Dimethylaminomicheliolide (DMAMCL) Inhibits Cell Proliferation and Increases Apoptosis and Ecacy of Gemcitabine via Annexin A2 in Pancreatic Adenocarcinoma

Background: Pancreatic adenocarcinoma is one of the highest malignant tumors in digestive tract with extremely poor survival rate. Dimethylaminomicheliolide (DMAMCL) is a clinical developing anti-cancer agent, however, little is known regarding its effects in pancreatic cancer, and the mechanisms of DMAMCL are still not fully understood. Methods: This study evaluated DMAMCL on three pancreatic cancer cell lines by cell viability assay, colony formation assay, and apoptosis assay. To identify the direct binding target of DMAMCL in pancreatic cells, a chemical proteomics approach, molecular docking and site-directed mutagenesis were performed. Results: DMAMCL inhibits proliferation and promotes apoptosis of pancreatic cancer cells. Interestingly, using a chemical proteomics approach, we identify ANXA2 as a direct binding target of DMAMCL in pancreatic cells. Molecular docking and site-directed mutagenesis conrm that Cys132 (C132) of ANXA2 is the binding site for DMAMCL. The knockdown of ANXA2 largely decreases the inhibition activity of DMAMCL, indicating that the effect of DMAMCL is mainly mediated by ANXA2 in pancreatic cancer. In addition, the combination regimen of gemcitabine and DMAMCL exhibits synergistic effect on pancreatic cancer cell lines at both proliferation and pro-apoptosis level. Conclusions: Thus, our ndings elucidate the mechanisms of DMAMCL and may provide a potential strategy to enhancing the ecacy of gemcitabine in pancreatic adenocarcinoma.

discovered that ANXA2 was a gemcitabine-resistant factor by comparing the protein pro ling of gemcitabine-resistant (GEM-) and wild type (WT-) MIA PaCa-2 cell lines. In addition, immunohistochemistry of pancreatic cancer tissues demonstrated that the overexpression of ANXA2 was signi cantly associated with rapid recurrence after gemcitabine adjuvant chemotherapy [16]. Therefore, ANXA2 could be a potential drug target for enhancing the sensitivity of gemcitabine in pancreatic cancer treatment.
DMAMCL is a novel anti-cancer agent and has been under clinical trials in Australia for treatment of glioma tumors (trial number: ACTRN12616000228482). It suppresses glioma by regulating PI3K/AKT/mTOR signaling pathway, NFkB signaling pathway, ROS, and rewiring aerobic glycolysis [17][18][19][20]. In addition, DMAMCL can also inhibit leukemia cell and liver cancer cell growth [21,22]. However, the potential effect of DMAMCL on pancreatic cancer and related mechanism has not yet reported.
In this study, we employed cell viability assay, colony formation assay, and apoptosis assay to demonstrate the effect of DMAMCL on cell viability and apoptosis in three pancreatic cancer cell lines (PANC-1, AsPC-1, BxPC-3). Chemo-proteomics, amino acid mutation, molecular modeling, and gene knock-down by RNAi were performed to verify that DMAMCL suppressed pancreatic cancer cells by targeting Cys132 (C132) of ANXA2. Moreover, we evaluate the synergistic effect between DMAMCL and gemcitabine on three pancreatic cancer cell lines. Overall, our results reveal that DMAMCL inhibits cell proliferation, promotes cell apoptosis, and increases the sensitivity of gemcitabine in pancreatic adenocarcinoma, which provides a novel strategy for the treatment of pancreatic cancer.

Methods
Chemicals DMAMCL and its active metabolic form micheliolide (MCL) were synthesized as previously reported [23].

Cell viability assay
For cell viability assay, human pancreatic cancer cell lines were seeded in 96-well-plates (4000 cells/well). Cells were treated with varying concentrations of DMAMCL or MCL for a certain time. After 48 or 72 h, 20 μL MTT reagent (Solarbio, M8180) (5 mg/mL) was added to each well. After 4-h incubation, the liquid in the wells was replaced by 150 μL dimethyl sulfoxide (DMSO). Cell viability was determined by optical density (OD) values at 570 nm.

