Id1 Transcription Inhibitor–Matrix Metalloproteinase 9 Axis Enhances Invasiveness of the Breakpoint Cluster Region/Abelson Tyrosine Kinase–Transformed Leukemia Cells

Introduction

Breakpoint cluster region/Abelson (BCR/ABL) oncogene is derived from the relocation of a portion of the c-abl gene from chromosome 9 to the portion of the bcr gene locus on chromosome 22 [t(9;22), Philadelphia (Ph) chromosome], and is expressed in chronic myelogenous leukemia (CML) and a subset of the acute leukemias from both myeloid and lymphoid lineages ( 1). CML typically first appears in a rather benign chronic phase (CML-CP) that, over time, undergoes blast crisis (CML-BC) transformation to an overt acute leukemia. The BCR/ABL gene–encoded BCR/ABL tyrosine kinase is constitutively expressed and active, and has been shown to stimulate numerous signaling molecules, including Ras, phosphatidylinositol 3-kinase, and signal transducer and activator of transcription 5 (STAT5). Continuous activation of these downstream effectors is essential for BCR/ABL–mediated leukemogenesis leading to genomic instability, proliferation in the absence of growth factors, protection from apoptosis in the absence of external survival factors, and facilitation of invasion of various organs (reviewed in refs. 2– 4).

Although studies with the use of knockout mice suggest that STAT5 might be dispensable during BCR/ABL–mediated transformation per se ( 5), other reports, including our own, implicate STAT5 in protection from apoptosis and stimulation of proliferation of leukemia cells ( 6– 9). We and others showed that STAT5-dependent stimulation of the Bcl-2 family members Bcl-xL and A1, and the proto-oncogene pim-1 contribute to protection from apoptosis and growth factor independence in BCR/ABL–transformed cells ( 10, 11).

Another STAT5 downstream effector is the basic helix-loop-helix (bHLH) Id1 protein ( 9, 12). Although most members of the family of bHLH contain a basic DNA-binding domain and act as transcription factors ( 13, 14), Id1 lacks the domain. It can, however, associate with other members of the bHLH family and impair their DNA-binding capacity ( 13). Therefore, the main function of Id1 is to regulate gene transcription by sequestering the ubiquitously expressed bHLH transcription factors.

With regard to function, Id1 can inhibit cell differentiation and stimulate proliferation, migration, and angiogenesis ( 13, 14). Enhanced expression of Id1 has been reported frequently in higher-grade, less-differentiated malignancies. The distinct role of Id1 in promoting tumor cell invasive and metastatic capacity are highlighted by the following observations: (a) in primary breast cancer cells, the expression of Id1 was associated with infiltrating, more aggressive tumors ( 15); (b) overexpression of Id1 causes mammary epithelial cells to invade the basement membrane ( 16); (c) constitutive expression of Id1 in nonaggressive breast cancer cell line confers a more invasive phenotype ( 15); and (d) Id1 expression is associated with the invasive behavior of endometrial carcinomas ( 17).

Id1-mediated enhancement of the invasive capacity of breast carcinoma cells was associated with the increased expression of 120 kDa gelatinase ( 16). It is tempting to speculate that this gelatinase consists of a 92 kDa matrix metalloproteinase-9 (MMP9) and a 25 kDa neutrophil gelatinase-associated lipocalin, which protects MMP9 from degradation ( 18). MMP9, a type IV collagenase, belongs to a family of over 20 MMPs that are characterized by their ability to degrade extracellular matrix and their dependence on Zn²⁺ binding for proteolytic activity ( 19). There is compelling evidence that MMPs can induce or enhance tumor cell proliferation, survival, invasiveness, angiogenesis, and metastatic capacity ( 20). MMP9 acts in part by dissolving connective tissue barriers composed of collagens I, IV, V, VII, and XI.

Our genome scale profiling of gene expression identified elevated expression of both Id1 and MMP9 in CML cells ( 21). However, the exact role of these two proteins in the BCR/ABL–mediated leukemogenesis remains unexplored.

Materials and Methods

Plasmids. pSRα-p210BCR/ABL retroviral construct was obtained from Dr. Charles Sawyers (University of California at Los Angeles, Los Angeles, CA) and used before ( 22). Id1 cDNA was obtained from DNAX Research Institute of Molecular and Cellular Biology, Inc. (Paolo Alto, CA) and cloned in the sense (S) and antisense (AS) orientation in the pMIG1 retroviral construct containing IRES-green fluorescent protein (GFP) cassette (gift of Dr. Warren Pear, University of Pennsylvania, Philadelphia, PA). Luciferase reporter construct containing full-length MMP9 promoter was obtained from Dr. J. Pagano (The University of North Carolina at Chapel Hill, Chapel Hill, NC).

Cells. 32Dcl3 murine hematopoietic cells (parental) and those expressing p210BCR/ABL, STAT5B dominant-active mutant (STAT5B-DAM), or p210BCR/ABL and STAT5B dominant-negative mutant (STAT5B-DNM) were characterized before ( 9, 23). FL5.12 murine hematopoietic cells and p210BCR/ABL transfectants were obtained from Dr. Warren Pear. BaF3 cells transformed with BCR/ABL and related fusion tyrosine kinases (FTK), such as TEL/ABL, TEL/JAK2, TEL/PDGFβR, NPM/ALK, and TEL/TRKC, were described before ( 24). Cells were maintained in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% fetal bovine serum (FBS), WEHI-conditioned medium as a source of interleukin-3 (IL-3; pretested concentrations adequate to support the proliferation of parental cells) and antibiotics. At least three clones per experimental group were used. Peripheral blood cells from two patients in myeloid and two patients in lymphoid CML-BC were obtained after informed consent, and CD34⁺ cells or mononuclear cells, respectively, were isolated as described ( 25). CD34⁺ mobilized peripheral blood progenitors were purchased from Cambrex BioScience, Walkersville, Inc. (Gaithersburg, NJ). Tk-ts13 cells were cultured in DMEM supplemented with 10% FBS and antibiotics.

