Takayuki Ikezoe contributed to the concept and design, interpreted and analyzed the data and wrote an article. Chie Nishioka performed all experiments and wrote an article. Mutsuo Furihata, Jing Yang, Satoshi Serada, and Tetsuji Naka, Sayo Kataoka and Atsuya Nobumoto performed the experiments. Akihito Yokoyama and Keiko Udaka provided important intellectual content
To identify molecular targets in leukemia stem cells (LSCs), this study compared the protein expression profile of freshly isolated CD34+/CD38− cells with that of CD34+/CD38+ counterparts from individuals with acute myelogenous leukemia (n = 2, AML) using isobaric tags for relative and absolute quantitation (iTRAQ). A total of 98 proteins were overexpressed, while six proteins were underexpressed in CD34+/CD38− AML cells compared with their CD34+/CD38+ counterparts. Proteins overexpressed in CD34+/CD38− AML cells included a number of proteins involved in DNA repair, cell cycle arrest, gland differentiation, antiapoptosis, adhesion, and drug resistance. Aberrant expression of CD82, a family of adhesion molecules, in CD34+/CD38− AML cells was noted in additional clinical samples (n = 12) by flow cytometry. Importantly, down-regulation of CD82 in CD34+/CD38− AML cells by a short hairpin RNA (shRNA) inhibited adhesion to fibronectin via up-regulation of matrix metalloproteinases 9 (MMP9) and colony forming ability of these cells as assessed by transwell assay, real-time RT-PCR, and colony forming assay, respectively. Moreover, we found that down-regulation of CD82 in CD34+/CD38− AML cells by an shRNA significantly impaired engraftment of these cells in severely immunocompromised mice. Taken together, aberrant expression of CD82 might play a role in adhesion of LSCs to bone marrow microenvironment and survival of LSCs. CD82 could be an attractive molecular target to eradicate LSCs.
Acute myelogenous leukemia (AML) is characterized by a cellular hierarchy, and is initiated and maintained by a subset of self-renewing leukemia stem cells (LSCs).1 To produce cure in individuals with AML, development of a novel treatment strategy targeting LSCs is urgently required. LSCs share some antigenic features with normal hematopoietic stem cells (HSCs). For example, both LSCs and HSCs express CD34 but not CD38. However, LSCs can be phenotypically distinguished from HSCs by several disparate markers, including CD117− and CD123+.1–3 LSCs exist in a quiescent state and are capable of self-renewal and differentiation, and are able to perpetuate leukemic cell growth in long-term culture assays and in the murine nonobese diabetic/severe combined immunodeficiency (NOD/SCID) model system.1–4 CD34+/CD38− AML cells were shown to fulfill the criteria for LSCs in vivo.5, 6 Although, recent studies employed more severely immunocompromised mice found that even CD34− or CD38+ AML cells in some cases were able to reconstitute AML.7, 8
The regulation of stem cell self-renewal and differentiation requires a specific microenvironment of surrounding cells known as the stem cell niche. The concept of the stem cell niche was first proposed for the human hematopoietic system in the 1970s.9 The HSC niche in mouse bone marrow (BM) is composed of an endosteal lining of stromal cells, extracellular matrix proteins, and osteoblasts.10–12 Specific adherens junction molecules such as N-cadherin mediate adhesion between HSCs and niche cells in the adult hematopoietic system.11
Recent work has shown that interaction between CXCR4 on leukemic cells and its ligand stromal cell-derived factor-1 (SDF-1) in the niche is necessary for proper homing and in vivo growth of leukemic cells.13 Moreover, interaction between LSCs and the niche mediated by adhesion molecule CD44 is required for maintenance of LSCs behavior.14 CD44 mediates adhesive cell-cell and cell-extracellular matrix interactions by binding its main ligand, hyaluronan, a glycosaminoglycan that is highly concentrated in the endosteal region.14, 15 All together, adhesion molecules play an important role in maintaining the characteristics of LSCs.
CD82/KAI-1, a member of the tetraspanin superfamily, was originally identified as an accessory molecule in T-cell activation.16 The most well-characterized function of CD82 in nonimmune cells is integrin-mediated cell adhesion to extracellular matrix.17 Forced expression of CD82 up-regulated tissue inhibitors of metalloproteinase 1 (TIMP1) and inactivated matrix metalloproteinases 9 (MMP9) in the H1299 human lung carcinoma cells, resulting in suppression of tumor invasion and metastasis.18 Cell adhesion to collagen I, which is one of the major proteins in the bone marrow (BM) niche, is mostly mediated by three integrin receptors α1β1, α2β1, and α11β1 expressed on cell surface of mesenchymal stem cells.19 Integrin may associate with CD82 in CD34+/CD38− AML cells to promote adhesion to the endosteal niche. However, the roles of CD82 in hematopoietic cells remain to be elucidated.
