A novel HDAC inhibitor suppresses extracellular vesicle VEGF-C-mediated lymphangiogenesis and pancreatic cancer early dissemination

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer characterized by early dissemination and poor drug response. Loss-of-function of dual specicity phosphatase 2 (DUSP2), a critical regulator of MAPKs signaling, is highly associated with cancer malignancies. Therefore, it may provide new therapeutic strategy if the actions of DUSP2 can be restored. The tumor suppressor role of DUSP2 was demonstrated via DUSP2 re-expression in the orthotopic mouse model of pancreatic cancer and knockout of Dusp2 in the pancreas by transgenic mouse model. Immunohistochemical staining and histology analysis was performed to evaluate tumor development and progression. Bioinformatic analysis was utilized to identify potential drug which mimics DUSP2 reexpression. Pancreatic cancer cell survival, migration ability, and the expression and function of extracellular vesicle (EV) associated vascular endothelial growth factor C (VEGF-C) was measured. The effect of the novel HDAC inhibitor on pancreatic cancer progression was evalulated by orthotopic mouse model. and an effective therapy the high priority to patients’ survival. In this we demonstrate DUSP2 acts as a tumor suppressor that exogenous is sucient to restrain tumor formation. Furthermore, we identied that novel HDAC inhibitor, B390 functions like DUSP2. In addition to abolishing cancer cell growth, B390 inhibits EV-associated VEGF-C expression and lymphatic invasion, by which DUSP2 regulates. Orthotopic mouse model demonstrates B390 not only promotes tumor apoptosis but also inhibits tumor spreading into the surrounding tissue. Together, our study suggests the anti-tumor function of the novel HDAC inhibitor in recapitulating the action of DUSP2 in suppressing PDAC progression.


Abstract Background
Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer characterized by early dissemination and poor drug response. Loss-of-function of dual speci city phosphatase 2 (DUSP2), a critical regulator of MAPKs signaling, is highly associated with cancer malignancies. Therefore, it may provide new therapeutic strategy if the actions of DUSP2 can be restored.

Methods
The tumor suppressor role of DUSP2 was demonstrated via DUSP2 re-expression in the orthotopic mouse model of pancreatic cancer and knockout of Dusp2 in the pancreas by transgenic mouse model. Immunohistochemical staining and histology analysis was performed to evaluate tumor development and progression. Bioinformatic analysis was utilized to identify potential drug which mimics DUSP2 reexpression. Pancreatic cancer cell survival, migration ability, and the expression and function of extracellular vesicle (EV) associated vascular endothelial growth factor C (VEGF-C) was measured. The effect of the novel HDAC inhibitor on pancreatic cancer progression was evalulated by orthotopic mouse model.

Results
Forced expression of DUSP2 abrogated tumor formation and loss of Dusp2 facilitated Kras-driven PDAC formation. Increased HDAC1 expression was found in PDAC and inhibition of HDAC showed similar gene pro le as Kras knockdown and DUSP2 re-expression. Treatment with B390 inhibited growth and migration abilities of PDAC cells, decreased EV-associated VEGF-C expression, and suppressed lymphatic endothelial cell proliferation. In vivo, B390 not only suppressed tumor growth by increasing tumor cell death, it also inhibited lymphangiogenesis and lymphovascular invasion.

Conclusions
Our data provide the proof-of-concept evidence to demonstrate the potential of using novel HDAC inhibitor in PDAC treatment which alleviates loss-of-DUSP2-mediated pathological processes.

