Human colorectal cancer cell lines (Caco-2, HCT116, HT29, SW480, and DLD-1) were obtained from the American Type Culture Collection and grown at 37°C in EMEM (Eagle’s Minimum Essential Medium) supplemented with 10% heat-inactivated FBS (fetal bovine serum; Sigma-Aldrich), GlutaMAX (ThermoFisher Scientific), and 1% penicillin-streptomycin (ThermoFisher Scientific) in a humidified cell culture incubator (ThermoFisher Scientific) with 5% CO₂.

Caco-2 cells (2 × 10⁴) were plated in duplicate in 12-well plates and cultured for 6 days. At 2 days intervals, cells were observed under an inverted phase-contrast microscope (IX71; Olympus). Total cell numbers were counted at 2 days intervals over a period of 6 days using a hemocytometer. The average value was recorded.

Cell extracts were separated by SDS-PAGE. The lysate proteins were transferred from the gel to a PVDF membrane and immunoblotted with anti-Pin1 (10495-1-AP, Proteintech), anti-cyclin D1 (MAI-39546; Invitrogen), anti-cyclin D2 (#3741; Cell Signaling Technology), anti-cyclin D3 (#2936; Cell Signaling Technology), anti-CDK4 (SC-23896; Santa Cruz Biotechnology), anti-CDK6 (SC-53638; Santa Cruz Biotechnology), anti-β-catenin (610154; BD Transduction Laboratories), anti-β-actin (AbC-2002, AbClon), or anti-α/β-tubulin (#2148S; Cell Signaling Technology) antibodies. Detection was performed by chemiluminescence using Western Lightning reagent (PerkinElmer Life Sciences).

For the colony formation assay, 2.5 × 10³ Caco-2, HCT116, HT29, SW480, or DLD-1 cells were plated in 12-well culture dishes and incubated with different concentrations of Juglone (Sigma-Aldrich) or KPT6566 (CSNpharm). After 5 days, colonies were stained with fix/staining solution [0.05% (w/v) Crystal Violet (Sigma-Aldrich), 1% formaldehyde (Sigma-Aldrich), 1% methanol (Sigma-Aldrich), and 1 × PBS] for 20 min. Colonies were counted using the ImageJ program (https://imagej.nih.gov/ij/) to quantify the reduction in CRC cell colony formation after Juglone or KPT6566 treatment. The number of colonies was compared with that in the control group.

Caco-2, HCT116, HT29, SW480, DLD-1, CD44⁺CD133⁺ tumor-initiating Caco-2, or ∆CD44⁺CD133⁺ non-tumor-initiating Caco-2 cells (1 × 10⁴) were plated in each well of 12-well plates, and Juglone or KPT6566 was added to the cell culture medium at different concentrations (0, 0.625, 1.25, 2.5, 5, or 10 μM Juglone; 0, 1.25, 2.5, 5, 10, or 20 μM KPT6566). Cell viability after treatment was measured using the ADAM MC Auto Cell Counter System (NanoEntek). The half-maximal inhibitory concentration (IC₅₀) of Juglone and KPT6566 was determined using the SoftMax Pro software (Molecular Devices).

Following treatment with Juglone or KPT6566, Caco-2 cells were fixed overnight at 4°C with fixation solution (70% ethanol), washed twice in 1 × PBS, resuspended in RNase solution (100 μg/ml RNase A in 1 × PBS), and incubated at 37°C for 30 min. The fixed Caco-2 cells were then stained with propidium iodide (PI) solution (33 μg/ml PI containing 10% NP-40) for 30 min. A histogram showing DNA content was obtained by flow cytometry (FACSCalibur; BD Biosciences).

Apoptosis assays were carried out by flow cytometry using a fluorescein isothiocyanate (FITC) Annexin V Apoptosis Detection Kit with PI (BioLegend). After treatment with Juglone or KPT6566, Caco-2 cells were washed twice with Cell Staining Buffer (BioLegend) and resuspended in Annexin V Binding Buffer, according to the manufacturer’s instructions. Caco-2 cells were stained with FITC-conjugated Annexin V and PI, and then analyzed with a FACSCalibur flow cytometer.

