AOM, crystal violet, Eagle's Minimal Essential Medium (EMEM), DMEM, hematoxylin-eosin (H&E), McCoy's 5A medium, protease inhibitor cocktail, and phenylmethylsulfonyl fluoride (PMSF) were purchased from Sigma-Aldrich (Merck KGaA). MTT was obtained from Calbiochem. DSS (36-50 kDa) was obtained from MP Biomedicals. Tissue Protein Extraction Reagent (TPER^®) was obtained from Thermo Fisher Scientific, Inc. and fetal bovine serum (FBS) from Gibco (Thermo Fisher Scientific, Inc.). All solvents (ethanol, methanol, dichloromethane, hexane and DMSO) were reagent or HPLC grade.

P. angulata calyces were collected at Loma de Arena, Bolívar, Colombia. A sample of the whole plant was sent to Herbarium of the Universidad de Antioquia, Medellín, Colombia and authenticated according to a voucher previously deposited (no. HUA213028) (6). Clean and dried calyces were used for preparation of PADF according to the procedure described in a previous article (6) (Data S1).

Colorectal adenocarcinoma (HT-29 and Caco-2) and healthy colonic epithelial cells (CCD 841 CoN) were obtained from the American Type Culture Collection (ATCC) and maintained at 37˚C and 5% CO₂ in McCoy's 5A medium (HT-29) or EMEM (Caco-2 and CCD 841 CoN) supplemented with 10% FBS and 1.5 g/l sodium bicarbonate (Sigma-Aldrich; Merck KGaA).

HT-29, Caco-2 or CCD 847 CoN cells were seeded (5,000-15,000 cells/well) in 96-well plates (2D culture) or Perfecta3D^® Hanging Drop Plate (3D culture; 3D Biomatrix) and incubated at 37˚C for 24 and 72 h, respectively, to allow attachment or spheroid formation, respectively. After that, cells were treated for 48 h at 37˚C with PADF (3.13, 6.25, 12.50 and 25.00 µg/ml), vehicle (MeOH), or Triton X-100 (1%) as a positive control. Cell viability was measured using MTT (0.25-5.00 mg/ml). After 4 h, formazan crystals were dissolved in DMSO, and the optical density (OD) at 550 nm was measured using a plate reader (Multiskan GO; Thermo Fisher Scientific, Inc.). Selective index (SI) was calculated as SI=IC₅₀ of PADF in CCD 847 CoN cells/IC₅₀ of the same compound in HT-29 or Caco-2 cells.

The replicative capacity of HT-29 cells was confirmed as previously described (15,16). Cells (750 cells/well) were cultured into a 6-well plate at 37˚C for 6 h and treated with PADF (3.13, 6.25, 12.50 and 25.00 µg/ml) or vehicle for 48 h. Subsequently, the culture medium was refreshed and colonies were allowed to form for 7 days at 37˚C. Finally, colonies were fixed with 7:1 acetic acid-methanol at 24˚C for 30 min, stained with 0.5% crystal violet at 24˚C for 2 h, and counted manually using light stereomicroscope (EZ4 HD, Leica Microsystems GmbH; magnification, x10). A colony was considered to contain at least 50 cells.

HT-29 cells (2x10⁵ cells/well) were grown with McCoy's 5A medium supplemented with 10% FBS (17,18), on µDish^{35 mm,high} plates (Ibidi) for 24 h (100% confluence). The insert was removed to generate a scratch in the monolayer, which was then treated with PADF (6.91 and 13.82 µg/ml) or vehicle. Images were captured at 24, 48 and 72 h with an inverted light microscope (Nikon Corporation; magnification x10). The open area in each image was measured using ImageJ 1.47v software (National Institutes of Health) and values were normalized to the open area at 0 h.

