Cordycepin disrupts leukemia association with mesenchymal stromal cells and eliminates leukemia stem cell activity

Maintaining stemness of leukemic stem cells (LSCs) and reciprocal interactions between leukemia and stromal cells support leukemic progression and resistance to chemotherapy. Targeting the niche-based microenvironment is thus a new approach for leukemia therapy. Cordycepin is an analogue of adenosine and has been suggested to possess anti-leukemia properties. However, whether cordycepin influences association of leukemia and mesenchymal stromal cells has never been investigated. Here we show that cordycepin reduces CD34+CD38− cells in U937 and K562 cells and induces Dkk1 expression via autocrine and paracrine regulation in leukemia and mesenchymal stromal/stem cells (MSCs). Cordycepin suppresses cell attachment of leukemia with MSCs and downregulates N-cadherin in leukemia and VCAM-1 in MSCs. Moreover, incubation with leukemic conditioned media (CM) significantly induces IL-8 and IL-6 expression in MSCs, which is abrogated by cordycepin. Suppression of leukemic CM-induced VCAM-1 and IL-8 by cordycepin in MSCs is mediated by impairing NFκB signaling. Finally, cordycepin combined with an adenosine deaminase inhibitor prolongs survival in a leukemic mouse model. Our results indicate that cordycepin is a potential anti-leukemia therapeutic adjuvant via eliminating LSCs and disrupting leukemia-stromal association.

Interactions between leukemia-stromal cells by VLA-4/fibronectin or VLA-4/VCAM-1 are associated with the overall survival of AML patients 21 and NFκ B-mediated chemoresistance 22 . The VCAM-1/VLA-4 axis triggers NFκ B activation and its consequent signaling which regulates cross-talk between leukemia and stromal cells 22 . In addition, BM derived MSCs preserve from elimination LSCs of CML when treated by a tyrosine kinase inhibitor through induction of N-cadherin and activation of wnt/β -catenin 23 . These results indicate that direct interaction of leukemia or LSCs with MSCs may contribute to enhanced survival and engraftment of LSCs as well as drug resistance in the BM.
LSCs respond to a number of supportive factors secreted by stromal cells in the BM microenvironment, including IL-6, IL-8 and stromal cell-derived factor-1 (SDF-1 or CXCL12) 24,25 . Among the secreted stromal factors, IL-6, IL-8 and are proinflammatory cytokines released by leukemia-associated stromal cells and are involved in maintaining the leukemic microenvironment for LSCs [24][25][26][27][28] . It has been shown that IL-6 controls leukemic progenitor cell fate and contributes to CML progression 29 . In addition, IL-8 released from BM-derived MSCs act as a mediator in the leukemic environment 26,27 . The IL-8-CXCR2 axis is overexpressed in LSCs and their expression is associated with a worse prognosis of AML and myelodysplastic syndromes patients (MDS) 28 . The IL-8/CXCR2 pathway expressed in leukemia-stromal cells is considered to be a potential prognostic factor and therapeutic target for leukemia and MDS 26 .
Dong-Chong-Xia-Cao is one of the commonly used traditional medicine which comprises the fungus (Cordyceps sinesis) and its infected larvae (Hepialus armoricamus) 30,31 . Cordycepin, also known as 3-deoxyadenosine, is a major active ingredient extracted from Dong-Chong-Xia-Cao and has been suggested to possess anti-leukemia properties [32][33][34][35][36][37] . However, cordycepin quickly loses its activity in animal models due to its short half-life 38 . As an adenosine analog, cordycepin is converted into inosine by adenosine deaminase (ADA) 39 . Co-administration of cordycepin with an ADA inhibitor (ADAi) or increasing cordycepin stability is considered a useful approach for leukemia therapy [40][41][42][43][44] . Cordycepin was reported to impair wnt/β -catenin signaling by inducing degradation of β -catenin in leukemia 35 . However, whether cordycepin suppresses leukemia progression via paracrine regulation has never been investigated. Our aim was to study the mechanism of how cordycepin contributes to suppressing leukemia's progression. Our results suggest that cordycepin is a potential agent or adjuvant for leukemia therapy by targeting LSCs and disrupting the leukemia-MSC association.

