Elena K. Siapati, PhD, Laboratory of Cell and Gene Therapy, Centre of Basic Research II, Biomedical Research Foundation of the Academy of Athens, 4, Soranou Efesiou Street, Athens 11527, Greece. E-mail: firstname.lastname@example.org
B-catenin is the central effector molecule of the canonical Wnt signalling pathway, which controls self-renewal of haematopoietic stem cells. Deregulation of this pathway occurs in various malignancies including myeloid leukaemias. The present study examined the functional outcome of stable β-catenin down-regulation through lentivirus-mediated expression of short hairpin RNA (shRNA). Reduction of the β-catenin levels in acute myeloid leukaemia (AML) cell lines and patient samples decelerated their in vitro proliferation ability without affecting cell viability. Transplantation of leukaemic cells with control or reduced levels of β-catenin in non-obese diabetic severe combined immunodeficient animals indicated that, while the immediate homing of the cells was unaffected, the bone marrow engraftment was directly dependent on β-catenin levels. Subsequent examination of bone sections revealed that β-catenin was implicated in the localization of AML to the endosteum. Examination of adhesion molecule expression before and after transplantation, revealed down-regulation of CD44 expression, accompanied by reduced in vitro adhesion. Gene expression analysis disclosed the presence of an autocrine compensatory mechanism, which responds to the reduced β-catenin levels by altering the expression of positive and negative pathway regulators. In conclusion, our study showed that β-catenin comprises an integral part of AML cell proliferation, cell cycle progression, and adhesion, and influences disease establishment in vivo.
B-catenin is the central effector molecule of the canonical Wnt signalling pathway, which governs cell fate and differentiation during embryogenesis. In the absence of signalling, β-catenin is phosphorylated/ubiquitinated through a complex of proteins namely APC (Adenomatous polyposis coli), Axin, Casein kinase I (CKI) and Glycogen synthase kinase-3 (GSK3-β). When the Wnt ligands bind to Frizzled and LRP5/6 receptors, the pathway is activated and a signalling cascade is initiated through Dishevelled that leads to inhibition of proteosomal degradation of β-catenin. Non-phosphorylated β-catenin accumulates and translocates to the nucleus where, together with T-cell factor (TCF) and lymphoid enhancer factor (LEF), it initiates transcription of numerous genes including MYC and CCND1 (Reya & Clevers, 2005).
In the haematopoietic system the role of Wnt signalling pathway is rather controversial. The pathway is active in haematopoietic stem cells (HSCs) and β-catenin -through enforced expression- was initially documented to maintain HSC numbers as well as their repopulating activity (Reya et al, 2003). More recent evidence, however, revealed that constitutive expression of β-catenin impairs multi-lineage differentiation and causes exhaustion of the HSC pool (Kirstetter et al, 2006; Scheller et al, 2006). These contradictory reports seem to indicate that a tight regulation of Wnt signalling is required for maintenance of stem cell phenotype, with excessive activation or inhibition of the pathway leading to HSC exhaustion (Suda & Arai, 2008). At the same time, conditional deletion of β-catenin and γ-catenin does not impair lymphopoiesis or haematopoiesis (Cobas et al, 2004; Koch et al, 2008), suggesting that the pathway may have a different impact in vivo where compensation by external factors or other signalling pathways may occur. An alternative explanation could be that the requirement for β-catenin in haematopoietic development may vary during embryonic life.
Deregulation of Wnt signalling has been implicated in the pathogenesis of many tumours including colon and breast cancer as well as leukaemia. In chronic myeloid leukaemia (CML), β-catenin has proved vital for the formation ofbcr-abl-induced leukaemias in vivo, supporting the notion that this protein is required for CML progression in vivo (Zhao et al, 2007).