In this study, we analyzed the protein expression profile of freshly isolated CD34+/CD38− AML cells from individuals with AML and compared it with the expression profile of their CD34+/CD38+ counterparts using isobaric tags for relative and absolute quantitation (iTRAQ) and found the aberrant expression of CD82 in CD34+/CD38− AML cells. This study also explored the function of CD82 in CD34+/CD38− AML cells in vitro as well as in vivo by utilizing NOD.Cg-Rag1tm1MomIl2rgtm1Wjl/SzJ mice.
Material and Methods
Sample collection and isolation of CD34+/CD38− AML cells and their CD34+/CD38+ counterparts
Leukemia cells were freshly isolated from AML patients (n = 18) with World Health Organization (WHO) classification system subtype minimally differentiated AML (case 6), AML without maturation (cases 1 and 10), AML with maturation (cases 2, 7, and 12), acute myelomonocytic leukemia (cases 4, 14, and 15), AML with myelodysplasia changes (cases 3, 5, 8, 9,16, 17, and 18), and therapy-related AML (cases 11 and 13) after obtaining informed consent with Kochi University Institutional Review Board approval (Supporting Information Table S1). The informed consent was obtained in accordance with the Declaration of Helsinki. CD34+/CD38− AML cells and CD34+/CD38+ counterparts were purified by magnetic cell sorting utilizing a CD34 MultiSort kit and a CD38 MicroBead kit (Miltenyi Biotec GmbH, Germany), as previously described (Supporting Information Fig. S1a).20
Chronic eosinophilic leukemia (CEL) EOL-1 cells were obtained from RIKEN BRC Cell Bank (Tsukuba, Japan). Imatinib-resistant EOL-1R cell line was established by culturing with increasing concentrations of imatinib (from 1 to 100 nM) for 6 months.21 Most of EOL-1R cells expressed CD34 (92 ± 9%) on their cell surface. On the other hand, CD34 was rarely detectable on cell surface of parental EOL-1 cells (0.1 ± 0.1%) (figure not shown).
Isolation and culture of primary mesenchymal stromal cells (MSCs)
MSCs were isolated from a BM of healthy donors. BM cells were subjected to centrifugation over a Ficoll-Hypaque gradient to separate mononuclear cells. These cells were resuspended in α-minimal essential medium (Gibco BRL, Rockville, MD) containing 20% fetal bovine serum (FBS) and plated at an initial density of 106 cells.22
Proteins were extracted using the complete mammalian proteome kit (539779, Calbiochem, Darmstadt, Germany), according to the manufacturer's instructions.
iTRAQ labeling of peptides
Each protein sample (100 μg) was digested with trypsin and labeled with iTRAQ reagents (Applied Biosystems, Framingham, MA) according to the manufacturer's instructions. Briefly, the proteins extracted from CD34+/CD38− AML cells were labeled with iTRAQ reagents 114 (case 1) or 116 (case 2), and proteins extracted from CD34+/CD38+ counterparts were labeled with iTRAQ reagents 115 (case 1) or 117 (case 2). Labeled peptide samples were mixed and fractionated as described previously.23
Mass spectrometric analysis
NanoLC-MS/MS analyses were performed on an LTQ-Orbitrap XL (Thermo Fisher Scientific, Waltham, MA) equipped with a nano-ESI source and coupled to a Paradigm MG2 pump (Michrom Bioresources, Auburn, CA) and autosampler (HTC PAL, CTC Analytics, Zwingen, Switzerland).23
iTRAQ data analysis
Protein identification and quantitation for iTRAQ analysis was carried out using SEQUEST (Bioworks version 3.3.1, Thermo Fisher Scientific) searching against the International Protein Index (IPI) human protein database (version 3.26).23 Relative protein abundances were determined by comparing the ratio of iTRAQ reporter ion intensities in the MS/MS scan.23
Quantitation of CD82-expressing cells using flow cytometry (FACS)
Leukemic peripheral blood (PB) (n = 3) and BM (n = 9) cells were collected from 12 AML patients (case numbers 1–12) after obtaining informed consent. Leukemic cells were stained with a fluorescein isothiocyanate (FITC)-conjugated monoclonal antibody (mAb) against CD34 (Beckman coulter, CA), a phycoerythrin (PE)-conjugated mAb against CD82 (Abcam, Cambridge, UK), and a PE Cy5-conjugated mAb against CD38 (BioLegend, San Jose, CA). Cells were stained for 30 min on ice. Isotypo-matched immunoglobulins were used as controls. Cells were then analyzed using flow cytometry (FACS Calibur, Becton Dickinson, San Jose, CA) following data analysis by FlowJo software (TreeStar, San Carlos, CA).