Results
Cordycepin reduces the ratio of LSCs and induces Dkk-1 expression via autocrine and paracrine regulation. The population of CD34 + CD38 − cells is considered to be LSC or leukemia initiating cells 45,46 . To investigate whether cordycepin influences the stemness of myeloid leukemia, U937 and K562 cells were treated with cordycepin (50 μ M) for 24 h and the percentage of CD34 + CD38 − cells were determined by flow cytometry analysis. Cordycepin significantly reduced the percentage of CD34 + CD38 − cells in both U937 (Fig. 1A, upper panel) and K562 (Fig. 1A, lower panel).
We postulated that MSCs are educated by leukemia. We therefore examined whether cordycepin influences β -catenin signaling of leukemia in a paracrine manner. MSCs were pretreated with conditioned media of MSCs (MSC-CM), U937 (U937-CM) and K562 (K562-CM) for 48 h, followed by cordycepin treatment for additional 24 h (procedure illustrated in Fig. 1B, upper panel). Expression of Dkk1, an antagonist of wnt/β -catenin signaling in leukemia and MSCs was determined by quantitative real-time PCR (Q-PCR). Cordycepin significantly induced Dkk1 expression in MSC-CM, U937-CM, and K562-CM treated MSCs, whereas cordycepin had no effect on Dkk1 expression in U937 and K562 cells (Fig. 1B, lower panel). These results indicate that cordycepin may attenuate wnt/β -catenin signaling in leukemia cells by both autocrine and paracrine regulation.

Cordycepin suppresses leukemia's attachment with MSCs and reduces expression of N-cadherin in leukemia and VCAM-1 in leukemic CM-treated MSCs.
To elucidate whether cordycepin affects association of leukemia with niche cells, we examined whether cordycepin influences leukemia's attachment to MSCs. U937 and K562 cells were plated on MSCs and consequently treated with different doses of cordycepin. Cordycepin significantly suppressed leukemia cell attachment to MSCs in a dose-dependent manner ( Fig. 2A). To further confirm the suppressive effect of cordycepin on leukemia's attachment with MSCs, U937 and K562 cells were pre-treated with cordycepin (50 μ M) for 24 h and subsequently plated on the MSCs for additional 24 h. Cordycepin significantly reduced U937 and K562 attached on MSCs (Fig. 2B). An earlier study has demonstrated that the leukemia-microenvironment association protects CML stem/progenitor cells from tyrosine kinase inhibitor treatment through N-cadherin signaling 23 . We therefore examined whether cordycepin modulates N-cadherin expression in leukemia and MSCs. U937 and K562 cells were co-cultured with MSCs and treated with or without 50 μ M cordycepin for 48 h. U937 or K562 cells were removed from MSCs and N-cadherin expression was determined by Western blot analysis (illustrated in Fig. 2C, upper panel). Co-culturing of U937 with MSCs induced but co-culture of K562 with MSCs had no significant effect of N-cadherin expression, whereas cordycepin markedly suppressed N-cadherin in U937 and K562 cells (Fig. 2C, lower panel). In contrast, neither co-culturing with leukemia nor cordycepin treatment influenced N-cadherin expression in MSCs (Fig. 2D).
A recent study has indicated that the leukemia-stroma VCAM-1/VLA-4 interaction is involved in chemoresistance of leukemia 22 . We examined whether cordycepin regulates VCAM-1/VLA-4 expression in leukemia-stromal cells. MSCs were incubated with MSC-CM, U937-CM and K562-CM for 48 h, followed by cordycepin treatment for an additional 24 h. MSCs and leukemia cells were harvested and expression of VCAM-1 was examined by Q-PCR and Western blotting (illustrated in Fig. 3A, upper panel). U937-CM and K562-CM induced VCAM-1 in MSCs (Fig. 3A, lower panel). These results were further confirmed by Western blot analysis (Fig. 3B). Intriguingly, cordycepin significantly reduced VCAM-1 expression in leukemic CM-treated MSCs but had no effect in leukemia cells (Fig. 3A, lower panel). We further determined VLA-4 expression by Q-PCR of Cordycepin abolishes leukemic CM-induced IL-8 and IL-6 in MSCs. IL-8 and IL-6 are proinflammatory cytokines involved in modulating the progression of myelogenous leukemia [24][25][26][27][28] . We examined whether MSCs are a source of stromal cell IL-8 and IL-6 secretion. We found that U937-CM and K562-CM abundantly induced IL-8 and IL-6 expression in leukemic CM-treated MSCs (Fig. 4A,B). We next incubated U937 and K562 cells in CM that was harvested from leukemic CM-incubated MSCs (illustrated in Supplementary Fig. 2A). CM from leukemic CM-treated MSCs promoted U937 and K562 cell proliferation ( Supplementary Fig. 2B). Additionally, we found that cordycepin suppressed IL-8 and IL-6 expression in leukemic CM-treated MSCs (Fig. 4A,B). In contrast, cordycepin did not affect IL-8 and IL-6 expression in U937 and K562 cells (Fig. 4A,B). We further harvested CM from leukemic CM-treated MSCs and determined IL-8 and IL-6 levels by ELISA (procedure illustrated in Fig. 4C, left panel). Cordycepin significantly abrogated leukemic CM-induced IL-8 and IL-6 ( Fig. 4C, right panel and 4D). As the IL-8/CXCR2 axis has been implicated as a therapeutic target against MDS and AML 28 , we examined CXCR2 expression by Q-PCR analysis in U937 and K562 cells which were co-cultured with MSCs. Neither the co-culture nor cordycepin significantly influenced CXCR2 expression which were analyzed by Q-PCR and flow cytometry (Supplementary Fig. 3A-D).

