BCR-ABL Tyrosine Kinase Activity Regulates the Expression of Multiple Genes Implicated in the Pathogenesis of Chronic Myeloid Leukemia¹

Over 90% of cases of CML⁴(1) and 10–25% of cases of ALL (2) are characterized by a reciprocal translocation between chromosomes 9 and 22. As a result, a BCR-ABL hybrid gene is formed on the derivative Ph chromosome. Depending on the location of the breakpoint in BCR, three types of fusion protein can be formed, all of which exhibit deregulated PTK activity compared to normal ABL(2, 3, 4). As a result, there is excessive tyrosine phosphorylation of many intracellular proteins including the BCR-ABL protein itself (5, 6). Although not all interactions of BCR-ABL with other proteins are phosphotyrosine dependent, it is clear from mutational analysis that the PTK activity is an absolute requirement for malignant transformation, and that it cannot be complemented by any downstream effector (7, 8). In contrast, it is, at present, less clear which of the various signaling pathways activated by BCR-ABL(such as RAS (9), Janus-activated kinase-signal transducers and activators of transcription (10, 11), and phosphatidylinositol 3′-kinase (12) is essential for transformation, and, indeed, redundancy is likely.

Basic mechanisms that have been attributed to BCR-ABL-positive cells, particularly in CML, are increased proliferation, increased resistance to apoptosis(13, 14, 15), and an alteration of their adhesion properties(16, 17). Given the fact that most of the pathways activated by BCR-ABL transmit signals to the transcriptional machinery of the transformed cell, we reasoned that the phenotype of BCR-ABL-positive cells should at least partially be explicable on the grounds of altered gene expression. In this scenario, BCR-ABL PTK activity would maintain the activation or suppression of a set of target genes that confer the malignant phenotype on a Ph-positive cell. Until now, only a few such downstream targets have been identified. Moreover, the data are mostly derived from cell lines transformed to growth factor independence by ectopic expression of BCR-ABL rather than from cells that acquired the translocation “naturally.”

To define downstream targets of BCR-ABL, we studied alterations in the profile of gene expression in BV173 cells (derived from a CML patient in lymphoid blast crisis; Ref. 18)exposed to STI571 (formerly known as CGP57148B), a specific inhibitor of BCR-ABL PTK (19, 20, 21). DD was used to compare mRNA expression in the cells after inhibition of BCR-ABL PTK for 24 h with untreated control cells.

We identified 12 differentially regulated mRNAs, 7 of which correspond to known genes, and 5 of which correspond to unknown genes. All known genes are implicated in cellular processes that are thought to be disturbed in Ph-positive leukemias.

All cell lines were grown in RPMI 1640 supplemented with 10% FCS,penicillin, streptomycin, and l-glutamine. Table 1 provides a list of the cell lines used. All experiments were performed on exponentially growing cells with viability of >90% as assessed by trypan blue exclusion. The tyrosine kinase inhibitor STI571 (kindly provided by Dr. Elisabeth Buchdunger; NOVARTIS, Basel, Switzerland) was added to the culture media at the required concentration. In one experiment, BV173 cells were cultured in the presence of a blocking antibody to the PDGFRβ [M4T9.22 (22); 10 μg/ml; a generous gift of Dr. Nick Landolfi (Protein Design Laboratories,Fremont, CA)].

Analysis of apoptosis in BV173 cells exposed to STI571 was done by assessment of DNA fragmentation on agarose gels as described previously(21).

Total RNA was extracted from BV173 cells treated with 1.0μ m STI571 for 24 h using the acid guanidinium thiocyanate method (23). Contaminating DNA was removed by digestion of 100 μg of RNA with RNase-free DNase (Boehringer Mannheim, Mannheim, Germany) in a 200 μl reaction containing 10 mm Tris (pH 8.3), 5 mm KCl, 1.5 mm MgCl₂, and 18 μl of DNase, 2μl of RNase inhibitor (Promega, Southampton, United Kingdom). Single-use aliquots (1 μg/μl) of DNA-free RNA were stored at−80°C. DD (24) was performed with the RNAimage kit(Gene Hunter, Nashville, TN), following the instructions of the manufacturer, with a few modifications. The cDNA synthesis was done with Superscript reverse transcriptase (Life Technologies, Inc.,Paisley, United Kingdom), and 1 μl of RNase inhibitor (Promega) was regularly added to the 40 μl reaction. PCR was performed with Taq polymerase from Boehringer Mannheim in 10 μl reactions. Thermocycling conditions were as recommended by Gene Hunter, except that the extension time was increased to 90 s. In several experiments,“long” 21-mer or 18-mer primers (a gift from Dr. Feilan Wang;Dana-Farber Cancer Institute, Boston, MA) were used, and the annealing temperature of the PCR was increased to 50°C. Reamplified PCR products were cloned into the TOPO-TA vector (Invitrogen, NV Leek, the Netherlands), according to the instructions of the manufacturer.