Anti-human CD44 (mouse, fluorescein-5-isothiocyanate (FITC)-conjugated; G44-26, BD Biosciences) and anti-human CD133 [mouse, phycoerythrin (PE)-conjugated; AC133, Miltenyi Biotech] antibodies were used for standard cell surface flow cytometry. Caco-2 cells were sorted using a FACSVantage SE flow cytometer (BD Biosciences) after staining with FITC-conjugated anti-CD44 and PE-conjugated anti-CD133 antibodies. Flow cytometry data were analyzed by CellQuest software (BD Biosciences).

TRIzol solution (Invitrogen) was used to isolate total RNAs from CD44⁺CD133⁺ and ∆CD44⁺CD133⁺ Caco-2 cells. Next, cDNAs were synthesized using a Superscript First-strand Synthesis System (Invitrogen), and RT-PCR reactions to amplify Pin1 were performed in triplicate. β-actin mRNA was used as an internal control. The following primers were used: Pin1: 5′-TCA​GGC​CGA​GTG​TAC​TAC-3′ and 5′-CGG​AGG​ATG​ATG​TGG​ATG-3′ (https://www.ncbi.nlm.nih.gov/nuccore/NM_006221.4) (Yang et al., 2013); and β-actin: 5′-GCT​CGT​CGT​CGA​CAA​CGG​CTC-3′ and 5′-CAA​ACA​TGA​TCT​GGG​TCA​TCT​TCT​C-3′ (https://www.ncbi.nlm.nih.gov/nuccore/NM_001101.5).

Quantitative real-time PCR was performed on a LightCycler 96 Real-Time PCR System (Roche) using BioFACT 2X Real-Time PCR Kit (With SFCgreen I) (BioFACT), according to the manufacturer’s instructions. Reaction conditions for optimal amplification of Pin1 were determined as described in the manufacturer’s manual. As an internal control, expression of mRNA encoding β-actin was also quantified. Relative expression of Pin1 was calculated and normalized to that of β-actin using the ∆C_T method (Schmittgen and Livak, 2008).

To measure tumor suppression by Juglone or KPT6566, CD44⁺CD133⁺ tumor-initiating Caco-2 cells were resuspended in PBS (1:1 ratio of PBS and Matrigel), and 1 × 10⁵ cells were injected into the left and right flanks of 8 week-old NSG mice using a 21-gauge needle. When tumors reached 15–25 mm³, 5 mg/kg Juglone or KPT6566 per NSG mouse was injected intraperitoneally once every 3 days for 30 days. An equivalent volume of vehicle was used as the control. All mice that developed tumors were monitored carefully and sacrificed 30 days after treatment started. Mice were weighed at 3 days intervals using a digital scale, and the tumor length and width were measured at 2 days intervals using a digital external caliper. Tumor volume was calculated using the following formula (Euhus et al., 1986; Tomayko and Reynolds, 1989; Kim et al., 2019):

Tumor volume = (L × W²)/2, where L = length (the greatest longitudinal diameter) and W = width (the greatest transverse diameter).

All experiments were carried out in accordance with the animal experimentation guidelines at Sogang University, and all mouse experiments were approved by the institutional committee (IACUCSGU 2020-18).

All experimental data are presented as the mean ± S.D. (standard deviation). An unpaired Student’s t-test was used to compare the means between two experimental groups. Statistical analysis was performed using Microsoft Excel (Microsoft Corporation). A p value < 0.05 was deemed statistically significant.

Overexpression of Pin1 has been reported in many human cancers and cancer cell lines (Lu, 2003). Also, Pin1 is overexpressed in human CRC patients (Kim et al., 2005; Kuramochi et al., 2006). However, little is known about whether targeting Pin1 suppresses the tumorigenic potential of human CRC. To investigate whether inhibition of Pin1 prevents CRC cell growth, we performed western blot analysis to measure expression of Pin1 in five representative colorectal cancer cell lines, namely, Caco-2, HCT116, HT29, SW480, and DLD-1. We detected high levels of Pin1 protein in all five human CRC cell lines (Figure 1A). Also, we tried immunocytochemistry analysis to assess the expression pattern of Pin1 in Caco-2 cells. The results showed that Pin1 protein was present in Caco-2 cells. Interestingly, the expression pattern of Pin1 in individual Caco-2 cells was not homogeneous. As shown in Supplementary Figure S1A, cell-level heterogeneity of Pin1 expression was observed in Caco-2 cells in each sampled region of a slide. We also performed immunocytochemistry experiments using the HCT116, HT29, SW480, and DLD-1 cell lines, and observed similar results (Supplementary Figures S1B–E).