RNA isolation and real-time reverse transcription-polymerase chain reaction (RT-PCR)
RNA isolation and cDNA preparation were performed as described previously.24 Real-time RT-PCR was carried out by using Power SYBR Green PCR Master Mix (Applied Biosystems, Warrington, UK) as described previously.24 Primers for PCR are shown in Table 1.
Table 1. PCR primers
Small interfering RNA
Control small interfering (si) RNA and an siRNA against CD82 were purchased from Santa Cruz Biotechnology and Sigma (Deisenhofen, Germany), respectively.
EOL-1 and EOL-1R cells were transiently transfected with either control or CD82 siRNA (300 nM) using an Amaxa Nucleofector II electroporator (Wako Pure Chemical Industries, Ltd., Osaka, Japan) with a Nucleofector Kit V (program U-001) as previously described.25 The preliminary experiments using the green fluorescence protein-expressing vector found that efficacy of transfection with this program was approximately 70% with nearly 70% cell viability, as measured by FACS and Annexin V/PI staining, respectively (figure not shown).
CD82 shRNA lentiviral vector, production and infection
The short hairpin (sh) RNA sequence used to target human CD82 corresponded to the following sequence on the human CD82 transcript variant 2, NCB I accession number NM_001024844. Lentiviral shRNA particles were produced using the viral power packaging system (Invitrogen, CA) with the 293FT packaging cell line (Invitrogen), and lentiviral CD82 shRNA particles (>108 titer unit (TU)/ml) were prepared using ultracentrifugation. 5 × 104 CD34+/CD38− AML cells were seeded in 24-well plates in 500 μl of Iscove's modified Dulbecco's medium (IMDM) (Invitrogen) containing 10% heat inactivated fetal bovine serum (FBS). After overnight incubation, 5 × 105 TU lentiviral CD82 shRNA particles and polybrene (10 μg/ml) were added per well with serum free medium containing IMDM. After overnight, 1 ml of full media supplemented with FBS, 2-mercaptoethanol, stem cell factor, granulocyte-macrophage colony-stimulating factor (GM-CSF), granulocyte-colony-stimulating factor (G-CSF), IL-3, and erythropoietin (EPO) was added and incubated for 7 days. The control and CD82 shRNA lentiviral vectors co-expressed green fluorescence protein (GFP). Quantification of GFP-positive cells using FACS analysis indicated that the lentiviral transduction efficiency was nearly 70% (Supporting Information Fig. S1b). GFP-positive cells were sorted using JSAN (Bay bioscience Co., Ltd., Kobe, Japan).
CD82 lentiviral vector
CD82 cDNA was purchased from Mammalian gene collection (BC000726) and was used as the template for PCR. PCR products were cloned into pLenti6.3/V5-TOPO vector (Invitrogen). CD82-transfected lentiviral particles were produced using the viral power packaging system (Invitrogen) with the 293FT packaging cell line (Invitrogen). The pLenti6.3/V5-TOPO vector was designed to co-express V5 epitope; 5 × 104 CD34+/CD38+ AML cells were seeded in 24-well plates in 500 μl of IMDM (Invitrogen) containing 10% FBS. After overnight incubation, 5 × 105 TU lentiviral CD82-transfected lentiviral particles and polybrene (10 μg/ml) were added per well. After overnight, supernatant was removed and 1 ml of full media was added and incubated for 7 days. FACS analysis utilizing an anti-V5 antibody (Invitrogen, R960-25) indicated that the efficiency of transduction into CD34+/CD38+ AML cells was nearly 80% (Supporting Information Fig. S1b).
Freshly isolated either CD34+/CD38− (5 × 105 cells) or CD34+/CD38+ AML cells (1 × 105 cells) were transduced by either CD82 shRNA or CD82 cDNA, respectively, and then seeded in the upper inserts with 3 μm pores coated by fibronectin (Cat. No. 354543, Becton Dickinson Biosciences, Bedford, MA) and mesenchymal stromal cells (MSCs) established from healthy donors, while the lower wells were filled with Iscove's Modified Dulbecco's Medium (IMDM) containing 10% heat inactivated FBS. Similarly, EOL-1 and EOL-1R cells were transiently transfected with either control or CD82 siRNA. After 48 hr, these cells (5 × 105 cells in 100 μl RPMI-1640) were seeded in the upper biocoat cell culture inserts coated by fibronectin, and the lower well was filled to the top with RPMI-1640 containing 10% heat inactivated FBS as a chemoattractant. After incubation for 48 hr, the supernatant was discarded and the cells that had adhered to the fibronectin were gently washed in phosphate-buffered saline (PBS). Cells were then fixed for 1 hr in 4% paraformaldehyde, washed twice in PBS, stained with 4′6-diamidino-2-phenylindole (DAPI), and counted under a microscope (OLYMPUS FV1000-D). The cells that had passed through the membrane filter were collected and the number of viable cells was counted under light microscope after staining with trypan blue.
