Growth, Differentiation, and Malignant Transformation of Pre-B Cells Mediated by Inducible Activation of v-Abl Oncogene¹

The Abelson murine leukemia virus (Ab-MuLV)⁵ expresses the v-Abl protein that induces pre-B cell malignant transformation (1). The v-Abl protein is a constitutively active nonreceptor tyrosine kinase that has been implicated in regulating proliferation and differentiation of pre-B cells (2, 3). Although v-Abl-transformed pre-B cell lines have been widely used as a model to study pre-B cell differentiation, including Ig gene rearrangement and signal transduction pathways (4, 5, 6), the molecular mechanisms that control differentiation and transformation have not been fully defined.

As reported by Rosenberg and colleagues (5, 6, 7), pre-B cell lines, transformed by a temperature-sensitive Ab-MuLV (ts-Ab-MuLV), exhibit temperature-dependent proliferation and differentiation. At the permissive temperature, active v-Abl protein kinase confers cell growth, yet inhibits Ig L chain (Ig-L) gene rearrangement. Upon inactivation of v-Abl, cells are arrested at the G₁ phase with many undergoing apoptosis (7). Thus, cell proliferation and differentiation are inversely regulated, and can be readily manipulated experimentally in this ts-Ab-MuLV model system.

Introduction of a bcl-2 gene into ts-Ab-MuLV-transformed cells (via transfection or transgenic mice) reduces the apoptosis associated with v-Abl inactivation (5, 8). The extended survival allows detailed analyses of pre-B cell differentiation that is induced by the absence of v-Abl kinase activity. In addition, the differentiation program and cell cycle inhibition of these cells can be halted by reactivation of v-Abl (5, 9). The cellular response to v-Abl reactivation in these cells, in some aspects, resembles the initial Ab-MuLV-mediated transformation of primary B cell precursors because both are triggered by a constitutively active v-Abl kinase. Whereas the latter has been extensively studied over the years (2, 3, 10, 11), little is known about the events associated with v-Abl reactivation in ts-Ab-MuLV cells. Analogous to the transformation of primary B cell precursors by Ab-MuLV, reactivation of v-Abl kinase activity in differentiated ts-Ab-MuLV cells may also induce mutations that lead to subsequent transformation. The bcl-2-expressing ts-Ab-MuLV cell lines offer a unique model to study mechanisms that control transformation vs differentiation without the extensive apoptosis observed in primary cells. In addition, the relatively homogenous population that can undergo reversible v-Abl activation/inactivation/reactivation makes it experimentally feasible to delineate molecular processes underlying transformation and differentiation.

We have established (8) ts-Ab-MuLV pre-B cell lines from bcl-2 transgenic SCID (SCID-ts) and wild-type (wt) heterozygous (s/+-ts) mice. As reported in our earlier studies (9), upon v-Abl inactivation followed by reactivation, SCID-ts cells remain locked in a G₀-G₁ state for an extended period of time. In this study, we demonstrate that the few SCID-ts cells that can escape the G₁ arrest are tumorigenic. These cells are altered in their responses to v-Abl inactivation, in terms of cell cycle arrest and IgL-gene rearrangement. We identify factors associated with v-Abl signaling that may determine the cell fate of either differentiation or transformation of SCID-ts pre-B cells. We found that v-Abl inactivation induces concurrent activation of Foxo3a and NF-κB/Rel, which are responsible for cell cycle arrest and initiation of Ig-κ gene rearrangement, respectively. However, this activation is attenuated in the SCID-ts-derived tumorigenic cells. The significance of our findings is discussed.

ts-Ab-MuLV-transformed pre-B cell lines were derived from bcl-2 transgenic SCID homozygous (SCID) mice by transformation of fetal liver with ts-Ab-MuLV, denoted as SCID-ts (8). S4 is a subclone of SCID-ts cells. 103/B4-ts, a ts-Ab-MuLV cell line derived from wt mice and transfected with the bcl-2 gene (5), denoted as wt-ts, was provided by Dr. N. Rosenberg (Tufts University, Boston, MA). Cells were maintained at 33°C. The v-Abl kinase is inactivated by culturing cells at 39°C for 2–3 days, and reactivated by returning cells to 33°C. Differentiation of Ab-MuLV cell lines can also be induced by incubating cells with 10 μg/ml LPS (Sigma-Aldrich) for 24 h. Cell cycle analysis was performed by propidium iodide analysis according to the procedure described before (7).