Background
Pancreatic ductal adenocarcinoma (PDAC) accounts for more than 90% of pancreatic malignancies and is notorious for the poor prognosis with a less than 6% ve-year survival rate. PDAC is rarely diagnosed in pre-metastatic state and lymphovascular invasion (LVI) is frequently observed in PDAC patients (1). As in other types of solid tumors, increased LVI and lymph node metastasis correlates with unfavorable outcome in PDAC (2). LVI can be the consequence that tumor cells stimulate the formation of blood/lymphatic vessels via secreting factors and gain of invasion ability. Since PDAC is highly metastatic, it is devastating to understand the mechanism of early metastasis and identify means to prevent it in order to improve patient survival.
KRAS mutation occurs in more than 90% of PDAC. Inhibition of MAPKs/ERK suppresses the growth of Kras driven pancreatic cancer growth in transgenic mouse model (3,4), indicating the necessity of MAPKs/ERK signaling. Dual speci city phosphatases (DUSPs) are protein phosphatases that can simultaneously dephosphorylate serine/threonine and tyrosine residues of the same substrate. DUSP2 belongs to the subgroup of MAPK phosphatase (MKPs) that antagonize the activation of MAPK cascades, mainly via negatively regulates ERK activation. As ERK is the convergent point of multiple signaling pathways, dysregulation of DUSP2 will lead to uncontrolled cellular responses and is associated with disease progression (5). Downregulation of DUSP2 has been found in leukemia and many solid tumors (6,7). It has been demonstrated that DUSP2 expression is suppressed by hypoxia and downregulation of DUSP2 promotes drug resistance and cancer stemness (6,8). In pancreatic cancer, loss of DUSP2 was observed at the early stage of pancreatic intraepithelial neoplasia (PanIN), and DUSP2 downregulation promotes LVI in the orthotopic mouse model (9). Considering the critical roles of DUSP2, identi cation of means to restore its function may be an alternative way to alleviate PDAC malignancy.
In this study, we aimed to investigate whether re-expression of DUSP2 is su cient to suppress PDAC progression. More importantly, to identify novel inhibitor that can mimic DUSP2's function. Our results indicate that HDAC inhibitor functions like DUSP2 and can be used to inhibit tumor growth and prevent lymphovascular invasion in mouse model of PDAC.

Cell culture
Pancreatic ductal adenocarcinoma cell line MIA PaCa-2 were maintained in RPMI 1640 complete medium containing 10% fetal bovine serum (FBS), glutamine, sodium pyruvate, and antibiotics for prevention of bacteria, fungi, and mycoplasma. HPDE-E6E7 was initially maintained in keratinocyte-serum free medium (50 mg/ml bovine pituitary extract) and gradually cultivated in RPMI 1640 complete medium. LECs were purchased (PromoCell) and cultured in endothelial cell growth medium MV2 with supplement (PromoCell). KPC cell line was isolated from a pancreatic tumor developed in KPC transgenic mouse model as reported (10) and maintained in RPMI 1640 complete medium. Cell lines were maintained at 37 °C, 5% CO 2 humidi ed incubator. MIA PaCa-2 and HPDE-E6E7 have been authenticated.
Proliferation and migration assay MIA PaCa-2 cells were plated on 24-well plate and transfected with GFP or DUSP2-GFP plasmids. After 48 hours, MTT was added and crystal was dissolved in DSMO, OD590nm was measured and quanti ed.
Isolated EV from control and B390 treated MIA PaCa-2 was resuspended in PBS and treated LECs for 48 hours, LECs were xed and stained for Ki67 and DAPI. MIA PaCa-2 cells were suspended in serum-free RPMI media and placed in transwell chamber with a pore size of 8.0 µm (PIEP12R48, Millipore). 10% FBS RPMI culture media was used as chemoattractant. After 16 hours of incubation, migrated cells were xed and stained with crystal violet. Cell numbers were counted and analyzed by ImageJ cell counter.

Extracellular vesicles isolation and analysis:
MIA PaCa-2 cells were plated on the rst day and medium was changed for serum-free RPMI medium containing ITS-M (Insulin-Transferrin-Selenium Mixture, Genedirex) for one more day. Conditioned medium (CM) was rst under centrifugation at 500 g for 10 min to remove cell debris and supernatant was further under centrifugation 16000 g for 30 min. Processed CM were sent to the Center for Micro/Nano Science and Technology at National Chung Kung University for nanoparticle tracking analysis (NTA). EVs were extracted from the cell culture medium by ExoQuick exosome precipitation solution (SBI system Biosciences) according to the manufacturer's protocol.

