Targeting Wnt pathway in mantle cell lymphoma-initiating cells

Background Mantle cell lymphoma (MCL) is an aggressive and incurable form of non-Hodgkin’s lymphoma. Despite initial intense chemotherapy, up to 50 % of cases of MCL relapse often in a chemoresistant form. We hypothesized that the recently identified MCL-initiating cells (MCL-ICs) are the main reason for relapse and chemoresistance of MCL. Cancer stem cell-related pathways such as Wnt could be responsible for their maintenance and survival. Methods We isolated MCL-ICs from primary MCL cells on the basis of a defined marker expression pattern (CD34-CD3-CD45+CD19-) and investigated Wnt pathway expression. We also tested the potential of Wnt pathway inhibitors in elimination of MCL-ICs. Results We showed that MCL-ICs are resistant to genotoxic agents vincristine, doxorubicin, and the newly approved Burton tyrosine kinase (BTK) inhibitor ibrutinib. We confirmed the differential up-regulation of Wnt pathway in MCL-ICs. Indeed, MCL-ICs were particularly sensitive to Wnt pathway inhibitors. Targeting β-catenin-TCF4 interaction with CCT036477, iCRT14, or PKF118-310 preferentially eliminated the MCL-ICs. Conclusions Our results suggest that Wnt signaling is critical for the maintenance and survival of MCL-ICs, and effective MCL therapy should aim to eliminate MCL-ICs through Wnt signaling inhibitors. Electronic supplementary material The online version of this article (doi:10.1186/s13045-015-0161-1) contains supplementary material, which is available to authorized users.


Background
Mantle cell lymphoma (MCL) is considered as an incurable subtype of non-Hodgkin's lymphoma that causes significant morbidity and early death presumably due to relapsed disease [1][2][3]. Despite apparent clinical remissions achieved with chemotherapy regimens (R-CHOP or R-hyperCVAD), MCL relapse rates hover around 50 % [4,5]. The relapse is considered to be due to chemoresistant cells that prevent complete elimination of MCL cells.
A small fraction of cells within tumors have tumorinitiating properties and are believed to be the source of relapsed cancer. These cells are referred to as cancer stem cells (CSCs) or tumor-initiating cells [6][7][8][9]. CSCs have been implicated in the growth, progression, and relapse of several tumor subtypes. The most current therapies target dividing tumor cells while sparing non-dividing and inherently chemoresistant CSCs; thus, they fail to provide long-term cures and result in tumor relapse [10,11].
CSCs and normal hematopoietic stem cells share Wnt, Notch, and Hedgehog signaling pathways, which are required for their growth and self-renewal [7]. Recent studies have suggested a role of Wnt signaling in MCL tumorigenesis [12][13][14]. The Wnt signaling pathway regulates development, and its dysregulation leads to oncogenesis [15][16][17]. Canonical Wnt signaling is initiated by the binding of Wnt ligands to their cognate Frizzled (FZD) receptors and its coreceptors, low density lipoprotein receptor related proteins 5/6 (LRP5/6). In the absence of Wnt signaling, β-catenin is phosphorylated and its interaction with GSK-3β and axin-1 leads to its ubiquitination and degradation [18]. Activation of the Wnt pathway prevents β-catenin phosphorylationinduced degradation, and stabilized β-catenin accumulates in the nucleus, where it forms active transcription complexes with the T cell factor/lymphoid enhancer binding factor (TCF/LEF) family of DNA-binding transcription factors [19][20][21]. Dysregulation of Wnt pathway can promote tumorigenesis [22,23]. Selective targeting of stem cell signaling pathways should eliminate CSCs [24].
MCL-initiating cells (MCL-ICs) have been recently identified based on a lack of CD19 marker (CD34-CD3-CD45 +CD19-cells) [25]. Two studies from different groups have shown that these MCL-ICs can repopulate tumor in mice [25,26]. As few as 100 of CD19-MCL-ICs have been found to produce whole tumor with both CD19+ and CD19-cells, while CD19+ MCL-non-ICs were incapable of tumor development at comparable limited dilutions in severe combined immunodeficiency (SCID) mice [25,26]. We suggest that the high relapse rates of human MCL arise from incomplete elimination of chemoresistant MCL-ICs [27]. Thus, in order to improve long-term survival of individuals with MCL, it is important to have a fuller understanding of the signaling pathways responsible for the chemoresistance and maintenance of MCL-ICs. In this study, we investigated the expression and importance of Wnt pathway in survival of MCL-ICs and explored ways to eliminate these cells.