To assess the potential antiproliferative effects of chemical Pin1 inhibitors in CRC, 2 × 10⁴ Caco-2 cells were seeded into 12-well plates and treated with 5 μM of Juglone (a chemical Pin1 inhibitor) for the indicated times. Cells were monitored for 6 days. Control Caco-2 cells showed epithelial-like cell morphology and formed colonies; however, 5 μM Juglone caused a time-dependent decrease in Caco-2 cell growth (Figure 1B).

To quantify suppression of proliferation by Juglone, the number of Caco-2 cells was counted at 2 days intervals for 6 days. Juglone treatment caused significant suppression of Caco-2 cell proliferation compared with control cells (Figure 1C).

Since the colony forming capability of cancer cells is used to gauge the transforming ability of cells responsible for tumorigenic potential, we wanted to test whether Juglone inhibits the clonal growth of CRC cells. To investigate whether Juglone affects colony formation by Caco-2 cells, 2.5 × 10³ cells were plated and treated with the indicated concentrations of Juglone (0–10 μM), and colonies were visualized 5 days later by Crystal Violet staining. Colony formation assays showed that colon cancer cells treated with different concentrations of Juglone were less able to form colonies than control colon cancer cells. As shown in Figure 1D, Juglone induced a dose-dependent decrease in the colony forming ability of Caco-2 cells (top panels). We also investigated whether Juglone suppresses the colony forming potential of other colorectal cancer cells. HCT116 (second panels), HT29 (third panels), SW480 (fourth panels), and DLD-1 cells (bottom panels) were treated with different doses of Juglone, and final measurements were taken 5 days later. Similar to Caco-2 cells, Juglone also decreased the colony forming ability of HCT116, HT29, SW480, and DLD-1 cells in a dose-dependent manner (control cells = 0 μM Juglone).

To determine the number of colonies, images of cell culture plates were analyzed using the ImageJ program after Crystal Violet staining. Treatment with 2.5, 5, or 10 μM Juglone decreased the number of Caco-2 cell colonies by 60, 99, and ∼100%, respectively (Figure 1E, first graph). While treatment with 2.5 μM Juglone had a marginal effect on the colony forming ability in HCT116 cells, 5 μM or 10 μM Juglone reduced colony formation by ∼70 and 99%, respectively (Figure 1E, second graph). We performed the same experiments using HT29, SW480, and DLD-1 colorectal cancer cell lines to further validate this observation. Treatment with 2.5, 5, or 10 μM Juglone decreased the number of HT29 colonies by 5, 20, and ∼99% (Figure 1E, third graph), the number of SW480 colonies by 7, 76, and ∼99% (Figure 1E, fourth graph), and the number of DLD-1 colonies by 10, 77, and ∼100% (Figure 1E, fifth graph), respectively.

To confirm these results, we evaluated the potential anti-colony forming effects of KPT6566. Similarly, Caco-2, HCT116, HT29, SW480, and DLD-1 cells were incubated with different concentrations of KPT6566 (0–20 μM). Colony formation assays revealed that CRC cells treated with increasing concentrations of KPT6566 were less able to form colonies than control (DMSO-treated) cells (Figure 1F). Treatment with 10 μM or 20 μM KPT6566 decreased colony formation by Caco-2 cells by ∼64 and 99% (Figure 1G, first graph), that by HCT116 cells by ∼26 and 83% (Figure 1G, second graph), that by HT29 cells by 79 and 85% (Figure 1G, third graph), that by SW480 cells by 85 and 99% (Figure 1G, fourth graph), and that by DLD1 cells by 49 and 63% (Figure 1G, fifth graph), respectively. Thus, Juglone and KPT6566 inhibit the colony forming ability of CRC cells, suggesting the potential of chemical Pin1 inhibitors to suppress tumorigenic properties. In addition, the CRC cell lines tested in this study were more resistant to KPT6566 than to Juglone.