Colony formation assay
For colony formation assay, human pancreatic cancer cell lines were seeded in 6-well-plate (2000 cells/well). Cells were treated with varying concentrations of DMAMCL or MCL for 14 days. After that, the culture medium was discarded, and cells were xed with ice methanol for 15 min. Fixed cells were stained with crystal violet (Solarbio, G1063) for 15 min and washed 3 times with PBS. Finally, the plates were photographed by a digital camera.

Apoptosis assay
The apoptosis assay was detected by FITC coupled with Annexin-V apoptosis detection kit (Beyotime, C1062). PANC-1 cells were seeded in 6-well-plates (2×10 5 cells/well). After cells adhered overnight, they were treated with different concentrations of DMAMCL or MCL, with or without gemcitabine. After 48 h treatment, cells were washed 3 times with PBS and digested with trypsin without EDTA (Gibco) in EP tubes, followed by centrifugation at 1000 g for 3 min. The cell pellets were washed twice with PBS, following by adding 100 μL 1×binding buffer to suspend the cells. For cell staining, 5 μL FITC Annexin V and 5 μL of PI were added to each tube and incubated at room temperature (keep in dark) for 15 min.

Chemo-proteomics to identify MCL-bound proteins
Pull-down experiment was carried out according to the method of literature [24]. Brie y, BxPC-3 cells were plated on 10 cm 2 cell culture dish and grown to con uence for 24 h. Cells were harvested and lysed in NP40 lysed buffer containing 1 mM PMSF and protease inhibitors. MCL-biotin (Probe) or MCL-S-biotin (NC probe) was incubated with cell lysates overnight at 4℃, then the prewashed streptavidin beads (Thermo, 20349) were added to each sample and incubated overnight at 4℃. On the second day, the beads ware washed six times with lysis buffer, and the bead-bound proteins were eluted and boiled in 2×SDS loading buffer. The bead-bound proteins were separated by SDS-PAGE and visualized by silver staining. The protein-containing band in the gel was excised, followed by in-gel digestion and analysis by LC-MS/MS.

Site-directed mutant assay
Site-directed mutant assay was constructed using QuikChange II XL site-directed mutagenesis kit (Stratagene, USA). In order to mutate the cysteine at position 132 of pETM3C-ANXA2 plasmid to glycine, primers were designed (forward, 5'-CTCATTGAGATCATCGGCTCCAGAAC-3'; reverse, 5'-CTCCTGAGAGAGTAA CTCTAGTAGC-3') and synthesized (GENEWIZ, China). According to the instructions of the site-directed mutagenesis kit, the mutated plasmid was obtained by PCR. Agarose gel electrophoresis was used to detect whether the target band of the PCR product is correct. If the target band was correct, added 1 μL of DMT enzyme to the PCR product and mixed well. The mixture was kept in a 37℃ water bath for 1 h to obtain digestion products. The digested product was puri ed by agarose gel DNA recovery kit (TIANGEN, DP209-03). The puri ed PCR product was transformed to E.coli strain DH5α, and the plasmid was extracted according to the plasmid extraction instructions, and then sent to the company for sequencing. The nally con rmed target plasmid was transformed into E. coli strain BL21 (DE3) for the expression and puri cation of protein.

Molecular docking
The molecular docking experiment was performed with Schrodinger software. The ANXA2 protein structure was downloaded from the UniProt datebase (https://www.uniprot.org), and the protein was optimized through Schrodinger software to obtain the best protein model. The three-dimensional structure of MCL was prepared using Chemdraw software. LigPrep tool was used to minimize the energy of the ligand to correct the coordinates, ionization, stereochemistry, and tautomeric substitution structure to obtain the proper conformation by adding or removing hydrogen bonds. Gliding agent ligand docking was used to dock MCL to the active site of ANXA2, and calculate the binding energy. All docking calculations used Glide XP (super precision) mode.