Retroviral infections. Cells were infected with pMIG1 retrovirus containing Id1(S)-IRES-GFP, Id1(AS)-IRES-GFP or IRES-GFP (empty plasmid) as described before ( 10). GFP⁺ cells were obtained by sorting after 72 hours of continuous infection, expanded in the presence of WEHI-conditioned medium and used for experiments. Id1 mRNA and protein expression in GFP⁺ cells was determined by semiquantitative reverse transcription-PCR (RT-PCR) and Western analysis, respectively.

Inhibition of BCR/ABL kinase. ABL kinase inhibitor STI571 (imatinib mesylate, Gleevec) was obtained from Novartis Pharma (Basel, Switzerland). Cells (10⁶/mL) were incubated for 8 or 24 hours with 1 μmol/L STI571 in the absence or presence of IL-3 (IL-6 was also added to lymphoid CML-BC cells), respectively, then washed and used for experiments.

Microarray analysis. The sample preparation and array protocol have been published previously ( 21).

Western analysis. Total protein cell lysates were obtained as previously described ( 10) and analyzed by SDS-PAGE followed by Western blotting with the use of anti-ABL (Ab-3, Oncogene Research Products, Cambridge, MA), anti-Id1 (C-20, Santa Cruz Biotechnology, Santa Cruz, CA), anti-Id2 (C-20, Santa Cruz Biotechnology), antiphosphotyrosine (PY20, Oncogene Research Products; and 4G10, Upstate Biotechnology, Lake Placid, NY), and antiactin (C-11, Santa Cruz Biotechnology) antibodies.

Semiquantitative RT-PCR. The reaction was done as described before ( 21) with modifications. Total RNA was extracted by RNeasy Mini kit 74104 (Qiagen, Valencia, CA). Reverse transcription reactions were done in the following conditions: 1 μg total RNA, 2 μg random primer p(dN)₆ (Roche Diagnostics GmbH, Sandhofer Strasse 116, D-68305 Mannnheim, Germany), 10 μmol/L deoxynucleotide triphosphates (Eppendorf AG, Hamburg, Germany), 5 mmol/L DTT (Fisher Scientific, Fair Lawn, NJ), 40 units of RNAsin (Promega, Madison, WI), and 20 units of reverse transcriptase avian myeloblastosis virus (Roche), total reaction volume 20 μL. Ten microliters of reverse transcription reaction were used as a template for 25 cycles of 50 μL PCR (initial denaturation 94°C/5 minutes, denaturation 94°C/45 seconds, annealing 60°C/45 seconds, polymerization 72°C/45 seconds, final polymerization 72°C/5 minutes) along with 5 units of Taq polymerase (Promega) to amplify simultaneously Id1 and GADPH in the same test tubes. The following primers were used: 10 pmol Id1 (primers: forward, ATGAACGGCTGCTACTCACG; reverse, GCGACACAAGATGCGATCGT; PCR product, 261 bp) and 1.3 pmol GADPH (primers: forward, ACCACAGTCCATGCCATCAC; reverse, TCCACCACCCTGTTGCTGTA; PCR product, 452 bp). GADPH gene product was detected as loading/performance control of the semiquantitative PCR. The products were resolved in 2% agarose gels containing 0.5 μg/mL ethidium bromide.

STAT5 activity. The DNA-binding activity of STAT5 was examined by electrophoretic mobility shift analysis using the FcγRI INF-γ-activated sequence motif as described previously ( 9).

MMP9 transactivation assay. MMP9 transactivation assay was examined by luciferase assay as previously described ( 9). In brief, Tk-ts13 hamster fibroblasts were cotransfected with Id1, BCR/ABL, BCR/ABL, and Id1AS, or insert-less expression vectors along with the luciferase reporter construct containing full-length MMP9 promoter and β-galactosidase expression vector. Thirty-six hours after transfection, cells were starved from serum in DMEM + 0.1% bovine serum albumin for 48 hours and harvested for the luciferase assay using Dual-Luciferase Reporter Assay System (Promega) according to the protocol of the manufacturer. For each transfection, luciferase activity was normalized using β-galactosidase activity as an internal control.

Zymography. Cells were incubated in serum-free IMDM at a concentration of 2 × 10⁶/mL for 12 hours at 37°C in 5% CO₂. Conditioned medium was concentrated with Amicon Ultra-15 centrifugal filter devices (Millipore Corporation, Bedford, MA) and MMP9 activity was assayed as described by Chemicon International (Temecula, CA) handbook with modifications. Briefly, 18 to 21 μL of the concentrated conditioned medium protein samples (∼30 μg) were mixed with 6 to 7 μL of the 4× Tris-glycine sample buffer without reducing agent, incubated at room temperature for 20 minutes, and electrophoresed in 10% polyacrylamide gel containing 0.1% gelatin (Sigma-Aldrich, St. Louis, MO) at 4°C. Then, the gel was washed in 2.5% Triton X-100 for 30 minutes at room temperature and incubated for 18 hours at 37°C in 50 mmol/L Tris-HCl buffer (pH 7.4) containing 0.15 mol/L NaCl, 10 mmol/L CaCl₂, and 0.02% NaN₃ while gently shaking. Finally, the gel was stained for 1.5 hours with 0.1% Coomassie brilliant blue R250 and destained in 50% methanol and 10% acetic acid in H₂O. The stained gel was photographed and processed in Adobe Photoshop.

