and FASA-treated group. Scattered fluorescence signals were detected in tumor tissues treated with free DOX or free 5-ASA+DOX combination. The fluorescence signals of the two markers were dramatically increased in FASA/DOX-treated tumors, likely due to the increased delivery of DOX by FASA (Fig. S17C-D).

Fig. S18-S20 show the results of preliminary toxicity evaluations. Mice treated with free DOX, alone or in combination with 5-ASA experienced a slight decrease in body weights on day 9 (Fig. S18). The serum levels of AST in the free DOX group (DOX or 5-ASA+DOX) were also significantly higher than those in the control group, suggesting a DOX-related hepatotoxicity (Fig. S19). Moreover, hepatocellular vacuolation was found in mice treated with free DOX, alone or in combination with 5-ASA (Fig. S20). On the other hand, both PASA/DOX and FASA/DOX were well tolerated in mice as manifested by normal body weights and minimal changes in blood levels of AST and ALT as well as liver and heart histology (Fig. S18-S20). These data suggest decreased toxicity of DOX following incorporation into 5-ASA-based nanocarrier.

10. Tumor immune environment evaluation

The lack of an obvious cytotoxic effect of 5-ASA or PASA in vitro but significant antitumor activity in vivo suggests that 5-ASA or PASA exerts its antitumor activity likely via a mechanism independent of its direct effect on tumor cells. PGE₂, a pro-inflammatory cytokine which is produced by COX, gives free rein to cancer immune evasion and immunotherapy resistance[8]. Therefore, we went to examine the impact of various treatments on the PGE₂ production in CT26 tumor tissues. Their impact on tumor immune microenvironment was also investigated. As shown in Fig. 5A, FASA treatment led to a drastic reduction in the level of PGE₂ in CT26 tumor tissues while free 5-ASA had minimal effect. Lack of any effect from free 5-ASA is likely due to its rapid elimination and thus limited accumulation at tumor tissue. It is also apparent that codelivery of DOX via FASA resulted in a further decrease in the tissue levels of PGE₂.

Fig. 5. Changes in CT26 tumor immune microenvironment after various treatments.

(A) PGE₂ levels in tumor tissues after different treatments. (B) Percentages of TAM population (M1, M2 and M1/M2 ratio) in tumor tissues. (C-F) T-cell subpopulations in tumors. Relative abundance of CD4⁺ (upper) and CD8⁺ (bottom) T cells (C), IFN-γ⁺ intratumoral CD4⁺ and CD8⁺ T cells (D), granzyme B⁺ CD8⁺ T cells (E) and FoxP3⁺ T regulatory cells (F) following different treatments. Bars represent means ± SEM. *p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001, ns, no significance.

Fig. 5B shows that treatment with FASA or FASA/DOX led to downregulation of M2 type macrophages, while M1 type macrophages and M1/M2 ratios were significantly increased, suggesting that the tumor infiltrating macrophages were polarized from a tumor-promoting to a tumor-suppressing phenotype.

There were minimal changes in the total numbers of CD4⁺ and CD8⁺ T cells after the different treatments (Fig. 5C); however, the numbers of IFN-γ⁺ CD4⁺ and IFN-γ⁺ CD8⁺ T cells were significantly increased following treatment of FASA or FASA/DOX, particularly the latter treatment (Fig. 5D). IFN-γ is a pleiotropic cytokine that can eliminate tumor cells directly and indirectly[26]. Therefore, FASA or FASA/DOX treatment increased the number of functional CD4⁺ and CD8⁺ T cells although the total number of CD4⁺ and CD8⁺ T cells were not affected. The numbers of GzmB⁺ CD8⁺ T cells were also increased following the different treatment, particularly FASA/DOX (Fig. 5E).

T_reg is immunosuppressive T cell and generally contributes to inhibition of effector T cells[27]. The number of T_reg was significantly decreased after treatment with FASA/DOX. However, all other treatments including FASA alone had no impact on the numbers of T_reg (Fig. 5F). Overall, we observed a significant improvement in the tumor immune microenvironment following treatment with FASA, particularly FASA/DOX.

