CaMKII-δ is selectively overexpressed in leukemia stem/progenitor cells and associated with poor prognosis in AML
CaMKⅡ is a multifunctional serine/threonine protein kinase family and there are four CaMKⅡ isoforms: α, β, γ, and δ in human cells (15), but it is unknown which CaMKⅡ isoforms are expressed in AML stem/progenitor cells. To address this issue, we initially analyzed the expression levels of CaMKⅡ isoforms α, β, γ and δ in AML data from The Cancer Genome Atlas (TCGA) dataset with 5,602 samples across 31 cancer types (http://gepia.cancer-pku.cn/detail.php). We found that CaMKII-δ but not isoforms α, β, and γ was significantly overexpressed in AML (Fig. 1a-d). Consistent with these results, CaMKII-δ abundance but not isoforms α, β, and γ was associated with overall survival of AML patients (Fig. 1e-h). CaMKII-δhigh AML patients had a significantly shorter overall survival than that of CaMKII-δlow patients (Fig. 1h, p = 0.029).
To examine whether aberrant activation of CaMKII-δ was limited in AML, we analyzed 30 other types of tumors for CaMKII-δ mRNA levels in TCGA dataset. We found that pancreatic adenocarcinoma (PAAD), thymoma (THYM), glioblastoma multiforme (GBM) and lymphoid neoplasm diffuse large B-cell lymphoma (DLBC) also exhibited significant increased CaMKII-δ mRNA levels (Fig.S1a-d), but no significant survival differences were observed between CaMKII-δhigh patients and CaMKII-δlow patients among these cancers (Fig.S1e-h). These results indicate that aberrant activation of CaMKII-δ occurs predominantly in AML but not in most other types of tumors. To confirm whether CaMKII-δ is aberrantly activated in AML LSCs, we next detected activated (pCaMKII-δ(T287)) and total CaMKII-δ (CaMKII-δ) protein levels in LSCs of various AML cell lines using Western blot. We found that p-CaMKII-δ levels of KG-1, KG-1a and Kasumi-1, which contain CD34+ leukemia stem/progenitor cells, were significantly higher than that of HL-60, NB4, MV4-11 and Molm-13 cell lines without CD34+ cells (Fig. 2a, b). To validate whether CaMKII-δ is aberrantly activated in primary AML LSCs and normal hematopoietic stem cells (HSCs), we examined pCaMKII-δ and CaMKII-δ protein levels in CD34+ cells (AML LSCs) from primary AML samples and HSCs from normal donors by Western blot. We observed that both CaMKII-δ and pCaMKII-δ levels were highly expressed in the CD34+ AML LSCs, but low in CD34+ HSCs (Fig. 2c). In addition, aberrant activation of CaMKII-δ was frequently observed in primary AML samples (Fig. 2d), but not in peripheral blood mononuclear cells (PBMCs) from healthy individuals (Fig. 2e). These findings indicate that CaMKII-δ is selectively activated in AML LSCs as well as bulk leukemia cells but not in normal HSCs.
To gain insight into the clinical importance of our findings, we then analyzed CaMKII-δ levels in relation with white blood cell (WBC) levels in various AML cases. We observed that CaMKII-δ levels correlated with hyperproliferative phenotype of AML in a retrospective cohort of 39 AML patients at diagnosis (Fig. 2f). AML patients with high level of CaMKII-δ (CaMKII-δhigh) expression had higher white blood cell numbers than CaMKII-δlow patients (median WBC: 46.9×109/L versus 16.5×109/L) (Fig. 2g, h, R = 0.4418), indicating that CaMKII-δ levels positively correlated with the malignant proliferative potential of leukemia cells in AML patients.