Adhesion to collagen IV. Adherence to collagen IV was examined as described before ( 22). Briefly, 2 × 10⁶ cells were suspended in 4 mL IMDM supplemented with 0.1% bovine serum albumin and incubated for 4 hours at 37°C in six-well plates coated with collagen IV (Becton Dickinson, Bedford, MA). Nonadherent cells were then removed and adherent cells were harvested after trypsinization and counted.

Invasion chamber assay. The assay was done as previously described ( 22). Briefly, 2 × 10⁶ cells of murine hematopoietic cell lines and 2 × 10⁵ primary cells suspended in 2 mL growth factor–free culture medium were placed in the upper chamber of Biocoat Matrigel invasion chambers (Becton Dickinson); the lower chamber was filled with 2 mL medium supplemented with IL-3 or IL-3 + IL-6 as chemoattractant for murine and human cells, respectively. MMP9-inactivating monoclonal antibodies (Ab-10, Oncogene Research Products; 5 μg/mL) or a gelatinase inhibitor ilomastat (GM6001, Chemicon International; 4 μg/mL) were added to the upper chamber; controls include nonimmune antibody and diluent, respectively. Viable cells able to migrate into the lower chamber and the percentage of living cells in the upper chamber were counted 24 hours later. Results were adjusted to represent the number of invasive cells present in the same number of living cells surviving the test.

Homing assay. In vivo invasion assay was done as described before ( 22). Briefly, cells were labeled with [5-³H]uridine (Moravec Biochemicals, Inc., Brea, CA; 16.2 Ci/mmol) for 24 hours at 37°C, washed, and injected i.v. into severe combined immunodeficient (SCID) mice (5 × 10⁶/mouse). Cells from bone marrow (both femurs), spleen, liver, kidney, lungs, and brain were collected after 24 hours and deposited onto glass microfiber filters (Whatman International Ltd., Maidstone, United Kingdom). Cell-associated radioactivity was measured in a β-scintillation counter.

Leukemogenesis in mice. Outbred SCID mice (Taconic Farms, Germantown, NJ) were injected i.v. with 10⁶ cells pretested for the invasive capability. The development of leukemia was examined 3 and 7 days after leukemia inoculation and also in visibly morbid animals as described before ( 9). In brief, internal organs (bone marrow, spleen, liver, brain, lungs, and kidneys) were harvested, fixed, embedded, sectioned, stained with H&E, and examined under the microscope as described ( 9). A chloroacetate esterase (Leder) staining confirmed myeloid origin of the leukemia cells in the selected tissue sections. Animal studies were approved by the Institutional Animal Care and Use Committee at Temple University.

Results

The expression of Id1 is regulated by BCR/ABL kinase and STAT5. BCR/ABL was able to stimulate STAT5 and up-regulate Id1 protein in 32Dcl3, BaF3, and FL5.12 hematopoietic cell lines in the absence of IL-3 ( Fig. 1A , compare group 3 with 1, 7 with 6, and 9 with 8). Inhibition of BCR/ABL kinase activity by STI571 abrogated this effect ( Fig. 1A, compare group 5 with 3). In addition, expression of a constitutively active form of STAT5B (STAT5B-DAM) elevated Id1 protein expression in 32Dcl3 cells ( Fig. 1A, compare group 2 with 1). Moreover, a dominant-negative mutant of STAT5 (STAT5-DNM) down-regulated Id1 in BCR/ABL–transformed 32Dcl3 cells ( Fig. 1A, compare group 4 with 3). Expression of STAT5B-DAM mutant in parental cells and STAT5B-DNM in BCR/ABL–transformed cells was associated with activation and inhibition, respectively, of STAT5 DNA-binding activity ( Fig. 1A, compare group 2 with 1, and 4 with 3, respectively).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

BCR/ABL kinase and STAT5 stimulate the expression of Id1. A, parental cells (1, 6, and 8) and cells expressing STAT5B-DAM (2), BCR/ABL (3, 7, and 9), BCR/ABL+STAT5B-DNM (4), and those expressing BCR/ABL and treated with STI571 (5) were starved from IL-3 for 8 hours. Expression of Id1 and expression and tyrosine phosphorylation of BCR/ABL was assessed by Western analysis. Equal protein loading was confirmed by detection of actin. STAT5 DNA-binding activity was assessed by electrophoretic mobility shift analysis using the INF-γ-activated sequence as a probe. Representative of three independent experiments. B, expression of Id1 in freshly isolated CML patient samples (CP, chronic phase; BC, blast crisis; BMC, bone marrow cells; PBC, peripheral blood cells) was compared with the samples obtained from healthy donors using Human Genome U95Av2 Array (Affymetrix, Santa Clara, CA) as described before ( 21). Columns, log₂ increase compared with normal samples; *, P = 0.02 compared with BC-BMC group. C and D, total cell lysates were obtained from myeloid (C) or lymphoid (D) CML-BC peripheral blood cells preincubated with 1 μmol/L STI571 for 24 hours in the presence of a threshold concentration of IL-3 (C) or IL-3 + IL-6 (D). Id1 and actin were detected by Western analysis (top). BCR/ABL protein expression and inhibition of BCR/ABL kinase activity by STI571 was confirmed by Western analysis with use of anti-ABL and anti–phosphotyrosine (P.Tyr) antibodies, respectively (bottom). Representative of four patients. E, Western analysis of Id1 and actin expression in BaF3 cells and the counterparts transformed by the indicated FTK: B/A, BCR/ABL; T/A, TEL/ABL; T/P, TEL/PDGFβR; T/J, TEL/JAK2; T/T, TEL/TRKC; N/A, NPM/ALK.