In AML, β-catenin expression is associated with enhanced blast clonogenicity and it predicts poor patient survival, constituting an independent adverse prognostic factor (Chung et al, 2002);(Ysebaert et al, 2006). AML translocation products, such as RUNX1-RUNX1T1 (AML1-ETO) and PML-RARA directly activate the Wnt pathway byup-regulating the levels of β and γ-catenin and thereby increasing TCF/LEF-dependent transcription (Muller-Tidow et al, 2004). This is also observed in patients carrying FLT3-internal tandem repeat (ITD) mutations (Tickenbrock et al, 2005). However, aberrant Wnt signalling has been observed in patients lacking the above genetic abnormalities, indicating that these mutations are not the only cause of deregulated Wnt signalling in AML. Another possible cause for the observed aberrant Wnt signalling in AML may be the epigenetic inactivation of negative regulators of the canonical Wnt pathway such as Wnt inhibitory factor (WIF1), Dickkopf (DKK) and secreted Frizzled-related proteins (SFRPs) (Chim et al, 2006);,(Jost et al, 2008). Alternatively, the constitutive expression of Wnt-1 and Wnt-2b ligands in some AML cases (Simon et al, 2005) indicate that the pathway may be regulated through an autocrine feedback loop, a mechanism that has also been observed in normal HSCs. Nonetheless, the exact mechanism by which Wnt signalling is implicated in the pathogenesis of AML remains to be clarified. Wnt ligands are also expressed by bone marrow (BM) stroma cells indicating the importance of the BM niche as a paracrine source of Wnt signalling, which maintains the self-renewal of HSCs (Fleming et al, 2008; Luis et al, 2009). The precise Wnt signalling interaction among AML blasts and BM stroma remains to be elucidated.
Evidence for direct involvement of β-catenin in the pathogenesis of leukaemias comes from studies in cell lines which over-express proteins that interfere with β-catenin nuclear signalling (Chung et al, 2002). However, the complex nature of transcription factor and signalling pathway crosstalk may complicate the interpretation of such experiments. We, therefore, performed specific down-regulation of β-catenin protein by RNAi technology in order to delineate the true effects of this protein on cell physiology, leukemogenesis and also elucidate the influence of β-catenin signalling in AML. We provide evidence that β-catenin is involved in AML cell proliferation, cell cycle progression and adhesion, partly through CD44. We show that establishment and progression of disease in vivo is dependent on functional Wnt signalling.
Materials and methods
Peripheral blood samples were obtained from AML patients with newly diagnosed or relapsed disease at G. Gennimatas General Hospital (Athens, Greece) after informed consent. The protocol was approved by the hospital ethics committee. Mononuclear cell (MNC) fractions were isolated by Ficoll-Paque (Histopaque®; Sigma-Aldrich, Athens, Greece) centrifugation. The leukaemic cell lines Fujioka, THP1, HL60, Kasumi-1 and U937 were obtained from Cancer Research UK Cell Bank (kind gift of Dr Bonnet). They were maintained in RPMI-Glutamax medium supplemented with 10% fetal calf serum (FCS) and 1% penicillin/streptomycin (all from Invitrogen, San Diego, CA, USA).
Lentivirus production and cell transduction
The β-catenin shRNA plKO.1 vectors were obtained from Sigma-Aldrich (Mission shRNA) and a puromycin N-acetyltransferase (PAC) gene was replaced for that of eGFP. As a control (NS shRNA) for shRNA production, a non-specific sequence was used as previously reported (Rangatia & Bonnet, 2006). Lentiviral vector production was performed by transient transfection of 293T cells using calcium phosphate as previously described (Siapati et al, 2005). Viral supernatant was concentrated 200-fold by ultracentrifugation. MNCs from AML patient samples were thawed on the day of transduction and in some cases the CD34+ fraction isolated using the CD34 isolation kit (Miltenyi Biotech, Cologne, Germany). Cells (MNCs or CD34+) were cultured in StemSpan serum-free medium (Stem Cell Technologies, Vancouver, BC, Canada) in the presence of 10% FCS, recombinant stem cell factor, Flt3-Ligand and thrombopoietin at 50 ng/ml, and interleukin 3 at 10 ng/ml (all cytokines were purchased from Peprotech, London, UK). Concentrated viral stock was added at a multiplicity of infection (MOI) of 50 and transduction was allowed to proceed for 18–20 h.