CaMKII-δ is essential for activation of STAT3, CDK6, and BCL-2
Given that STAT3, CDK6 and BCL2 are aberrantly activated in AML and essential for viability, cell cycle and anti-apoptosis of AML LSCs (11–13), to address whether CaMKII-δ regulated these factors, we used CRISPR/Cas9 mediated CaMKII-δ-KO(Knock out) to abrogate CaMKII-δ expression and then determined levels of these factors using Western blot. The results showed that expression levels of STAT3, pSTAT3(Tyr705), CDK6, and BCL-2 were rapidly lost upon CRISPR-mediated CaMKII-δ-KO in Molm-13 cells (Fig. 3a). Similar results were also observed in CD34+ Kasumi-1 cells (Fig. 3b). To further assess the role of CaMKII-δ enzyme activity in regulating co-activation of LSC-related factors and cell proliferation, we transfected Molm-13 or Kasumi-1 cells with an expression vector harboring a K43M mutant CaMKII-δ, a conserved residue near the ATP-binding site of this enzyme. This mutation results in a “kinase-dead” enzyme, and expression of this mutant inhibits endogenous CaMKII-δ activity in a dominant-negative manner. We found marked decreases of pSTAT3(Tyr705), CDK6 and BCL-2 in AML stem/progenitor cells (Fig. 3c, d). To further validate these observations, we transfected Molm-13 and Kasumi-1 with an expression vector harboring T287D mutant CaMKII-δ, which resulted in a constitutively activation of this kinase. As expected, this mutant CaMKII-δ co-activated pSTAT3(Tyr705), CDK6, and BCL-2 in MOLM-13 and Kasumi-1 cell lines (Fig. 3e, f). In addition, to further confirm the effects of CaMKII-δ in regulating STAT3, CDK6 and BCL-2, we treated AML cell lines with CaMK inhibitor KN93 and KN92 (the inactive structural analog of KN93), and examined the proliferation. The results showed protein level of pSTAT3(Tyr705), CD6 and BCL-2 was decreased by KN93 in a time-dependent manner (Fig.S2f-h). And KN93 showed significantly inhibition of all three AML cell lines in a dose-dependent manner (Fig.S2j). Taking together, these results indicate that CaMKII-δ is a critical regulator for activating STAT3, CDK6 and BCL-2, suggesting that CaMKII-δ may be an essential protein kinase in maintaining survival and anti-apoptosis of AML stem/progenitor cells as well as bulk cancer cells of AML.
CaMKII-δ is required for growth and anti-apoptosis of AML stem/progenitor and bulk cells
To determine whether CaMKII-δ is essential for AML cell growth, we performed genetic silence of CaMKII-δ in a panel of human AML cell lines using doxycycline (Dox)-inducible CRISPR/Cas9 mediated CaMKII-δ knock out (CaMKII-δ-KO) system, and examined the effects of CaMKII-δ silence on proliferation and viability of AML cells. We observed that CaMKII-δ silence (Fig. 4a, e) dramatically suppressed cell proliferation (Fig. 4b, f). In parallel, we evaluated the effect of CaMKII-δ silence on apoptosis induction in these cells. The Molm-13(39.4% versus 9.1%, p < 0.001) and Kasumi-1 (59.5% versus 13.4%, p < 0.001) cells showed potent apoptosis, respectively, as shown by staining for Annexin V, after CaMKII-δ-KO (Fig. 4c, d, g, h). Consistently, colony forming assay results showed that CaMKII-δ-KO significantly decreased colony size (Fig. 4i, j), numbers and sphere diameter of AML cells (Fig. 4k, l). In addition, after CaMKII-δ silencing induced by Dox in 3 and 5 days (Fig. 2a), we also showed light microscopic images (Fig.S2b) which consistent with the results of the MTT assays. Western blotting analysis showed that apoptosis-related molecules cleaved Caspase-3 and cleaved-PARP were clearly increased in Molm-13 cells after silencing induced by Dox (Fig.S2c), and qPCR results showed the increased expression of apoptosis associated factors (Fig.S2d, e). In contrary, cells with CaMKII-δ over expression accelerating leukemic cell proliferation (Fig.S3a-d). To further confirm whether CaMKII-δ enzyme activity is essential in regulating cell proliferation, we transfected Molm-13 or Kasumi-1 cells with an expression vector harboring a K43M mutant CaMKII-δ, we observed significant growth inhibition of AML cells (Fig.S3e, f). Conversely, we transfected Molm-13 or Kasumi-1 cells with an expression vector harboring T287D mutant CaMKII-δ, and observed a significant growth promotion of AML cells (Fig.S3g, h). In addition, we also examined cell-cycle progression and observed that silence of CaMKII-δ resulted in a decrease of S phase cells (37.53% versus 48.41%, P = 0.0027) and an accumulation of AML cells in G0/G1 phase (52.92% versus 43.40%, P = 0.0024) as compared with controls (Fig.S3i, j). These results indicate that CaMKII-δ enzyme activity is essential for survival and proliferation of AML stem/progenitor cells as well as cell cycle G1 phase progression and G1/S transition.