The sensitivity of human colon cancer cell lines to Juglone was determined by calculating the 50% inhibitory concentration (IC₅₀). To analyze the cytotoxic effects of Juglone in CRC cells, we investigated its effects on the viability of the five colorectal cancer cell lines. Growth of Caco-2 (Figure 2A, first panels), HCT116 (second panels), HT29 (third panels), SW480 (fourth panels), and DLD-1 cells (fifth panels) after Juglone treatment was examined by microscopy. As shown in Figure 2A, Juglone decreased the number of Caco-2, HCT116, HT29, SW480, and DLD-1 cells, and inhibited cell growth in a dose-dependent manner (Figure 2A).

Caco-2 and DLD-1 cells were more sensitive to Juglone than the other CRC cell lines. The IC₅₀ of Juglone for inhibiting Caco-2 and DLD-1 cell growth was 1.85 and 1.79 μM, respectively. In addition, the IC₅₀ of Juglone for HCT116, HT29, and SW480 cells was 2.77, 2.63, and 2.51 μM, respectively, which was slightly higher than that for Caco-2 and DLD-1 cells (Figure 2B).

We also analyzed the effect of KPT6566 on cell viability. The growth of Caco-2 (Figure 2C, first panels), HCT116 (second panels), HT29 (third panels), SW480 (fourth panels), and DLD-1 cells (fifth panels) after KPT6566 treatment was examined by microscopy. Addition of KPT6566 also inhibited proliferation of all CRC cells in a dose-dependent manner (Figure 2C).

In the case of KPT6566, Caco-2 and HCT116 cells were more sensitive than other CRC lines (Figure 2D). The IC₅₀ of KPT6566 for inhibiting Caco-2 and HCT116 cell growth was 7.45 and 9.46 μM, respectively, whereas the IC₅₀ for HT29, SW480, and DLD-1 cells was 13.8, 11.1, and 10.7 μM, respectively, which was higher than that for Caco-2 and HCT116 cells (Figure 2D). In general, the CRC cells studied here were more sensitive to Juglone than to KPT6566. Based on this initial screening result, we selected the Caco-2 cell line for further experiments designed to test the hypothesis that Pin1 inhibitors are efficacious against tumorigenesis of CRC.

Next, we investigated the effects of Juglone or KPT6566 on G1-phase cell cycle-regulating proteins in Caco-2 cells. For this, cells were treated 10 μM Juglone for 12, 24, 36, and 48 h. Juglone caused a time-dependent decrease in expression of cyclin D1, cyclin D2, cyclin D3, CDK4, and CDK6 (Figure 3A). We also found that β-catenin levels in Caco-2 cells were downregulated after treatment with Juglone. These results indicate that Juglone effectively inhibits D-type cyclins and CDK4/6 in Caco-2 cells.

The effects of KPT6566 on Caco-2 cells were also examined. We found that KPT6566 inhibited expression of D-type cyclins and CDK4/6 proteins, but appeared to be less potent than Juglone (Figure 3B). These data suggest that Juglone and KPT6566 control Caco-2 cell proliferation by downregulating expression of G1-phase cell cycle regulatory proteins. However, although KPT6566 targets Pin1 for degradation through covalent binding (Campaner et al., 2017), treatment with KPT6566 did not affect the level of Pin1 protein in Caco-2 cells significantly (Figure 3B, top panel).