Untransduced or cell lines transduced with BCATshRNA vector containing a puromycin N-acetyltransferase (PAC) gene were electroporated with TCF-eGFP reporter plasmids (TOP-eGFP or FOP-eGFP) (310 V, No resistance, 975 μF). Cells were assayed by flow cytometry for eGFP expression, 24–48 h later.
At various time points after transduction cells were assayed for proliferation using the CellTiter 96® Aqueous One Solution (Promega, Manheim, Germany) according to the manufacturer’s recommendations. Cells were plated at 104 per well in a 96-well plate, the solution added after 24–48 h and the optical density at 490 nm (OD490 nm) was read 2–4 h later.
Cell adhesion assay
The method was adapted from Avigdor et al (2004). Briefly, non-TC treated plates were coated with 150 μg/ml Hyaluronic Acid (Sigma-Aldrich) overnight and then blocked with 1% bovine serum albumin (BSA) for 1 h at 37°C. Cells were added at 105 per well in the presence of 125 ng/ml stromal cell-derived factor 1α (Peprotech) and allowed to adhere for 2 h. The non-adhered cells were removed by gentle washing and CellTiter 96® Aqueous One Solution (Promega) was added to estimate the live adherent cells. The OD490 nm was read 2–4 h later.
Non-obese diabetic severe combined immunodeficient (NOD/SCID) animals were a kind gift from Dr Bonnet (Cancer Research UK, London) and were maintained in sterile conditions at BRFAA animal facility. Prior to cell inoculation, animals 8–12 weeks of age were conditioned with two intraperitoneal 20 mg/kg doses of Busilvex (Pierre Fabre, Athens, Greece) on days -2 and -1. AML cell lines transduced with β-catenin shRNA or NSshRNA vectors were intravenously injected and sacrificed 3–4 weeks later by cervical dislocation to check for engraftment in BM and spleen. All animal experiments were performed according to institutional guidelines and were approved by the National Veterinary Board.
Fluorescence-activated cell sorting (FACS) analysis was performed on Cytomics FC500 (Beckman-Coulter, Fullerton, CA, USA) and cell sorting on Aria™ (BD Biosciences, Athens, Greece). Analysis of cell apoptosis was performed using Annexin-V-PE (BD Biosciences) and 7-AAD according to the manufacturer’s protocol. Following AML transplantation the haematopoietic organs were analysed using CD45-PECy5 and CD44-PE (BD Biosciences). Expression of adhesion molecules prior and after transplantation was done using CD49d (VLA-4), CD49e (VLA-5), CD29, VCAM-1, CXCR4, LFA-1, ICAM-3 and N-cadherin antibodies (all from BD Biosciences). Dead cell discrimination was done using 7-AAD (Sigma-Aldrich).
Recipient femurs were isolated from transplanted animals, fixed in 4% paraformaldehyde for 16–20 h, rinsed in dH2O and decalcified using Decalcifying Solution Lite (Sigma-Aldrich) for 3 h. They were then mounted in paraffin at longitudinal orientation and 5 μm sections were cut on poly-lysine-coated slides. Immunohistochemical staining was performed using standard deparaffinization and rehydration techniques, blocking with H2O2 and 2% BSA and overnight incubation with anti-eGFP antibody (1:200; Millipore, Billerica, MA, USA). Secondary anti-mouse biotinylated antibody (Dako, Glostrup, Denmark) was used followed by ΑBC Reagent (Vector Labs, Burlingame, CA, USA) and staining with DAB chromogen. Images were obtained on a Leica DFC500 microscope using Leica Application Suite acquisition software (Leica Microsystems GmbH, Wetzlar, Germany).