Cells were either maintained at 33°C (i.e., noninduced) or cultured at 39°C for 2–3 days (i.e., induced). A modified limiting dilution assay was used to assess growth potential of induced and noninduced cells (12). Various concentrations of both noninduced cells and induced cells were seeded in 96-well flat-bottom plates containing RPMI 1640 supplemented with 20% FBS. The plates were cultured at 33°C for 12 days with medium changes every 3–5 days, and scored for cell growth (CG). Limiting dilutions were performed such that cellular expansion is readily observed for noninduced cells at 1 and 5 cells/well and 10 and 20 cells/well for induced cells. The plates with the lowest seeding number of cells (i.e., cells/well) that show CG were taken into the count. The percentage of CG was derived from the following equation, and is presented as the mean of two independent experiments: CG (percentage) = total number of positive wells with CG divided by the product of (cell number seeded per well × 96 × number of plates) × 100%. The growth potential of induced cells is represented by a normalized CG (percentage), i.e., the CG of induced cells divided by the CG of noninduced cells, as shown in Table I.

SCID mice were maintained in a pathogen-free environment at the animal research center of Arizona State University according to the Institutional Animal Care and Use Committee protocol. Noninduced S4 and 103/B4-ts cells were maintained at 33°C, whereas induced cells were cultured at 39°C for 3–5 days followed by their return to 33°C for 4–14 days before injection. SCID mice were injected with 10², 10⁴, or 10⁵ induced or noninduced cells s.c. and monitored for tumor growth. Mice were sacrificed when tumors reached >1 cm³. Cell lines S4C2-1 and 3-4-1 were isolated from solid s.c. tumors derived from independent inductions of the parental S4 cell line. Different culture conditions were tested that could induce tumorigenic transformation of S4 cells. Although S4C2-1 was derived from S4 cells independently by incubation at 39°C for 5 days (S4C2-1) followed by 2 wk at 33°C, 3-4-1 is derived from S4 cells cultured at 39°C for 3 days followed by 4 days at 33°C (see Table I). To ensure that tumorigenic changes are induced during in vitro manipulation, but not caused by an in vivo environment, independent subclones were derived by culturing S4 (SCID-ts parental) cells at 39°C for 3 days followed by seeding them in 96-well plates at 1 cell/well immediately before returning to 33°C. The clone, 015-3, was derived from this experiment. Upon injection of 015-3 cells, tumor formation was observed.

The rearranging status of Ig κ L chain (Ig-κ) was analyzed by Southern blot of BamHI-digested genomic DNA, isolated from S4 and S4C2-1 cell lines, as well as BALB/c thymus, according to the established procedure (13). The blot was hybridized with pECκ that spans both Jκ and Cκ regions (14).

Recombination signal ends were analyzed by ligation-mediated (LM) PCR, as described previously (8). The expression of various genes was assessed by RT-PCR (15) with primers described below: YC51, 5′-GAATGGGAAGCCGAACATACTG-3′ and YC52, 5′-TGCTGATCACATGTCTCGATCC-3′ for β₂ microglobulin as a control; MYC-F, 5′- CCGACGCTTGGCGGGAAA-3′ and MYC-R, 5′-GGAGTCGTAGTCGAGGTCATAG-3′ for c-myc; MB70, 5′-CCGGATCCACGCATGCTTGGAGAGGGGTT-3′ and MB130, 5′-ATGGATCCAGTTGGTGCAGCATC-3′ for κ_o-germline transcript; YC109, 5′-CCAAGCTGCAGACATTCTAGCA-3′ and YC110, 5′-CAACATCTGCCTTCACGTCGAT-3′ for RAG-1; andYC84, 5′-CACATCCACAAGCAGGAAGTAC-3′ and YC85 5′-GGTTCAGGGACATCTCCTACTA-3′ for RAG-2. LM PCR and RT-PCR products were then analyzed by Southern blot with ³²P-labled probes derived from purified individual PCR products. The intensity of PCR was quantified using Molecular Dynamics phosphorimaging and analyzed by ImageQuant software.

