Matrix Metalloproteinase Production by Bone Marrow Mononuclear Cells from Normal Individuals and Patients with Acute and Chronic Myeloid Leukemia or Myelodysplastic Syndromes¹

Leukemia can be defined as the uncontrolled proliferation or expansion of hematopoietic cells that do not retain the capacity to differentiate normally to mature blood cells (1). In addition, AML⁴ and CML in blast crisis are characterized by an excessive egress of leukemic cells from bone marrow into peripheral blood, followed by an infiltration of organs, skin, or mucous membranes. These movements, which are normally restricted to functional mature leukocytes, include the necessity for the cells to cross matrix barriers and penetrate blood vessel walls, which depends on the catalytic modification of ECM and basement membranes.

The MMP multigene family of structurally related enzymes currently includes more than 15 members that are collectively capable of digesting almost all components of ECM. Two of them, the M_r 72,000 type IV collagenase (MMP-2) and the M_r 92,000 type IV collagenase (MMP-9), are subclassified as gelatinases, which preferentially degrade denatured collagens (gelatin) and intact collagen type IV. Under physiological conditions, MMP expression, secretion, and activity are highly regulated. A variety of biochemical stimuli, including cytokines, growth factors, hormones, or components of the ECM, modulate the transcription of MMPs (2). Once released from the cell, proteolytic activity is controlled by activation of the latent proenzymes and inhibition of the activated proteinases by a family of naturally occurring and highly specific inhibitor proteins termed TIMPs, frequently cosecreted from MMP-producing cells. Pro-MMP activation is catalyzed by various extracellular proteinases such as plasmin, urokinase, or elastase as well as intracellular furin in certain cases. Together, these enzymes as well as their inhibitors are thought to build up a complex proteolytic cascade generating activated forms of MMPs (3). For MMP-2, a characteristic cell surface activation mechanism is provided by interaction with membrane-type MMP (MT1-MMP), a specialized MMP that is located in the plasma membrane of normal and neoplastic cells acting both as MMP-2 receptor and activator (4).

Due to their effective matrix modulating capacities, MMPs are implicated not only in numerous normal but also in pathological processes, such as tumor invasion and metastasis (2, 5, 6). Correlative evidence is abundant for the involvement of MMP-2 and MMP-9 in the invasive phenotype of various malignant neoplasias, including tumors of the colon, breast, ovary, kidney, or skin (7, 8, 9, 10, 11, 12, 13). Direct evidence for a critical role of MMP-9 in tumor cell invasion has been obtained by transfection of tumorigenic but noninvasive rat cells with a MMP-9 expression vector that conferred metastatic behavior to these cells (14). Recently, soluble MMP-2 has been reported to be able to bind to the surface of invasive cells in vitro and in vivo by interaction with the integrin receptor αvβ3, thereby promoting the invasive capacities of these cells (15). The significance of gelatinases for invasive processes is thought to be due to their capacity to degrade collagen type IV, the major component of basement membranes, which normally represents a nonpermeable barrier for migrating cells.

Mature leukocytes, such as neutrophil granulocytes, monocytes/macrophages, and T lymphocytes, are potent producers of MMP-2 or MMP-9 (16, 17, 18, 19). The capacity for gelatinase production is thought to enable leukocytes to cross ECM barriers to reach their target tissue at the sites of inflammation (20). Accumulating data from recent studies indicate the involvement of MMPs and TIMPs in growth and progression of lymphoid neoplasias (21, 22, 23, 24, 25). In contrast, comprehensive analysis of MMP or TIMP expression in cells of the myeloid system and their implication in hematological malignancies has not been carried out. It can be hypothesized that leukemic cells have illicitly acquired the biochemical machinery for trafficking as used by their mature counterparts. Genetic alterations and regulatory dysfunctions in MMP production may facilitate leukemic cells to prematurely leave the bone marrow and to invade peripheral tissues (26). We previously reported the constitutive secretion of MMP-9 from HL-60 and NB4 leukemic cells to be driven by an autocrine stimulatory mechanism (27, 28, 29). Recently, leukemic blast cells purified from the peripheral blood of patients with AML have been documented to continuously release MMP-2 and MMP-9 (30). Due to the limited accessibility of bone marrow and its complex cellular composition, no data exist about MMP expression in hematopoietic cells isolated from the bone marrow. The purpose of this study was to examine the production of MMP-9, MMP-2, MT1-MMP, and TIMPs in mononuclear cells from the bone marrow of healthy individuals compared to that of patients with different leukemic or preleukemic diseases.