LiCl treatment and western blotting
AML cell lines (1–2 × 106 cells) transduced with NSshRNA or BCATshRNA were incubated in 10 mmol/l LiCl for 16 h, washed in phosphate-buffered saline and lysed in Protein Lysis Buffer (20 mmol/l Tris–HCl, pH 7·6, 150 mmol/l NaCl, 5 mmol/l EDTA, 1% Nonidet P-40, 1 mmol/l phenylmethylsulfonyl fluoride, protease inhibitor cocktail). The entire protein extract was subjected to 10% sodium dodecyl sulphate polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane overnight at 4°C (Biorad, Herts, UK). The >50 kDa part of the membrane was incubated with anti-non-phosporylated β-catenin antibody (1:300 clone 8E4; Santa Cruz Biotechnology, Heidelberg, Germany or 1:550 clone 8E7; Millipore) followed by anti-mouse horseradish peroxidase (HRP) antibody (Millipore). The <50 kDa part of the membrane was used to verify equal protein loading with anti-human β-actin HRP (1:2500; Santa Cruz).
Real-time PCR was performed using 2xSYBR Green PCR Master Mix (Fermentas, St. Leon-Rot, Germany or Agilent Technologies, Santa Clara, CA, USA). All samples were analysed in duplicate 25 μl reactions on an ABI PrismR7000 Sequence Detector (Applied Biosystems, Foster City, CA, US). Primers were as follows:
The Student’s t-test or the non-parametric Mann–Whitney test was used to compare the effect of β-catenin down-regulation on cell physiology, gene expression and engraftment in vivo. The Kruskal Wallis analysis of variance (anova) test was used to estimate significance in gene expression changes following β-catenin down-regulation.
Gene expression using the SuperArray™ platform
Total RNA was obtained from sorted BCATshRNA- or NSshRNA-transduced Fujioka cells using TRI Reagent® (Molecular Research Center Inc, Cincinnati, OH, USA). Further RNA purification and reverse transcription were performed according to SA Biosciences’ recommendations using the RT2 qPCR-Grade RNA Isolation kit and RT2 First Strand kit, respectively. Gene expression analysis in three biological replicates was done using the Wnt pathway Super Array platform (SA Biosciences, Frederick, MD, USA).
B-catenin is implicated in cell cycle progression and proliferation in AML cell lines
To study the role of β-catenin in AML we have obtained pLKO.1 lentiviral shRNA constructs generated by the RNAi consortium and replaced the puromycin gene with that of eGFP reporter to be able to identify and sort the transduced cells by FACS. The efficiency of β-catenin down-regulation by the different shRNA vectors was investigated by testing viral stocks on leukaemic cell lines, namely U937 (myelomonocytic), THP1 (monocytic), HL60 (promyelocytic), Fujioka (monoblastic) and Kasumi-1 (RUNX1-RUNX1T1 positive). TRC3845 shRNA proved the most efficient of five different shRNA oligonucleotides tested in down-regulating CΤΝΝΒ1 mRNA by >70% (data not shown) and was selected for further studies (will be referred to as BCATshRNA from here onwards). Cell lines transduced with β-catenin shRNA or a control vector carrying a non-specific shRNA sequence (NSshRNA) (Rangatia & Bonnet, 2006) were sorted for eGFP to obtain a pure population of transduced cells and were analysed by Western blotting for non-phosphorylated β-catenin, which represents the ‘active’ form of the protein responsible for gene transcription. As shown in Fig 1A, active β-catenin could be detected in cells transduced with the vector carrying a NSshRNA but was significantly reduced or even absent from cells transduced with BCATshRNA. When cells were cultured in the presence of the GSK3β inhibitor, LiCl, NSshRNA-transduced cells up-regulated the levels of non-phosphorylated β-catenin while BCATshRNA-transduced cells did not. This clearly indicated that the canonical Wnt pathway could not be activated in cells that expressed β-catenin shRNA. In order to test the in vitro functional outcome of β-catenin silencing, we electroporated untransduced or cell lines transduced with BCATshRNA-puromycin vector with TCF-eGFP reporter plasmids (TOP-eGFP or FOP-eGFP) which contain active or mutant TCF binding sites respectively, upstream of an eGFP reporter gene (Reya et al, 2003). Reduced reporter activity (average 60%) was observed in BCATshRNA-transduced cells compared to controls (Fig 1B), as estimated by the percentage of eGFP+ cells (P < 0·005). This indicated that decreasing the pool of active β-catenin protein had a negative effect on transcription by TCF/β-catenin complexes. In addition, the expression of known β-catenin transcriptional targets, such as MYC and CCND1, was also reduced by an average of 60% and 38% respectively, in cells expressing BCATshRNA compared to control cells (P < 0·02; Fig 1C).