Whole cell extracts were prepared by lysing cells in Nonidet P-40 lysis buffer (50 mM Tris (pH7.5), 150 mM NaCl, 0.05% SDS, 1% Nonidet P-40, and 2 mM EDTA) with protease inhibitors mixture and phosphatase inhibitor mixture I/II (Sigma-Aldrich) and spun to collect supernatants. Cytoplasmic and nuclear proteins were prepared as described previously (15), and their concentrations were determined using the BioRAD assay (Pierce). Fifty micrograms of cellular extracts was separated on a 10% SDS-PAGE. For phosphotyrosine assays, 1 × 10⁷ cells were loaded per lane. Gels were then transferred onto Nitrobind membrane (Osmonics). The membranes were incubated with specific Ab and revealed by chemiluminescence (Pierce) as recommended by manufacturer. Abs used in this study include the following: rabbit anti-p65 (C-20), goat anti-actin (I-19), rabbit anti-cyclin E (M20), and rabbit anti-cdk4 (C-22) (all obtained from Santa Cruz Biotechnology); rabbit anti-IκBα and rabbit anti-HMG2 (BD Biosciences); rabbit anti-phosphotyrosine (4G10), rabbit anti-FKHRL1 (Upstate Biotechnology); and rabbit anti-p50 (Ab-2), rabbit anti-p27 (Ab-2), and mouse anti-cyclin D1 (Ab-4) (Neomarkers).

EMSA was performed according to the procedure previously described (6), with some modifications. Annealed oligonucleotides are NF-κB-Eκ (YC268, 5′-CATCCAACAGAGGGGACTTTCC-3′ with YC269, 5′- CATCCCTCTCGGAAAGTCCCCTCTGTTG-3′) or the mutated site κ (NFM) (YC289, 5′-CATGCAACAGAGCTCACTTTATGAGAGG-3′ with YC290, 5′-CATGCCTCTCATAAAGTGAGCTCTGTTG-3′). Oligonucleotides were labeled with [α-³²P]-dCTP and Klenow (Stratagene). Seven nanograms of substrate was incubated with 12 μg of nuclear extract for 20 min at room temperature in binding buffer (20 mM HEPES (pH 7.5), 50 mM NaCl, 0.5 mM EDTA, 1 mM DTT, 5% glycerol, 1 μg poly(I:C), and protease inhibitor mixture) and then separated on a Tris-glycine gel.

The ts-Ab-MuLV-transformed cells cultured at the nonpermissive temperature (39°C) can undergo both recombination activation and cell cycle arrest at the G₁ phase, which resembles the differentiation transition from pre-B-I to pre-B-II stage in developing B cells (5, 6). After their return to the permissive temperature, the majority of SCID-ts cells remain locked in the G₁ phase, yet a few cells were found to be able to return into the cell cycle (our unpublished observation). To determine the frequency of the cells that can reenter the cell cycle and continue their growth, we performed a limiting dilution assay to estimate clonogenic potential of the cells before and after their induction at 39°C. A ts-Ab-MuLv clone (103-4B-ts), derived from wt mice (5), was included as a control. As summarized in Table II, ∼0.3% of the S4 cells can grow out after their return from 39°C to 33°C, consistent with the early observation of their prolonged cell cycle arrest at G₁ phase (9).

It is possible that these few cycling S4 cells underwent a selection for a transforming mutation analogous to the v-Abl-mediated transformation of primary pre-B cells, whereby only a few cells escape G₁ arrest and/or apoptosis to become established tumorigenic cells (10, 16, 17). To address this issue, we investigated the tumorigenic potential of these proliferating S4-derived cells by injecting them into SCID mice. Comparisons were made between S4 cells maintained at 33°C (i.e., noninduced) and those cultured at 39°C followed by return to 33°C (i.e., induced), as well as 103-4B-ts cell lines under both induced and noninduced conditions. More than six independent inductions and injections of induced S4 cells consistently lead to the formation of tumors in mice within 2–3 wk. Injection of 100- and 1,000-induced S4 cells resulted in an average of 60 and 89% tumor formation, respectively (Table II). In contrast, noninduced S4 cells that were only cultured at 33°C did not form tumors even after injecting 100,000 cells, nor did the wt ts-cells develop tumor under either condition (Table II). These results demonstrate that v-Abl inactivation followed by its reactivation promotes malignant transformation of SCID-ts cells.