Wnt pathway genes are overexpressed in MCL-ICs
Analysis from previous studies using unfractionated MCL cells have implicated the Wnt pathway in the pathogenesis of mantle cell lymphoma [12][13][14]. Therefore, we first investigated Wnt3 expression in unfractionated MCL. Our observations suggest that 9 out of 20, nearly 45 % MCL samples, overexpress Wnt3. We next investigated the expression of Wnt3 in MCL-ICs isolated from MCL samples expressing high and low Wnt3 levels. Our results showed that MCL-ICs were enriched in Wnt3 compared to MCLnon-ICs and B-cells, irrespective of total tumor Wnt3 expression (Fig. 3a). We observed differential up-regulation of Wnt ligands and their FZD receptors in MCL-ICs compared with MCL-non-ICs (Fig. 3b, Table 1), using B-cells as a reference. To show other evidence of enhanced Wnt signaling, we performed immunostaining for β-catenin. Higher cellular and nuclear levels of β-catenin were observed in MCL-ICs than in MCL-non-ICs (Fig. 3c, Additional file 1: Figure S1) whereas B-cells did not show detectable β-catenin levels (Additional file 1: Figure S1). Activation of Wnt signaling in MCL-ICs was confirmed by the elevated expression of the Wnt target genes encoding ID2 and TCF4 (both >100-fold) compared with MCLnon-ICs (Fig. 3d). Thus, by 3 independent methods, we show that the Wnt pathway is differentially up-regulated in MCL-ICs.