We next examined the effect of Juglone and KPT6566 on the cell cycle distribution of Caco-2 cells. Caco-2 cells treated for 48 h with 10 μM Juglone or KPT6566 were stained with PI, and the cell cycle distribution was monitored by flow cytometry. Treatment with Juglone or KPT6566 significantly increased the percentage of Caco-2 cells in the sub-G₁ phase compared with the control group (Figure 4A). Juglone at 10 μM caused a statistically significant increase in the percentage of cells in the sub-G₁ phase, from 0.7% (Control) to 72.3%, with a concomitant decrease in the percentage of cells in S phase, from 22.9 to 10.0% (Figure 4A, first and second panels). We also observed a marked increase in the percentage of cells in sub-G1 phase, from 0.7 to 63.2%, with a concomitant decrease in the percentage of cells in S phase, from 22.9 to 13.4%, following KPT6566 treatment (Figure 4A, first and third panels). Figure 4B shows the percentage of Caco-2 cells in sub-G₁ phase, and percentage of cells in each phase, after treatment with Juglone (orange bars) or KPT6566 (green bars).

To explain whether the increased cell death induced by Juglone or KPT6566 was linked to apoptosis, we double-stained cells with Annexin V-FITC/PI. Annexin V binds phosphatidylserine (PS) in apoptotic cells. During apoptosis, the cell membrane changes the distribution of PS from the inner leaflet to the outer leaflet of the lipid bilayer, where it can be reached by Annexin V (Andree et al., 1990; Crowley et al., 2016). After exposure to 10 μM Juglone or KPT6566, we found a substantial increase in the percentage of apoptotic Caco-2 cells (Figure 4C). The rate of Caco-2 cell apoptosis (percentage of Annexin V-positive cells) was 88.5% after exposure to Juglone and 78.1% after exposure to KPT6566, suggesting that both Juglone and KPT6566 induce apoptosis, and that Juglone is more potent. Quantification of Annexin V-positive Caco-2 cells after exposure to Juglone or KPT6566 is shown in Figure 4D. These results strongly indicate that Juglone and KPT6566 not only reduce the number of surviving cells, but also increase the number undergoing apoptosis.

Recent studies suggest that tumor-initiating cells are enriched after chemotherapy; this is because tumor-initiating cells can resist and proliferate despite the fact that many classic chemotherapeutic drugs kill the bulk of the tumor (Dean et al., 2005; Tredan et al., 2007; Phi et al., 2018; Li et al., 2021). Based on these observations, we postulated that tumor-initiating Caco-2 cells might be more resistant to Juglone or KPT6566 than non-tumor-initiating Caco-2 cells. To investigate whether Juglone or KPT6566 can eliminate tumor-initiating cells in CRC, Caco-2 cells were divided into a tumor-initiating subpopulation (CD44⁺CD133⁺-positive Caco-2 cells) and a non-tumor-initiating subpopulation (CD44⁺CD133⁺-negative Caco-2 cells; tumor-initiating cell-depleted Caco-2 cells). As we reported recently (Kim et al., 2021), unsorted parental Caco-2 cells comprised three representative subpopulations (non-tumor-initiating CD44⁻CD133^- cells, non-tumor-initiating CD44⁻CD133⁺ cells, and tumor-initiating CD44⁺CD133⁺ cells); however, a CD44⁺CD133⁻ Caco-2 fraction was barely detectable (Figure 5A top panel). To evaluate whether the CD44⁺CD133⁺ subpopulation of tumor-initiating Caco-2 cells was isolated successfully by FACS, we monitored these cells by flow cytometry analysis immediately after the primary sorting. In the same way, the ΔCD44⁺CD133⁺ non-tumor-initiating cell population was also assessed to evaluate if the CD44⁺CD133⁺ tumor-initiating Caco-2 cells were eliminated successfully. As shown in Figure 5A, the fraction of CD44⁺CD133⁺ tumor-initiating Caco-2 cells in the sorted CD44⁺CD133⁺ subpopulation was about 87% (middle panel), and less than 2% of the isolated ΔCD44⁺CD133⁺ subpopulation was tumor-initiating CD44⁺CD133⁺ cells (bottom panel), indicating that both populations of Caco-2 cells were isolated successfully.