The culture supernatant as well as whole cell proteins of EOL-1 and EOL-1R cells were harvested. Gelatinolytic activities were carried out by utilizing a gelatin-zymography kit (Primarycell, Hokkaido, Japan). Each lane was loaded with 30 μg of whole protein lysates or 20 μl of supernatant.
Colony forming assay
The colony-forming assay was performed with methylcellulose medium H4034 (StemCell Technologies, Vancouver, BC, Canada), as previously described.20
Bone marrow transplantation and engraftment assay
NOD.Cg-Rag1tm1MomIl2rgtm1Wjl/SzJ mice (Stock number: 007799) were purchased from the Jackson Laboratory for experimental animals (Bar Harbor)26 and bred in a pathogen-free environment in accordance with the guidelines of the Kochi University School of Medicine. The 6-week-old mice were utilized for experiments. CD34+/CD38− AML cells (1 × 104 cells) transfected with either scrambled control or CD82 shRNA were injected to each mouse intravenously via the tail vein. At 9 weeks after transplantation, mice were euthanized and BM were removed. BM cells were flushed from the femurs using 25-gauge needles (Becton Dickinson Biosciences) and then fixed in formalin. The human cell engraftment was analyzed by using flow cytometry after staining of spleen cells with human CD45 PE Cy5-conjugated mAb (Dako, Glostrup, Denmark) and human CD33 PE-conjugated mAb (Becton Dickinson Biosciences).
Single cell RT-PCR
A single CD34+/CD38− AML cell was isolated by BD FACS AriaII (Becton Dickinson Biosciences) and subjected to RT-PCR by AmpliSpeed slide cycler (Beckman Coulter, Munich, Germany) and ABI StepOnePlus (Applied Biosystems) to measure the levels of CD82 and MMP9.
Immunohistochemistry of CD82 in BM sections
Immunohistochemical staining of CD82 was performed with a Ventana DISCOVERYTM autostainer system (Ventana Japan, Osaka, Japan) as previously described.25 The anti-CD82 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was used.
Cell cycle analysis by flow cytometry (FACS)
Cell cycle distribution of CD34+/CD38− AML cell was measured as previously described after transduction of either scrambled control or CD82 shRNA. Briefly, the cells were stained with Ki-67 (Santa Cruz Biotechnology) and propidium iodide and subjected to FACS.20
The human AML cells isolated from patients were treated with anti-human CD82 (Santa Cruz Biotechnology) or control IgG (eBiosScience, San Diego, CA) antibody on ice for 1 hr. This anti-CD82 antibody worked as a neutralizing antibody (Nishioka et al. unpublished data). Cells were washed gently with PBS and were injected to each mouse intravenously via the tail vein. At 16 hr after transplantation, mice were euthanized and BMs were removed and analyzed by flow cytometry with anti-human CD45 and CD34 antibodies. We acquired in the range of 1 × 106 to 3 × 106 events per sample.
When comparing two groups, Student's t-test was used. For demonstration of association, the Pearson's correlation coefficient test was applied. All statistical analyses were carried out using SPSS software (Version 11.03; spss, Tokyo, Japan) and the results were considered to be significant when the p value was <0.05, and highly significant when the p value was <0.01.
Protein expression profiles of CD34+/CD38− AML cells and CD34+/CD38+ counterparts
Four samples (two from CD34+/CD38− AML cells lysates and two from CD34+/CD38+ counterparts lysates, approximately 10 million cells per sample) were trypsinized and labeled with a specific isobaric iTRAQ reagent. To compensate for extreme sample complexity, each iTRAQ sample was separated into 24 fractions using strong cation-exchange chromatography.23 A total of 2,537 and 2,506 proteins were identified with >95% confidence for each of the biological replicates. Since iTRAQ internal replicates typically yield high confidence results,27 differences greater than threefold or less than 0.5-fold are considered significant. We listed the proteins whose expression was greater than 3-fold or less than 0.5-fold in CD34+/CD38− AML cells as compared with their CD34+/CD38+ counterparts in Table 2 and Supporting Information Table 2. Either 481 or 700 proteins were differentially expressed in case 1 and case 2, respectively (Table not shown), and only 104 proteins were overlapped in both cases (Table 2, Supporting Information Table S1).The expression of 98 of these proteins increased, while the expression of six proteins decreased in both cases (Table 2, Supporting Information Table S1). Two of the identified proteins are involved in differentiation, two play a role in the cell cycle, six play a role in adhesion, five are involved in DNA replication and repair, and eight are involved in apoptosis and anti-apoptosis (Table 2, Supporting Information Table S1). In addition, 6 nuclear transcription factors, 17 enzymes involved in drug-resistance, 6 human leukocyte antigens, 18 ribosome nucleoproteins, and 10 histone and histone-binding proteins were differentially expressed between CD34+/CD38− AML and CD34+/CD38+ counterparts (Supporting Information Table S1). Additional differentially expressed proteins are also listed in Supporting Information Table S1. Other studies identified aldehyde dehydrogenase activity (ALDH), B-cell lymphoma/leukemia 11A (BCL11A), ILK, and HLA-DR as highly expressed proteins in leukemia stem cells.28–31 These proteins were also overexpressed in CD34+/CD38− AML cells in the present study (Table 2), indicating an acceptable sensitivity of the current study.