To test the role of β-catenin in cell physiology, we investigated the cell proliferation, viability and cell cycle following β-catenin down-regulation. We observed that BCATshRNA-transduced AML cell lines showed decelerated in vitro cell proliferation by 20–30% compared to cells with the non-specific sequence (Fig 1D). To investigate if the reduced cell proliferation was due to cell death, the levels of apoptosis were measured in cells with reduced β-catenin levels at various time points after transduction by flow cytometry. The percentage of 7AAD− Annexin-V+ cells was only marginally increased compared to control cells, indicating that the reduced cell proliferation was not due to cell death (Fig 1E). We subsequently analysed the cell cycle status of NSshRNA- and BCATshRNA-transduced AML cell lines and observed a slight increase in the proportion of cells in G1 (from 44·7% to 51%) in accordance with the reduction of CCND1 levels. Reducing the β-catenin levels in AML cell lines also conferred a statistically significant drop in the proportion of cells in S phase (41% vs. 52% in control cells; P < 0·02) and an increase in G2 phase (8% compared to 3·6% in control cells) (P < 0·04) (Fig 1F). Delineation of cell cycle-related genes revealed a reduction in the levels of CCNE1 which peaks at G1/S transition, suggesting a smaller proportion of BCATshRNA-transduced cells lay in this phase (Fig 1G). This corroborated with the cell cycle data. No significant changes in expression of the cell cycle- dependent kinase CDK4 were observed while expression of CDK2 and CCNA2 was variable. Taken together, our data suggest that β-catenin is implicated in the cell cycle progression and proliferation of AML cells.
B-catenin is required for efficient in vivo engraftment of AML cell lines
In order to examine the effect of down-regulated β-catenin in the homing and engraftment of AML cells we used the NOD/SCID xenotransplantation mouse model. The Fujioka and HL60 cell lines engraft in these animals primarily in the spleen and BM and manifest with an AML-like phenotype, splenomegaly and animal weight loss. Initially we investigated the homing potential of NSshRNA- and BCATshRNA-transduced cells, 6–8 h following intravenous administration into NOD/SCID animals that had been conditioned with busulfan. FACS analysis of BM and spleen cells revealed similar percentages of eGFP+ cells, suggesting that cells transduced with either NSshRNA or BCATshRNA had comparable homing capacity (Fig 2A) and β-catenin did not influence the homing potential of AML cells. In order to assess the role of β-catenin in AML in vivo progression, mice that had been transplanted with Fujioka or HL-60 cells were analysed 3–4 weeks later for engraftment in the spleen and BM. Mice that received control NSshRNA-transduced cells presented with splenomegaly and twofold higher average levels of eGFP+ cells (27%) in the spleen compared to cells with down-regulated β-catenin, which engrafted in the spleen at an average 12% (Fig 2B). In the BM, where the Wnt signalling is of prime importance for maintenance of normal HSCs, a more pronounced 35-fold difference in engraftment was seen between NSshRNA- and BCATshRNA-transduced cells (P < 0·03), with the latter only comprising an average 0·4% of the total BM population at 3–4 weeks post-injection (Fig 2C). This finding clearly demonstrated the importance of functional canonical Wnt signalling in engraftment and in vivo progression of AML. We also examined the ability of AML cells with reduced β-catenin levels to competitively engraft in vivo when co-transplanted with untransduced AML cells. Following BM analysis we found that BCATshRNA-transduced cells engrafted at twofold lower levels (36%) compared to NSshRNA-transduced cells (68%) (Fig 2D). Collectively, our data indicated thatβ-catenin is essential for leukaemia establishment and progression in vivo.