To confirm that the Id1 gene represents a BCR/ABL target also in patient-derived leukemia cells, its expression was analyzed in CML patient cells and normal counterparts. Using the database described before, we detected a modest up-regulation of Id1 mRNA expression in CML-CP and CML-BC patient cells harvested from bone marrows; interestingly, overexpression of Id1 in CML-BC cells in peripheral blood was more abundant ( Fig. 1B). Western blot analysis clearly shows that primary leukemia cells of myeloid (CD34⁺ cells) and lymphoid (mononuclear cells) origin express Id1 ( Fig. 1C and D, respectively). Furthermore, this expression is dependent at least partially on the BCR/ABL kinase activity because it was profoundly diminished upon cell treatment with STI571. Similar results were obtained after preincubation of CML-BC myeloid cells for 48 hours without growth factors to eliminate normal cells, which may potentially contaminate the population of leukemia cells (data not shown).

Other members of the BCR/ABL–related family of FTKs, such as TEL/ABL, TEL/PDGFβR, TEL/JAK2, NPM/ALK, but not TEL/TRKC, were also associated with stimulation of Id1 protein expression in BaF3 murine hematopoietic cells in the absence of IL-3 ( Fig. 1E).

Role of Id1 in BCR/ABL–mediated invasiveness. Prevention of cell apoptosis, induction of growth factor independence, and modulation of cell-extracellular matrix interactions are among the major functions of BCR/ABL ( 2, 3). To examine the role of Id1 in these processes, Id1 expression was up-regulated in the parental cells and down-modulated in the BCR/ABL counterparts by infection with pMIG1 retroviral particles carrying Id1 in sense and antisense orientation. Elevated or down-regulated expression of Id1 mRNA and protein levels in GFP⁺ cells was confirmed by semiquantitative RT-PCR and Western analysis, respectively ( Fig. 2A and B, left ).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Role of Id1 in BCR/ABL–mediated invasion. 32Dcl3 (A) and FL5.12 (B) experimental models were used. Western blot (WB) analysis (left, A and B, top boxes) of Id1, Id2, BCR/ABL, and actin in parental cells (1); in cells overexpressing Id1 (2); in cells expressing BCR/ABL (3); in cells expressing BCR/ABL and Id1 antisense (4); and in cells expressing BCR/ABL and A1 antisense (5) after 8-hour starvation from IL-3. Id1 and GADPH mRNAs were detected by semiquantitative RT-PCR in the same conditions (left, A and B, bottom boxes). Matrigel invasion test (middle, A and B) shows the number of cells that passed the matrix membrane while migrating toward IL-3. Adhesion test (right, A and B) shows the number of cells attached to collagen IV–coated surface. Columns, mean from three independent experiments; bars, SD. P < 0.01 compared with group 1 (*) and group 4 (**).

Enhancement of Id1 expression in 32Dcl3 and FL5.12 parental cells to the level seen in the BCR/ABL–positive cells was achieved by infection with pMIG1 retroviral particles containing Id1 cDNA ( Fig. 2A and B, left, compare group 2 with 1). Overexpression of Id1 in parental cells did not mediate any protection from apoptosis in the growth factor (IL-3)–free medium (data not shown); however, greatly increased invasive capability of the infected cells in the in vitro Matrigel invasion test without affecting their ability to adhere to collagen IV ( Fig. 2A and B, middle and right, respectively, compare groups 2 with 1). In concordance, down-regulation of Id1, but not Id2 protein expression, by infection with pMIG1 retrovirus containing Id1 antisense cDNA ( Fig. 2A and B, left, compare group 4 with 3) did not sensitize BCR/ABL–positive cells to apoptosis upon the growth factor withdrawal (data not shown), but reduced their invasiveness without significant modification of the adhesion to collagen IV ( Fig. 2A and B, middle and right, respectively, compare groups 4 with 3). Cells transfected with Id1 antisense cDNA displayed similar levels of BCR/ABL as those transfected with empty plasmid ( Fig. 2A and B, left, compare groups 4 and 3). Moreover, Id1 was not affected in BCR/ABL cells infected with pMIG1 retrovirus carrying A1 antisense-IRES-GFP ( 10), an antisense construct of similar length and base composition to Id1 ( Fig. 2A, left, compare group 5 with 3). Interestingly, Id2 expression seems not to be regulated by BCR/ABL, at least during 8 hours of starvation ( Fig. 2A, left, compare group 4 with 2). Id3 and Id4 are not expressed in these cells ( 26). Altogether, these control experiments confirm the specificity of Id1 antisense effect.

The percentage of apoptotic cells during the invasion test was usually <5% in BCR/ABL–transformed population; Id1 antisense did not change this number as expected. The apoptotic index in parental cells and cells expressing Id1 was usually ∼10% to 15%. This limited apoptosis was probably associated with the protective effect of IL-3 diffusing from the lower chamber to the upper chamber via the pores in the Matrigel membrane.