Table 2. Protein expression profiles of CD34+/CD38− and CD34+/CD38+ AML cells
CD82 is overexpressed in CD34+/CD38− AML cells
We focused on CD82 because this protein functions as an adhesion molecule that is important to maintain the character of LSCs. We attempted to validate these results in other CD34+/CD38− AML cells isolated from patients (11 from BM, 5 from PB, cases 1–14, 17, 18) by FACS. In 15 of 16 cases (94%), the relative expression levels of CD82 were significantly higher in CD34+/CD38− AML cells (68 ± 27%) as compared with their CD34+/CD38+ counterparts (30 ± 19%) (p < 0.01, Fig. 1a, Supporting Information Fig. S2a). On the other hand, a mean 35 ± 19% of CD34+ hematopoietic stem/progenitor cells isolated from healthy volunteers (n = 6) were positive for CD82 staining (Supporting Information Fig. S2b). In addition, we found that imatinib-resistant EOL-1R cells which stayed on a dormant state and possessed the immature character with aberrant expression of CD34 (data not shown) expressed a greater amount of CD82 on their cell surface (96 ± 1%) than parental EOL-1 cells (47 ± 4%) (Fig. 1a).
The MMPs enzymatic activity
Aberrant expression of CD82 was associated with inactivation of matrix metalloproteinase 9 (MMP9) in the H1299 human lung carcinoma cells.14 We therefore examined the relationship between CD82 and MMPs in CD34+/CD38− AML cells. Real-time RT-PCR found that the levels of MMP9 were significantly lower in CD82 over-expressed CD34+/CD38− AML cells than their CD34+/CD38+ counterparts (n = 7, p < 0.01) (Fig. 1b). On the other hand, levels of MMP2 in CD34+/CD38− AML cells were almost identical to those in CD34+/CD38+ counterparts (Fig. 1b). We also found that the levels of both MMP-9 and -2 were down-regulated in imatinib-resistant EOL-1R cells as compared with parental EOL-1 cells (Fig. 1b). To explore a potential link between CD82 and MMPs in leukemia cells, EOL-1R cells were transiently transfected with either scrambled control or CD82 siRNA (Fig. 1c), which efficiently decreased levels of CD82 in these cells (from 96 ± 1% to 41 ± 1%, Fig. 1c). The MMPs enzymatic activity in these cells was determined by performing gelatin zymography with the culture supernatant as well as whole cell proteins extracted from EOL-1 and EOL-1R cells (Fig. 1d). Interestingly, when CD82 was down-regulated in EOL-1R cells by an siRNA, enzymatic activity of MMP9 was dramatically increased (Fig. 1d), suggesting that CD82 negatively regulated MMP9. Real-time RT-PCR found that down-regulation of CD82 by an siRNA increased levels of MMP9 by nearly twofold in EOL-1R cells (Fig. 1e). On the other hand, levels of MMP2 were not affected by down-regulation of CD82 (Fig. 1e). Moreover, to explore the function of CD82 in freshly isolated CD34+/CD38− AML cells, we genetically down-regulated CD82 in these cells. An shRNA targeting CD82 decreased expression of CD82 in four cells (case 1, from 85 ± 2% to 47 ± 3%; case 2, from 75 ± 1% to 13 ± 2%; case 6, from 47 ± 7% to 21 ± 3%, p = 0.066; case 14, from 43 ± 1% to 18 ± 1%, Fig. 1f). Real-time RT-PCR found that the levels of MMP9 were significantly increased after down-regulation of CD82 in CD34+/CD38− AML cells (n = 3, cases 1, 2, and 6, p < 0.05, Fig. 1g). We next exposed CD34+/CD38+ AML cells to the CD82-expressing lentiviral particles, which increased levels of CD82 (n = 4, case 1; from 56 ± 7% to 88 ± 3%, case 2, from 23 ± 1% to 58 ± 1%; case 14; from 6 ± 4% to 85 ± 4%, case 15; from 43 ± 7% to 89 ± 2%, Fig. 1h). As expected, the levels of MMP9 were decreased by half in these cells (Fig. 1i).