Adhesion molecule expression following β-catenin silencing in AML
To further delineate the underlying mechanism of the observed impaired BM engraftment of BCATshRNA-transduced cells, we analysed the expression of adhesion molecules involved in HSC heterotypic interaction with stroma. In vitro cultured AML cell lines transduced with NSshRNA or BCATshRNA were stained for VLA-4, VLA-5, N-cadherin, ICAM-3, CD44, LFA-1 and CXCR4. We could detect a significant (close to 25%) reduction in CD44 mean fluorescence intensity (MFI) in HL-60 and THP-1 cells with reduced β-catenin levels compared to control cells (Fig 3A) but no differences in any other of the adhesion molecules tested (data not shown). To test the functional importance of this observation we analysed the in vitro adhesion to the CD44 substrate Hyaluronic Acid; we observed that the lower CD44 levels conferred reduced adhesion proving the functional significance of β-catenin down-regulation (Fig 3B). We subsequently analysed the expression of adhesion molecules of AML cells (Fujioka or HL60) in the BM of NOD/SCID animals 3–4 weeks following engraftment. Interestingly, there was a 43% reduction in the CD44 MFI in BCATshRNA-transduced cells in the BM compared to cells that had been transduced with NSshRNA vector (P < 0·05)(Fig 3C). No differences in expression of CD49d, CD49e, CD29 or the chemokine receptor CXCR4 were observed between engrafted NSshRNA and BCATshRNA-transduced cells (data not shown). Our findings showed that the Wnt signalling pathway partly regulates CD44 expression and AML adhesion.
We then proceeded to analyse the localization of engrafted AML cell lines in bone sections of transplanted animals. Immunohistochemical staining for GFP revealed that control cells had adhered and proliferated close to the endosteum of the long bones but they were equally detected throughout the BM cavity (Fig 3D,E). In contrast, AML cells with low β-catenin levels were mainly localized in the central part of the bone sparing the endosteum (Fig 3H,J). Examination of the trabecular bone indicated that control cells were capable of abutting the bone next to osteoblasts (Fig 3F,G), whereas cells with reduced β-catenin levels had failed to localize in the endosteal clefts of the trabecular bone (Fig 3J,K). This finding indicated that β-catenin silencing, and hence, abrogation of functional Wnt signalling, affected the localization of AML cells in the BM and, most probably, subsequent proliferation and engraftment.
Down-regulation of β-catenin in AML patient samples reduces their in vitro proliferation
B-catenin comprises an independent prognostic factor for poor survival in AML patients. In order to study the precise role of β-catenin in primary AML samples, we obtained peripheral blood MNCs (PBMNCs) from AML patients at diagnosis and quantified their CTNNB1 mRNA levels by Quantitative Taqman polymerase chain reaction (PCR). When compared to normal donors, patient PBMNCs expressed approximately 100-fold higher CTNNB1 mRNA levels (Fig 4A). To investigate whether our findings in the AML cell lines were reproduced in primary AML blasts, we transduced MNCs with BCATshRNA and NSshRNA lentiviral vectors and measured CTNNB1 mRNA levels by Taqman PCR (Table I)(Fig 4B). We observed a 50% reduction in CTNNB1 mRNA levels (P < 0·05) and a 30% down-regulation of MYC expression, a β-catenin transcriptional target, indicating a functional reduction in the β-catenin levels (Fig 4C). CCND1 expression was more variable among AML patient samples following β-catenin silencing with an average 13% drop in expression compared to NSshRNA-transduced cells.
Table I. AML patient characteristics.
FAB, French-American-British type.