Expression and puri cation of recombinant ANXA2
The pETM3C-ANXA2 plasmid was transformed into BL21 competent cells, coated on a kana-resistant solid medium plate, and placed in a 37℃ incubator for 12-14 h. Monoclonal strains were picked into 5 mL of LB medium with kana resistance and shaken overnight at 37℃ and 180 rpm constant temperature shaker. The bacterial liquid obtained above was transferred to 250 mL of LB medium with kana resistance, and expanded for 4 h at 37℃ and 180 rpm constant temperature shaker. The bacterial solution was added to 25 μL IPTG (1 mmol/L), and it was induced for 20 h at 18℃ and 180 rpm constant temperature oscillator. The bacterial solution was collected, centrifuged at 4000 rpm, 4℃ for 30 min, and the supernatant was discarded. The pellet was resuspended in an appropriate volume of 1× bacteriolytic buffer and stored at -80℃ for the next step of puri cation. The bacterial pellet was ultrasonically disrupted for 15 min, centrifuged at 4000 rpm, 4℃ for 30 min, and the supernatant was collected. The supernatant was added to a nickel column equilibrated with 10 mM imidazole in advance, and repeated 2 times. Contaminated proteins were eluted with 30 mM imidazole until the collection solution detected by Coomassie Brilliant Blue G250 was not blue. The target protein was eluted with 250 mM imidazole and collected into 1.5 mL centrifuge tubes on ice until the collection solution detected by Coomassie Brilliant Blue G250 was not blue. 20 μL of eluate were drawn from each tube, and 5 μL of 5×loading buffer was added to the suction uid, mixed, and boiled to obtain protein samples. The protein purity was detected by SDS-PAGE and Coomassie Brilliant Blue G250 copying method. The high-purity protein collection solution was used to remove imidazole by ultra ltration, and then the protein solution was obtained by adding an appropriate volume of physiological saline.
The same procedure was used to express and purify the pETM3C-ANXA2-C132G mutant.

Lentiviral transduction
293T cells were seeded in a 6-well plate (3.5×10 5 cells/well) and cultured overnight in a humidi ed incubator with 5% CO 2 at 37℃.The medium was changed to serum-free medium (1 mL/well) before cell transfection. According to the instructions of JIKAI GENE, the lentiviral packaging systems corresponding to the three plasmids of ANXA2-RNAi (16427-1), ANXA2-RNAi (16423-1) and ANXA2-RNAi (16424-1) plasmids were obtained, and the system volume of each is 50 μL. The lentivirus packaging system was added to 293T cells and incubated in a humidi ed incubator with 5% CO 2 at 37℃ for 6 h. After 6 h, the mediums in the cells were replaced with new complete mediums (2 mL/well). After the cells were cultured in a humidi ed incubator with 5% CO 2 at 37℃ for 48 h or 72 h, shRNA virus was obtained, and the shRNA virus was centrifuged to take the supernatant for use or stored at -80℃. For lentiviral transduction, AsPC-1 cells were cultured in serum-free medium containing lentivirus for 6 h, then replaced by new complete medium for 24 h. For generating stable cell lines, infected cells were selected with 1 μg/mL puromycin (Sigma, 540411) for 72 h. Western blot experiments veri ed which plasmid has the best knockdown effect and used it for the next knockdown experiment.

Western blot analysis
Cell lines were lysed in NP40 lysis buffer with protease inhibitors and the total protein concentration was quanti ed by BCA assay (Thermo Fisher Scienti c, 23227). The normalized samples were analyzed by SDS-PAGE and western blot using standard protocols and the following primary antibodies: anti-ANXA2 Human pancreatic cancer cell lines were seeded in 96-well plates (4000 cells/well). The cells were treated with DMAMCL, gemcitabine and DMAMCL combined gemcitabine (concentration ratio 1:3) for 72 h, respectively. After 72 h, 20 μL MTT reagent (Solarbio, M8180) (5 mg/mL) was added to each well. After 4 h incubation, the liquid in the wells was replaced by 150 μL dimethyl sulfoxide (DMSO). Cell viability was determined by optical density (OD) values at 570 nm and the inhibition rate of each treatment was calculated. The inhibition effect and drug combination index of DMAMCL and gemcitabine was calculated by the Compusyn software (Dr. Dorothy Chou) using constant drug concentration ratio.