Fifteen μg of total RNA or 2 μg of poly(A) RNA (isolated directly from cells with the MACS mRNA isolation kit; Miltenyi Biotech, Bergisch Gladbach, Germany) were separated on a 0.8% agarose gel and blotted onto nylon membrane (Hybond N; Amersham, Amersham, United Kingdom). Hybridizations were carried out at 42°C in 50% formamide, 10%dextran sulfate (Pharmacia, Uppsala, Sweden), 1 m NaCl, 1%SDS, and 0.2 mg/ml sonicated salmon sperm DNA (Sigma, Paisley, United Kingdom).

Probes were prepared by PCR amplification of the vector inserts from flanking primers (M13 standard primers), restriction enzyme digestion of the plasmid, or PCR amplification of the target gene with specific primers (sequences are available on request). The purified probes were labeled with ³²P (Megaprime; Amersham). Specific activities of more than 5 × 10⁸cpm/μg DNA were routinely achieved.

As a control for RNA loading, the blots were stripped and reprobed withβ-actin cDNA (kindly provided by Dr. Philip Mason, Imperial College School of Medicine, London, United Kingdom). Densitometric analysis of the autoradiographs was performed using GELBLOTPRO software(Ultra Violet Products, Cambridge, United Kingdom).

In several instances, no signal could be obtained even on Northern blots prepared from poly(A) RNA after exposure of the filters for up to 7 days. In these cases, the cloned PCR products were sequenced, and sequence-specific primers were designed (sequences are available on request). PCR conditions were optimized for each individual fragment. Amplification of glucose-6 phosphate dehydrogenase (25)served as a control for equal amounts of cDNA. The number of cycles was chosen so that the reaction was in its logarithmic phase. PCR products were resolved on an ethidium bromide-stained 2% agarose gel and transferred to a nylon membrane (Hybond N; Amersham), as described previously (26). The filter was then probed with an internal oligonucleotide.

In one case (oncostatin β receptor), the 5′ rapid amplification of cDNA ends-PCR kit (version 2.0; Life Technologies, Inc.) was used to extend the initially unknown fragment derived from DD into the published sequence.

Lysates from 2.5 × 10⁶ cells were prepared as described by Kabarowski et al.(27). The protein concentration was determined with the Dc protein assay (Bio-Rad, Hercules, CA). One hundred μg of protein were separated on 6% SDS-PAGE gels, blotted onto polyvinylidene difluoride membranes (Millipore, Bedford, MA), and probed with an anti-phosphotyrosine antibody (PY99; Santa Cruz Biotechnology,Santa Cruz, CA). Bands were visualized by enhanced chemiluminescence(Amersham) and subsequently quantified by densitometry. To control for equal protein loading, the blots were stained with Coomassie Blue and/or stripped and reprobed with anti-ABL antibody (Abl-3; Calbiochem,Nottingham, United Kingdom). All experiments were done at least twice,with similar results.