Bone marrow aspirates were obtained from 5 healthy volunteers and 49 patients with de novo or secondary AML, AML in remission, CML, or MDS at the time of analysis. The criteria for the diagnosis were based on clinical symptoms, differential cell counts (blast count in bone marrow >30% for primary and secondary AML), cell morphology after differential cell staining and the FAB classification in the case of AML and MDS. This study was approved by the Institutional Review Board of Munich University Medical School. All samples were collected at the University Hospital of Munich with consent of the patients. BM-MNCs were purified from aspirates by the use of Ficoll-Hypaque (Biochrom KG, Berlin, Germany) density gradient centrifugation. To control the enrichment of BM-MNC achieved by this method, we performed flow cytometry analysis on selected samples using CD45, CD15, and CD14 as markers. After purification with Ficoll, one portion of BM-MNCs from each individual was directly used for RNA isolation, and the other cell portion was washed twice and cultured at a cell density of 1.5 × 10⁶ cells/ml under serum-free conditions in RPMI 1640 (Sigma Chemical Co., Munich, Germany) supplemented with Nutridoma-SP (Boehringer Mannheim GmbH, Mannheim, Germany) in the presence of antibiotics. After 96 h, conditioned media were harvested by centrifugation and subsequently stored at −20°C until use.

Zymography was used as a sensitive method for the determination of actually secreted gelatinases in cell culture supernatants, as described previously (27). In brief, SDS-PAGE was performed on horizontal ultrathin gradient gels (Pharmacia, Uppsala, Sweden) that contained gelatin type I as a substrate at a final concentration of 1.5 mg/ml. Ten μl of the BM-MNC-conditioned medium sample were mixed with 10 μl of loading buffer (5% SDS) and run under nonreducing conditions without prior boiling. After electrophoresis, gels were washed in 2% Triton X-100 to remove SDS and allow proteins to renature. Then the gels were immersed in buffer containing 50 mm Tris (pH 7.5), 5 mm CaCl₂, 1 μm ZnCl₂, and 0.01% (w/v) NaN₃ for 24 h at 37°C. To inhibit metalloproteinase activity, we added EDTA at a final concentration of 5 mm to the incubation buffer. The zymograms were stained with 0.4% (w/v) Coomassie Blue R-250 (Sigma) and destained in 35% ethanol-10% acetic acid. Clear zones of gelatin lysis against a blue background stain indicated enzyme activity. For qualitative assessment of gelatinolytic activity in BM-MNC-conditioned medium, we did not dilute samples prior to zymography, whereas purified gelatinase standards samples had to be diluted appropriately for comparison. For the determination of the molecular masses of gelatinolytic enzymes in zymograms, commercially available purified M_r 72,000 type IV collagenase (MMP-2) and M_r 92,000 type IV collagenase (MMP-9; both from Boehringer Mannheim) or MMP-2 from foreskin fibroblasts and MMP-9 from NB4 promyelocytic leukemia cells (29) as well as standard molecular mass marker proteins (Bio-Rad, Munich, Germany) were used. Enzyme activity determined by zymographic analysis actually represents the total amount of secreted MMP-2 and MMP-9 because this technique allows the gelatinases to be separated from potentially cosecreted inhibitors during electrophoresis, and subsequent treatment with Triton X-100 leads to a regeneration of proteolytic activity together with an activation of latent proenzymes in the gel.

SDS-PAGE was performed as described for zymography with the modification, that the gel was polymerized on Net-Fix for PAG (Serva, Heidelberg, Germany) in the absence of gelatin. BM-MNC-conditioned medium samples were concentrated ∼3-fold for MMP-2 detection by the use of microsep ultrafiltration devices (Pall Filtron, Dreireich, Germany). Samples (40 μl) were mixed with loading buffer and separated either under reducing conditions in the presence of dl-DTT with prior boiling or under nonreducing conditions without boiling. After electrophoresis, proteins were transferred to polyvinyl difluoride membranes (Pall Filtron) using a semidry blotting apparatus (Pharmacia) and probed with mouse monoclonal antibodies to MMP-2 (0.4 μg/ml) or MMP-9 (0.2 μg/ml), followed by incubation with peroxidase-labeled secondary antibodies (all antibodies were purchased from Amersham, Braunschweig, Germany). Detection was performed by the use of an chemiluminescence system (Amersham) according to the manufacturer’s instructions.