To evaluate the effects of CaMKII-δ on survival and growth of leukemia cells as well as STAT3, CDK6 and BCL-2 in vivo, we established a subcutaneous xenograft mouse model in nude mice using AML Molm-13 cells with Dox-inducible CaMKII-δ-KO vector or control vector. After xenograft tumors reached 100mm3, the mice received Dox to initiate CaMKII-δ-KO via oral gavage for 16 days and then euthanized (Fig. 5a). Tumor tissues were processed for CaMKII-δ/ pCaMKII-δ(T287), CDK6, STAT3/pSTAT3(Tyr705), and BCL2 protein analysis using Western blotting. Consistent with the in vitro results, CaMKII-δ-KO induced a significant growth inhibition of leukemia cells in nude mice as compared with controls. A 4.2-fold decrease of tumor weight was observed with CaMKII-δ-KO compared with control at 16 days after CaMKII-δ-KO initiation (Fig. 5b-d). Consistently, concomitant decreases of CDK6, STAT3/pSTAT3(Tyr705) and BCL2 were also observed in CaMKII-δ-KO-mediated xenograft (Fig. 5e). Next, we established human refractory AML orthotopic model in NSG mice using AML Molm-13-luciferin cells with Dox-inducible CaMKII-δ-KO or control vector. Consistent with the subcutaneous xenograft mouse model and in vitro results, a significant reduction in tumor signal was observed with CaMKII-δ silencing compared with control on the 14th day after CaMKII-δ-KO initiation (Fig. 5f). These observations indicate that CaMKII-δ-KO inhibits growth of AML cells in vivo via targeting CaMKII-δ/STAT3/CDK6/BCL2 signaling pathways that regulates AML stem/progenitor cells.
Transcriptomic analysis predicts CaMKII-δ as a key regulator of DHCR24 associated with cholesterol metabolic pathway
To better understand the role and mechanisms of CaMKII-δ in AML, we used CRISPR/Cas9 mediated CaMKII-δ-KO and RNA-Seq approach to identify its potential target genes in pathogenesis of human AML cells (Fig. 6a). We found 1324 genes were downregulated and 1237 genes were upregulated in volcano plot through gene expression analysis (Fig. 6b). We also observed that a range of genes involved in cell cycle, and apoptosis were markedly down-regulated after CRISPR-mediated CaMKII-δ-KO in Molm-13 cells (Fig.S4a, b). We next focused on cancer-associated gene ontology (GO) analysis to define consensus pathways that may be affected by silencing of CaMKII-δ. We found that CaMKII-δ-downregulated genes were predominantly associated with metabolism and cholesterol biosynthetic process (Fig.S4c-f), whereas CaMKII-δ-upregulated were mainly associated with protein processing in endoplasmic reticulum and metabolism (Fig.S4d, e). Consistently, we also found that metabolic pathway and LSC pathway show the remarkable significance among the top 10 canonical pathways (Fig. 6c, d). Several LSC representative genes such as CD123, IL1-RAP, CLL-1, FLT-3 and KIT were confirmed by qRT-PCR (Fig. 6f). These results are consistent with the metabolic functions of STAT3, CDK6 and BCL-2 in cancer cells (21–23). Surprisingly, we found that two cholesterol metabolism-related enzymes: 3β-Hydroxysteroidδ24 reductase (DHCR24) and acetyl-CoA acetyltransferase 2(ACAT2) were located at the top range among CaMKII-δ downregulated genes in metabolic pathways (Fig. 6e). Consistent with these results, both qRT-PCR and Western blot results confirmed that CaMKII-δ-KO reduced DHCR24 levels as well as other LSC-related genes (Fig. 6g).
It is intriguing that DHCR24 is a cholesterol-synthesizing enzyme and has the ability of anti-apoptotic induced by oxidative stress or endoplasmic reticulum stress (ERS) as reactive oxygen species (ROS) scavenger(24–26), whereas ACAT2 is involved in cholesterol esterification (27). To determine whether DHCR24 is involved in AML cells and the correlation with CaMKII-δ, we firstly measured the CaMKII-δ and DHCR24 protein levels in 12 primary AML samples by WB, the results showed that DHCR24 highly expressed in AML patients, and patients with high level of CaMKII-δ expression usually associated with higher DHCR24 express. (Fig. 6h, I, r2 = 0.57, p = 0.0028), indicating that CaMKII-δ levels positively correlated with the expression of DHCR24 in AML cells. Next, DHCR24 was further analyzed in AML cell lines and in both primary AML LSC and normal HSC samples using Western blot and found that DHCR24 protein levels were significantly higher in AML cell lines and LSCs instead of other cell lines and normal HSCs (Fig.S5a, b). And we also found DHCR24 mRNA expression was higher in AML cell lines especially in the CD34 + cell lines KG-1 and KG1a (Fig.S6a). In addition, we analyzed DHCR24 levels in different subtypes of AML patients and found that DHCR24 levels in M3 and M5 were significantly higher compared with normal blood cells (Fig.S5c). And AML with t (8;21), t (15;17), inv (16), MLL gene, normal and complex karyotype demonstrated highly DHCR24 expression than healthy bone marrow cells (Fig.S5d). Intriguingly, we found DHCR24 expression was higher in FLT3-ITD positive AML than FLT3-wild type AML (Fig.S5e, f). And the TCGA data shows that DHCR24 highly expressed AML patients show poor prognosis (Fig.S5g). As DHCR24 expression could be induced by insulin through STAT3, which directly binds to the promoter elements of DHCR24 (28). Consistently, the majority of functionally defined LSCs are characterized by relatively low levels of reactive oxygen species (termed ‘‘ROS-low’’), whereas ROS-low LSCs aberrantly overexpress BCL-2(13). These findings suggest that there may be a CaMKII-δ/DHCR24 axis that plays a critical role in metabolic function of AML stem/progenitor cells via regulating STAT3 and BCL-2.