The role of β-catenin in leukemia hijacks
MSCs. An earlier study has indicated that cordycepin induces β -catenin degradation through activating GSK-3β and that a pharmacological inhibitor of GSK-3β significantly restored cordycepin-reduced β -catenin 35 . As β -catenin is a crucial factor for maintaining the stemness of LSCs, we investigate whether β -catenin is involved in regulating the leukemia-stromal association. U937 and K562 cells were treated with cordycepin or/and with the GSK-3β inhibitor (SB 216763) for 24 h. U937-CM and K562-CM were harvested and were added into MSCs for 48 h followed by Q-PCR analysis (procedure illustrated in  Cordycepin suppresses VCAM-1 and IL-8 expression through attenuation of NFκB activation in leukemic CM-treated MSCs. NFκ B has been implicated in regulating the leukemia-stromal interaction [47][48][49] . We decided to examine whether NFκ B is involved in cordycepin's disruption of the leukemia-MSC association. We found that U937-CM and K562-CM induced expression of NFκ B's subunit p65 in leukemic CM-incubated MSCs whereas cordycepin abolished leukemic CM-induced p65 expression (Q-PCR analysis in Fig. 6A and Western blot analysis in Fig. 6B). We further performed a knockdown of p65 by transfection with p65 siRNA. MSCs were transfected with p65 siRNA for 24 h, followed by the addition of leukemic and control CM (experimental procedure illustrated in Fig. 6C, upper panel). Reduced expression of p65 by siRNA was confirmed by Q-PCR analysis (Fig. 6C, lower panel). Silencing p65 attenuated leukemic CM-induced VCAM-1 (Fig. 6D) and IL-8 (Fig. 6E) expression but had no significant effect on IL-6 (Fig. 6F). These results suggest that leukemic CM induced VCAM-1 and IL-8 expression is through NFκ B activation in MSCs. Cordycepin suppressed NFκ B expression, thereby inhibiting VCAM-1 and IL-8 expression in MSCs.  Supplementary Fig. 6A Cordycepin prolongs survival of U937 and K562-inoculated mice. To investigate whether cordycepin suppresses leukemia progression, we performed in vivo experiments with U937 and K562-inoculated NOD-SCID mice. U937 and K562 cells were pre-treated with or without cordycepin for 24 h and were consequently inoculated into NOD-SCID mice by intravenous injection. Pretreatment of cordycepin prolonged the survival rate in U937 and K562-inoculated mice ( Supplementary Fig. 4A,B).
Cordycepin has a short half-life in vivo due to degradation by adenosine deaminase (ADA) 39 . To examine the in vivo effect of cordycepin on leukemia suppression, we administered U937 cells intravenously to NOD-SCID mice. U937-inoculated mice were intraperitoneally injected with cordycepin with or without pentostatin, an ADA inhibitor (ADAi). Cordycepin combined with ADAi significantly prolonged the survival rate of mice whereas cordycepin or pentostatin alone did not (Fig. 7A).