The homing ability of 32Dcl3 parental cells, 32Dcl3-BCR/ABL cells cotransfected with the control empty pMIG1 (BCR/ABL), pMIG1-Id1 antisense cDNA [BCR/ABL+Id1(AS)], or pMIG1-A1 antisense cDNA [BCR/ABL+A1(AS)] plasmids was determined in SCID mice injected with ³H-labeled cells as described before ( 22). As expected, radioactivity in the spleen and bone marrow from mice injected 24 hours earlier with BCR/ABL and BCR/ABL+A1(AS) cells was ∼2.5-fold higher than that detected in the corresponding organs on mice injected with the parental cells; however, this effect was reduced ∼2-fold in animals injected with BCR/ABL cells displaying down-regulation of Id1 protein ( Fig. 3 ). Interestingly, a strong reduction of cell-associated radioactivity was detected in the brain and lungs, but not in the liver and kidney of the latter mice compared with those injected with BCR/ABL and BCR/ABL+A1(AS) cells. Therefore, it seems that down-regulation of the Id1 protein impairs the ability of BCR/ABL leukemia cells to infiltrate hematopoietic tissues and invade some (brain, lungs) nonhematopoietic organs. The lack of the effect of Id1 down-regulation on homing in the liver and kidneys may be masked by nonspecific entrapment of injected cells in these organs. This speculation is supported by the fact that similar cellular associated radioactivity was detected in the livers and kidneys of mice inoculated with parental, BCR/ABL, and BCR/ABL+A1(AS) cells ( Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Infiltration of hematopoietic and nonhematopoietic organs by BCR/ABL leukemia cells is regulated by Id1. Left, single-cell suspensions were prepared from the indicated tissues of mice 24 hours after injection with [³H]uridine-labeled 32Dcl3 parental cells, BCR/ABL cells, BCR/ABL cells expressing Id1 antisense [BCR/ABL+Id1(AS)], and with those expressing BCR/ABL and A1 antisense [BCR/ABL+A1(AS)]. Tissue-specific, cell-associated radioactivity is shown as the percentage of total cell–associated injected radioactivity. Columns, mean from three mice in each group; bars, SD. *, P < 0.05; **, P < 0.001 compared with BCR/ABL and BCR/ABL+A1(AS) groups. Right, early organ involvement by the BCR/ABL leukemia cells (H&E stains). A, small clusters of the leukemic cells infiltrating gray matter were detected in brain tissue (× 40 magnification). B, framed fragment (×100) illustrates that the infiltrate is parenchymal and contains blood vessels. C, lung tissue showing an infiltrate by the leukemic cells with large, hyperchromatic, and irregular nuclei (×100).

To provide additional evidence that the above described effect of Id1 on homing of leukemia cells is due to invasion of the parenchyma, but not to nonspecific intravascular binding, a rigorous histologic analysis was done shortly after leukemia cell inoculation. Small clusters of blast-type cells were noted in both lung and brain parenchymas in two of five and three of five mice injected with BCR/ABL leukemia cells 3 and 7 days earlier, respectively (total 10 of 20 leukemia-positive organs). At the same time, none of five, and only one of five mice injected with BCR/ABL+Id1(AS) leukemia cells displayed parenchymal involvement in lung and brain (total 2 of 20 leukemia-positive organs). The clusters of leukemia cells were clearly interstitial (extravascular) as determined by their infiltrative nature, lack of surrounding vessel wall, and no admixed red cells ( Fig. 3, right).

Id1 is required for BCR/ABL–mediated leukemogenesis. To determine whether Id1-dependent homing/infiltration activity plays a role in the BCR/ABL–mediated leukemogenesis in vivo, SCID mice were injected with either GFP+ 32Dcl3-BCR/ABL cells cotransfected with the control empty plasmid (BCR/ABL) or the Id1 antisense cDNA [BCR/ABL +Id1(AS)]. All mice inoculated with the BCR/ABL cells succumbed to leukemia in 3.3 ± 0.5 weeks ( Fig. 4, top ). Terminally ill animals contained elevated WBC counts (41.2 ± 20.9 × 10³/mL); >90% of myeloid cells were GFP+. Histologic examination of the internal organs (bone marrow, spleen, liver, lung, kidney, and brain) in the terminally ill animals revealed the presence of myelogenous leukemia in essentially all organs. The extent of involvement ranged from focal to diffuse, the latter frequently with the architectural effacement of the organ and, occasionally, infiltration of the adjacent tissues. There was considerable variability in the degree of cell maturation among the involved sites with the hematopoietic-type organs (bone marrow, spleen, and liver) typically displaying more differentiated cells. Myeloid origin of the blasts was confirmed by the chloroacetoesterase staining. As compared with the BCR/ABL cell-injected mice, animals inoculated with the BCR/ABL+Id1(AS) cells displayed twice as long survival (6.5 ± 1.4 weeks, P < 0.05). Elevation of WBC counts (36.5 ± 8.3 × 10³ cells/mL) was detected in these terminally ill recipients; again, >90% of myeloid cells were GFP+.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Role of Id1 in BCR/ABL–mediated leukemogenesis. Top left, SCID mice (20 per group) were injected i.v. with 10⁶ BCR/ABL cells (▴) and BCR/ABL+Id1(AS) cells (○). Survival of the animals was monitored weekly. Top right, Western analysis of GFP+ leukemia cells sorted from spleen (SPL), bone marrow (BMC), and liver of terminally ill animals injected with BCR/ABL cells (E) or BCR/ABL+Id1(AS) cells (AS). Bottom, histology of the central nervous system (A and B) and lungs (C and D) harvested from the mice injected with BCR/ABL or BCR/ABL+Id1(AS) cells (H&E stains). A, note an extensive involvement (arrows) of the meninges and brain parenchyma by the leukemic-type cell infiltrate (×5 magnification). Inset, morphologic detail of the infiltrating, poorly differentiated myeloid cells (×100). B, focal perivascular brain involvement (arrow; ×5) by acute leukemic-type cells. Inset, close-up of the infiltrate (arrow; ×40). C, note the extensive peribronchial involvement of the lung tissue by acute leukemic-type cellular infiltrate (arrows). D, focal peribronchial involvement of the lung parenchyma by acute leukemia-type cellular infiltrate (arrows). Inset, close-up of the foci (×40).