Inhibiting of DHCR24 effectively suppresses the growth of AML stem/progenitor cells
To confirm the role of DHCR24 in AML cells, we constructed genetic silence of DHCR24 in AML cell lines Molm-13 and THP-1 using doxycycline (Dox)-inducible CRISPR/Cas9 mediated DHCR24 knock out (DHCR24-KO) system, and examined the effects of DHCR24-KO on proliferation and viability of AML cells. We observed that DHCR24-KO (Fig. 7a, e) dramatically suppressed cell proliferation (Fig. 7b, f). In parallel, we evaluated the effect of DHCR24 silence on apoptosis induction in these cells. The Molm-13(19.5% versus 7.6%, p < 0.001) and THP-1 (18.4% versus 3.2%, p < 0.01) cells showed potent apoptosis, respectively, as shown by staining for Annexin V, after DHCR24-KO (Fig. 7c, d, g, h). And the western blotting analysis showed that apoptosis-related molecules cleaved Caspase-3 and cleaved-PARP were clearly increased in MOLM-13 cells after DHCR24-KO (Fig.S6b). Consistently, colony forming assay showed that DHCR24-KO significantly decreased colony numbers of Molm-13 and Kasumi-1 cells (Fig.S6c-f). To further evaluate in vivo the effect of DHCR24-KO on survival and growth of leukemia cells. We established human refractory AML orthotopic model in NSG mice using AML Molm-13-luciferin cells with Dox-inducible DHCR24-KO or control vector. Consistent with the results in vitro, a significant reduction in tumor signal was observed with DHCR24-KO compared with control on the 21th day after DHCR24-KO initiation (Fig. 7i, j). Meanwhile, the mouse weight between the control and DHCR24-KO group showed no significance (Fig. 7k). However, the survival time of mice injected by DHCR24-KO leukemic cells was longer than the control group (Fig. 7l, p = 0.0025).
Novel CaMKII-δ/DHCR24 axis inhibitor hesperadin suppresses the growth of AML cells in vitro and in vivo
Zhang et al. recently reports that CaMKII-δ plays critical role in cardiac ischemia/reperfusion (I/R) injury, and they identified hesperidin, an Aurora B kinase inhibitor, which can bound to CaMKII-δ and exert dual functions of cardioprotective and antitumor effects(29). However, the effects of hesperadin in hematopoietic malignancies had been never reported. Here, we firstly demonstrate hesperadin can inhibits leukemia cell proliferation in vitro and in vivo via targeting CaMKII-δ/DHCR24 axis. We treated human AML cell lines Molm-1, Mv4-11, Kasumi-1, KG-1, KG-1a, murine C1498 cell Line, and normal spleen and bone marrow cells from C57BL/6 with hesperadin at various concentrations for 72 h and then collected for cell viability analysis using MTT assay. The IC50 values of hesperadin for above-mentioned cell lines were 3.820nM, 4.262nM, 1.798nM, 26.74nM, 8.122nM, 48.17nM, 287.4nM and 155.0nM, respectively (Fig. 8a). To further evaluate in vivo the effect of hesperadin on survival and growth of leukemia cells. we established a subcutaneous xenograft mouse model in nude mice with AML Molm-13 cells. which grouped by 10mg/kg, 20mg/kg with hesperadin and the vehicle. After xenograft tumors reached 50-100mm3, the mice were randomly divided into 3 groups to receive various doses of hesperadin (10mg kg− 1 and 20mg kg− 1) via peritoneal injection for 12 days and then euthanized. Tumor tissues were processed for CaMKII-δ, pCaMKII-δ(T287), DHCR24, and BCL-2 protein analysis by western blotting. Consistent with the in vitro results, hesperadin induced a significant growth inhibition of leukemia cells in nude mice as compared with the vehicle (Fig. 8b). Moreover, dose-dependent decreases of tumor weight and tumor volume were observed in hesperidin treated mice (Fig. 8c, d). Meanwhile, concomitant decreases of pCaMKII-δ(T287), DHCR24 and BCL2 were also observed in hesperadin-mediated groups (Fig. 8e). Altogether, these results indicate that hesperadin suppresses the growth of AML cells through targeting CaMKII-δ/DHCR24 axis.