Despite the overexpression of Pin1 by Caco-2 cells as demonstrated in Figure 1A, it is unclear whether CD44⁺CD133⁺ tumor-initiating Caco-2 cells express Pin1. To confirm this, total RNA from the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations of Caco-2 cells were purified, and RT-PCR was performed. PCR products derived from Pin1 mRNA were observed in tumor-initiating Caco-2 cells and in non-tumor-initiating Caco-2 cells (Figure 5B). We sequenced these RT-PCR products and confirmed them to represent the reported human Pin1 cDNA sequence (data not shown). These results demonstrate the existence of Pin1 mRNA in the CD44⁺CD133⁺ tumor-initiating cell subpopulation of Caco-2 cells.

We were also able to quantify Pin1 mRNA levels using quantitative real-time PCR. Interestingly, human Pin1 mRNA was about 2-fold more abundant in the CD44⁺CD133⁺ subpopulation than in the ΔCD44⁺CD133⁺ subpopulation (Figure 5C).

The amounts of Pin1 protein in the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations of Caco-2 cells were determined by western blotting. As shown in Figure 5D, the Pin1 protein was expressed at similar or slightly higher levels in the CD44⁺CD133⁺ subpopulation than in the ΔCD44⁺CD133⁺ subpopulation.

To investigate whether Juglone affects the colony forming potential of CD44⁺CD133⁺ tumor-initiating Caco-2 cells, FACS-isolated CD44⁺CD133⁺ tumor-initiating Caco-2 cells and ΔCD44⁺CD133⁺ non-tumor-initiating Caco-2 cells were treated with different amounts of Juglone. Final measurements were made 5 days after treatment. When comparing the inhibitory effect of Juglone on the colony formation ability of CD44⁺CD133⁺ cells and ΔCD44⁺CD133⁺ Caco-2 cells, we noticed that Juglone repressed the colony forming ability of CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ cells in a concentration-dependent manner (Figure 5E). A colony formation assay of both subpopulations showed that the number of colonies was significantly lower after Juglone treatment (Figure 5F), indicating that Juglone might be an effective inhibitor of CD44⁺CD133⁺ tumor-initiating Caco-2 cell-mediated tumorigenesis.

The growth potential of CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ cells after Juglone treatment was also monitored by phase-contrast microscopy. Juglone meaningfully suppressed the growth of both CD44⁺CD133⁺ tumor-initiating Caco-2 cells and ΔCD44⁺CD133⁺ non-tumor-initiating Caco-2 cells in a concentration-dependent manner (Figure 5G). Interestingly, the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations of Caco-2 cells displayed similar sensitivity to Juglone, although tumor-initiating cells generally show resistance, or lower sensitivity, to chemotherapeutic drugs (Phi et al., 2018; Li et al., 2021). The IC₅₀ values of Juglone for inhibiting cell growth were 1.38 μM for CD44⁺CD133⁺ tumor-initiating Caco-2 cells and 1.21 μM for ΔCD44⁺CD133⁺ non-tumor-initiating Caco-2 cells (Figure 5H), suggesting Juglone is a small-molecule compound targeting CD44⁺CD133⁺ cells to inhibit their tumorigenic tendency.

The anti-tumorigenic potential of KPT6566 was also analyzed in a clonogenic assay and by growth curve analysis of CD44⁺CD133⁺ tumor-initiating Caco-2 cells. Interestingly, KPT6566 also significantly affected colony formation by CD44⁺CD133⁺ Caco-2 cells. As presented in Figure 5I, KPT6566 inhibited colony formation by the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations of Caco-2 cells in a concentration-dependent manner. Quantification of the colony formation assays for both groups showed that the number of colonies was significantly lower after KPT6566 treatment (Figure 5J), indicating that KPT6566 might also be an effective inhibitor of CD44⁺CD133⁺ tumor-initiating Caco-2 cells.

The proliferative properties of CD44⁺CD133⁺ cells and ΔCD44⁺CD133⁺ cells after KPT6566 treatment were monitored by microscopy. KPT6566 also suppressed the growth of the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations in a concentration-dependent manner (Figure 5K). Similar to Juglone in Figure 5H, the CD44⁺CD133⁺ and ΔCD44⁺CD133⁺ subpopulations of Caco-2 cells showed similar sensitivity to KPT6566. The IC₅₀ values of KPT6566 for inhibiting cell growth were 5.76 μM for CD44⁺CD133⁺ tumor-initiating Caco-2 cells and 5.39 μM for ΔCD44⁺CD133⁺ non-tumor-initiating Caco-2 cells (Figure 5L), suggesting KPT6566 as another small-molecule compound targeting CD44⁺CD133⁺ cells. Taken together, these data suggest that targeting the Pin1 protein might be a preferential and effective strategy for killing tumor-initiating CRC cells.