Discussion
The high rate of MCL relapse after initial apparent clinical remissions achieved with conventional chemotherapy suggests incomplete elimination of MCL cells and implicates a-c qRT-PCR performed using the total cellular RNA isolated from MCL-ICs (n = 4) for a stem cell transcription factors (Nanog, Oct4, Sox2, Klf4), b ALDH isoforms and antioxidant enzymes SOD2 and MT1b, and c chemoresistance-associated genes encoding ABCC3, ABCC6, and CD44. Differences between MCL-ICs and MCL-non-ICs were significant (P < 0.05) for ALDH1, ALDH2, SOD2, MT1b, Nanog, Oct4, Klf4, ABCC3, ABCC6, and CD44. d Cell cycle analysis of isolated MCL-ICs, MCL-non-ICs, and total MCL cells by flow cytometry. e ALDH activity in freshly isolated MCL-ICs from apheresis samples evaluated using ALDEFLUOR kit  Figure S1. d Relative expression levels of Wnt target genes encoding ID2 and β-catenin-interacting transcriptional factor TCF4 in MCL-ICs (n = 4) and MCL-non-ICs (n = 4). Differences between MCL-ICs and MCL-non-ICs were significant (P < 0.05) for both genes a role for chemoresistant MCL-ICs in relapse. Here we showed that MCL-ICs have functional properties of cancer stem cells: high expression of ALDH, antioxidant enzymes, chemoresistance-associated genes, and stem cellassociated transcription factors, while still retaining t (11;14) (q13; q32) and overexpression of cyclin D1. Our analysis showed that MCL-ICs overexpress a subset of Wnt ligands and FZD receptors and that Wnt signaling is activated in MCL-ICs. Treatment of primary MCL cells with Wnt inhibitors preferentially eliminated MCL-ICs, which was not achieved with the current chemotherapy agents vincristine, doxorubicin, or even with the recently FDA-approved agent ibrutinib [28]. Burton tyrosine kinase (BTK) has been shown to be a negative regulator of Wnt signaling [29]. Therefore, it is not surprising that ibrutinib (a BTK inhibitor) probably resulted in inducing Wnt signaling rather than inhibiting it and thereby could not eliminate MCL-ICs. Our results suggest that the inability of conventional chemotherapy to kill MCL-ICs can be overcome by adding inhibitors of Wnt signaling.
A recent study showed that cobble stone area-forming cells (CAFCs) that developed from MCL co-cultured with human mesenchymal stem cells (hMSCs) are the morphologic equivalent of MCL-ICs with the CD34-CD3-CD45 +CD19-CD133+ marker phenotype and manifested their tumor-initiating capacity in NOD/SCID mice [26]. CAFCs were also resistant to bortezomib, fludarabine, and doxorubicin and expressed stem cell transcription factors Nanog and Oct4 but not Sox2 [26]. The CD34-CD3-CD45+CD19-MCL-ICs characterized in our study are identical to the CAFCs; they were also CD133+ and exhibited the same characteristics. However, our study has further extended the characterization of MCL-IC, by identifying a hyperactive Wnt signaling pathway, crucial for their maintenance and survival.
Our results showed up-regulated expression of canonical ligand Wnt3 [30] but not of the non-canonical ligands such as Wnt4, Wnt5, and Wnt11 [31] in MCL-ICs compared to MCL-non-ICs. Immunostaining results also confirmed the differentially higher staining of active unphosphorylated βcatenin in MCL-ICs, which is required for canonical but not for non-canonical Wnt pathway [32]. In addition, the FZD6 receptor, which is associated with inhibition of canonical Wnt signaling pathway [33], was not differentially expressed in MCL-ICs. These results clearly indicate the presence of activated canonical Wnt signaling pathway in MCL-ICs.
Other investigators have revealed that Wnt is an important pathway in primary MCL tissues and have implicated this pathway in the pathogenesis of MCL [13]. Analysis of unfractionated MCL shows only a threefold up-regulation compared to B-cells in the previous study [13]. Our study showed a 23-fold enhanced Wnt3 expression in MCL-ICs compared to B-cells. These results clearly show the importance of Wnt signaling in MCL-ICs.
