Pin1 inhibition exerts potent activity against acute myeloid leukemia through blocking multiple cancer-driving pathways

Bone marrow samples (n = 150) were collected from healthy donors and patients treated in the Hematology Department of the Fujian Medical University Union Hospital from September 2012 to February 2015. Of these, 107 samples were taken from patients with acute leukemia (AL) and 43 were taken from healthy donors. AL patients after the first visit included 61 male and 46 female patients in age of 13~76 years old with median age of 45 years old. According to FBA classification for the diagnosis and classification of AML [35], there were 6 patients with M0, 9 with M1, 18 with M2, 7 with M3, 1 with M4, 41 with M5, 3 with M6, 1 with M7, 21 cases with other types of AL, 43 healthy controls included 20 male, and 23 female donors in age of 20~45 years old with median 32 years old. All of the patients and healthy donors signed informed consent.

All cells were cultured in RPMI1640 with 10 or 20% FBS. HL60 (acute myelogenous leukemia cell line), U937 (histiocytic lymphoma cell line), KG-1a (acute myeloid leukemia cell line), NB4 (acute promyelocytic leukemia cell line), Nalm-6 (acute B acute lymphoblastic leukemia cell line), Molt-4 (acute T acute lymphoblastic leukemia cell line), Kasumi-1 (acute myelogenous leukemia cell line), and K562 (chromic granulocytic leukemia cell line) were cultured at the Fujian Institution of Hematology. HL-60, U937, and KG-1a cell lines were cultured at the BIDMC and were gifts from Dr. Daniel G Tenen. Antibodies against various proteins were obtained from the following sources: Pin1 was previously described [36]; β-actin from Sigma; Cyclin D1 (DCS-6) from Biolegend; NF-κB/p65 (D14E12, 8242) from Cell Signaling Technology; β-catenin from BD biosciences; and Rab-2A (15420-1-AP) from Proteintech Group. All-trans retinoic acid (ATRA) were purchased from Sigma, 1,25-dihydroxyvitamin D3 were purchased from Cayman Chemicals, and ATRA-releasing pellets were from Innovative Research of America.

Total RNA from bone marrow mononuclear cell and cell line was extracted by TRIzol (Invitrogen), and cDNA was transcribed with reverse transcription kit (Thermo Fisher Scientific) in accordance with the manufacturer’s protocols. PCR reactions were performed using quantitative PCR kit (Roche) and ABI7500 fluorogenic quantitative PCR instrument (Applied Biosystem Company). β-actin was used as an internal control of samples. The following primers were used:

In line with the operation steps of the quantitative PCR kit (Roche), the reaction system contains SYBR Green Master (ROX) 10 μl, upstream primer (10 pmol/μl) 0.15 μl, downstream primer (10 pmol/μl) 0.15 μl, cDNA 1 μl, and double distilled water 8.7 μl. The reaction was conducted in ABI7500 fluorogenic quantitative PCR instrument (Applied Biosystem Company) under conditions: at 50 °C for 2 min and 95 °C for 10 min, and then at 95 °C for 15 s and 60 °C for 1 min as 1 cycle for 40 times in total. Then, melting curve reaction was conducted under conditions: at 95 °C for 15 s, 60 °C for 1 min, 95 °C for 15 s, 60 °C for 15 s, 1 cycle for once. Three ventral orifices were set in every sample. With β-actin as an internal reference, the relative expression quantity of mRNA was represented with RQ value by 2^−ΔCT method: RQ = 2^−ΔCT. The final results are presented as the fold change of Pin1 expression in a target sample relative to a reference sample, normalized to β-actin.

Recombinant lentiviral particles were produced in 293FT cells by co-transfecting pRevTRE plasmid containing Pin1 shRNA sequence, along with helper plasmids, including pCMV-VSVG and pCMV-dR8.91. The virus-containing medium was harvested at 48 h after transfection and filtered using a 0.45-μm filter. For infection, the collected virus medium was added to the rtTA-expressing cells with polybrene. The cells were selected with G418 and puromycin. Pin1 shRNA were induced by doxycycline, and Pin1 protein level was analyzed by immunoblot. Pin1 shRNA sequence is CCACCGTCACACAGTATTTAT.

Cells were seeded on 96-well plates at a density of 5000 cells per well. Cells were stained with Trypan blue and counted every day. In two additional groups, cells were treated with DMSO or ATRA. After 72 h, cell viability was measured by CCK-8 assay (DOJINDO) and the number of cells was determined by CellTiter-Glo® 2.0 Assay (Promega, Madison, WI) according to the manufacturer’s instructions.