To further investigate the effect of the Wnt signalling on the cell cycle status of AML primary samples, we measured the mRNA levels of various genes involved in the cell cycle following β-catenin silencing. No overall differences were seen in CDK4, CDK2 and CDKN1B expression while CCNE1 and CCNA2 expression was either up- or down-regulated in BCATshRNA-transduced cells compared to control (Fig 4D). The observed CCNE1 levels indicate that cells with lower β-catenin levels accumulated in early G1 or in the G1/S transition phase. The altered expression of CCNA2 may be explained by a higher proportion of cells in late G2/M, an observation also made in AML cell lines. In addition, we observed a 20% reduction in cell proliferation (P < 0·05) by day 7–10 in AML patient samples transduced with BCATshRNA vector compared to NSshRNA-transduced control cells (Fig 4E), a finding that was not attributed to apoptosis (Fig 4F). Down-regulation of β-catenin also conferred a 30% reduction in the CD44 mRNA levels compared to control AML cells (P < 0·02) (Fig 4G) recapitulating the phenomena observed in AML cell lines.
The canonical Wnt pathway is tightly regulated in AML
In order to elucidate how down-regulation of β-catenin may affect Wnt signalling in AML, we performed SuperArray analysis of the human Wnt signalling pathway in sorted Fujioka cells as described in the Materials and methods. Overall, there was tight regulation of the canonical Wnt signalling pathway in AML with minor changes of the pathway components following reduction of β-catenin levels by twofold. BCATshRNA-transduced cells showed an increase in the levels of the following positive regulators of β-catenin signalling: Ccd1 (DXDC1: +1·87), FRAT1 (+1·49) and casein kinase I alpha (CSNK1A1: +1·43) (Table II). In addition, the levels of the canonical Wnt lignands WNT3 and WNT10A were alsoup-regulated by 1·63- and 1·96-fold respectively, alongside the Frizzled 2 receptor (FZD2: +1·91). Collectively, these data suggested that the cell responds to the reduction of β-catenin levels by up-regulating proteins that are involved in the canonical Wnt signalling upstream of β-catenin. In agreement with the above observation, the mRNA levels of the Wnt pathway negative regulator IDAX, which interacts with Dvl preventing the downstream signal transmission, was slightly down-regulated (CXXC4: −1·42), while others remained unaffected (NKD1: −1·03; PPP2CA: +1·03). Components of the β-catenin degradation complex were also analysed: APC expression decreased by 1·4-fold, while other factors (DVL2: +1·3; GS3KB: −1·06; AXIN1: +1·02) and receptors (LRP5: +1·2; FZD1: +1·01) remained unaffected. Of the secreted inhibitors only SFRP4 was detected in these cells and its expression didn’t change upon β-catenin reduction (+1·15). In summary, gene expression analysis of the Wnt pathway following down-regulation of β-catenin in AML indicated that there is a compensatory mechanism in place which mobilizes specific components to ‘correct’ the fluctuations in β-catenin levels.
Table II. Gene expression analysis in Fujioka cells following β-catenin down-regulation.
*Accession numbers were taken from NCBI RefSeq database.
The Wnt signalling pathway plays a major role in the maintenance of stem cell phenotype and is highly active during organogenesis and body patterning. Aberrant Wnt signalling is also observed in many types of cancer, including myeloid leukaemias. Although β-catenin has come to comprise an independent prognostic factor for AML survival, its precise role in AML cell physiology and disease establishment in vivo remains undetermined. Our study describes the delineation of the role of β-catenin in AML through specific down-regulation of the protein with an shRNA-expressing lentiviral vector.
Quantitation of CTNNB1 levels in PBMNC from AML patients showed nearly 100-fold higher mRNA levels compared to normal controls. Transduction of AML cell lines and AML patient samples with a BCATshRNA-expressing vector resulted in functional down-regulation of CTNNB1 levels and concomitant reduction in expression of the known β-catenin transcriptional targets MYC and CCND1. In addition, using TCF/LEF reporter constructs we showed that, in AML cell lines, TCF-mediated reporter transcription was reduced following CTNNB1 silencing. In agreement with other reports we found that β-catenin is involved in AML cell proliferation but, contrary to what is known for normal human CD34+ progenitors (Simon et al, 2005), down-regulation of β-catenin did not affect AML cell viability. We also demonstrated that reduction in the levels of β-catenin caused an increase the proportion of cells in G1 and G2 phases of the cell cycle while lowering the percentage of cells in S phase, similarly to what has recently been reported for multiple myeloma (Dutta-Simmons et al, 2009).