The possibility that the tumorigenic potential of SCID-ts cells is due to a mutation in the ts-v-Abl gene cannot be disregarded. To entertain this possibility, we analyzed cell cycle progression and v-Abl activity of several independent tumor cell lines (see Materials and Methods for a description of individual tumor lines) under different culture temperatures, i.e., 33°C, 39°C, and 39°C to 33°C. Upon culture at 39°C, virtually all S4 cells are arrested at the G₁ phase (95–97%), and only 70–80% of tumor lines were in G₁ (Fig. 1,A). As expected, the S4 cells remain in G₁, whereas the tumor cells are able to enter the cell cycle immediately upon return to 33°C (Fig. 1 A). Thus, although their response is altered, these tumor cells still maintain some level of temperature-dependent cell cycle arrest.

Inactivation of ts-v-Abl inhibits autophosphorylation and phosphorylation of downstream tyrosine residues (7). To verify temperature-dependent regulation of v-Abl activity, we examined the tyrosine-phosphorylation status of cells cultured at various conditions. Cytoplasmic extracts prepared from tumor and S4 cell lines were analyzed by Western blot using anti-phosphotyrosine Abs. As shown in Fig. 1 B, all cell lines exhibit temperature-dependent v-Abl phosphorylation profiles, i.e., high in cells maintained at 33°C, low upon culture at 39°C, and high again after cells are returned to 33°C. Thus, these tumor cells still exhibit a temperature-dependent tyrosine phosphorylation pattern, and are likely to retain a nonmutated ts-v-Abl gene, which was confirmed by our sequence analyses (data not shown). Taken together, these results indicate that malignant transformation is not due to mutations in the ts-v-Abl gene, which might convert the temperature-dependent kinase into a constitutively active form. Instead, the different patterns in cell cycle progression between S4 and tumor cell lines may be the result of alterations in pathways downstream or independent of v-Abl. Thus, upon v-Abl inactivation and reactivation, these tumor cells might have acquired some mutations that promote cell cycle progression and proliferation, which presumably contributes to oncogenic transformation. We attempted to determine whether the mutation might reside in p16INK/p19ARF and p53 genes because mutations in these genes were found to facilitate v-Abl-mediated transformation of primary pre-B cells (11, 18). However, no apparent abnormality was found in the tumor lines derived from SCID-ts cells (our unpublished observation).

To determine the molecular basis of the altered cell cycle profile, the expression of several G₁-S cell cycle-related proteins was compared between S4 and one tumor line, S4C2-1. This tumor line is shown as a representative, because many independently derived tumor lines had the same profiles as S4C2-1 (data not shown). Specifically, nuclear protein extracts were prepared from cells cultured at 33°C, 39°C, or cells down-shifted from 39°C to 33°C for various days, and analyzed by Western blot. As shown in Fig. 2,A, significant reductions in cyclin D1, cyclin E, and cdk4 were observed in S4 cells incubated at 39°C, whereas p27 expression was increased in comparison to the cells at 33°C, which is consistent to the previous report (19). A similar profile of these proteins was found in S4C2-1 at 39°C, but with a less significant change in protein levels (Fig. 2 A, lanes 2 and 7). Thus, the incomplete cell cycle arrest of S4C2-1 was likely due to the continued presence of cyclin D1, cyclin E, and cdk4, in the absence of significant p27 levels. Upon return to 33°C, S4C2-1 showed prompt elevation in cyclin D1 and cyclin E and reduction in p27, leading to rapid G₁ exit. In contrast, S4 cells failed to quickly up-regulate cdk4, cyclin D1, and cyclin E proteins, whereas the level of p27 remained elevated even 4 days after return to 33°C. These results indicate that the prolonged cell cycle arrest in S4 cells results from both the down-regulation of G₁-S phase cdk-cyclins and the persistent elevation of p27, which are apparently lost in S4C2-1 cells, allowing them to proliferate.