The effects of CD82 on migration of CD34+/CD38− AML cells
We next examined the function of CD82 in CD34+/CD38− AML cells. When levels of CD82 were down-regulated in CD34+/CD38− AML cells after lentiviral transduction of CD82 shRNA (n = 4. cases 1, 2, 6 and 14, Fig. 1f), their migration was significantly stimulated (Fig. 2a). Moreover, we enhanced expression of CD82 in CD34+/CD38+ AML cells by using CD82-expressing lentiviral particles (n = 3, cases 1, 14 and 15, Fig. 1h). Forced-expression of CD82 in CD34+/CD38+ AML cells dramatically increased number of cells adhered to the insert in parallel with a decrease in the number of migrated cells (Fig. 2b). Similarly, after down-regulation of CD82 in EOL-1R cells, adhesive cells decreased by approximately half (Fig. 2c). In parallel, population of EOL-1R cells migrated to the lower well increased by 10-fold after down-regulation of CD82 in these cells (Fig. 2c).
The effect of CD82 on colony forming ability of CD34+/CD38− AML cells
We first examined whether CD82 regulated proliferation of CD34+/CD38− AML cells (cases 1, 6, and 14) by using colony forming assay (Fig. 3a). Down-regulation of CD82 by an shRNA (from 59 to 27%) inhibited their colony forming ability by approximately 50% (Fig. 3a). Likewise, down-regulation of CD82 by an siRNA (from 96% to 48%) inhibited colony forming ability of CD34+ EOL-1R cells which mimic LSCs by mean 50% (Fig. 3b). On the other hand, forced-expression of CD82 in CD34+/CD38+ AML cells (cases 1 and 14) by transduction of CD82-expressing lentiviral particles increased levels of CD82 from 18 to 89% and stimulated their colony forming ability by mean 1.7-fold (Fig. 3c). These observations suggested that CD82 plays a role in survival of CD34+/CD38− AML cells.
AML engraftment was inhibited by down-regulation of CD82 in vivo
We next examined the function of CD82 in vivo. CD34+/CD38− cells isolated from three different AML patients (cases 1, 6 and 14) were transduced with either scrambled control or CD82 shRNA. We transplanted these cells into NOD.Cg-Rag1tm1MomIl2rgtm1Wjl/SzJ mice via the tail vein. Transplanted mice were sacrificed at 9 weeks after transplantation and analyzed the human engraftment in their spleens and BM by quantifying the population of positive cells for human CD45 and CD33 antigens (Figs. 4a and 4b). Transplantation of CD34+/CD38− AML cells transduced by scrambled control shRNA resulted in the mean human engraftment either 14 ± 6% or 15 ± 5% in spleen and BM, respectively (n = 8, Fig. 4a). These cells expressed CD33antigen on their cell surface (4% in both spleen and BM) (Fig. 4b). On the other hand, when these cells were transduced with CD82 shRNA, human engraftment was significantly impaired (7 ± 4% or 6 ± 2% in spleen and BM, respectively, p < 0.01) (n = 6, Fig. 4a). In addition, population of cells expressing CD33 on their cell surface was decreased to either 0.1% or 1% in spleen and BM, respectively (Fig. 4b). These observations suggested that down-regulation of CD82 impaired AML engraftment as well as AML reconstitution in an immunodeficient mice. In addition, we assessed levels of CD82 and MMP9 in isolated CD34+/CD38− AML cells (n = 9) by using single cell real-time RT-PCR (Fig. 4c). Inverse correlation was noted between levels of CD82 and MMP9 (r = −0.58, Fig. 4c). Moreover, we examined the levels of CD82 in transplanted human AML cells as well as localization of these cells in murine BM by immunohistochemistry (Figs. 4d–4g). Notably, human AML cells expressing CD82 were localized in endosteal region of BM at 9 weeks after transplantation of scrambled control shRNA transduced CD34+/CD38− AML cells (Figs. 4d and 4f). On the other hand, human AML cells were not detectable in this region of murine BM, that were transplanted with CD82-depleted CD34+/CD38− AML cells (Figs. 4e and 4g). To examine longer-term reconstituting capability, we carried out secondary transplantation of CD34+/CD38− AML cells recovered from primary recipient mice. At 9 weeks post-transplantation, human AML engraftment was 18% as assessed by quantification of human CD45 expressing cells in PB isolated from the secondary recipients by FACS. These observations suggested that functional properties of CD34+/CD38− AML cells were maintained in the BM microenvironment of recipient mice and CD34+/CD38− AML cells utilized in this study fulfill the criteria for LSCs in vivo.
Moreover, we examined whether CD82 affected homing of AML cells to BM. AML cells isolated from three patients (cases 6, 14, 17) were treated with anti-CD82 or control IgG antibody. These cells were transplanted per mouse, and homing of the cells was analyzed in the BM of mice after 16 hr of transplantation. In the control group (n = 6), an average of 0.39% in human CD34+ AML cells were detected in the mouse BM compared with 0.77% in the CD82 antibody-treated group (Supporting Information Fig. S6).