To elucidate the underlying mechanism for the 5-ASA-mediated improvement in the tumor immune microenvironment, we conducted RNA-Seq of CT26 tumor following treatment with FASA and FASA/DOX, respectively (Fig. S21). Preliminary analysis shows upregulation of Cxcl10 and downregulation of Ccl22 by both FASA and FASA/DOX. Cxcl10 has been shown to be involved in the recruitment and activation of Th1 cells[28], while Ccl22 is involved in the recruitment of T_reg[29]. More studies are needed to further establish the roles of Cxcl10/Ccl22 in 5-ASA-induced changes in tumor immune microenvironment.

11. Effect of inhibition of PGE₂ on the expression of PD-1 in T cells

Despite the favorable changes in various immune cell subsets as described above, treatment with FASA or FASA/DOX dramatically increased the expression of PD-1 on the surface of CD4⁺ and CD8⁺ T cells (Fig. 6A-B). Upregulation of PD-1 was also observed on CD8⁺ T cells after treatment with free 5-ASA+DOX combination, although less dramatically compared to treatment with FASA or FASA/DOX. We isolated CD4⁺ and CD8⁺ T cells from BALB/c mice and treated them with PGE₂ at different concentrations. PGE₂ inhibited PD-1 expression in a dose dependent manner (Fig. 6C). The expression of PD-1 on T cells was dramatically decreased at a PGE₂ concentration as low as 10 nM. It is possible that upregulation of PD-1 on T cells following FASA or FASA/DOX treatment is attributed, at least partially, to the attenuation of the repression effect of PGE₂ on PD-1 expression of T cells.

Fig. 6. PD-1 expression on CD4⁺ and CD8⁺ T cells and in vivo synergistic antitumor activity of PD-1 blockade with FASA/DOX treatment in CT26 tumor model.

(A-B) Percentage of PD-1⁺ CD4⁺ (A) and PD-1⁺CD8⁺ (B) T cells in CT26 tumor tissues after treatment. (C) PD-1 expression on CD4⁺ and CD8⁺ T cells after PGE₂ treatment. (D-E) Average (D) and individual (E) tumor growth curves in control and treated group (n = 5). All data are means ± SEM. *p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001.

We also examined the PD-L1 expression in CT26 tumor tissues and cultured CT26 cells after different treatments. There are varied reports on the impact of DOX on the PD-L1 expression on tumor cells. While DOX decreased the expression in MDA-MB-231 cell line[30], it was shown to induce PD-L1 expression in HCT116 cell line[31]. Fig. S22 shows that FASA alone had no effect on the expression of PD-L1 in cultured CT26 cells or CT26 tumor in vivo. However, there was an obvious upregulation of PD-L1 expression at both mRNA and protein levels following different DOX treatments including free DOX, 5-ASA+DOX and FASA/DOX (Fig. S22). In addition, PD-L1 upregulation after different DOX treatments was seen with both cultured CT26 cells and CT26 tumors in vivo (Fig. S22).

12. Combination therapy of FASA/DOX and anti-PD-1

PD-1 is a protein on the surface of cells that prevent the immune system from killing cancer cells[32]. This prompted us to examine the potential of combining FASA/DOX with anti-PD-1 antibody to further improve the overall therapeutic efficacy. In this study, treatment was initiated when the tumors reached a relatively large size of 100 mm³. In addition, FASA/DOX was given at a reduced DOX dose of 2.5 mg/kg. Anti-PD-1 antibody was given at a dose of 5 mg/kg once every three days for a total of three treatments. As shown in Fig. 6D, anti-PD-1 or FASA/DOX alone showed a modest antitumor activity. Combination of both led to a drastic improvement in the overall therapeutic efficacy. The growth of tumors was well controlled following the 1^st treatment. In addition, 4 out of 5 tumors completely regressed at day 18 following the 1^st treatment on gross examination, clearly demonstrating the therapeutic benefit of combining the two treatments (Fig. 6E). The histological analysis showed large nuclei in the tumor tissue with PBS treatment, while shrunk nuclei were observed in the tumor tissues with other treatments, especially the combination group (Fig. S25). All treatments were well tolerated as manifested by minimal changes in body weights and blood levels of AST and ALT, as well as normal histological morphology (Fig. S23-S25), indicating negligible toxicity of combination of FASA/DOX with anti-PD-1 treatment.