Cells were seeded on 24-well plate at a density of 500 cells (KG-1a is 1000 cells). Methylcellulose (Sigma) in final concentration of 0.8% was added and mixed. After 7 days (KG-1a is 14 days), cells were stained with 0.5 ml 1 mg/ml p-Iodonitrotetrazolium Violet. The number of colonies and the total area were calculated by ImageJ analysis.

The PPIase activity on GST-Pin1 in response to ATRA and 1,25-(OH)₂ vitamin D3 were determined using the chymotrypsin-coupled PPIase activity assay with the substrate Suc-Ala-pSer-ProPhe-pNA (50 nM) in buffer containing 35 mM HEPES (pH 7.8) and 0.1 mg/ml BSA at 10 °C as described, with the exception that the compounds were preincubated with enzymes for 12 h at 4 °C [37].

To assess cell surface expression of CD11b, cells were washed with PBS, harvested by non-enzymatic cell dissociation solution, and resuspended in blocking solution (Ca2+, Mg2+-free PBS containing 2% FCS). Cells were then incubated with CD11b-FITC (Biolegend) for 15 min at 4 °C in the dark. Cells were washed and analyzed by using a Beckman Coulter’s Gallios flow cytometry.

Cells from each group were collected, and cell lysis buffer (Roche Company) was added to extract total protein. The protein concentration was detected by the Bradford method. The loading sample of equivalent protein was transferred to a nitrocellulose membrane (Bio-Rad, cat. No. 162-0115) through SDS-PAGE electrophoresis and immunoblotting.

For xenograft experiments, 5 × 10⁵ HL-60 and U937 cells stably expressing Tet-On shPin1 were injected subcutaneously into flank of 7-week-old BALB/c nude mice (Taconic Laboratories) and fed with normal or doxycycline-containing diet. For ATRA treatment, 5 × 10⁵ U937 cells were injected subcutaneously into flank of 7-week-old BALB/c nude mice. After 5 days when tumor growth was just about notable by sight, the mice were randomly divided into placebo group or ATRA slow-releasing pellet group. Tumor growth was monitored twice a week until sacrifice criteria were met in the first mice. Tumor sizes were recorded twice a week by a caliper and tumor volumes were calculated using formula L × W² × 0.52, where L and W represent length and width, respectively. Tumors were cut into small pieces and quickly stored in liquid nitrogen, partly for protein extraction. All BALB/c nude mice were housed in laminar flow cabinets with free access to food and water. Animal work was carried out in compliance with the ethical regulations approved by the Animal Care Committee, Beth Israel Deaconess Medical Center, Boston, USA.

Experiments were routinely repeated at least three times, and the repeat number was increased according to effect size or sample variation. Statistical analyses were performed using SPSS 23.0 and GraphPad Prism 5 software package. All data are presented as the means ± SD/SEM, followed by determining significant differences using the two-tailed student’s t test or analysis of variance (ANOVA) test where *p < 0.05, **p < 0.01, and ***p < 0.001.

To investigate the potential clinical significance of Pin1 in leukemia, we examined the relative expression of PIN1 mRNA in bone marrow mononuclear cells from 107 newly diagnosed leukemia patients and 43 healthy bone marrow donors. We found that PIN1 mRNA expression was significantly higher in bone marrow cells of acute leukemia (AL) patients compared to those of healthy controls (Fig. 1a, lane 2 versus lane 1). Among these samples, 86 samples were diagnosed with AML patients, which also showed significantly higher PIN1 mRNA expression, as compared with healthy controls (Fig. 1a, lane 3 versus lane 1). Further analysis of the relationship between the expression of PIN1 mRNA and FAB (French–American–British classification system) subtypes of AML showed that PIN1 mRNA was upregulated in most subtypes of AML. In the case of M4 and M6, low sample sizes precluded sufficient power for statistical analysis (Fig. 1b). Consistent with the results of PIN1 mRNA, Pin1 protein levels were higher in AML patient samples compared with healthy controls (Fig. 1c, d; Additional file 1: Figure S1a). A positive correlation was found between PIN1 mRNA level and Pin1 protein level (Fig. 1e). In addition, we used different colors for individual patients in western blot analysis. These results together demonstrate that Pin1 is highly overexpressed in AML patients.

We also examined Pin1 expression in human leukemia cell lines, including six AML cell lines (Kasumi-1, U937, K562, NB4, HL-60, and KG-1a) and two ALL cell lines (Nalm-6 and Molt-4). Both PIN1 mRNA (Fig. 1f) and Pin1 protein levels (Fig. 1g, h; Additional file 1: Figure S1b) were found to be significantly higher in human leukemia cell lines as compared with bone marrow cells of healthy control (Fig. 1f–h). A positive correlation of PIN1 mRNA and Pin1 protein levels was also found in leukemia cell lines (Fig. 1i). These results further confirm that Pin1 is overexpressed in leukemia cells including AML cells.