Total RNA was isolated as described (29). Briefly, up to 5 × 10⁶ BM-MNCs were centrifuged and lysed in 500 μl of lysis buffer containing 4 m guanidinium thiocyanate, 25 mm sodium citrate (pH 7.0), 0.5% (w/v) sarcosyl, and 14.5 mm DTT. RNA was then extracted with phenol and precipitated with isopropyl alcohol. The RNA pellet was washed in ethanol and stored at −20°C.

RNA (10 μg/sample) was glyoxylated and run on a 1% agarose gel. Then the RNA was transferred to nylon membranes (Hybond-N; Amersham) and immobilized by baking at 80°C for 2 h. The membranes were prehybridized with denatured salmon sperm DNA at 48°C for 2 h. The following specific DNA probes were used in this study: a 2.5-kb XbaI cDNA fragment of MMP-9 (kindly provided by Dr. G. Goldberg, Washington University School of Medicine, St. Louis, MO), a 1.1-kb EcoRI fragment for MMP-2 (PH3A kindly provided by Dr. W. Stetler-Stevenson, National Cancer Institute, Bethesda, MD), a 4.05-kb genomic XbaI fragment for TIMP-1 (ATCC 59666) and a 780-bp XbaI/PstI fragment of GAPDH (ATCC 57090). These probes were radiolabeled with [α-³²P]dCTP by the random priming method (Amersham). The membranes were hybridized to these probes overnight and washed as described (29).

For the quantitative evaluation of relative levels of MMP-9 and TIMP-1 mRNA expression, a Fuji Bas 1000 phosphoimager and appropriate evaluation software were used. Values were given in PSL. To allow comparison of signals between different Northern blots, we included identical RNA samples on each blot as a standard. Equivalence of RNA loading was achieved by standardization against the corresponding GAPDH signal after stripping the blot in 5 mm Tris (pH 8) containing 2 mm EDTA and 0.1× Denhardt’s solution at 70°C for 1 h and subsequent rehybridization with the GAPDH probe.

cDNA was synthesized from 1 μg total RNA at 40°C for 1 h using 0.2 μg of oligo(dT) primers and 20 units of Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc., Eggenstein, Germany), 50 mm Tris-HCl (pH 8.3), 75 mm KCl, 3 mm MgCl₂, 10 mm DTT, 1 unit of RNasin, and 1 mm each dNTP in a total volume of 20 μl. PCRs were performed using specific primers for MMP-2, TIMP-2, MT1-MMP, and GAPDH, as described by Onisto et al.(31) and Yamamoto et al.(32).

The primer sequences were as follows: for MMP-2, sense, 5′-CCA CGT GAC AAG CCC ATG GGG CCC C-3′, and antisense, 5′-GCA GCC TAG CCA GTC GGA TTT GAT G-3′; for TIMP-2, sense, 5′-TGC AGC TGC TCC CCG GTG CAC-3′, and antisense, 5′-TTA TGG GTC CTC GAT GTC GAG-3′; for MT1-MMP, sense, 5′-CCC TAT GCC TAC ATC CGT GA-3′, and antisense, 5′-TCC ATC CAT CAC TTG GTT AT-3′; and for GAPDH, sense, 5′-CGG GAA GCT TGT GAT CAA TGG-3′, and antisense, 5′-GGC AGT GAT GGC ATG GAC TG-3′. These primer pairs amplified PCR fragments for MMP-2, TIMP-2, MT1-MMP, and GAPDH of 450, 590, 550, and 350 bp, respectively.

PCR was carried out in a 50-μl volume containing 2 μl of cDNA template, 1× PCR buffer, 0.5 μm each primer, 0.2 mm dNTPs, 1.5 mm MgCl₂, and 1.25 units of Taq polymerase (Promega, Mannheim, Germany). After a 5-min period of denaturation at 94°C, reaction mixtures were subjected to 35 cycles of PCR amplification in a Perkin-Elmer GeneAmp 2400 Thermal Cycler (Applied Biosystems GmbH, Weiterstadt, Germany). Each cycle consisted of 1 min of denaturation at 94°C, a primer-specific annealing temperature and period (60°C for 1.5 min for MMP-2; 60°C for 1 min for TIMP-2; 58°C for 1 min for MT1-MMP; and 60°C for 1 min for GAPDH), and primer extension at 72°C (1.5 min for MMP-2 and TIMP-2 and 1 min for MT1-MMP and GAPDH). The PCR products were visualized as single bands after electrophoresis on a 2% agarose gel and staining with ethidium bromide. Total RNA isolated from the human fibrosarcoma cell line HT1080 was used as a positive control for amplification of MMP-2, TIMP-2, and MT1-MMP.