One of the important characteristics of tumor-initiating cells is high expression levels of ATP-binding cassette (ABC) transporters (Phi et al., 2018; Li et al., 2021), therefore, we performed RT-PCR to detect expression of representative multidrug resistance genes. As shown in Supplementary Figure S2, mRNAs encoding ABCG2 and ABCB1 were more abundant in CD44⁺CD133⁺ tumor-initiating Caco-2 cells than in ΔCD44⁺CD133⁺ non-tumor-initiating Caco-2 cells (i.e., the CD44⁻CD133^- and CD44⁻CD133⁺ fractions). However, the expression levels of ABCC1 and ABCC2 drug efflux transporters were very similar between the non-tumor-initiating subpopulation (CD44⁻CD133^- and CD44⁻CD133⁺ Caco-2 cells) and the tumor-initiating subpopulation (CD44⁺CD133⁺ Caco-2 cells).

Finally, to investigate the effects of Juglone or KPT6566 treatment on tumor-initiating cell-induced tumorigenesis, the in vivo anti-tumorigenic properties of Juglone and KPT6566 were examined in NSG mice bearing CD44⁺CD133⁺ Caco-2 cell xenografts. Approximately 1 × 10⁵ CD44⁺CD133⁺ tumor-initiating Caco-2 cells were suspended in 100 μl of injection solution and inoculated subcutaneously in the flanks of NSG mice. When tumors had grown to 15–25 mm³, the mice were arbitrarily separated into three groups and treated with Juglone (5 mg/kg), KPT6566 (5 mg/kg), or vehicle (0.1% DMSO) once every 3 days for 30 days. We first evaluated the adverse effects of Juglone or KPT6566 in tumor-bearing NSG mice. To check the health status of mice treated with Juglone or KPT6566, mean body weight of mice was measured at 3 days intervals throughout the experimental period. The body weight of NSG mice treated with Juglone or KPT6566 was similar to that of vehicle-treated NSG mice. These results suggest that Juglone or KPT6566 themselves had no significant toxic effects in these experiments (Figure 6A).

Next, we investigated the antitumor effects of Juglone or KPT6566. Consistent with the cell-based experiments (Figure 5), tumors in mice treated with Juglone or KPT6566 were significantly smaller than those in control mice (Figure 6B). These results indicate that Pin1 is critical for CD44⁺CD133⁺ tumor-initiating cell-mediated tumorigenesis, and that suppression of Pin1 function significantly impairs tumor growth in vivo. Compared with the vehicle group (1431.1 ± 409.4 mm³) at the end of treatment, Juglone (199.7 ± 67.1 mm³) or KPT6566 (367.0 ± 76.9 mm³) reduced tumor volume by 86.0% (n = 5, p < 0.01) and 74.4% (n = 5, p < 0.01), respectively.

The weight of the tumors in mice treated intraperitoneally with Juglone or KPT6566 was much smaller than that of tumors in the vehicle-treated group (Figure 6C). Although the average tumor weight in the control group was 1.52 ± 0.33 g, that in the Juglone-treated group was 0.20 ± 0.05 g, and that in the KPT6566-treated group was 0.48 ± 0.11 g at 30 days after treatment (Figure 6D). Compared with the control group at the end of treatment, Juglone or KPT6566 treatment decreased tumor weight by 86.8% (p < 0.01) and 68.4% (p < 0.01), respectively. Taken together, these results suggest that Pin1 activity is essential for CD44⁺CD133⁺ tumor-initiating Caco-2 cell-induced tumorigenesis in vivo, and that Juglone or KPT6566 effectively inhibit tumor-initiating Caco-2 cell-mediated tumor growth.