It has been reported (20, 21) that p27 expression is regulated by the Foxo transcription factors. In particular, Foxo3a nuclear protein functions to reduce cyclin D1 and enhance p27 expression through transcriptional and posttranscriptional mechanisms (20, 21). During culture of S4 at 39°C, there was a significant increase in the level of both nuclear Foxo3a and p27, which remained elevated for several days even after cells were returned to 33°C (Fig. 2,B). The nuclear levels of Foxo3a protein was only marginally increased when S4C2-1 were cultured at 39°C, and returned to basal levels after v-Abl reactivation (Fig. 2 C). Thus, the nuclear protein profile of Foxo3a transcriptional factors parallels the p27 protein pattern.

c-myc is a transcriptional factor that directly controls the G₁-to-S transition by facilitating activation of cyclin D1 or cyclin E and reducing p27 levels (22, 23). Given the different expression profiles of cyclin D1, cyclin E, cdk4, and p27 between S4 and S4C2-1, we reasoned that c-myc mRNA levels might also differ in these two cell types. The c-myc transcript was reduced in S4 cells cultured at 39°C, as compared with cells maintained at 33°C, which is consistent with previous reports (24, 25). Interestingly, the level of c-myc mRNA remained low in S4 cells after their return to 33°C, and eventually reached its basal level 4 days later (Fig. 2 D). In contrast, S4C2-1 cells have comparable levels of c-myc, regardless of culture temperature. Thus, the c-myc expression in S4 and S4C2-1 cells is not directly correlated with their v-Abl activation status. S4 cells fail to up-regulate c-myc expression upon v-Abl reactivation, whereas S4C2-1 cells constitutively express c-myc even during v-Abl inactivation.

Given the fact that S4C2-1 cells have failed to undergo cell cycle arrest, one of the two events induced by v-Abl inactivation, i.e., cell cycle arrest and Ig-L gene rearrangement, we wished to determine whether S4C2-1 cells are also hindered in their ability to rearrange their Ig-L genes. To address this issue, we examined recombination initiation of Ig-L genes in both S4C2-1 and S4 cells by analyzing recombination signal ends made at the κ-locus (8). Virtually no Jκ-signal ends were found in S4C2-1 cells during their culture at 39°C (Fig. 3,A), whereas they were readily detectable in S4 cells under the same condition, as expected (8). The lack of attempted recombination is not due to lack of Jκ-locus because a genomic Southern blot analysis indicated the presence of one germline Jκ allele in S4C2-1 (Fig. 3,B). Thus, v-Abl inactivation fails to induce attempted κ-gene rearrangement in S4C2-1 cells. To determine whether the lack of recombination cleavage in these cells is due to inactive recombination machinery or inaccessibility of the recombination locus, we examined expression of RAG1, RAG2, and κ-germline (κ₀) transcripts, which have been well characterized to be responsible for activation of κ-gene rearrangement (26). Interestingly, whereas the levels of RAG1 and RAG2 mRNA were elevated in both cell types cultured at 39°C, κ₀ transcripts could only be detected in S4 but not S4C2-1 cells (Fig. 3, C and D). This finding indicates that the κ-locus in S4C2-1 cells is not actively transcribed and therefore inaccessible to the recombinases. Hence, unlike S4, the recombination activity in S4C2-1 cells is only partially induced by v-Abl inactivation, i.e., inducing RAG1/2 expression, but without making their κ-locus accessible.