Discussion

COX has been well studied in terms of its role in tumorigenesis and progression including its impact on tumor immune microenvironment. However, the role of tumor-derived COX in COX inhibitors-mediated antitumor activity has not been well elucidated. To answer this question, we examined the cytotoxic effect of 5-ASA, alone or in combination with DOX, on several human and murine cancer cell lines with different expression levels of COX-1/2. We further examined the antitumor activity of 5-ASA+DOX in two syngeneic murine cancer models (4T1.2 and CT26) in vivo. To facilitate effective codelivery of 5-ASA and DOX in vivo, PASA polymer was designed based on 5-ASA structure (Fig. 2). The amino group in 5-ASA rendered it easy for conjugation to the polymer. As a 5-ASA prodrug, PASA could slowly release 5-ASA over a prolonged period to achieve sustained inhibition of COX. In addition, PASA could self-assemble to form a micellar carrier to co-deliver another drug such as DOX. Many carriers have been reported for delivery of DOX or codelivery of DOX and another drug[33, 34]. One major advantage of our nanocarrier lies in the unprecedented high DOX loading capacity (42.28%) (Fig. 2). This is likely attributed to strong interactions between PASA and DOX, including ionic interaction as well as hydrophobic/hydrophobic and ∏-∏ interactions (Fig. 2). These strong interactions led to the formation of highly compact nanoparticles as evidenced by the significantly reduced particle size following incorporation of DOX into PASA micelles (197.6 nm vs. 67.9 nm). These strong PASA/DOX interactions also explain a very slow release of DOX from DOX-loaded PASA micelles compared to many reported DOX micellar formulations. The excellent stability of DOX-loaded micelles together with their small sizes (~70 nm) contributed significantly to the effective accumulation of DOX in tumor tissue after systemic administration.

5-ASA, at concentrations effective in inhibiting PGE₂ synthesis, showed minimal cytotoxicity on all human and murine cancer cell lines tested (Fig. 1). No synergistic effect in cytotoxicity was seen in combination with DOX in vitro. On the contrary, an obvious combination effect was observed for two murine cancer models (4T1.2 and CT26) in vivo, particularly upon codelivery using PASA-based micellar carrier (Fig. 4). The improved therapeutic efficacy of PASA/DOX may be largely attributed to the enhanced delivery of 5-ASA and DOX to tumor tissues. We also noticed a further improvement in antitumor activity for both PASA and PASA/DOX following conjugation with folate. Folate receptor, particularly FRα, has been reported to be overexpressed in various types of human and murine cancers[35]. In addition, tumor macrophages, particularly M2 macrophages overexpress FR, mainly FRβ[36]. Our data suggest that folate ligand can facilitate cellular uptake of FASA by M2 macrophages (Fig. S9). Thus, the enhanced antitumor activity of FASA or FASA/DOX could benefit from targeting of both tumor cells and macrophages, particularly considering that tumor M2 macrophages also express a high level of COX-2. It should be noted that a more drastic effect was observed in CT26 tumor model that has a highest level of COX expression despite a comparable targeting efficiency of PASA/FASA in 4T1.2 and CT26 tumor models (Fig.4). Our data suggest that 5-ASA largely inhibits the tumor growth in vivo via a mechanism that is independent of direct effect on tumor cell proliferation. The fact that CT26 tumor model responded more dramatically to 5-ASA treatment suggests that a higher level of COX might play a more oncogenic role in CT26 tumor compared to 4T1.2 tumor model. These data are consistent with previous clinical research that NSAIDs reduce the risk of colorectal cancers that overexpress COX-2 but have minimal impact on the colorectal cancers with weak expression of COX-2[37]. It is possible that mechanisms other than COX inhibition also contribute to the different responses to 5-ASA treatment among the two tumor models.