Statistical analysis
Results are representative examples of three individual experiments. Statistical analysis and graphical presentation were using GraphPad Prism 5.0. P values were determined by a two-tailed Student's t test.

DMAMCL inhibits proliferation and promotes apoptosis of pancreatic cancer cells
To investigate the effect of DMAMCL on the proliferation of pancreatic cancer cells, we performed cell viability assays on AsPC-1, BxPC-3 and PANC-1 cells (Table 1) were 18.79 ± 1.27 μM, 11.73 ± 1.07 μM, and 6.08 ± 0.78 μM, respectively. Compared to AsPC-1 and BxPC-3, PANC-1 behaves more sensitive to DMAMCL and MCL. Furthermore, we investigated the effect of DMAMCL on the anchorage-independent growth of pancreatic cancer cells through colony formation assays. It was observed that the inhibition of DMAMCL on the foci formation of AsPC-1 and PANC-1 was enhanced in a dose-dependent manner, and MCL also has a similar effect ( Figure 1A). These results indicated that DMAMCL could inhibit the proliferation of pancreatic cancer cells.
In order to explore the effect of DMAMCL on the apoptosis of pancreatic cancer cells, cell apoptosis was detected by ow cytometry analysis and further con rmed by western blot. PANC-1 cells were treated with different concentrations of DMAMCL or MCL for 48 h. PI-Annexin V analysis ( Figure 1B) showed that DMAMCL and MCL promoted PANC-1 cell apoptosis in a dose-dependent manner. Compared to DMSO treatment group, both 20 μM DMAMCL and 10 μM MCL improved 20.4 % and 45.4 % apoptosis respectively. Western blotting ( Figure 1C) showed that DMAMCL treatment promoted the activation of caspase-9 and the release of cytochrome C in PANC-1 cells, which is consistent with the study of DMAMCL-induced apoptosis in hepatocellular carcinoma by Shunnan Yao et al [25]. The above results indicate that DMAMCL may promote apoptosis of pancreatic cancer cells in a caspase-dependent manner.

MCL directly binds to Cys132 of ANXA2 in BxPC-3 cell
To identify the cellular targets of DMAMCL, we performed chemical proteomics assay in BxPC-3 cell. The main active metabolite of DMAMCL is MCL, which contains a typical α-methylene-γ-lactone group that can react with -SH of cysteine in accessible proteins, and induce a covalent modi cation [26]. Thus, using biotin-conjugated MCL (MCL-biotin, Probe) and biotin-conjugated inactive MCL (MCL-S-biotin, NC probe) as probes (Figure 2A), we explored the targets of DMAMCL in BxPC-3 cells. Cellular lysates were rstly incubated with Probe or NC probe, and the mixture was pulled down with streptavidin-coated agarose beads. Then precipitated proteins on beads were resolved by SDS-PAGE and detected by silver staining. Special proteins bands in Probe treatment was excised and con rmed by LC-MS/MS. Peptide mass ngerprinting data analysis revealed that 13 peptides of ANXA2 were detected in the mass spectrometry, which con rmed ANXA2 was one target of MCL. Immunoblotting was further used to monitor the presence of ANXA2 in the precipitates. The results showed that a single band with a molecular weight of about 38 kDa was clearly precipitated with Probe instead of NC probe, indicating ANXA2 was present in the precipitate ( Figure 2B). Furthermore, the amount of ANXA2 in precipitate increased with the probe in concentration-dependent manner ( Figure 2C). Using 20-fold excess MCL as a competitor, ANXA2 was disappeared in pull down precipitate of Probe ( Figure 2D), which further suggested that MCL directly binds to ANXA2.
ANXA2 is highly expressed in pancreatic cancer cells and has four cysteine residues (Cys-8, Cys-132, Cys-261, and Cys-334) [7]. To further explore the site of MCL binding to ANXA2, molecular docking experiments and site-directed mutation experiments were performed. Molecular docking predicted that Cys132 (C132) of ANXA2 could be the binding residue for MCL with the lowest binding free energy as -3.084 ( Figure 2E). Then we constructed plasmid containing C132G mutation of ANXA2 and expressed it in E. coli. Probe incubation and western blot assay showed that recombinant ANXA2 with C132G mutation displayed less binding signal with Probe ( Figure 2F), which con rmed Cys132 was the main binding site of MCL.