Our xenotransplantation experiments with AML cell lines demonstrated the importance of β-catenin in disease establishment and maintenance in vivo. Our data showed that reducing the levels of β-catenin in HL-60 and Fujioka cell lines caused a decrease in AML engraftment. This was more prominent in the BM where the canonical Wnt signalling has an established role in the maintenance of normal haematopoiesis (Kirstetter et al, 2006; Scheller et al, 2006). A similar requirement for β-catenin has been reported for CML where disease onset by bcr-abl-transduced β-catenin null HSCs was significantly delayed compared to cells expressing functional β-catenin (Zhao et al, 2007). The engraftment impairment of cells with lower β-catenin was not due to reduced homing ability neither due to altered expression of CXCR4. In competitive repopulating assays, AML cells transduced with BCATshRNA exhibited lower engraftment ability compared to control cells. This indicated a competitive disadvantage possibly in the docking and subsequent proliferating ability of the cells with lower β-catenin. When we examined the femurs of transplanted animals, cells with lower β-catenin levels were preferentially localized in the long bone cavity, sparing the endosteal clefts of the trabecular bone, in sharp contrast to what we observed with control cells. Osteoblast-rich areas of the endosteum constitute the ‘niche’ of leukaemic stem cells (Ishikawa et al, 2007) and the canonical Wnt pathway has proven essential for normal HSC by BM stroma (Kim et al, 2009; Nemeth et al, 2009; Nygren et al, 2009). Our data indicate that the canonical Wnt pathway is also involved in the maintenance of AML cells in vivo.
To further delineate the mechanism of Wnt-mediated AML establishment in vivo, we examined the expression of various adhesion molecules that are involved in HSC interaction with the BM niche. We observed down-regulation of CD44 in AML patient samples and cell lines transduced with BCATshRNA, in vitro as well as in AML cells engrafted in vivo. CD44 is a known TCF target (Wielenga et al, 1999) and we show that its expression in AML (Bendall et al, 2000) is partly regulated by aberrant Wnt signalling. Moreover, CD44 is a known leukaemic cell marker and its blocking with antibodies abrogates leukaemia establishment in vivo (Jin et al, 2006). Collectively, our data suggest that the canonical Wnt pathway, directly or indirectly through CD44, is implicated in AML adhesion and is required for disease establishment in vivo.
Various reports indicate that an autocrine or paracrine feedback loop may be operating in the Wnt pathway in both normal and malignant haematopoietic cells (Simon et al, 2005). In order to investigate whether such a mechanism is operating in AML we analysed the effect of β-catenin down-regulation on expression of Wnt pathway components. Overall, we observed minor changes indicating tight regulation of the Wnt pathway in AML. The cells, nonetheless, ‘sensed’ the reduced β-catenin levels and responded to revert to their original state by boosting the expression of positive regulators (Ccd1, FRAT1, CKI) and decreasing the levels of negative regulators (CXXC4). The cellular response to the reduced β-catenin levels included an increase in the expression of the canonical ligands Wnt3 and Wnt10A alongside the Frizzled 2 receptor. These findings indicate the presence of an autocrine compensatory mechanism in AML cells, which acts to correct any fluctuations in the components of the Wnt pathway in order to maintain steady-state signalling. Our data suggest that such an autocrine loop operates in cells with abnormal steady-state signalling, indicating that both normal and malignant cells respond to fold changes of β-catenin regardless of its absolute levels(Goentoro & Kirschner, 2009).
Similarly to what we know about normal haematopoietic cell maintenance, our data strongly argue for the requirement of a functional Wnt signalling pathway to establish and maintain leukaemia in vivo. We demonstrate that the canonical Wnt signalling through β-catenin is implicated in AML cell proliferation, cell cycle progression and adhesion. Our findings further contribute to our understanding on how a deregulated canonical Wnt pathway is implicated in AML pathogenesis and suggest that molecular targeting of this pathway should be considered for therapeutic intervention.
Conflict of interest
The authors declare that there are no competing financial interests in relation to the work described in this paper.