The χ² test was used to compare the presence of MMP-2 in BM-MNC with the presence of leukemic blasts in bone marrow or peripheral blood. The potential of zymographic MMP-2 determination for the discrimination of myeloid leukemias in acute phases characterized by blasts counts in bone marrow of >30% from that with <30% blasts in marrow was assessed by calculating its sensitivity, specificity, positive predictive value, and diagnostic benefit. The latter was calculated by the difference between the prior test probability and the posterior test probability. All calculations were performed using StatView Version 4.51 (Abacus Concepts Inc., Berkeley, CA).

In the course of the study, we examined bone marrow samples of 5 healthy individuals and of a total of 49 patients with primary AML M1–M5 (n = 11), secondary AML (n = 8), AML in remission (n = 5), CML (n = 17), and MDS (n = 8; Tables 1,2,3,4). For the analysis of gelatinase secretion, cell culture supernatants from BM-MNC of all patients and normal persons were examined by zymography. We detected three major gelatinolytic activities of M_r ∼72,000, 80,000, and 92,000, respectively, as well as numerous minor activities of molecular masses between 130 and 200 kDa in these samples (Fig. 1). Addition of EDTA to the incubation buffer during the development of the zymogram inhibited all of these activities (data not shown), indicating them to belong to the metalloproteinase subclass of enzymes.

Due to their molecular mass and the comigration with purified enzymes from other cell species in zymograms, we suspected that the M_r 72,000 and 92,000 gelatinases secreted from BM-MNCs represent MMP-2 and MMP-9, respectively (Fig. 1). To verify this assumption, we performed Western blotting using monoclonal antibodies to MMP-2 and MMP-9. In representative samples from patients with AML, MDS, and CML, each containing both the M_r 72,000 and 92,000 gelatinolytic activity, we identified these enzymes as the proenzyme forms of MMP-2 (Fig. 2,A) and MMP-9 (Fig. 2 B), respectively. The identities of the other gelatinolytic activities present in culture supernatants of BM-MNCs remain to be determined.

MMP-9 was found to be present in BM-MNC conditioned media of all 49 patients as well as the 5 healthy persons, as determined by zymography (Tables 1,2,3,4).

In contrast to MMP-9, MMP-2 activity was absent in healthy individuals (Table 1). In AML, MMP-2 was detected in 8 of 11 (73%) patients with FAB M1–M5 at the time of analysis, 7 of 8 (88%) patients with a secondary AML, and only 1 of 5 (20%) AML patients in clinical remission (Table 2). The presence of MMP-2 and MMP-9 in BM-MNC preparations from patients with blast counts in marrow of up to 92% additionally enriched by purification with Ficoll, indicates that both gelatinases are released from the leukemic blast cells in AML. In CML, only 2 of 17 (12%) cases were MMP-2 positive (Table 3). One of the MMP-2-positive CML patients (patient 41) had already entered blast crisis at the time of analysis, and the other patient (patient 40) went into blast crisis 2 months after analysis, whereas the MMP-2-negative CML patients showed no acceleration of their disease within this period. MMP-2 was significantly correlated with the diagnosis of primary or secondary AML (15 of 19, 79%) and CML in blast crisis (1 of 1, 100%) as compared to AML in remission (1 of 5, 20%) and CML in chronic phase (1 of 16, 6%; P < 0.0001). In MDS, 3 of 8 (38%) patients were positive for MMP-2 (Table 4). From these three individuals one patient (patient 48) developed a secondary AML 5 months after analysis, another patient (patient 47) underwent transition from RAEB to RAEB in transformation after 12 months. No progression of disease could be observed in the five MMP-2-negative persons with MDS.