ANXA2 is upregulated in cancer patients and affects the inhibition activity of DMAMCL in pancreatic cancer cells
ANXA2 is a member of vertebrate annexins. To clarify its role in cancer, the expression of ANXA2 in different types of cancer patients was examined through the HPA database (https://www.proteinatlas.org). Among seventeen types of cancer examined, patients with pancreatic cancer has the highest expression of ANXA2 ( Figure 3A). When compared the expression levels of ANXA2 between three primary tumors (glioma, colorectal cancer, pancreatic cancer) and corresponding normal tissues, ANXA2 were up-regulated in all three examined tumors ( Figure 3B). By correlating ANXA2 with overall survival in cancer patients, we founded that high levels of ANXA2 was associated with a low survival probability, whereas low ANXA2 expression predicted a relatively high survival probability ( Figure  3C). These results indicated that ANXA2 could be a potential marker and drug target for cancer treatment.
In order to verify whether DMAMCL or MCL inhibited pancreatic cancer proliferation by targeting ANXA2, we performed dose inhibition assays on AsPC-1-WT and AsPC-1-sh ANXA2 cells ( Figure 3D). After treatment with MCL for 48 h, AsPC-1-shANXA2 cells showed less sensitivity to MCL, indicating that knockdown of ANXA2 partially abrogate the inhibition effect of MCL in AsPC-1 cells. Thus, ANXA2 was considered as one of the crucial drug targets for DMAMCL in pancreatic cancer cells.

DMAMCL enhances the inhibition effects of gemcitabine in pancreatic cancer cells
Gemcitabine is one of the rst-line chemotherapeutic agent for pancreatic cancer and its low clinical response is related to the high expression of ANXA2 [14,16,27]. Considering DMAMCL binds to ANXA2, we performed cell viability assays of gemcitabine with or without DMAMCL on three pancreatic cancer cells. The cells were treated with different concentrations of DMAMCL, gemcitabine or DMAMCL+ gemcitabine (mixed in proportion) for 72 h and then for MTT assay. Compared with DMAMCL or gemcitabine alone, DMAMCL combined gemcitabine signi cantly inhibited the proliferation of pancreatic cells ( Figure 4A). In order to verify whether DMAMCL and gemcitabine produce a synergistic effect, we used CompuSyn software to calculate the combination index (CI) and dose reduction index (DRI) ( Figure  4B, Table 2 and Table 3). The results showed that the CI value of DMAMCL and gemcitabine at ED 50 , ED 75 and ED 90 were 0.63, 0.69 and 0.77 in BxPC-3 cells, which exhibits the best synergistic effects in terms of reduction of cell viability based on the Chou-Talalay analysis. For MCL and gemcitabine, the best synergistic effect was found in PANC-1 cells, which is 0.43, 0.46 and 0.89 at ED 50 , ED 75 and ED 90 .
Furthermore, under the treatment of DMAMCL at 5 μM, the IC 50 of gemcitabine was found decreased from 4.21 to 2.85 μM in AsPC-1 cells and from 2.33 to 0.75 μM in PANC-1 cells, indicated that DMAMCL could increase the e cacy of gemcitabine on pancreatic cancer cell proliferation (Table 4 and Figure 5).
DMAMCL enhances the pro-apoptosis effect of gemcitabine in pancreatic cancer cells In order to further verify whether DMAMCL enhance the pro-apoptosis effect of gemcitabine in pancreatic cancer cells, the early and late apoptosis was analyzed by ow cytometry analysis. Dual-labeled uorescence activated cell sorting (FACS) analysis (Annexin V and PI) was used to measure early and late apoptosis. Annexin V detects early apoptosis, whereas both PI and Annexin V detect late apoptosis. By 48 h treatment, compared with single treatment (20 μM DMAMCL, 10 μM MCL, 2.5 μM gemcitabine), combination regimen (20 μM DMAMCL + 2.5 μM gemcitabine ; 10 μM MCL + 2.5 μM gemcitabine) signi cantly increased cancer cell apoptosis, especially the early apoptosis ( Figure 6A).