Previous studies showed that the LSCs-niche interaction was important to maintain the stemness of leukemia cells.15 In this study, iTRAQ technique identified adhesion molecule CD82 as an overexpressed protein in CD34+/CD38− AML cells (Supporting Information Table S1). Additional experiments utilizing FACS confirmed aberrant expression of CD82 in freshly isolated CD34+/CD38− AML cells (n = 16) (Fig. 1a). In addition, this study found that down-regulation of CD82 by an shRNA increased levels of MMP9 in CD34+/CD38− AML cells (Fig. 1g) and stimulated their migration (Fig. 2a). Meanwhile, forced expression of CD82 decreased levels of MMP9 mRNA in CD34+/CD38+ AML cells (Fig. 1i) and enhanced adhesion of these cells to fibronectin and MSCs, a kind of artificial BM niche (Fig. 2b). Single cell RT-PCR also demonstrated that down-regulation of CD82 by an shRNA increased levels of MMP9 mRNA in CD34+/CD38− AML cells, and reverse correlation was noted between levels of MMP9 and CD82 in CD34+/CD38− AML cells (Fig. 4c). The levels of MMP9 mRNA in CD34+/CD38− AML cells shown in Figures 1g and 4c were inconsistent, although the cells utilized in these studies were isolated from same populations. In these studies shown in Figure 4c, we used a single cell immediately after isolation from murine BM. On the other hand, cells used in these studies shown in Figure 1g were incubated for 7 days in full media to be transduced by shRNA. As a result, the levels of MMP9 mRNA in leukemia cells might be down-regulated in these cells. Notably, down-regulation of CD82 in CD34+/CD38− AML cells (n = 3) by an shRNA impaired engraftment of these cells in the BM as well as spleen in NOD.Cg-Rag1tm1MomIl2rgtm1Wjl/SzJ mice (Fig. 4a). These observations suggested that CD82 played an important role in adhesion of CD34+/CD38− AML cells to BM microenvironment via down-regulation of MMP9. Other investigators showed that MMP9 in human mononuclear phagocytes was inhibited by IL-10.32 Similarly, IL-10 activated the tissue inhibitors of expression of metalloproteinases (TIMP-1/2) and down-regulated the levels of MMP2 and MMP9 in human prostate cancer cells.33 Thus, down-regulation of MMP9 by IL-10 may augment adhesion of HSCs to BM osteoblastic niche and exogenous administration of IL-10 may be useful to promote the repopulating ability of HSCs and engraftment of HSCs to BM niche. We also found that exposure of CD34+/CD38− AML cells to IL-10 (5 ng/ml) down-regulated levels of MMP9 mRNA in these cells (Supporting Information Fig. S5). On the other hand, down-regulation of IL-10 in these cells (n = 2, cases 2 and 6) by an shRNA increased levels of MMP9 by twofold (Supporting Information Fig. S5). Further experiments found that down-regulation of CD82 in CD34+/CD38− AML cells by an shRNA potently down-regulated levels of IL-10 (in preparation for publication). Notably, forced expression of CD82 down-regulated levels of MMP9 mRNA, resulting in inactivation of MMP9 in leukemia cells (Figs. 1d and 1e). We therefore hypothesize that CD82 may inactivate MMP9 via IL-10 signaling in LSCs. MMP9 promotes mobilization of HSCs from the BM osteoblastic niche by release of soluble Kit-ligand (sKitL),34 which increases the motility of HSCs and progenitors within the BM. In addition, MMPs cleave integrin β4 and β1 in cultured human corneal epithelial cells and mouse epidermal keratinocytes.35 CD82 associates with various integrins including α3β1, α4β1 and α6β1.36, 37 Amongst them integrin α4β1 (VLA4) plays an important role in adhesion of LSCs to BM microenvironment.38 We found that CD34+/CD38− AML cells expressed a greater amount of integrin α4β1 (VLA4) than their counterparts, as measured by real time RT-PCR (n = 5, Supporting Information Fig. S7a). The hematopoietic cells adhere to stromal endothelial cells through the VLA4/vascular cellular adhesion molecule-1 (VCAM-1) pathway.39 Thus, activation of MMP9 by down-regulation of CD82 may be able to mobilize CD34+/CD38− AML cells from BM via disruption of interaction between VLA4 and VCAM-1 on cell surface of stromal cells. Moreover, to assess if CD82 interacts with VLA4 molecules, we utilized an anti-integrin β1 (CD29) Ab (Beckman coulter, CA, 6603113) which blocks adhesion of leukemia cells to fibronectine. Forced-expression of CD82 stimulated an adhesion of CD34+/CD38+ AML cells (case 15) to the artificial niche, which was hampered when these cells were treated with an anti-integrin β1 Ab (Supporting Information Fig. S7b). Furthermore, we examined interaction between CD82 and integrin α4 by utilizing Immunoprecipitation assay and found that CD82 directly interacted with integrin α4 (Supporting Information Fig. S7c). These observations suggested that VLA4 interacted with CD82 and played a role in adhesion of these cells to the artificial niche.