The zymographic detection of MMP-2 release from BM-MNC in primary/secondary AML (patients 1–19) and CML (patients 25–41) provides a method for the discrimination of acute clinical phases characterized by blast counts in bone marrow of >30% from that <30%, with a positive predictive value of 94%, a sensitivity of 80%, a specificity of 93% and a diagnostic benefit of 39% (P < 0.0001). For a correlation between the detection of MMP-2 in bone marrow and the presence of malignant blasts (>0%) in peripheral blood, we compared all patients with AML, CML, or MDS, where the blast count in peripheral blood was known (n = 46). In 15 of 21 (71%) MMP-2-positive cases but also in 13 of 25 (52%) MMP-2-negative cases, leukemic blasts were found in the circulation. The chromosomal location of the MMP-2 gene is 16q21. We did not observe a correlation between cytogenetic chromosomal abnormalities in BM-MNC of AML patients and a constitutive MMP-2 release in these cells.

MMP-9 and TIMP-1 production was assayed by semiquantitative Northern analysis in BM-MNCs directly after isolation from marrow aspirates. MMP-9 mRNA could be detected in all samples of normal persons and patients with AML, CML, or MDS, which is in accordance with the results obtained by zymography (Tables 1,2,3,4). The MMP-9 expression rates in BM-MNC from healthy persons and patients with CML were higher than those of AML or MDS patients (Tables 1,2,3,4). This correlates to the enzyme activities in zymograms, as obtained by semiquantitative visual determination. The TIMP-1 transcription level was low or moderate in normal individuals and patients with AML, CML, or MDS without any correlation to the type or stage of disease (Tables 1,2,3,4). MMP-2 mRNA could not be detected in any of the samples by Northern hybridization. This failure may be explained by low MMP-2 transcription levels in BM-MNCs, which is confirmed by the comparatively weak activity bands to be observed in zymograms (Fig. 1, B–D).

When we used RT-PCR for mRNA amplification, MMP-2 transcripts became detectable in the BM-MNC fractions of 20 of 22 representative patient samples and in all healthy donors (Table 5; Fig. 3). TIMP-2 was also expressed in the majority of the samples from healthy persons (4 of 5) and patients (18 of 22; Table 5; Fig. 3). MT1-MMP transcripts were found in BM-MNC from healthy individuals (four of five), primary AMLs (three of five), MDS (three of four), and both patients with AML in remission (two of two), but in none of the samples from patients with secondary AML (zero of six) or CML (zero of five; Table 5; Fig. 3).

This study represents the first analysis of MMP production in human ex vivo bone marrow cells. Examination of gelatinase secretion revealed a constitutive expression and release of MMP-9 in normal BM-MNC from healthy individuals. Due to the heterogeneity in the cellular composition of the BM-MNC fraction, the hematopoietic cell population(s) responsible for the production of this enzyme are unknown. Continuous secretion of MMP-9 can be observed in mature monocytes/macrophages and T-lymphocytes as well as malignant B-lymphocytes (17, 18, 33). Neutrophil granulocytes also synthesize MMP-9. Enzyme production in these cells starts at the stage of myelocyte/metamyelocyte differentiation. However, MMP-9 is stored in granules and released from mature neutrophils only after appropriate stimulation (16, 34). Hitherto, no information is available about MMP secretion in normal hematopoietic cells isolated from bone marrow.

We detected MMP-9 expression and secretion not only in all BM-MNC samples from normal individuals but also in samples from all 49 patients investigated during this study. The MMP-9 transcription level in these cells showed no correlation to the blast count in marrow or blood and FAB classification but was lower in patients with AML compared to healthy persons and patients with CML in chronic phase. This may be explained by the presence of MMP-9-releasing mononuclear phagocytes and lymphocytes in the BM-MNC fraction of healthy individuals and patients with CML compared to that of AML patients consisting predominantly of leukemic blasts.

In addition to MMP-9, we detected MMP-2 enzyme release in BM-MNC from various patients with myeloid leukemias and preleukemias but not in samples from healthy individuals. Cell fractions from AML patients with up to 92% blasts in marrow consisted almost exclusively of leukemic blasts after additional enrichment by purification with Ficoll. The presence of MMP-2 and MMP-9 in these samples strongly suggests that both enzymes were released from the leukemic blast cell population. This is in accordance with findings on AML blasts isolated from peripheral blood of patients with de novo AML. These cells have also been shown to continuously secrete MMP-9 and MMP-2 (30). The assumption that MMP-2 is produced by leukemic cells in bone marrow is confirmed by the lack of this enzyme in samples from healthy persons and AML patients being in clinical remission. Although immunoreactive MMP-2 was found to be released from leukemic blasts during cultivation, we failed to detect specific mRNA in the freshly isolated BM-MNC fractions by Northern blotting but were successful by the use of RT-PCR, indicating the high sensitivity of zymography for MMP-2 enzyme detection.