ANXA2 mediated PI3K/AKT inhibition contributes to the synergistic effect of DMAMCL and gemcitabine
The phosphoinositide 3-kinase/Akt signalling pathway is a recognized key parameter in numerous cellular processes such as proliferation, cell cycle and angio-genesis, and is frequently activated in pancreatic cancer and especially in gemcitabine resistance [28]. Akt kinase can be activated by phosphorylation on Thr 308 or Ser 473 and active-Akt will promote cell growth and survival to apoptotic insults [29,30]. It was also proved that DMAMCL has the ability to inhibit of PI3K/AKT and induce apoptosis in hepatocellular carcinoma [25]. To investigate whether the effect of DMAMCL on ANXA2 further in uence the PI3K/AKT signal pathway, we detected the expression and the phosphorylation of Akt on ser473. Western blot results demonstrated that compared with DMAMCL or gemcitabine alone, DMAMCL + gemcitabine exhibited more potent inhibition of AKT signaling pathway ( Figure 6B), which suggested that the synergistic effect of DMAMCL and gemcitabine could be partly mediated by ANXA2-AKT pathway.

Discussion
Gemcitabine is the standard treatment for advanced pancreatic adenocarcinoma, but the e cacy of this reagent is still limited. ANXA2, as a tumor-associated protein, has been proved promoting cancer progression including proliferation, invasion, and metastasis in various cancer types [8][9][10][11][12][13]. In addition, ANXA2 was also reported to mediate gemcitabine resistance in pancreatic cancer [9,16,31,32]. Thus, ANXA2-targeted agent is anticipated to be a novel treatment for pancreatic cancer.
DMAMCL has been shown to exert potent anticancer properties on multiple cancers, but has not yet in pancreatic cancer. Herein, we explored its effect on three pancreatic cancer cell lines and identi ed ANXA2 was a novel target for its property. Meanwhile, we found that gemcitabine in combination with DMAMCL could inhibit tumor growth and induce apoptosis more effectively than gemcitabine or DMAMCL alone.
To clarify the mechanisms of gemcitabine resistance induced by ANXA2 overexpression, Kagawa et al. analyzed the signaling pathways up-regulated in the gemcitabine-resistant cell lines with overexpressed ANXA2. Bio-Plex phosphorylation protein assay showed up-regulation of p-Akt in GEM-MIA PaCa-2 cells in which ANXA2 is highly expressed. Inhibition of p-Akt through mTOR inhibitor canceled gemcitabine resistance in GEM-MIA PaCa-2 cells [28], indicated ANXA2-mediated activation of Akt accounting for the low response of gemcitabine [15,[33][34][35]. Our previous work has demonstrated that DMAMCL can signi cantly inhibit the level of p-AKT in different cancer cells [20,25]. In this study, we also observed that the inhibition of DMAMCL on p-AKT was signi cantly reduced in AsPC-1 (sh-ANXA2) cells. Thus, we supposed that the synergistic mechanism of enhanced e cacy of gemcitabine may be ANXA2-AKT pathway inhibition induced by DMAMCL.
In conclusion, our work revealed that DMAMCL could suppress pancreatic cancer by targeting ANXA2. The binding of DMAMCL on C132 of ANXA2 might down-regulate p-AKT that offset the activation of p-AKT by gemcitabine. DMAMCL in combination with gemcitabine exhibits a synergistic cytotoxic effect, which may be used as a more effective treatment for pancreatic cancer.   Table 2 The combination index (CI) and dose reduction index (DRI) of DMAMCL and gemcitabine in AsPC-1, BxPC-3 and PANC-1 cells.  Table 3 The combination index (CI) and dose reduction index (DRI) of MCL and gemcitabine in AsPC-1, BxPC-3     The protein levels of AKT/p-AKT in PANC-1 cells analyzed by western blot. Error bars, mean ± s.e.m., n = 3 biological replicates. *P < 0.05, **P < 0.01, ***P < 0.001.