The impact of PADF on cell cycle distribution was determined by propidium iodide (PI) staining (cat. no. ab139418; Abcam) and analyzed using a FACS Aria III flow cytometer (BD Biosciences) and FlowJo X10.0.7 software (FlowJo LLC). HT-29 cells (2x10⁵ cells/ml) were exposed to PADF (13.8, 18.0 and 27.6 µg/ml) for 24-48 h at 37˚C. Cells were harvested by trypsinization, fixed with 66% ethanol for 2 h at 4˚C, stained with PI and incubated with RNase A for 30 min at 37˚C in the dark. Doxorubicin (Doxo) was employed as a positive control.

HT-29 cells (1-2x10⁵ cells/ml) were exposed to PADF (0, 13.8, 18.0 and 27.6 µg/ml) for 24-48 h at 37˚C. In the first assay, cells were stained with Annexin V-FITC (cat. no. BMS500FI-100; eBioscience) and analysis was performed using a flow cytometer (FACS Aria III, BD Biosciences) and the images were analyzed using FlowJo X10.0.7 software. For the second assay, DNA electrophoresis was used to detect DNA fragmentation using the apoptotic DNA ladder detection kit (cat. no. ab66093; Abcam) following the manufacturer's instructions.

ΔΨm was monitored by JC-1 Mitochondrial Membrane Potential kit (Item No. 10009172; Cayman Chemical Company). Cells were incubated at 37˚C with JC-1 staining solution for 15 min and rinsed according to the manufacturer's instructions. Measurement was performed using a microplate fluorometer (Infinite 200 PRO, Tecan Group Ltd.) to quantify the fluorescence of J-aggregates (excitation 535 and emission 595 nm) and -monomers (excitation 485 and emission 585 nm).

Female Balb/c mice (mean weight, 19.41±0.13 g) were obtained from the Instituto Nacional de Salud (Bogotá, Colombia). Animals were housed in filtered-capped polycarbonate cages and kept in a controlled environment (19.48±0.10˚C, 68.63±1.15% humidity under a 12/12-h light/dark cycle) with access to food and water ad libitum. All experiments were designed and conducted following local and international regulations [European Union regulations (CEC council 86/809), EU Directive 2010/63/EU, protocols of the Organisation for Economic Cooperation and Development] and approved by the Committee of Ethics in Research of the University of Cartagena (Project Approval Minute No. 81 from August 13, 2015).

The total experiment duration was 77 days (11 weeks). In brief, 5-6-week-old animals were randomized into control (vehicle), AOM + DSS and treatment groups (PADF 10 and 20 mg/kg, 6/group). Carcinogenesis was induced by intraperitoneal (IP) injection of AOM 10 mg/kg. One week later, animals received 2-4% DSS in drinking water for 7 days. To promote chronic inflammation-driven tumor progression, three DSS cycles at 2 weeks intervals (normal drinking water) were performed (19,20). Two doses of PADF (10 and 20 mg/kg, IP), selected according to previous study (6), and the acute toxicity test (Data S1), were administered for 3 weeks after the final DSS cycle, while control groups were treated with vehicle (saline). To check the safety of PADF, toxicity control groups administered PADF daily (10 and 20 mg/kg/day) without AOM or DSS. Body weight was monitored daily, while macroscopic parameters such as colon length and weight/length ratio and tumor progression (tumor number, size and load) were measured following sacrifice. Tumor load value was calculated as the sum of the diameters of all the tumors in each animal. A portion of the colon was rolled from the proximal to distal end. Samples of colonic tissue containing tumors and adjacent normal control tissue were also harvested for further analysis. The experiments had two sets of replications using 197 experimental animals in total. Sacrifice was performed by cervical dislocation following anesthesia by 2% inhaled sevoflurane. Animal death was verified by loss of heartbeat, reflexes and breathing. Humane endpoints were excessive weight loss (>4 g), diarrhea or hemorrhage. However, no animal was euthanized according to these endpoints.