RT-PCR also revealed the presence of MMP-2 transcripts in the BM-MNC fractions from healthy persons and patients with CML, where MMP-2 activity was not detectable by zymography. This may be explained by the capacity of RT-PCR to identify only a few specific mRNA molecules deriving from MMP-2-expressing cells, which are present in these BM-MNC fractions but do not necessarily secrete detectable amounts of this enzyme. Due to the heterogeneous cellular composition of BM-MNCs from healthy persons as well as patients with CML or MDS, the subset of hematopoietic cells responsible for MMP-2, TIMP-2, and MT1-MMP expression in these collectives remains to be determined.

The release of MMP-9 from AML bone marrow blasts occurred independently of the grade of leukemic cell differentiation because it was observed in all cases from FAB M1–M5. This is consistent with findings on established leukemic cell lines of different stage of maturation such as HL-60, NB4, U-937, and THP-1 that are also known to permanently release MMP-9 (27, 29, 35, 36). The capacity for continuous MMP-2 production in ex vivo AML blasts, however, may be connected to the stage of cellular differentiation as this enzyme was not detected in three cases with AML-M1 and -M2 but was present in eight cases with AML-M2–M5. In normal leukocytes, MMP-2 secretion has been reported to occur only in mature macrophages and T-lymphocytes in response to immunological stimuli, whereas neutrophil granulocytes are not able to synthesize this enzyme (17, 18, 19, 37). More detailed studies on different subsets of purified normal and malignant hematopoietic progenitors are required to answer the question of whether gelatinase release is a consequence of the malignant transformation in leukemic cells.

We found the MMP-2 release from BM-MNC to be associated with a more aggressive phenotype in myeloid leukemias because its presence significantly correlated with high blast counts in bone marrow which is characteristic for primary and secondary AML or blast crisis in CML but was absent in patients with AML in remission, CML in chronic phase, and most of the MDS cases. Moreover, in CML, only 1 of the 16 patients in chronic phase was MMP-2 positive (patient 40). This patient underwent acceleration into blast crisis 8 weeks after our analysis. Similar to CML in chronic phase, MDS can also progress to acute leukemia. We found that 2 of 3 MMP-2-positive individuals with MDS RAEB transformed to RAEB in transformation (patient 47) or developed a secondary AML (patient 48) within months. Possibly, zymographic analysis allows the detection of a MMP-2 producing blast cell clone in bone marrow prior to its expansive proliferation leading to a progression of disease. However, although 49 samples of patients with different hematological malignancies have been analyzed, due to the heterogeneity of these disorders the potential prognostic value of MMP-2 release from BM-MNC has to be considered premature and therefore requires further investigation.

One physiological role of MMP-2 and MMP-9 in normal leukocyte function arises from the ability of these enzymes to degrade collagen type IV, the main component of basement membranes and is thought to be associated with the transmigration and degradation of ECM structures of tissue and blood vessels (20, 38). Studies using monoclonal antibodies to MMP-9 in interleukin 8-stimulated rhesus monkeys have indicated the involvement of this enzyme in the process of hematopoietic progenitor cell mobilization from bone marrow into peripheral blood (39, 40). Thus, it can be hypothesized that the premature production of gelatinases in leukemic cells may be involved in their dissemination from bone marrow. Actually, we found that 71% of all patients that were MMP-2 positive in bone marrow had leukemic blasts in their peripheral blood. On the other hand, also in 52% of the MMP-2 negative patients blasts were present in the circulation, albeit in lower counts. One explanation for this observation might be that MMP-2 is, indeed, released from the blast cells, but most of the MMP-2 negative cases in our study are patients with CML or MDS that characteristically display low blast counts in bone marrow producing too small amounts of MMP-2 for zymographic detection, but enough for them to enter circulation. Matsuzaki and Janowska-Wieczorek (30) recently reported the constitutive secretion of MMP-2 and MMP-9 in primary cultures of leukemic blast cells isolated from peripheral blood of patients with de novo AML. This is consistent with our present findings on AML blasts from bone marrow. The authors demonstrated invasive capacities of the peripheral blasts in an in vitro assay using reconstituted basement membrane as migration barrier and showed a reduction of the cellular invasiveness in the presence of metalloproteinase inhibitors, suggesting that these enzymes may be involved in leukemic cell invasion. In addition to their capacity for matrix degradation, gelatinases have been found to have some atypical substrates. For example, active tumor necrosis factor-α is liberated from its inactive precursor form, and Fas ligand is shed from the leukocyte cell surface by the action of metalloproteinases (41, 42). Thus, the constitutive release of gelatinases we found in BM-MNC under both normal and pathological conditions may also influence hematopoietic cell growth and behavior by the processing of regulatory proteins in marrow.