GFP+ BCR/ABL+Id1(AS) cells sorted from single cell suspensions of spleen, bone marrow, and liver displayed inhibition of Id1 expression compared with BCR/ABL counterparts ( Fig. 4, top right); in vitro growth rate of both cell populations placed in liquid culture medium was similar (data not shown).

Histologically, there was a similar degree of cell maturation and organ involvement with the striking exception of the central nervous system and, possibly, the lungs. Ten of 11 BCR/ABL+E–injected mice developed central nervous system (CNS) involvement, in contrast to only 4 of 12 BCR/ABL+Id1(AS) mice. Furthermore, in seven of the BCR/ABL mice, the involvement was extensive and clearly infiltrative, whereas all four BCR/ABL+Id1(AS) mice displayed only focal, mostly meningeal involvement ( Fig. 4, bottom, A and B). Lungs were involved in the leukemia process in all injected animals. However, only 6 of 12 mice injected with BCR/ABL+Id1(AS) cells displayed extensive infiltration of the lungs, compared with 8 of 11 of those injected with BCR/ABL cells ( Fig. 4, bottom, C and D).

Higher CNS infiltration in mice injected with BCR/ABL cells observed in this work compared with that published before ( 9) may depend on the differential disease stage; samples were collected from the moribund animals and from those developing leukemia, respectively.

Id1 induces MMP9 expression and enhances invasiveness of the BCR/ABL–transformed cells. To determine the role of Id1 in the BCR/ABL–mediated induction of MMP9 expression ( 27, 28) and functional activity, two types of experiments were done: transactivation studies with the use of MMP9 promoter and MMP9 gelatinase assay. Ectopic expression of Id1 or BCR/ABL greatly enhanced transactivation of the MMP9 promoter, whereas expression of Id1 antisense cDNA abrogated BCR/ABL–dependent transactivation ( Fig. 5A ). The zymography assay done to determine the MMP9 enzymatic activity was conducted using conditioned medium from the above-described 32Dcl3 cell populations: parental and those transfected with Id1 cDNA, BCR/ABL–positive, or BCR/ABL–positive and transfected with Id1 antisense cDNA (see Fig. 2A). As shown in Fig. 5B, MMP9 gelatinase activity was stimulated by the enhanced expression of Id1 (compare group 2 with 1) or the expression of BCR/ABL (compare group 3 with 1). In turn, inhibition of Id1 in BCR/ABL–positive cells decreased the activity of MMP9 by ∼5-fold (compare group 4 with 3). Gelatinase activity of another metalloprotease, MMP2, was not detectable, in accordance with other studies indicating that CML cells display this activity rarely and at a low level ( 27, 28).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

BCR/ABL→Id1 pathway induces MMP9. A, transactivation of the MMP9 promoter was measured by luciferase assay in Tk-ts13 cells transiently transfected with empty plasmid (1), or constructs carrying Id1 (2), BCR/ABL (3), or BCR/ABL + Id1(AS) (4) along with the luciferase reporter gene driven by the MMP9 promoter. Luciferase activity is expressed in arbitrary units. *, P < 0.05 compared with group 1; **, P < 0.05 compared with group 1 and 4. Id1 and actin proteins were detected by Western blotting. Id1 and GADPH mRNAs were measured by semiquantitative RT-PCR. B, MMP9 activity in the conditioned medium from 32Dcl3 parental cells (1) and in cells expressing Id1 (2), BCR/ABL (3), and BCR/ABL + Id1(AS) (4) was assessed by zymography. Control (C) includes conditioned medium from HT-1080 fibrosarcoma cells as a standard. Representative of three independent experiments.

To examine the role of MMP9 in BCR/ABL-Id1–dependent stimulation of cell invasion, a gelatinase inhibitor, ilomastat, and the anti-MMP9 neutralizing antibodies were used. Ilomastat, a broad-spectrum inhibitor of MMPs, inhibited the invasion of myeloid 32Dcl3-BCR/ABL and lymphoid BaF3-BCR/ABL cells through the Matrigel membrane ( Fig. 6A and B , respectively, left, see group 2). Moreover, ilomastat reduced the invasive capability of 32Dcl3 parental cells overexpressing Id1 ( Fig. 6A, left, group 3). To confirm that MMP9 is responsible for BCR/ABL–dependent stimulation of invasion through Matrigel membrane, we used next an inhibitory antibody specific for MMP9. Similar to ilomastat, anti-MMP9 antibody also inhibited the BCR/ABL– and Id1-dependent cell invasion ( Fig. 6A and B, right, see group 2 and 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

MMP9 stimulates the invasiveness of BCR/ABL leukemia cells. The invasive abilities of cells were examined using Matrigel invasion test in the absence (control, gray columns) or presence (overlaying black columns) of gelatinase inhibitor ilomastat or neutralizing anti-MMP9 antibodies, as indicated. Controls contained diluent and nonimmune antibodies, respectively. A, 32Dcl3 parental cells (1), BCR/ABL cells (2), and Id1 overexpressing cells (3). B, BaF3 parental (1) and BCR/ABL cells (2). C, CD34⁺ cells from normal donors (1), two myeloid CML-BC patients (2 and 3), and peripheral blood mononuclear cells from two lymphoid CML-BC patients (4 and 5) were used. Columns, mean number of cells that passed the matrix membrane while migrating toward IL-3 (A and B) and IL-3 + IL-6 (C) in three independent experiments; bars, SD. *, P < 0.05 and **, P < 0.01 compared with the untreated group.