Discussion
Targeting the association of cancer stem cells with their surrounding niche is a potential therapeutic approach or adjuvant treatment for malignancies, particularly drug-resistant tumors. Bone marrow microenvironment hijacking by LSCs has been implicated in leukemia development and chemoresistance [17][18][19] . Wnt/β -catenin signaling is crucial for LSC development 1-9 and paracrine Dkk1 secreted from MSCs inhibits leukemia and cancer cell proliferation 10,20 . An earlier study has indicated that cordycepin suppresses β -catenin and cell proliferation in U937, K562 and THP-1 cells 35 . In this study, we found U937-CM and K562-CM did not reduce Dkk-1 expression in MSCs (Fig. 1B). However, cordycepin induced Dkk1 expression in MSC-CM, U937-CM and K562-CM treated MSCs (Fig. 1B). These results suggest that Dkk-1 expression can be induced by cordycepin in either normal or leukemia-CM treated MSCs. Cordycepin has autocrine and paracrine anti-leukemia effects via attenuating wnt/β -catenin signaling in leukemia and associated stromal cells.
Cellular N-cadherin and VCAM-1 are involved in the leukemia-stroma interaction and protect leukemia cells from tyrosine kinase inhibitors or chemotherapy 22,23 . Our findings indicate that co-culturing leukemia with MSCs induces N-cadherin expression in U937 cells (Fig. 2C) and the incubation of MSCs with leukemic-CM (U937-CM and K562-CM) enhances VCAM-1 expression in MSCs (Fig. 3A,B). Expressions of N-cadherin in leukemia cells and VCAM-1 in MSCs were significantly impaired by cordycepin treatment (Figs 2C,3A and B). Moreover, cordycepin has been reported to possess suppressive effect on cell proliferation. In this study, we found that cordycepin reduced attached cell number of U937 and K562 on MSCs. This effect may due to suppress N-cadherin, VCAM-1 or other adhesive molecules in leukemia and MSCs. However, it may be possible caused by reduced cell proliferation or viability. In addition, IL-6, IL-8 and several other pro-inflammatory cytokines have been shown to mediate the leukemia microenvironment in myeloid leukemia [24][25][26][27][28] . We found an induction effect of IL-6 and IL-8 by paracrine regulation in U937-CM and K562-CM treated MSCs which was significantly abrogated by cordycepin (Fig. 4A-D). In this study, we show for the first time that cordycepin disrupts leukemia-MSC association through downregulation of N-cadherin in leukemia cells and attenuation of leukemia-CM-induced VCAM-1, IL-6 and IL-8 in MSCs. Taken together, cordycepin is a potential anti-leukemia agent that disrupts the leukemia-MSC association and eliminates leukemia stem cell activity (Fig. 7B). Cordycepin suppresses β -catenin via GSK-3β signaling activation and treatment with a pharmacological inhibitor of GSK-3β restored cordycepin-reduced β -catenin in leukemia 35 . We then examined whether leukemic GSK-3β /β -catenin signaling involves the paracrine regulation for VCAM-1, IL-6 and IL-8 expression in MSCs. CM harvested from U937 and K562 cells pre-treated with cordycepin or/and SB 216763 has no significant or only partial effect on Dkk-1, VCAM-1, IL-6 and IL-8 expression in MSCs (Fig. 5B-D). These results suggest that β -catenin signaling in leukemia is not a major factor in the paracrine regulation of the leukemia niche. Previous studies have implicated leukemia-secreted factors or exosomes in regulating crosstalk between leukemia and bone marrow stromal cells 26,27 . We found that leukemia-CM may educate MSCs thereby inducing VCAM-1, IL-6, IL-8 and other effectors on promoting leukemia progression. Cordycepin has a profound effect on attenuating the expression of N-cadherin in leukemia and VCAM-1, IL-8 and Il-6 in MSCs. However, the detailed mechanism and the involved secreted factors in hijacking the leukemia niche needs further investigation.
Cordycepin is the major ingredient of the caterpillar fungus Dong-Chong-Xia-Cao and is an analogue of adenosine. Cordycepin has a short in vivo half-life in mice due to ADA degradation 38,39 . We injected cordycepin and/or ADAi into U937-inoculated NOD-SCID mice. Injection of cordycepin or ADAi did not extent mice survival whereas a combination of cordycepin and ADAi significantly prolonged the survival rate of U937-inoculated  (N = 4). Scale bars: mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001. Actin was used as loading control. mice (Fig. 7A). These results echo the suggestion found in earlier studies that stability issue of cordycepin should be considered in future clinical applications [38][39][40][41][42][43][44] . Cordycepin combined with ADAi or developing stable cordycepin derivative compounds is therefore a potential adjuvant used for leukemia therapy.