MMPs are synthesized and secreted as latent zymogens, which require activation to gain full activity (43). Because activation of MMPs is based on proteolytic truncation resulting in a reduction of the molecular mass, the additional M_r 80,000 gelatinase present in many BM-MNC-conditioned media most probably corresponds to an activated form of MMP-9. The precise identity of this activity, however, remains to be determined. Extracellular net proteolytic activity of MMPs depends on the balance between active enzyme and its specific inhibitor. Similar to MMP-9, we found a constitutive expression of TIMP-1 in all BM-MNC fractions of normal or malignant origin. This is not surprising because MMP-9 is frequently found to be coexpressed and secreted from the cells together with TIMP-1 (44). In contrast to MMP-9, TIMP-1 transcription in BM-MNC did not vary considerably among samples and patient groups. In addition to its regulatory function on MMP activity, TIMP-1 is known to act as a factor that stimulates growth and differentiation in a wide range of cells, including erythroid precursors, erythroleukemia cells, Burkitt lymphoma cells, and HL-60 leukemia cells (25, 45, 46, 47). To what extent the pleiotropic effects of TIMP-1 produced by BM-MNC influence normal hematopoiesis and the development of hematological malignancies is unknown.

The presence of MMP-2 in culture supernatants of AML blasts leads to the question of whether these cells additionally synthesize TIMP-2 and MT1-MMP, factors that are closely involved in MMP-2 activation at the cell surface of normal and neoplastic cells. We found TIMP-2 transcripts to be present in most of the MMP-2-positive cell fractions. This is in accordance with findings in various other cell systems that MMP-2 release is frequently associated with TIMP-2 production, which modulates the MMP-2 activation process and balances MMP-2 net proteolytic activity. In contrast to TIMP-2, MT1-MMP transcription was not detected in the majority of patients with AML or CML. Thus, the presence of TIMP-2 together with the lack of MT1-MMP synthesis in MMP-2-secreting AML blast cells suggests that these cells cannot activate MMP-2 under culture conditions. This hypothesis is confirmed by the absence of activated MMP-2 in AML blast cell-conditioned medium. Although the activation mechanisms of MMPs are still unclear in detail (3), under more physiological conditions when stimulatory cytokines, growth factors, or stromal cells are present that may induce or provide MMP activating enzymes, it is not unlikely that leukemic blast cells achieve MMP activation to use constitutively secreted MMPs for their invasive behavior. The functional role of gelatinase release for leukemic blast cell dissemination in organism, however, remains to be investigated.

Taken together, we have shown the constitutive production of MMP-9 and its inhibitor TIMP-1 in mononuclear cells isolated from bone marrow and demonstrated leukemic blast cells to additionally express and secrete MMP-2. Our findings suggest that MMP-2 represents a marker of malignant transformation in AML with a potential prognostic significance for the progression of disease in CML and MDS. The permanent release of MMP-2 and MMP-9 from leukemic blasts may contribute to the leukemic cell dissemination by local digestion of ECM barriers similar to the invasion and metastasis processes of tumor cells. Currently, MMP-2 and MMP-9 are promising targets in an adjuvant therapy of solid tumor spread (48). Synthetic MMP inhibitors such as Batimastat and Marimastat have been shown to be effective in various models for tumor metastasis and are currently used in clinical studies on human cancers (49). Recently, another synthetic MMP inhibitor was reported to prevent lethal acute graft-versus-host disease in mice (50). Thus, future effort should be directed to the investigation of MMP production in the various subsets of hematopoietic cells in bone marrow that will shed more light on the pathological biology of myeloid leukemia and possibly open new avenues of therapeutic intervention.

We thank Dr. Rainer Bartl, Rolf Baumgart, Marion Janocha, Birgit Herzum, Bettina Wunderlich, and Vera Brumme for their support in the collection of patient material and clinical data. We are grateful to Dr. Martin Schreiber for statistical advice.