CD34⁺ cells from peripheral blood of two myeloid CML-BC patients ( Fig. 6C, groups 2 and 3) and mononuclear cells from two lymphoid CML-BC patients ( Fig. 6C, groups 4 and 5) displayed higher invasive capability compared with CD34⁺ cells from peripheral blood of normal healthy donors ( Fig. 6C, group 1, P < 0.05 compared with the cumulative results in groups 2-5). Importantly, ilomastat and MMP9-neutralizing antibody reduced the invasion of primary cells through Matrigel membrane ( Fig. 6C, compare black columns with corresponding gray columns).

Discussion

BCR/ABL confers growth factor independence and resistance to apoptosis ( 1– 3). Furthermore, the BCR/ABL–expressing cells accumulate not only in the patient hematopoietic organs but, over time, also infiltrate other tissues such as the CNS, testes, skin, and gastrointestinal tract. We have shown previously that BCR/ABL–mediated activation of a small GTP-binding protein Rac plays an important role in invasiveness ( 29). It has been proposed by others that Rac may work through modulation of the cytoskeletal proteins and integrins ( 30), which are important for BCR/ABL leukemogenesis ( 31, 32). In addition, reduced expression of α2-integrin in cells expressing BCR/ABL mutant lacking SH3 domain (BCR/ABLΔSH3) was associated with decreased interaction with collagen IV and laminin, and invasion through Matrigel membrane ( 22). Therefore, α2-integrin and/or Rac-mediated regulation of cell invasiveness seem to depend on modulation of the cell-extracellular matrix adhesion and cell plasticity because inhibition of cell adhesion to the basement membrane components results in reduced tissue invasion by the malignant cells.

Here, we identified another mechanism stimulating extramedullar spread of BCR/ABL–positive leukemias. BCR/ABL–STAT5 signaling pathway is responsible for the enhanced expression of a helix-loop-helix transcription regulator Id1 leading to the overexpression of MMP9 metalloproteinase and enhanced capability to digest the basement membrane components. Altogether, it seems that BCR/ABL can stimulate cell invasiveness by at least three mechanisms: (a) α2-integrin affecting binding to Matrigel substrates ( 21); (b) Rac1-mediated modulation of the cytoskeleton and integrins, perhaps changing the cell plasticity and/or binding capability ( 29); and (c) Id1-MMP9–mediated digestion of the basement membrane components (this article). The fact that cells expressing BCR/ABLΔSH3 mutant have impaired invasion ability ( 21) while expressing Id1 (data not shown) suggests that α2-integrin–mediated adhesion to the Matrigel membrane may play an important role in the subsequent Id1-MMP9–directed invasion.

Id1 protein has been described previously as a downstream effector of STAT5 ( 9, 12) and its role in tissue invasion/metastasis by malignant cells is well documented ( 15– 17). BCR/ABL–dependent increase in Id1 mRNA and protein expression in BCR/ABL–transformed cell lines and CML patients was detected using genome-wide screen ( 21) and Western analyses (this work), respectively. Down-regulation of Id1 in cytokine-independent BCR/ABL–positive 32Dcl3 cells and overexpression of Id1 in parental cells did not affect their response to growth factor withdrawal. Thus, it seems unlikely that Id1 may be responsible for the growth factor independence in BCR/ABL–transformed cells. This function is likely carried on by two other STAT5 downstream effectors: A1 and pim-1 ( 10). However, Id1 may exert yet unknown function(s) essential for early stages of BCR/ABL–mediated transformation. This speculation is based on the observation that we have been unable to stably transform bone marrow cells harvested from Id1−/− mice (ref. 33; obtained from Dr. R. Benezra, Memorial Sloan-Kettering Cancer Center, New York, NY) with BCR/ABL (data not shown). Because the loss of Id1 promotes cellular senescence associated with increased expression of the cell cycle inhibitor p16/Ink4a ( 34), we hypothesize that expression of BCR/ABL kinase may generate a stress signal, thus accelerating cell senescence and/or inducing apoptosis. Freshly transformed BCR/ABL–positive cells may not achieve growth factor independence initially ( 35); conversely, they sometimes display apoptosis ( 36). Therefore, “oversensitive” Id1−/− cells may be unable to survive a “crisis” stage associated with the ABL kinase–mediated transformation ( 37) and develop into fully transformed cells.

Although down-regulation of the Id1 protein expression neither affects growth factor independence nor adhesion to collagen IV, it impairs the invasive capacity of the BCR/ABL–transformed cells. The process of invasion consists of two events: adhesion to the basement membrane and actual penetration through the membrane ( 38). The first phenomenon is dependent on adhesion molecules, whereas the second phenomenon should require the Id1-MMP9 axis. In accordance, down-regulation of Id1 did not disturb adhesion of BCR/ABL+Id1(AS) cells to collagen IV while inhibiting invasion through the Matrigel membrane. On the other hand, overexpression of Id1 increased the invasive capability of cells; again, adhesion to collagen IV was not affected. These results support the notion that Id1 regulates invasion but not adhesion.