We have previously shown that STI571 inhibits the proliferation of BV173 cells in a dose-dependent manner (21). Analysis of phosphotyrosine blots showed that STI571 reduced BCR-ABL autophosphorylation with IC₅₀ and IC₉₀ values of approximately 0.25 and 1.0μ m, respectively (Fig. 1). Little DNA laddering above background levels was detectable in BV173 cells incubated with STI571 for 24 h, whereas extensive laddering was seen after 48 h (Fig. 2). The cultures could be rescued after up to 36 h of exposure if STI571 was thoroughly washed out, whereas longer exposure times invariably led to cell death. Based on these observations, we chose an exposure time of 24 h to avoid detection of changes of gene expression associated with the general metabolic breakdown that accompanies the final stages of apoptosis. To establish the optimal dose of STI571, a series of pilot experiments was performed that showed no appreciable difference in the pattern of bands obtained by DD between cells incubated with 0.25 and 1.0 μmSTI571, respectively. Therefore, 1.0 μm(corresponding to the IC₉₀ for inhibition of BCR-ABL tyrosine phosphorylation in BV173 cells) was chosen for the experiments. This is also the concentration that discriminates maximally between CML and normal progenitor cells in clonogenic assays(21).

An estimated 8000 bands were displayed from RT-PCR amplifications with 114 primer combinations. The regulation pattern seen on the DD gels was confirmed for 12 cDNA fragments. In nine cases, this was done by Northern blot analysis (Fig. 3). Three fragments gave no signal at all, even after hybridization to poly(A)-selected RNA and prolonged exposure of the autoradiographs. For these fragments, sequence-specific primers were designed, and the expression was analyzed by semiquantitative RT-PCR. In all three cases,the regulation pattern expected from the original DD gel (data not shown) was confirmed (Fig. 4). Of the 12 differentially expressed mRNAs, 7 represent known genes,and 5 are novel sequences (Table 2). Known genes found to be up-regulated on inhibition of BCR-ABL PTK were MPP1 (28), BCL-6, and R-PTPκ (29). Genes whose expression decreased were integrinα₆, cyclin D2, CSCP, and OSMRβ. Four of the five novel sequences correspond to expressed sequence tags, whereas one (DD-W) does not match any sequence in public databases. These genes are currently under investigation.

Time course experiments showed that the up- or down-regulation of the differentially expressed genes occurred after various times of treatment with STI571 (Fig. 5). Cyclin D2, integrin α₆, and OSMRβwere strongly down-regulated within 3 h, and BCL-6 was strongly up-regulated within 6 h, and their level of expression did not change significantly thereafter. In the case of the other genes, the up- or down-regulation followed a more protracted course. However, a change in the level of expression was usually evident after 3 h. Given the limitations of semiquantitative PCR, no attempt was made to quantify the expression of the respective genes over time.

We then asked whether the pattern of regulation seen in BV173 cells would be reproducible in other Ph-positive cell lines and, conversely,be absent in Ph-negative lines. Generally, the regulation pattern could not be reproduced in Ph-positive myeloid cell lines (Fig. 6). The expression of the genes in question was either below the threshold of sensitivity of total RNA Northern blots or unaffected by STI571 (data not shown). In contrast, in many cases, the pattern observed in BV173 cells was reproducible in other Ph-positive lymphoblastic cell lines (Table 1), with the notable exception of SD1. This cell line is unique because it is EBV-transformed(30). In particular, cyclin D2 was consistently down-regulated and the unknown gene DD221 was consistently up-regulated in all Ph-positive lymphoblastic cell lines treated with STI571. Similarly, there was a consistent pattern of up-regulation for BCL-6,although the extent was variable, and the gene is not expressed in all of the cell lines. Integrin α₆ was down-regulated in all EBV-negative B-cell lines, whereas its expression seemed to increase in CML-T1 cells on treatment with STI571. The level of expression was below the threshold of detection in TOM1 cells. Up-regulation of MPP1 and DDI was observed in the two B-cell lines(BV173 and NALM1) that originated from CML in blast crisis rather than from ALL. In some cases (OSMRβ, CSCP, and DDQ), the expression of the mRNA in question appeared to be restricted to BV173 cells at the given level of sensitivity. Importantly, none of the changes was detectable in any of the Ph-negative cell lines exposed to STI571.

R-PTPκ, HRF-5, and DD-W showed highly variable levels of expression in cell lines other then BV173, although they were detectable in all cases under optimal conditions for RT-PCR amplification. Thus, given the limitations of semiquantitative PCR, no attempt was made at quantifying their expression in cell lines other than BV173.