Given the overexpression of Pin1 in AML, including biopsied bone morrow leukemia cells and established human AML cell lines, the question remains as to whether Pin1 plays any role in leukemogenesis. Based on the “two-hit model,” the ability of a molecule to promote cell proliferation and colony formation is required in order to initiate leukemogenesis [38]. We investigated the effects of Pin1 on leukemogensis-related traits. To this end, multiple AML cell lines, including HL-60, U937 and KG-1a, were infected with lentiviruses carrying validated Pin1-specific shRNA [26] or control shRNA with Puro^r. Using puromycin selection, stable Pin1-knockdown (Pin1 KD) cell lines were established. Immunoblot analysis confirmed a dramatic decrease in endogenous Pin1 in cells carrying Pin1-shRNA compared with cells carrying the control shRNA (Fig. 2a).

The rates of cell growth of different pairs of AML Pin1-shRNA and control-shRNA cells were monitored by counting cell number to demonstrate the effects of Pin1 downregulation on cell proliferation. In all three AML cell lines examined, Pin1 KD resulted in a significant reduction in cell proliferation (Fig. 2b).

To test if Pin1 KD would impair the leukemogenic potential of AML cells to grow as clonogenic colonies, HL-60, U937, and KG-1a expressing Pin1 shRNA or control shRNA were plated at a low density and observed for colony formation. A significant decrease both in the number (Fig. 2c, d) and size (Fig. 2c, e) of colonies was observed in the Pin1 KD group compared with controls. Thus, our data demonstrates that Pin1 plays a positive role in leukemogenesis-relevant events in vitro. As a major regulator of oncoproteins, Pin1 amplifies oncogenic pathways by activating more than 43 oncogenic molecules and suppressing at least 20 tumor-suppressing molecules, including many of which have well-established roles in CSCs [17, 21, 39]. To further characterize the underlying mechanisms of Pin1-mediated leukemogenesis, we analyzed the expression of several leukemogeneis-related signaling molecules that have been reported to be Pin1 substrates. Wnt/β-catenin and NF-κB pathways are well-known signaling pathways in leukemogenesis through regulating cell proliferation, differentiation, or apoptosis and have been shown to be Pin1 substrates in other cancer cells [26, 40‐46]. AML Pin1 KD cells also downregulated the expression of Pin1 oncoprotein substrates, including β-catenin and NF-κB (Fig. 2f), as with solid tumors [26, 40‐46]. Besides these key regulators in leukemogenesis, Pin1 also transcriptionally regulated some stem cells-related molecules, including Rab2A [47] and CyclinD1 [48, 49]. In AML Pin1 KD cells, these molecules were downregulated as well (Fig. 2f) (Additional file 2: Figure S2). Therefore, these biochemical analyses further support the role for Pin1 in the regulation of cell proliferation and clonogenicity through deregulation of multiple oncogenic pathways.

To further confirm and evaluate the effects of Pin1 inhibition on the tumorigenesis of AML, we generated stable tetracycline-inducible Pin1 knockdown HL-60 and U937 cells using a Tet-On system [28, 50]. We infected HL-60- and U937-rtTA cells with lentivirus containing either Pin1 shRNA or control shRNA with Puro^r. Stable tetracycline-inducible Pin1 KD and control HL-60- and U937-Tet ON cells were established using puromycin selection. In the presence of doxycycline, Pin1 protein levels were dramatically reduced in both cell lines (Fig. 3a, b). Furthermore, the clonogenic assay showed a significant decrease in the number (Fig. 3c, d) and size of colonies (Fig. 3e, f) after doxycycline treatment as compared with the vehicle control. These results show that our Tet-On Pin1 KD system is highly inducible and could be used to confirm the role of Pin1 in leukemogenesis, as shown in constitutive Pin1 KD cells (Fig. 2).

To examine whether inducible Pin1 inhibition would affect tumorigenesis of human AML cells in vivo, we subcutaneously implanted HL-60- and U937-Tet ON cells into the flanks of nude mice, then fed the mice with a normal or a doxycycline-containing diet, respectively. Tumor growth was monitored twice a week until sacrifice criteria were met in the first mice. Our results showed that doxycycline-induced Pin1 KD significantly suppressed tumor growth in both HL-60 (Fig. 4a) and U937 Pin1 KD cells (Fig. 4b) with reductions of both tumor weight and tumor volume exceeding 50% (Fig. 4c–f). Thus, inducible Pin1 downregulation suppressed tumor growth in AML xenograft model in vivo. To assess inactivation of Pin1-regulated cancer pathways in primary xenograft tumors in vivo, we extracted total proteins from xenograft tumor samples and then subjected to perform western blot analyses on Pin1 and its downstream oncoproteins, including β-catenin, NF-κB, cyclinD1 and Rab-2A, whose protein levels have been shown to be upregulated by Pin1 [26, 40‐46, 51]. The protein levels of these oncoproteins were decreased with Pin1 downregulation (Fig. 4g, h), implying that Pin1 KD suppresses tumorigenesis through downregulation of oncogenic signaling pathways.