Immunohistochemistry (IHC)
Tumors from animals were xed in 4% paraformaldehyde, embedded in para n, and cut into 5-µm sections and sections were stained with H&E or IHC. Primary antibodies against Lyve-1 (Angiobio, 11-034) and CD31 (abcam, ab28364) were used for immunohistochemistry staining of lymphangiogenesis and angiogenesis. Cleaved caspase 3 (cell signaling, 9661T) was used for the staining of apoptosis.
HDAC-1 (cell signaling, #5356 s) was used for staining of pancreas and tumor from mice.
Male SCID mice (6-to 8-weeks of age) were used for pancreas orthotopic injection. 1 × 10 6 MIA PaCa-2 cells were suspended in 100 µl of RPMI/matrigel mixture and injected into the pancreas of mice. For drug experiment, mice were randomly assigned to experimental or control groups after two weeks of inoculation. In the treatment group, mice received vehicle or B390 (12.5 mg/kg) by oral gavage with exible feeding tubes, ve days/week for two consecutive weeks (8). Mice were sacri ced three days after the last treatment and tumors were excised.

Statistical analysis
GraphPad Prism 5.00 was used for statistical analysis. Two-tailed Student's t-test and one-way ANOVA with Tukey posttests were employed for comparing two or three groups, respectively. Error bars represent standard error of the mean from independent experiments. Asterisks denote signi cant difference from control group *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Expression of DUSP2 inhibits pancreatic cancer tumor formation
Our previous study demonstrated that DUSP2 is downregulated in PDAC (9); therefore, we aimed to investigate whether re-expression of DUSP2 can diminish tumor burden. MIA PaCa-2 cells, which expressed very low level of DUSP2, were transfected with plasmids containg GFP or DUSP2-GFP.
Exogenously overexpression of DUSP2 decreased basal level of ERK activation in MIA PaCa-2 cells (Fig. 1A) and suppressed cell growth (Fig. 1B). To determine the effect of DUSP2 on tumor formation in vivo, control and DUSP2 expressing MIA Paca-2 cells were orthotopically injected into mouse pancreas and tumor growth was followed by luciferase imaging. Tumor formation was abrogated when DUSP2 was expressed (Fig. 1C). Indeed, when the animals were sacri ced, no tumor was observed in the DUSP2expressing group (Fig. 1C). Because there is no tumor developed in the DUSP2 overexpressing group, we can not evaluate how DUSP2 inhibited tumor growth. Therefore, another pancreatic orthotopic tumor model was generated by injecting control or DUSP2 knockdown (KD) PANC-1 cells. Immunohistochemical staining showed knckdown of DUSP2 increased Ki67-positive cells and markedly reduced number of apoptotic cells (Fig. 1D). These results indicate that DUSP2 function as a tumor suppressor and loss of DUSP2 promotes proliferation and survival of tumor cells during PDAC tumorigenesis.

Loss of Dusp2 promotes Kras driven PanIN progression in transgenic mouse model
Activating mutation of KRAS occurs in more than 90% of PDAC; however, mice carry Kras mutation in the pancreas develop pre-invasive PanIN but seldom progresses to invasive cancer (11,12). Therefore, we aimed to evaluate the effect of Dusp2 loss in the background of Kras mutation. For this purpose, KDC mice (Pdx-1Cre, LSL-Kras G12D/+ , Dusp2 / ) were generated ( Fig. 2A). Pancreas of KDC mice started to show histological abnormalities at 2 months of age (Fig. 2B). However, to evaluate the additional effect of Dusp2 loss, we compared the histology of pancreas from wild type, KC (Pdx-1Cre, LSL-Kras G12D/+ ), and KDC mice at 6 months of age. Pancreas of KC mice showd regions of acinar to ductual metaplasia (ADM), which was considered as the pre-neoplastic lesions of PDAC (13). PanINs were also observed in 6 months old KC pancreas (Fig. 2C, c-d). Of note, KDC mice not only showed more severe ADM and PanIN phenotypes but alo developed carcinoma surrounded by dense stroma in pancreas (Fig. 2C, e-f). Together, these results indicate that loss of Dusp2 speed up the development and progression of Kras-induced precancerous neoplasia.