To exclude the possibility that insufficient inhibition of BCR-ABL tyrosine kinase activity was responsible for the differences between the lymphoid cell lines, we prepared phosphotyrosine blots and determined the respective IC₅₀ values (Fig. 1;Table 3). With the exception of CML-T1 cells, tyrosine phosphorylation of BCR-ABL was reduced in a dose-dependent manner in all cell lines,although the IC₅₀ values differed slightly from cell line to cell line.

Intensive research during the past decade has elucidated many of the signal transduction pathways activated by BCR-ABL. In contrast, few data are available that address BCR-ABL-dependent gene expression. Genes reported to be overexpressed in v-abl- or BCR-ABL-positive cells include MYC(31), BCL-2 (32), the melanoma-related antigen PRAME (33), and the RAS-like gene KIR (34). A more recent study compared the gene expression profile in Mo7 cells transfected with a P210^BCR-ABL expression vector using DNA arrays and suggested differential expression of multiple genes(35). Confirmation by Northern analysis has not yet been presented.

We show that inhibition of BCR-ABL tyrosine kinase by a specific inhibitor (STI571) alters the expression of multiple genes in BV173 and in other lymphoid cell lines. Hence, these genes can be considered as downstream targets of BCR-ABL. We chose BV173 cells for the initial screening for several reasons: (a) they do not express normal ABL, which excludes effects related to ABL PTK(36); (b) they exhibit a dose-dependent growth inhibition on exposure to STI571 that is correlated to inhibition of BCR-ABL autophosphorylation; and (c) unlike other CML cell lines, they readily engraft in SCID-NOD mice (37),suggesting that they may be closer to the clinical disease.

We found 12 genes (7 previously known genes and 5 unknown genes) whose expression is dependent on BCR-ABL PTK activity in BV173 cells. All known genes reflect pathogenetic principles thought to be relevant in CML.

Integrin α₆ (VLA-6), CSCP, and MPP1 are genes involved in the organization of the cell membrane and its adhesion properties. It is therefore conceivable that their aberrant regulation contributes to the well-described adhesion defect of CML cells (38). Abnormal function and expression of α integrins in BCR-ABL-positive cells have been reported (16, 35, 39). Similarly, adhesion of hematopoietic progenitor cells to fibronectin appears to require the coordinated function of chondroitin sulfate proteoglycans, CD44 andα ₄β integrins (40, 41). It is thus conceivable that BCR-ABL perturbs the ordered expression of CSCP with resulting alterations in the composition of proteoglycans, which, in turn, may impact on homing and adhesion to stroma. Surprisingly, we failed to detect a CSCP message of identical size to that in BV173 cells in any of the other cell lines tested, and we are currently investigating whether this message represents a splice variant that is peculiar to BV173 cells. The interaction of MPP1 with protein 4.1 and glycophorin C is vital for the stability of the erythrocyte membrane, but the wide expression of glycophorin C in nonerythroid cells indicates that the system may have additional functions. With respect to CML, it is noteworthy that talin, a cytoskeletal protein related to protein 4.1, is tyrosine-phosphorylated by BCR-ABL (42) and that abnormal processing of glycophorin C has been demonstrated in leukemic cells(43). Moreover, loss of MPP1 exon 5 due to exon skipping has recently been described in a patient with CML in blast crisis(44). Thus, repression of MPP1 may be yet another way by which BCR-ABL PTK activity alters the properties of the cell membrane.

BCL-6 functions as a transcriptional repressor (45, 46)and is essential for germinal center formation (47). It is absent in B cells before and after germinal center formation(45). The up-regulation observed on inhibition of BCR-ABL PTK could indicate that BCR-ABL inhibits the expression of BCL-6, which, in turn, prevents the execution of a differentiation program at a stage prior to germinal center formation.

Cyclin D2 and D3 mediate the G₁-S-phase transition in hematopoietic cells and,if overexpressed, allow for G₁-S-phase progression under conditions of growth-factor deprivation(48). Thus, it is likely that BCR-ABL provides a mitogenic signal that results in overexpression of cyclin D2 and facilitates the G₁-S-phase transition. Our observation is in line with a recent report that showed cooperation between cyclin D1 and BCR-ABL in the transformation of fibroblasts and murine B cells(49).