MCL is believed to be driven by enhanced cyclin D1 expression due to t (11;14) (q13; q32) present in >90 % of MCL [34,35]. A minority of MCLs do not express cyclin D1 [36]. However, other isoforms of cyclin D are overexpressed in cyclin D1-negative MCLs, which suggests an indispensable requirement for the expression of at least one isoform of cyclin D in MCL [37]. Thus, it appears that mechanisms other than t (11;14) (q13; q32) are responsible for the overexpression of at least one other cyclin D in cyclin D1-negative MCL. It is of interest that Wnt signaling could potentially fulfill this role [38][39][40] as it was shown to upregulate cyclin D2 expression [41,42]. Wnt3 was also noted to be overexpressed in cyclin D1-negative MCL that also lacked t (11;14) (q13; q32) [36]. Therefore, enhanced Wnt signaling may be the underlying driver of overexpression of cyclin D in all MCL regardless of t (11;14) status.
Overexpression of Wnt ligands and their cognate FZD receptors in MCL-ICs points to the existence of an autocrine signaling loop. Immunostaining of β-catenin and the elevated expression of Wnt target genes encoding ID2 and TCF4 clearly confirmed higher Wnt activity in MCL-ICs than in MCL-non-ICs. However, inhibitors of Wnt signaling acting upstream of β-catenin degradation ; destruction complex involving APC, Axin-1, and GSK-3β interacts with and phosphorylates β-catenin leading to its degradation. Binding of Wnt ligands to their cognate Frizzled (FZD) receptors and its co-receptors, low density lipoprotein receptor related proteins 5/6 (LRP5/6), activates the Wnt pathway (right) leading to sequestration and degradation of Axin-1 and phosphorylation and degradation of GSK-3β. Degradation of destruction complex components leads to accumulation of β-catenin and its subsequent translocation into the nucleus where it interacts with TCF4 to promote transcription of Wnt target genes. IWR1 stabilizes Axin-1 and promotes formation of destruction complex and degradation of β-catenin (red arrows). IWP2 inhibits Porcupine-mediated acylation and subsequent secretion of Wnt ligands. XAV939 inhibits Tankyrase-mediated degradation of Axin-1 and thus promotes formation of destruction complex (red dashed arrow). CCT036477, iCRT14, and PKF118-310 target β-catenin-TCF4 transcription complex step had no effect on MCL-ICs. On the other hand, downstream Wnt inhibitors targeting the β-catenin-TCF4 transcription complex (CCT036477, iCRT14, and PKF118-310) preferentially eliminated MCL-ICs over MCL-non-ICs. These findings can be explained either by the very low expression of proteins involved in β-catenin degradation such as GSK-3β and axin-1 (target of XAV939, IWR1-endo, IWP2 compounds) observed in MCL-ICs or by a contribution of redundant and additional pathways to β-catenin activation, such as autocrine fibroblast growth factor receptor signaling [43,44]. Further experiments will be required to delineate the major contributors to β-catenin activation in MCL-ICs.
We found elevated levels of FZD1 and its ligand Wnt3 in MCL-ICs as compared to MCL-non-ICs. Other researchers have demonstrated that targeting FZD1 reverses multidrug resistance in neuroblastomas and breast cancer cells [45,46]. Thus, Wnt3-FZD1 signaling may be one of the reasons for chemoresistance in MCL [47,48].
In summary, we have outlined an enhanced Wnt/β-catenin signaling in MCL-ICs and shown that inhibition of the Wnt pathway effectively eliminates MCL-ICs, which are implicated in MCL relapse. We have also demonstrated that present therapy approaches for MCL, including recently approved drug ibrutinib, do not address the killing of MCL-ICs. Thus, we anticipate that current rates of MCL relapse will not decrease substantially with current therapies. A detailed examination of selectively enhanced signaling in MCL-ICs may be a good starting point to expose pathways important for MCL tumor stem cell survival. Presenting clinical and MCL features at time of initial MCL lymphoma presentation do not identify a priori, who are the patients who will relapse from those will attain cures. Perhaps, studying the MCL-IC and their response to targeting agents will be a key to identify patients who will relapse. Our results clearly show that Wnt signaling inhibitors targeting β-catenin-TCF4 interaction can eliminate MCL-ICs. However, blocking the Wnt pathway exclusively in tumor cells will be challenging, as Wnt signaling also has a role in the self-renewal of non-malignant tissues such as intestinal crypts, and bone growth plates [49,50]. Nevertheless, our results point to the important and actionable targets of Wnt signaling in MCL pathogenesis and its potential usefulness as a target for therapy to eliminate MCL-ICs and reduce the risk of MCL relapse.