We found that blockade of CD82 on cell surface of AML cells by an antibody stimulated BM homing of CD34+ AML cells in these mice (Supporting Information Fig. S6). Similar to the present study, other investigates also showed that cord blood (CB) CD34+ cells treated with stem cell factor enhanced expression of MMP2/MMP9 and increased BM homing of human CB CD34+ cells in NOD/SCID mice.40, 41 Thus, blockade of CD82 may increase levels of MMP9, resulting in enhanced migration of AML cells into the BM. However, we hypothesize that these cells could not fully adhere to the BM microenvironment and survive to develop AML.
Interestingly, the CXCR4 antagonist AMD3100 effectively mobilized AML cells without inducing their proliferation.42 Preclinical studies showed that treatment of leukemic mice with a chemotherapeutic agent in combination with AMD3100 resulted in decreased tumor burden and improved their overall survival compared with mice treated with a chemotherapeutic agent alone. These observations provided a proof-of-principle for directing therapy to the critical tethers that promote AML-niche interactions42 and supported our hypothesis that inhibition of CD82 could mobilize LSCs from BM niche and sensitize these cells to chemotherapeutic agents. Further studies are clearly required to test our hypothesis in vivo.
Another idea to sensitize LSCs to chemotherapeutic agents related to stimulation of cell-cycling of dormant LSCs. Recent studies showed that CD34+/CD38− AML cells were induced to enter the cell cycle by treatment with G-CSF in vivo. G-CSF significantly enhanced apoptosis of CD34+/CD38− AML cells mediated by cell cycle-dependent chemotherapeutic agents and eliminated CD34+/CD38− AML cells from mice.43 The additional experiments found that down-regulation of CD82 was not able to stimulate cell cycling of CD34+/CD38− AML cells (cases 1 and 6) and EOL-1R cells (Supporting Information Fig. S3), suggesting that CD82 was not involved in the maintenance of dormancy in these cells.
CD82 inhibited the receptor tyrosine kinase human mesenchymal-epithelial transition factor (c-Met) activity,44 which promoted proliferation and migration of cancer cells.45, 46 c-Met was shown to mediate G-CSF-induced mobilization of hematopoietic progenitor cells (HPCs) via reactive oxygen species (ROS) signaling.47 Aberrant expression of CD82 in CD34+/CD38− AML cells may inactivate c-Met and cause engraftment of these cells to BM niche. We found that levels of CD82 in CD34+ hematopoietic stem/progenitor cells isolated from healthy volunteers (n = 6) were lower than those in CD34+/CD38− AML cells (35% vs. 59%, p = 0.02, Supporting Information Fig. S4). Importantly, down-regulation of CD82 in CD34+ hematopoietic stem/progenitor cells by an shRNA did not significantly inhibit their colony forming ability (Supporting Information Fig. S4).
CD82 expression was strongly correlated with the tumor suppressor gene p53.48 On the other hand, other studies showed that levels of CD82 did not correlate with the expression of p53 in human hepatocellular carcinoma.49 We also examined the correlation between CD82 and p53 mRNA levels in CD34+/CD38− AML cells (n = 6) and their CD34+/CD38− counterparts by utilizing real-time RT-PCR and found that the correlation coefficient was 0.54 (figure not shown). Thus, we think that expression of CD82 is not related with p53 in CD34+/CD38− AML cells.
Overexpressed CD82 in AML cells may render these cells to adhere to BM niche and regulate maintenance of leukemia stem cells within BM niche. On the other hand, down-regulation of CD82 in AML cells may stimulate circulating of these cells from BM niche to PB.
Taken together, our data suggested that CD82 negatively regulated MMP9 and played an important role in CD34+/CD38− AML cells to adhere to BM microenvironment. In addition, CD82 was involved in survival of CD34+/CD38− AML cells. CD82 might be an attractive molecular target to eradicate LSCs in AML patients. Further studies are warranted to evaluate the function of CD82 in LSCs in vivo.
This work was supported in part by The Kochi University President's Discretionary Grant (to T.I.), Setsuro Fujii Memorial, The Osaka Foundation for Promotion of Fundamental Medical Research (to T.I.) and Certificate of Kochi Shin-kin/Anshin-tomo-no-kai Prize (to C.N.). C.N. is grateful for a JSPS Research Fellowship for Young Scientists from the Japan Society for the Promotion of Science.