The BCR-ABL-dependent expression of the OSMRβ appears to be limited to BV173 cells, at least at the given level of sensitivity. OSMRβ is the high-affinity receptor for OSM, a cytokine with pleiotropic biological activities. The biological effects of OSM depend on the cellular context: it inhibits the proliferation of many cancer cell lines (50) but can also act as a potent growth factor, e.g., for the spindle cells in AIDS-related Kaposi’s sarcoma (51). Because OSM mRNA is detectable in BV173 cells by RT-PCR (data not shown), it is possible that an OSM autocrine loop exists in BV173 cells that is maintained by the BCR-ABL-induced expression of OSMRβ.

The role of protein tyrosine phosphatases such as PTP1B(52), SYP (53), SHP-2 (54), and SHPTP1 (55) in the transformation by BCR-ABL is complex. For example, PTB1B expression is induced by BCR-ABL PTK activity, and it has been suggested that this may represent a compensatory mechanism by which the cell attempts to limit the effects of BCR-ABL. Down-regulation of R-PTPκ in BV173 cells would be expected to increase rather than decrease the impact of BCR-ABL by reducing the activity of a potential antagonist. We are currently investigating this possibility.

It is interesting to note that not all genes show a consistent pattern of up- or down-regulation within the lymphoid cell lines used in this study. Even phenotypically very similar cell lines such as BV173 and NALM1 show significant differences that cannot be explained by variations in IC₅₀ values among the individual cell lines. In some cases (OSMRβ, DDQ, and CSCP), the expression of the gene in question appears to be restricted to BV173 cells, at least at the given level of sensitivity. Thus, the set of genes regulated by BCR-ABL is partly dependent on the cell line studied. This is most strikingly illustrated by the EBV-transformed SD1 cell line that expresses a largely different set of genes and is very resistant to STI571. Moreover, the results were generally not reproducible in Ph-positive myeloid cell lines. It should be noted that the difference in BCR-ABL-regulated gene expression profile appears to be dependent on the actual cell lineage (lymphoid versus myeloid) rather than the type of BCR-ABL fusion protein (i.e., p210 versus p190) because both p210- and p190-producing cell lines were included in the lymphoid collection examined. Taken together, our data underline the importance of the cellular context in the pathogenesis of Ph-positive leukemias. One might speculate that this is also reflected in the diverse clinical behavior of individual cases of CML and Ph-positive ALL.

The issue of specificity is of great importance in any study that investigates biological changes induced by a chemical agent. STI571 is known to inhibit the tyrosine kinases of KIT and PDGFβ receptors at concentrations comparable to ABL PTK. In our system, effects related to inhibition of ABL can be excluded because it is not expressed in the BV173 cell line. Effects of STI571 related to inhibition of PDGFRβPTK can be ruled out because blockage of the PDGFRβ with a specific antibody (22) did not induce any of the changes seen on incubation with STI571 (Fig. 7). In contrast, we cannot exclude effects due to inhibition of KIT because the latter may be activated by BCR-ABL in a tyrosine kinase-dependent manner, at least in myeloid cells (56). Similarly, potential effects of STI571 on other kinases downstream of BCR-ABL cannot be excluded with certainty because not all possible candidates have been tested (19, 20). On the other hand,nonspecific effects not related to the expression of BCR-ABL are unlikely because none of the effects was seen in Ph-negative cell lines of similar phenotype.

Bearing these caveats in mind, our approach has two advantages over systems that use ectopic expression of BCR-ABL in factor-dependent cell lines: (a) BCR-ABL is expressed from its natural promoter at physiological levels; and (b) cell transformation to growth factor independence does not reflect all aspects of BCR-ABL-mediated malignant transformation. It will be important to investigate whether the gene expression changes identified in this study are also observed in primary CML or Ph-positive ALL. Preliminary results of competitive RT-PCR tests suggest that expression of the MPP1 gene, for example,follows an up-regulation pattern on BCR-ABL inhibition in CML CD34+progenitors similar to that found in BV173 cells (data not shown). Detailed investigations on the functional significance of the changes detected in each gene are currently in progress.

In summary, our results demonstrate that BCR-ABL PTK activity regulates the expression of specific genes that partly explain the abnormal biological behavior of Ph-positive cells. They further indicate that genes hitherto not implicated in the pathogenesis of Ph-positive leukemias could be important in the transformation process and may constitute targets for future therapeutic intervention.