Restoration of DUSP2-mediated functions by novel histone deacetylase inhibitor
Due to the di culty in restoration of a tumor suppressor protein in reality, we aimed to identify molecules which may exert similar functions. By cross-referencing our previously published microarray dataset (8) with connectivity map (14), we identi ed a list of inhibitors that function like DUSP2 (Fig. 3A). Next, the CLUE Plateform (COMMAND) was utilized to search the mechanism of action of these inhibitors and identi ed that HDAC inhibitor functions similarly to KRAS knockdown (Fig. 3B), an indications of DUSP2 reactivation. Moreover, a public dataset contains transcriptome pro ling of two class I/II HDAC inhibitors (sodium butyrate and Trichostatin A) (15) was utilized to cross with our own dataset of DUSP2 overexpression. 44 up-regulated genes and 21 down-regulated genes in both HDAC inhibitor datasets show signi cant fold change in DUSP2 dataset (Fig. 3C). Interestingly, DUSP2 was found up-regulated in both HDAC inhibitors treatment after 12 and 48 hours, implying HDAC inhibitor may exert similar tumor suppressive functions via reinforcement of DUSP2 expression. As the expression of HDACs in PDAC is largely uninvestigated, we analyzed microarray datasets from Oncomine and identi ed that the expression of some HDACs, especially HDAC1, is signi cantly increased in multiple datasets (Fig. 3D). We further performed immunohistochemistry on mouse pancreatic tumor developed in KPC transgenic mouse model (LSL-Kras G12D ; TP53R 172H ; Pdx-1-Cre), which has been demonstrated to resemble features of human PDAC (16). As compared to normal epithelial ducts (littermate control), increased HDAC1 can be detected in PanINs and carcinoma (Fig. 3E). These results suggest that HDAC1 may participate in the progression from normal duct to neoplasia.
The effects of novel HDAC inhibitors (B369 and B390) had recently been demonstrated in colon cancer which abolished hypoxia mediated cancer stemness (8). Thus, we rst determined the dose response of normal and cancer cells to B369 and B390. Normal human pancreatic ductal epithelial cell line, HPDE-E6E7, show no response to both B369 and B390 (Fig. 4A) while the growth of MIA Paca-2 was signi cantly inhibited after 48 hours treatment, spe cially by B390 (Fig. 4B). Importantly, treatment of B369 and B390 increased the expression of DUSP2 in MIA Paca-2 cells (Fig. 4C), indicating novel HDAC inhibitors selectively kill cancer cells via re-expression of DUSP2. We also evalulated the effect of B390 on primary cancer cells isolated from pancreatic tumor developed in KPC mice, which is also sensitive to B390 induced death similar to MIA PaCa-2 cells (Fig. 4D). Next, migration ability, another attribute that DUSP2 regulates in pancreatic cancer cells was also measured. Pre-treatment of MIA Paca-2 cells with B390 for 24 hours signi cantly diminished migration ability compared to control (Fig. 4E).
Since we have recently demonstrated that DUSP2 knockdown promotes extracellular vesicles (EV)-VEGF-C secretion and lymphangiogenesis (9), we thus investigated the effect of B390 on these processes. Reduced levels of EV-VEGF-C was observed (Fig. 4F) and further study revealed that B390 inhibited VEGF-C expression at the transcriptional level (Fig. 4G), thus reducing its loading to EVs. Furthermore, in vitro assay showed that the proliferation of LECs was signi cantly inhibited when treated with EVs isolated from B390-treated MIA PaCa-2 cells (Fig. 4H and 4I). Together, B390 shows inhibitory effects similar to DUSP2 expression in pancreatic cancer cells.