Given the oncogenic role of Pin1 in AML, we wondered whether targeting Pin1 with chemical inhibitors would show any therapeutic benefit in the treatment of AML [28]. ATRA has been reported to inhibit and degrade active Pin1, resulting in the blockade of multiple Pin1-regulated oncogenic pathways in APL, breast cancer, and liver cancer [28, 30, 52]. However, the effect of ATRA on AML cells is unknown. Therefore, we treated multiple AML cells, HL-60, U937, and KG-1a, with different concentrations of ATRA. ATRA can degrade Pin1 and its downstream oncoproteins in a dose-dependent manner (Fig. 5a). Notably, in both cells, among these oncoproteins, while NF-κB was slightly upregulated in low concentrations of ATRA likely due to ATRA-induced differentiation, they were downregulated once ATRA concentration was high enough to induce Pin1 degradation. This further supported that ATRA could downregulate multiple oncoproteins through degrading Pin1 in AML.

Since ATRA induces differentiation in various leukemia cell lines, it is unclear whether differentiation would affect Pin1 levels. To address this question, we treated cells with other differentiation-inducing reagent − 1-α, 25-(OH)₂ vitamin D₃ (1,25-D3), the active form of vitamin D₃, which has been shown to induce differentiation in various AML cell lines [53‐56]. After 72 h of incubation of 1, 25-D3 or ATRA, differentiation was evaluated by flow cytometry using a general myeloid marker, CD11b. As shown in Figure S3a (Additional file 3; Figure S3a), the percentages of CD11b-positive cells were increased in a dose-dependent manner in 1,25-D3-treated HL60 and U937, but not in KG-1a, likely due to low basal level of PKCβ [57]. The differentiation state of 1,25-D3 was similar to that of ATRA in HL-60 and U937 (Additional file 3: Figure S3b). Using these models, we further analyzed Pin1 stability and activity. The results showed that 1,25-D3 neither reduced Pin1 levels in cells (Additional file 3: Figure S3c) nor inhibited Pin1 PPIase activity in vitro (Additional file 3: Figure S3d). We also checked the stability of Pin1 downstream targets in U937 (Additional file 3: Figure S3e). As shown in Additional file 3: Figure S3e, differentiation-unrelated Rab2A did not show obvious changes. NF-κB and β-catenin showed a slightly upregulated in high doses, and cyclinD1 was slightly downregulated at high doses, which could likely be due to proliferation inhibition. These results show that differentiation-inducing drug per se could not downregulate Pin1 stability and activity, confirming that ATRA-induced Pin1 degradation and downregulation of Pin1 substrates is differentiation-independent. Furthermore, ATRA treatment also significantly inhibited cell proliferation and colony formation of all three leukemia cells (Fig. 5b, c). To explore if ATRA could be effective for those normal cells that do not overexpress Pin1, we treated two immortalized normal blood cells (N1 and N5 cells) with ATRA. First, in both normal cells, Pin1 protein levels were extremely low (Additional file 4: Figure S4a), as we have shown in normal human bone marrow cells (Fig. 1c, g). Importantly, these two normal blood cell lines were completely resistant to ATRA (Additional file 4: Figure S4b). These results are consistent with our previous findings that ATRA selectively targets active Pin1 in breast cancer cells, but not normal breast cells [58]. Thus, besides genetic inhibition, chemical inhibition of Pin1 by ATRA can also inhibit AML leukemogenesis through downregulation of multiple oncogenic pathways.

Given the effects of ATRA on cell growth, leukemogenesis, Pin1 protein levels, and the downstream in AML in vitro, a critical question is whether ATRA suppresses AML tumor growth in vivo. We thus examined ATRA in mouse models xenografted with U937 cells. ATRA is extremely light-sensitive and can be metabolized quickly in the liver, with a 45-min half-life in humans. To improve the activity of drug and maintain a constant drug level in the blood, we implanted slow-releasing ATRA pellets, as described in previous studies [28, 52]. As expected, ATRA blocked the tumor growth (P = 0.0037) (Fig. 5d) and tumor size and weight (P = 0.0057) (Fig. 5e, f). Furthermore, the protein level of Pin1 and Pin1’s oncogenic downstream (Fig. 5g) was downregulated in xenograft tumors from nude mice treated with slow-releasing ATRA. These results demonstrate that ATRA has potent anti-leukemia activity through targeting Pin1 and multi-cancer driving pathway.