Colon samples were fixed with 4% buffered formalin for 30 days at 24˚C and embedded in paraffin; 5 µm sections were cut and stained with H&E and examined by two blinded researchers using light microscopy (cat. no. DM500, Leica Microsystems GmbH; magnification x10). Severity of mucosal inflammation, dysplasia and cancer were assessed. The epithelial damage and cellular infiltration were scored from 0-4 as previously described (21). Dysplastic proliferation was graded as low- or high-grade and scored considering the area of mucosal surface affected as absent (0), low- (1), medium- (2) or high-grade (3). Cancer was scored as absent (0), early invasive (1) or advanced (2).

RAW 264.7 macrophages (ATCC; cat. no. TIB-71) maintained in DMEM with 10% FBS at 37˚C and 5% CO₂, were cultured in 24-well plates (200,000 cells/well). After 48 h, macrophages were treated with 12.5 µg/ml PADF for 1 h and then stimulated for 6 h with LPS (1 µg/ml) or IL-4 (40 ng/ml) to induce an inflammatory M(LPS) or alternative M(IL-4) profile, respectively. Culture supernatant was collected and added to HT-29 cells seeded in 96-well plates (10,000 cells/well), which were incubated for 48 h at 37˚C. HT-29 cell viability was measured using the MTT method as aforementioned.

Colonic tissue and HT-29 cells were homogenized using TPER^® extraction reagent supplemented with protease inhibitor cocktail and PMFS for western blotting (Data S1) or cOmplete tablets Mini, EDTA-free (Roche Diagnostics GmbH) for ELISA. The extracted proteins were quantified with Bradford assay (Bio-Rad Laboratories, Inc.) using BSA as standard. The levels of IL-1β (cat. no. 88-7261-88), IL-6 (ref 88-7066-88; both Invitrogen), and TNF-α (cat no. 430204; BioLegend) in supernatant were measured using mouse ELISA kits following the manufacturer's instructions.

All data are presented as the mean ± SEM of ≥2 independent experiments. The half-maximal inhibitory concentration 50 (IC₅₀) was calculated using non-linear regression. Statistical analysis was performed by one-way ANOVA followed by Dunnett's multiple comparisons test. Statistical analyses were performed using GraphPad Prism 6 demo version (GraphPad Software, Inc.; Dotmatics). P<0.05 was considered to indicate a statistically significant difference.

A total of 3.15 g PADF was obtained from 114 g dried calyces (2.76% yield) of P. angulata. Based on the information previously reported (6), four types of metabolites (withanolides, sucrose esters, phenolic compounds and flavonoids) were quantified for PADF chemical characterization. Only sucrose derivatives, most likely sucrose esters, were detected in significant amounts (Table SI; Data S1). There were low levels of withanolides and phenolic compounds and flavonoids were not detected.

Fig. 1A shows the inhibitory effect of PADF on the viability of colorectal adenocarcinoma cell lines. HT-29 and Caco-2 cells were notably sensitive to PADF, exhibiting IC₅₀ values of 13.82±0.42 and 29.56±3.73 µg/ml, respectively. By contrast, normal colon epithelial cells (CCD 841 CoN) were not significantly affected by PADF at cytotoxic concentrations (12.5 µg/ml), displaying an IC₅₀ of 44.80±1.77 µg/ml and selectivity index of 3.24 (HT-29) and 1.52 (Caco-2).

Due to the higher susceptibility of HT-29 cells, this cell line was selected to study the effect of PADF. Initially, its cytotoxic effect was tested at different exposure times; PADF exerted significant cytotoxicity at 6 h, which increased to a maximum at 48 h (Fig. S1A; Data S1). Due to the ability of HT-29 cells to form spheroids, 3D culture was also employed to mimic the cancer microenvironment better since it reflects physiological relevant cell-cell interactions (22-24). PADF showed a concentration-dependent effect, preserving the cytotoxic effect (IC₅₀=13.31±4.50 µg/ml; Fig. 1B) observed in the 2D model. Thus, this fraction demonstrates its ability to affect HT-29 cell viability in a tumor environment.