5-ASA treatment led to an improvement in the tumor immune microenvironment (Fig. 5). It has been reported that PGE₂ secreted by tumor cells is one of the principal mediators allowing tumor cells to escape immunosurveillance[38]. PGE₂ was shown to induce differentiation of macrophages from an M1 to an M2 phenotype and the production of pro-inflammatory factors, such as CXCL1 and IL-6[38, 39]. Accordingly, myeloid cells could be stimulated by PGE₂ secreted from mouse melanoma tumor cells to produce CXCL1, IL-6 and G-CSF[3]. Moreover, PGE₂ potently suppressed NK cell activity, which could be recovered by depletion of tumor-derived PGE₂[40]. Fig. 5 shows that 5-ASA treatment was associated with increases in M1/M2 ratio and the number of functional CD4⁺ and CD8⁺ cells. Our data are consistent with the published literature. Our data also show that combination of 5-ASA and DOX resulted in a further improvement in tumor immune microenvironment (Fig. 5). In addition to direct killing of tumor cells, DOX can elicit antitumor immunity through induction of immunogenic cell death (Fig. S17)[24]. 5-ASA that was slowly released from the PASA and FASA helped to further improve and sustain an active tumor immune microenvironment. The detailed mechanism for the 5-ASA-mediated improvement in tumor immune microenvironment is unclear at present. Our preliminary RNA-Seq data showed upregulation of Cxcl10 and downregulation of Ccl22 by both FASA and FASA/DOX (Fig. S21). Cxcl10 has been shown to be related to the recruitment and activation of Th1 cells[28], while Ccl22 is involved in the recruitment of T_reg[29]. It remains to be tested whether changes in the expression levels of Cxcl10/Ccl22 are responsible for the increased functional CD8⁺ T cells and a concomitant decrease in the number of T_reg following treatment with FASA and FASA/DOX.

It is interesting to notice that the expression levels of PD-1 on CD4⁺ and CD8⁺ cells were significantly upregulated following treatment with free 5-ASA+DOX combination, FASA, and FASA/DOX, respectively, particularly the latter two treatments (Fig. 6). Upregulation of PD-1 could be secondary to the increased production of IFN-γ as a result of an enhanced antitumor immunity. However, a role of tumor-derived PGE₂ in regulating the expression of PD-1 on immune cells was suggested by the study of Bottcher et al. showing a high expression level of PD-1 in COX knockout tumors[11]. Our preliminary study also showed that treatment with PGE₂ led to decreased expression of PD-1 on CD4⁺ and CD8⁺ cells (Fig. 6). The underlying mechanism is unclear at present and requires more studies in the future. Nonetheless, this pointed to a potential combination of FASA/DOX with anti-PD-1 antibody. Indeed, a drastic improvement in the overall antitumor activity was observed for the combination therapy including regression of some established tumors at a suboptimal dose of FASA/DOX.

Conclusion

In summary, we have confirmed and extended previous literature that tumor-derived COX plays an important role in promoting an immunosuppressive tumor microenvironment. We have also developed a novel dual functional nanocarrier (FASA) that is highly effective in codelivery of 5-ASA, a COX inhibitor, and DOX that induces tumor immunogenic cell death. Delivery of DOX via FASA, alone or in combination with immune checkpoint blockade, may hold promise as a new and effective immunochemotherapy for various types of cancers, particularly those COX-overexpressing cancers.

Data availability

The authors declare that all data supporting the findings of this study are available within the paper and its supplementary information files. Correspondence and requests for materials should be addressed to S.L.