Inhibition of tumor progression by B390 in orthotopic mouse model of pancreatic cancer
To determine the in vivo effect of B390, MIA PaCa-2 cells were injected into the pancreas of mouse and animals received vehicle or B390 for 3 consecutive weeks. Treatment with B390 decreased the size of tumors measured by IVIS imaging (Fig. 5A). We further performed histology analysis on the tumors. Multinodular tumors were observed in the control group while B390-treated tumors are smaller and with necrosis ( Fig. 5B-C). In addition, enhanced cell death was observed in B390-treated tumors as stained by antbodies against cleaved caspase-3 ( an apoptotic marker) (Fig. 5D).
Since loss of DUSP2 promotes tumor-associated lymphangiogenesis, we thus determined the levels of angiogenesis and lymphangiogenesis in control and B390 treated mice. As shown, both angiogenesis (CD31) and lymphangiogenesis (Lyve-1) are signi cantly inhibited in B390 treated group (Fig. 6A-B). Histology examination showed two out of ve mice displayed vascular invasion ( Fig. 6C and Fig. 6 Da-b) and one out of ve shows tumor cells seeding in the fat tissue in the control group (Fig. 6Dc). In contrast, no invasion was observed in B390 treated tumors (right, n = 7). Together, these results suggest that novel HDAC inhibitors decrease VEGF-C expression and exert multiple effects in inhibiting cancer progression, partially mimicking the effect of DUSP2 re-expression.

Discussion
Pancreatic cancer is a very aggressive type of tumor that is often diagnosed with locally advanced stage or metastatic spread. Survival duration for patients with non-resectable tumors is only 4-8 months and chemotherapy shows limited effects (17). While gemcitabine has been used to treat metastatic PDAC, resistance develops rapidly. Therefore, seeking an effective therapy is the high priority to improve patients' survival. In this study, we demonstrate DUSP2 acts as a tumor suppressor that exogenous expression is su cient to restrain tumor formation. Furthermore, we identi ed that novel HDAC inhibitor, B390 functions like DUSP2. In addition to abolishing cancer cell growth, B390 inhibits EV-associated VEGF-C expression and lymphatic invasion, by which DUSP2 regulates. Orthotopic mouse model demonstrates B390 not only promotes tumor apoptosis but also inhibits tumor spreading into the surrounding tissue. Together, our study suggests the anti-tumor function of the novel HDAC inhibitor in recapitulating the action of DUSP2 in suppressing PDAC progression.
DUSP2 belongs to a subclass of dual speci city phosphatase and is inducibly expressed in the immune system (18). Loss of DUSP2 expression correlates with cancer malignancies and constitutive ERK activation (6)(7)(8)19). Recently, we demonstrated that loss-of-DUSP2 promotes pancreatic cancer early dissemination, likely through lymphovasculature (9). In this study, we show that re-expression of DUSP2 is su cient to prevent tumor formation in the orthotopic mouse model of pancreatic cancer. Also, we found that knockout of Dusp2 facilitates the progression of Kras-mediated PDAC development, suggesting the critical gatekeeper role of DUSP2. Therefore, we reason that identifying a way to restore the expression or function of DUSP2 may be able to inhibit PDAC progression.
By using bioinformatic analysis and cross-referencing with online databases, we found that DUSP2 functions similar to several inihibitors. No surprise that action of mitogen-activated protein kinase (MEK) inhibitors was identi ed with the top score, re ecting that DUSP2 overexpression works similar as MEK inhibition. However, since KRAS mutation tumors show intrinsic resistance to MEK inhibitor (20); therefore, we seeked for different drugs. Interestingly, we found that HDAC inhibitors exert similar functions like DUSP2, which prompted us to consider whether administration of HDAC inhibitor can restore DUSP2 expression or mimic DUSP2's function in suppressing cancer progression. Since overexpression of HDACs has been found in different types of cancer and is associated with tumor grade, prognosis, and chemoresistance (21); therefore, targeting HDAC has always been considered as an anticancer approach. Indeed, there are many recruiting or active clinical trials using HDAC inhibitors or in combination with other drugs. In general, HDAC inhibitors exert anti-cancer effects via reactivating scilenced tumor suppressor genes by epigenetic regulation or inducing cancer cell death. For example, trichostatin A inhibits pancreatic cancer cell growth via inducing cell-cycle arrest and apoptosis (22), or induces the re-expression of tumor suppressor gene such as transforming growth factor beta type II receptor (23). In this study, we identify unrevealed functions of HDAC inhibitor, which advances our understandings about the actions of HDAC inhibitor's anti-cancer mechanism.
Apoptosis is one of the major mechanisms that HDAC inhibitors selectively kill cancer cells. However, there are still some caveats of how HDAC inhibitors promote apoptosis. In this study, we observed the massive activation of caspase-3 in B390-treated tumors, which is similar to DUSP2 overexpression phenotype ( Fig. 1 and reference (6)). As DUSP2 is frequently suppressed by hypoxic stress in solid cancers (6), loss-of-DUSP2 warrants the cancer cells to grow without triggering the proliferation-induced apoptosis switch. The inactivation of apoptosis plays a central role in cancer development. By orthotopic mouse model, we observed that loss of DUSP2 showed less apoptosis in the tumor regions. These results indicate that DUSP2 may be responsible for stress induced cell death and thus loss of function promotes survival of tumor cells and facilitates cancer progression. The anti-cancer function of HDAC inhibitors may be mediated through re-activation of DUSP2 or DUSP2-regulated downstream effectors. By using bioinformatic analysis of our previous published microarray data and online databases, we found HDAC inhibitors and DUSP2 regulate many common biological processes. Since cancer cells treated with B390 showed increased expression of DUSP2, implying the anti-cancer effects of novel HDAC inhibitors may be due to the restoration of DUSP2 expression. However, the detail mechanism of how loss of DUSP2 overlaps with HDAC1/2-mediated functions warrants further invesitagtion.
The most exciting nding in this study is that we identi ed a novel mechanism of HDAC inhibitor and evaluated rarely investigated but highly critical pathophysiological processes in cancer progression including EV function and lymphangiogenesis. Our previous data showed that knockdown of DUSP2 increased EV-associated VEGF-C in PDAC, which facilitated lymphangiogenesis and ultimately results in pancreatic cancer cells early dissemination (9). Since previous study showed that treatment with HDAC inhibitors rescued hypoxia mediated DUSP2 downregulation in colon cancer, we hypothesized that B390 may abrogate loss-of-DUSP2-induced EV-VEGF-C secretion and lymphangiogenesis. Our results suggest that treatment with B390 signi cantly inhibited the mRNA expression of VEGF-C, thus resulting in the reduction of EV-VEGFC in pancreatic cancer cells. Importantly, B390 suppressed tumor cells invaded into lymphovasculature and the surrounding tissue, a phenomenon particularly critical for PDAC malignancy.
These data reveal the previously uncharacterized novel effects of HDAC inhibitor in regulating cancer early dissemination and malignancy.