HT-29 cells formed multiple colonies after 7 days (Fig. 1C). This was inhibited by PADF treatment in a concentration-dependent manner. Similarly, PADF decreased gap closure of treated cells in comparison with control cells. (Fig. 1D). Therefore, PADF affected not only the clonogenicity but also the migration of CRC cells.

Flow cytometry was employed to detect changes in cell cycle distribution. PADF induced cell cycle arrest in G2/M phase and significantly increased the proportion of sub-G1 cells, suggesting apoptosis induction (Fig. 2), which was corroborated by ΔΨM, Annexin V-staining and detection of DNA fragmentation. Annexin V-staining demonstrated that PADF induced a significant increase in the proportion of apoptotic cells (early and late), especially at the highest concentration (Fig. 3A). Consistently, this effect was accompanied by a significant reduction of ΔΨM, expressed as JC-1 fluorescence ratio, compared with control cells (Fig. 3B), as well as the degradation of DNA in PADF-treated HT-29 cells (Fig. 3C). To verify if apoptosis was associated with induction of oxidative stress by PADF, induction of reactive oxygen species (ROS) was measured; no change in ROS production was detected (Fig. S1B; Data S1).

AOM-DSS produced a notable decrease in body weight on day 56; this loss of weight was significantly improved by the treatment with PADF at 10 mg/kg/day, while the higher dose (20 mg/kg/day) did not reverse this (Fig. 4A). Macroscopic assessment showed that 100% of animals in the AOM + DSS developed tumors (mean, 10.82±0.96 tumors in the middle and distal colon); by contrast, vehicle group showed no tumors (Fig. 4B). PADF (10 and 20 mg/kg/day) markedly reduced the total tumor load, specifically diminishing the number of tumors >1 mm in diameter. Furthermore, PADF significantly reduced the tumor size and tumor load by 35-85% (Fig. 4C-E). To assess the severity of the disease, colon weight/length ratio was measured; this was reduced in animals treated with PADF. Histological examination of colon sections stained with H&E of the AOM + DSS group confirmed large adenomas or adenocarcinomas inside the lumen and prominent extension of dysplasia (both high- and low grade) accompanied by slight inflammation of the epithelium and cellular infiltration (Fig. 5A and B). Conversely, PADF (10 and 20 mg/kg) promoted the recovery of tissue architecture by reducing tumor growth and dysplasia, thus lowering the histology score from 3.48±0.35 for AOM + DSS to 2.21±0.19 for PADF 20 mg/kg group (Fig. 5B).

During the study, the toxicity control group was monitored to verify that the administration of PADF did not alter the body weight of healthy mice; no signs of toxicity were detected at any dose (Table SII; Fig. S2; Data S1). Moreover, a detailed post-mortem analysis revealed no notable changes during necropsy and hematological analysis (Tables SII and SIII; Data S1). Likewise, PADF did not induce genotoxic effects, as established by the Ames test, micronucleus assay and comet assay (Fig. S2; Data S1), indicating the safety of PADF at effective dosage levels.

PADF significantly reduced the levels of pro-inflammatory cytokines (TNF-α, IL-6, and IL-1β) in colon tissue compared with the AOM + DSS group (Fig. 5C). Based on the immunomodulatory effect on macrophages of PADF (6), it was assessed whether PADF inhibited viability of CRC cells treated with macrophage-conditioned media. Treatment with supernatants from LPS-stimulated macrophages decreases viability (Fig. 5D). By contrast, treatment with supernatants from IL-4-stimulated macrophages increased viability. However, when PADF (12.5 µg/ml) was administered, the viability of HT-29 cells was always significantly reduced, regardless of the conditional media.

Western blot analysis of colonic samples showed that PADF (10 and 20 mg/kg/day) significantly reduced the expression of proliferating cell nuclear antigen (PCNA) while enhancing that of phosphorylated MAPK in comparison with the AOM + DSS group (Fig. 6A and B). The decreased levels of PCNA and increased of p-p38 were confirmed in human CRC cells by western blotting (Fig. 6C and D).