Interestingly, genome-wide screening revealed that CML-BC peripheral blood cells express higher levels of Id1 mRNA than those in the bone marrow, whereas CML-CP and CML-BC bone marrow cells display similar modestly up-regulated levels of Id1. Elevation of Id1 in CML-BC peripheral blood cells depended on the BCR/ABL kinase activity (see Fig. 1C and D), but was not accompanied by an increase in BCR/ABL expression (data not shown), implicating additional mechanism(s), e.g., deregulated protein degradation. Altogether, we speculate that whereas elevation of Id1 is a feature of CML as a whole, its overexpression in peripheral blood leukemic blasts seem to be associated with gain-of-function rather than progression from chronic phase to blast crisis. In concordance with this hypothesis, inhibition of Id1 expression affected the ability of BCR/ABL cells to penetrate through the Matrigel membrane in vitro and infiltrate hematopoietic (bone marrow, spleen) and some nonhematopoietic (brain, lung) organs. This effect was associated with delayed leukemogenesis in mice. BCR/ABL–transformed 32Dcl3 cells rather than bone marrow cells were used here to avoid the potential negative effect of down-regulation of Id1 on the proliferation of primary cells (as discussed before). In addition, mice that succumbed to leukemia induced by BCR/ABL cells expressing low levels of Id1 displayed reduced disease involvement in the brain and lungs compared with those inoculated with BCR/ABL cells expressing high levels of Id1. Thus, Id1 may play an important role in leukemogenesis by regulating the ability of leukemia cells to penetrate through basement membrane and infiltrate various organs. In addition, by promoting extramedullar localization of leukemia cells, Id1 can reduce the effectiveness of antileukemia drugs, e.g., imatinib mesylate ( 39). Because Id1 expression is increased mainly in CML-BC cells from peripheral blood, the above speculations may be most relevant to these cells.

GFP+ BCR/ABL+Id1(AS) leukemia cells harvested from bone marrow, spleen, and liver of terminally ill animals showed sustained inhibition of Id1 expression. This observation implicates that other mechanisms, in addition to Id1-MMP9 axis, play a role in the invasiveness of BCR/ABL leukemia cells. For example, decreased adhesion to stroma and fibronectin but increased adhesion to the basement membrane components laminin and collagen IV may underlie abnormal trafficking of BCR/ABL leukemia cells ( 40). Thus, a combination of mechanisms regulating adhesion and penetration through basement membranes and connective tissue components may determine the invasive properties of leukemia cells.

The role of Id1 in regulating invasiveness may not be limited to Ph chromosome–positive leukemias. We show here that BCR/ABL–related FTKs, such as TEL/ABL, TEL/JAK2, TEL/PDGFβR, NPM/ALK, but not TEL/TRKC, also stimulated Id1 expression. Low level of Id1 in cells transformed by TEL/TRKC is accompanied by the lack of activation of STAT5 and stimulation of its downstream effectors, such as Bcl-xL and RAD51 ( 24, 41).

Id1-mediated stimulation of the invasive capacity of breast carcinoma cells was associated with overexpression of the 120 kDa gelatinase ( 16), which may consist of MMP9 and associated lipocalin neutrophil gelatinase–associated lipocalin ( 18). Noteworthy, we show for the first time that Id1 induces transactivation of the MMP9 promoter leading to increase of the gelatinase activity. In concordance, other investigators showed MMP9 expression and activity in CML cells ( 27, 28, 42). Interestingly in this context, our earlier studies detected higher levels of MMP9 mRNA in CML-BC peripheral blood cells compared with the ones from bone marrow ( 21), concomitant with high levels of Id1 in the former cells. Hence, MMP9 may facilitate the release of leukemia cells from bone marrow to peripheral blood, in accordance with its role in tumor cell invasion ( 20). In support of this hypothesis, we show that MMP9 is responsible for enhanced invasion of the basement membrane by CML-BC myeloid and lymphoid peripheral blood cells and BCR/ABL–transformed cell lines compared with normal counterparts. MMP9 was also implicated in the penetration of normal CD34⁺ cells from the bone marrow to peripheral blood ( 43, 44).

BCR/ABL–transformed myeloid and lymphoid cell lines and patient cells displayed activation of STAT5 ( 45), overexpression of Id1, and MMP9-dependent invasion of the Matrigel membranes (this article). In addition, expression of a constitutively active STAT5B mutant or overexpression of the Id1 protein enhanced invasive properties of hematopoietic cell lines (data not shown and this work, respectively). Therefore, it is conceivable that BCR/ABL-STAT5-Id1-MMP9 pathway plays a role in the leukemia cell trafficking. This capacity, along with a variety of BCR/ABL–mediated aberrations of the cell adhesion/migration (e.g., decreased adhesion to stroma and fibronectin combined with increased adhesion to the basement membrane components laminin and collagen IV) and an increase in spontaneous motility, may greatly contribute to leukemia course and outcome ( 22, 29, 31, 40, 46– 48). Accordingly, the integrin-dependent increase of adhesion to the basement membrane followed by Rac-modulated cell plasticity and MMP9-mediated digestion of the membrane components could represent an important mechanism in dissemination of the BCR/ABL–transformed cells.