Conclusions
Our results reveal the tumor suppressor function of DUSP2 in pancreatic cancer and unravel a novel mechanism of HDAC inhibitor in mouse model of PDAC. B390 not only induces apoptosis of cancer cells but also inhibits tumor invasion through lymphavascular system. Since dissemination through lymaphtic system represents a novel and previously not well-characterized route for cancer cells, especially in PDAC, our study shed light on the potential use of B390 or similar HDAC inhibitors as anti-cancer drugs in inhibiting pancreatic cancer progression. Availability of data and materials:

Abbreviations
The datasets during and/or analysed during the current study available from the corresponding author on reasonable request.     quantitative result (right, n=3) show migration assay in MIA PaCa-2 cells pre-treated with DMSO or B390 (1μM) for 24 hours. (F) Serum free conditioned medium (CM) of control and B390 treated MIA PaCa-2 cells was collected and EV was isolated by Exoquick-TC. Western blotting was performed to detect VEGF-C expression. CD9 and HSP70 were used as markers for EV. (G) RT-qPCR of VEGFC in MIA PaCa-2 cells treated with B369 and B390 (1μM) for 24 hours. CYCLOPHILIN B (PPIB) was used as internal control. (H) A selected confocal image shows LECs uptake EV from MIA PaCa-2 cells. MIA PaCa-2 cells were labeled with uorescence dye PKH67 (left) and CM was collected. EV was isolated from CM to treat lymphatic endothelial cells (LECs) for 6 hours (right). (I) Representative images (left) and quantitative result (right, n=3) show proliferation (Ki67 staining) of LECs treated with EV from control or B390 treated MIA PaCa-2 cells for 48 hours. * P < 0.05; ** P < 0.01;***P < 0.001.