Conclusions
Our results clearly demonstrate the differential activation of Wnt pathway in MCL-ICs. Not all steps in the Wnt signaling pathway are amenable to effective blocking in MCL. We show that blocking of Wnt signaling at the βcatenin-TCF4 transcription complex effectively blocks signaling in MCL-IC and preferentially kills the MCL-IC cells, which harbor chemoresistance. This study results shows identification of effective agents in MCL-IC that would not had been possible by studying whole MCL cells. Since inhibition of Wnt pathway resulted in preferential elimination of MCL-ICs, we conclude that Wnt pathway should be targeted to eliminate MCL-ICs and reduce the risk of relapsed MCL.

Patients and agents
Cells and clinical information from MCL patients described in this manuscript (Additional file 4: Table S2)

Isolation of normal B-cells
Peripheral blood B-cells were isolated from healthy donors' blood obtained from the Gulf Coast Blood Center (Houston, TX) by using CD19-positive magnetic beads and were released with the competitive CD19 DETACHaBEAD according to the manufacturer's instructions (Invitrogen-Life Technologies, San Diego, CA). All procedures were performed under The University of Texas MD Anderson Cancer Center IRB-approved clinical protocol LAB08-0190. ALDH activity assay ALDH activity in cells was determined by using an ALDE-FLUOR kit according to the manufacturer's protocol (STEMCELL Technologies, Vancouver, Canada). Briefly, 1 × 10 6 cells were resuspended in a 1 ml assay buffer with 5 μl of ALDEFLOUR reagent. DEAB was used as inhibitor of ALDH activity. A 500-μl aliquot of the reagent mixed cells was transferred to an Eppendorf tube containing 5 μl of DEAB as a control. Samples were incubated at 37°C for 45 min. Green fluorescence intensity was measured with a BD Fortessa flow cytometer (Becton Dickinson, San Jose, CA) and evaluated with FlowJo software (Tree Star, Ashland, OR).

Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was extracted from cells using RNAqueous kit according to the manufacturer's protocol (Ambion-Life Technologies, Austin, TX). First-strand cDNA was synthesized using a Superscript III reverse transcriptase kit according to the manufacturer's protocol (Invitrogen-Life Technologies, San Diego, CA). Samples were analyzed on 96-well microtiter plates using the StepOnePlus real-time PCR System (Applied Biosystems, Grand Island, NY). qRT-PCR was performed using SYBR green dye and primers specific for selected human genes (Additional file 5: Table  S1) as described earlier [51,52]. PCR was performed with 40 cycles of 95°C for 15 s and 60°C for 1 min.
Step-One software version 2.1 was used to analyze the qRT-PCR data.

Growth and treatment of MCL cells
Primary MCL cells were seeded onto a layer of human bone marrow mesenchymal stromal cells (hMSCs) at a MCL to a stromal cell ratio of 10:1 and grown in hMSC medium supplemented with mesenchymal cell growth factors and glutamine (Lonza, Allendale, NJ) at 37°C in 5 % CO 2 as described previously [53]. MCL cells were harvested, resuspended in 50 % fresh and 50 % conditioned medium from hMSC cultures, and incubated with the indicated agents for 6-48 h. Cells were either stained with propidium iodide for cell cycle analysis as described earlier [54] or had RNA isolated for further analysis. The percentage of MCL-ICs was determined using procedures described in the "Isolation of MCL cells and MCL-ICs" section above.

Statistical analysis
Experimental data are reported as means or medians with standard deviation or error of mean, unless otherwise indicated. Differences between groups were calculated using the two-tailed Student's t test (GraphPad Prism, GraphPad Software, Inc, La Jolla, CA). P < 0.05 was considered statistically significant.

Additional files
Additional file 1: Figure S1. Wnt signaling pathway is active in MCL-ICs. Detection β-catenin expression and localization by immunofluorescence and confocal microscopy in (a) B-cells from healthy donors and in (b) freshly isolated MCL-ICs, and MCL-non-ICs.
Additional file 3: Figure S3. Hedgehog and Notch signaling pathways in MCL-ICs. Expression of mRNA encoding (a) Hedgehog signaling pathway transcription factors Gli1, Gli2, Gli3 and (b) Notch signaling target genes Hes1, Hes5 and Hey1 in freshly isolated MCL-ICs, and MCL-non-ICs relative to B-cells from healthy donors. Horizontal lines represent median for each group. Differences between MCL-ICs and MCL-non-ICs were significant (P < 0.05) for Gli3, and Hey1. (c) Percentage of MCL-ICs evaluated by immunostaining and flow cytometry (as shown in Fig. 1a) of primary MCL cells (n = 3) treated with Hedgehog inhibitors LDE225 (5 μM), Cyclopamine (5 μM), GANT61 (5 μM) and Notch inhibitor RO4929097 (5 μM) for 48 h. *Differences between treated and control group were significant P < 0.05. Overexpression of Gli3, a repressor of hedgehog pathway [55,56] in MCL-ICs and inability of hedgehog and notch signaling pathway inhibitors to decrease percentage of MCL-ICs, suggest that these pathways may not be effective targets for reducing the percentage of MCL-ICs.