Methods
Cell culture and preparation of CMs. Human MSCs were isolated from the bone marrow of normal donors described previously 52,53 after signing informed consent and all experiments and protocols were performed in accordance with guidelines and regulations approved according to the procedures of the Institutional Review Board of National Taiwan University Hospital (NTUH), Taiwan. MSCs were maintained and cultured in low-glucose Dulbecco's modified Eagle's medium (DMEM, Gibco) supplemented with 10% fetal bovine serum (Gibco) and penicillin-streptomycin (100 U/ml, 100 μ g/ml; Gibco). The U937 and K562 cells were cultured in Roswell Park Memorial Institute (RPMI) medium (Gibco) supplemented with 10% FBS and penicillin-streptomycin. The U937-CM and K562-CM were harvested from medium incubated with 5 × 10 4 cell/ml for 48 h followed by centrifugation at 1500 rpm for 5 min. CMs were stored at − 80 °C.  U937 and K562 cells. U937 and K562 cells were treated with 50 μ M cordycepin for 24 hours followed by incubation on ice with an Fc receptor binding inhibitor (eBiosciences) for 15 min. Cells were incubated for an additional 60 min at 4 °C with the antibodies against CD34, CD38 or CXCR2 in the dark. The cell were washed with PBS three times then fixed with fixation buffer (Biolegend). Flow cytometric acquisition was performed using an FACS Calibur (BD Biosciences) and data were analyzed by CellQuest software (BD Biosciences).

Co
Western blot analysis. Cells were lysed with RIPA (Millipore) buffer containing cocktail protease inhibitors (Roche). The lysates were centrifuged at 15,000 rpm at 4 °C for 10 min. The supernatant was collected and total protein was determined using a Bio-Rad protein assay kit (Bio-Rad Laboratories). Equal amounts of total protein from each sample were loaded into the gradient SDS-PAGE gel, and the proteins were then transferred onto PVDF membranes. The membranes were blocked with 5% non-fat milk and probed with specific antibodies against N-cadherin (BD), NFκ B p65 (Santa Cruz), V-CAM1 (Santa Cruz), β -catenin (BD) and β -Actin (sigma). The membranes were immersed in horseradish peroxidase-conjugated secondary antibody and immunoreactive bands were exposed on an X-ray film.

Real-Time Quantitative PCR (Q-PCR). Total RNA was isolated by RNAzol ® RT (Molecular Research
Center) and cDNA was synthesized using an ABI RT Kit (ABI Applied Biosystems). RNA expression levels were analyzed using quantitative real-time PCR (Q-PCR) with specific primers (Supplementary Table 1). PCR reactions were performed using SYBR Green (Kapa Biosystems) in an ABI Prism 7900 System (ABI Applied Biosystems). The quantity of the specific genes was normalized with GAPDH of the same sample. Fold change was determined as the ratio of the each control sample. All PCR reactions were performed in triplicate.
ELISA. Levels of IL-6 and IL-8 in medium were measured using ELISA kits purchased from eBioscience (San Diego, CA) according to the manufacturer's instructions. The capture antibody was coated on the ELISA plates and incubated overnight at 4 °C. After washing with PBS containing 0.1% Tween-20 (PBST), the plates were treated with a blocking buffer at room temperature for 1 h. Diluted samples or standards of IL-6 or IL-8 were added and incubated overnight at 4 °C. The detection antibody was incubated for 1 h and Avidin-HRP was incubated for 30 min at room temperature. Absorbance was measured at 450 nm.
Transfection of siRNA. MSCs were seeded in 6 well plates the day before transfection. Transfection was performed using Lipofectamine ® RNAiMAX Reagent (Invitrogen). 15 pmole of scramble or siRNA for human NFκ B p65 (Sequences: GCCCUAUCCCUUUACGUC and GACGUAAAGGGAUAGGGC, Santa Cruz) was added into the transfection mixture. Transfection efficiency was determined by Q-PCR.
U937-inoculated SCID mice model. The SCID mice as a model for U937 inoculation was previously established 54 . The protocol of this study for U937-inoculated SCID mice was performed in accordance with the guidelines and regulations approved by the Institutional Animal Care and Use Committee of the National Health Research Institutes. NOD-SCID mice (8 weeks old) were purchased from the National Laboratory Animal Center and housed in microisolator cages at a specific pathogen-free facility at the Laboratory Animal Center, National Health Research Institutes of Taiwan. NOD-SCID mice received of 300 cGy X-ray irradiation to eliminate their hematopoietic systems. After 24 h, 5 × 10 6 U937 cells were intravenously injected into the tail vein. After one week, the mice were divided into four groups: control (N = 8), cordycepin (N = 6), ADAi (N = 5) and cordycepin combined with ADAi (N = 8). U937-inoculated mice were intraperitoneally injected with 5 mg/kg cordycepin and/or 1 mg/kg pentostatin (ADAi) (Santa Cruz) in the lower-right area of the abdominal cavity. The treatment was given once every two days for three times.
Statistical analysis. Data are presented as mean ± SEM. Student's t-test was used to analyze differences between 2 experimental groups. Mantel-COX test was used to analyze the survival of the mice. A P-value of less than 0.05 was considered statistically significant. *, ** and *** Indicate P-value < 0.05, 0.01, and 0.001, respectively.