Germline κ₀-transcription is regulated by the NF-κB/Rel transcriptional factors (27). NF-κB/Rel transcription factors are sequestered in the cytoplasm by interaction with their inhibitor, IκBα, which upon various stimuli is targeted for degradation to allow NF-κB/Rel nuclear localization and activation (28). To directly assess NF-κB/Rel activation status, we analyzed NF-κB (p50/p52), RelA (p65), and IκBα proteins in both cell types under various conditions by Western blot. As shown in Fig. 4,A, S4 cells increased both cytoplasmic and nuclear levels of p50/52 and p65, but decreased IκBα upon incubation at 39°C. In contrast, S4C2-1 cells lacked significant elevation in NF-κB/Rel nuclear protein levels, which was also found in other tumor lines (Fig. 4,B). Unexpectedly, S4C2-1 cells had an elevated level of IκBα upon culture at 39°C, which is in striking contrast to the reduced IκBα levels in S4 cells (Fig. 4,A, lanes 3 and 9), as well as to those reported in other ts-Ab-MuLV cell lines (6, 29). To determine the functional activity of the NF-κB/Rel complex, the NF-κB/Rel DNA-binding activity was analyzed by EMSA with nuclear proteins prepared from S4 and S4C2-1 cells. Consistent with the level of nuclear NF-κB/Rel proteins detected in these two types of cells, we found a high level of NF-κB/Rel DNA binding in S4, but a low level in S4C2-1 cells (Fig. 4 C, lanes 3 and 6). Taken together, these data suggest that the signal transduction pathway from v-Abl inactivation to IκBα degradation and NF-κB/Rel activation is dysfunctional in these tumor cell lines, thus inhibiting κ-gene rearrangement.

To determine whether the lack of NF-κB/Rel activation in these tumor cells is unique to v-Abl inactivation or results from a general defect in NF-κB/Rel activation, we treated cells with a well-characterized NF-κB1/RelA activator, LPS (30). We examined the levels of κ₀ transcripts and nuclear NF-κB/Rel proteins in response to LPS stimulation. Clearly, very low levels of κ₀ transcripts and nuclear NF-κB/Rel proteins were detected in S4C2-1, whereas both were elevated in S4 cells, thus resembling their responses to v-Abl inactivation (Fig. 4, D and E). Therefore, S4C2-1 cells are unable to induce NF-κB/Rel activity in response to both v-Abl inactivation and LPS stimulation. This finding suggests that the mutations acquired by S4C2-1 cells are likely to reside downstream of where v-Abl and LPS-mediated signal transduction pathways converged.

We took advantage of the bcl-2-expressing ts-Ab-MuLV cell model system to demonstrate that inactivation followed by reactivation of v-Abl in SCID-ts cells can lead to malignant transformation and impaired differentiation of an inducible pre-B cell model. SCID-ts cells are defective in the DNA repair protein, catalytic subunit of DNA-dependent protein kinase (DNA-PKcs) (31). Although the defective DNA-PKcs has been associated with increased vulnerability of SCID mice to oncogenic transformation induced by ionizing radiation or p53-mutation (32, 33), its role in v-Abl-mediated transformation is unclear. It is possible that v-Abl inactivation leads to accumulation of unresolved V(D)J recombination intermediates in SCID-ts cells, which in turn increases their chances to acquire additional mutation for tumor formation. Alternatively, other cellular alterations caused by faulty DNA-PKcs may also interact with v-Abl signaling cascades during v-Abl reactivation to accelerate the transformation process. DNA-PKcs has been implicated to participate in signal transduction cascades that are not directly related to V(D)J recombination or dsDNA break repair (34, 35, 36). Thus, unresolved V(D)J recombination intermediates, additional cellular stresses due to defective DNA-PKcs, or both may make SCID-ts more susceptible than wt-ts to oncogenic selection.

Transformation of primary pre-B cells with Ab-MuLV has been shown to place cells in a state of crisis leading to either apoptosis or oligoclonal selection and proliferation (10, 16, 17). It has further been shown that this malignant transformation of v-Abl-transformed cells is enhanced by selection for mutations in tumor suppressors such as p16INK/p19ARF and p53 (10, 18). No abnormality of these genes, however, was found in our induced tumor lines (our unpublished observation). Because cell cycle progression by v-Abl occurs at the G₁-to-S transition through regulation of cyclin D1 and p27, it is possible that these genes would instead be altered in the SCID-ts model, especially given that the regulators of these genes are often the targets of oncogenic mutations (37, 38). Indeed, the malignantly transformed SCID-ts cells have only a modest increase in p27 and an incomplete cell cycle arrest upon v-Abl inactivation (Fig. 2), suggesting that the new transforming mutations in these tumor lines may prevent p27 elevation.

Several intertwined pathways may account for the altered regulation of p27 in SCID-ts cells during v-Abl inactivation and reactivation. Cyclin:cdk complexes, c-myc, and Foxo3a have been shown to modulate p27 at transcriptional and posttranslational levels in several cell models (20, 21, 22, 23). The ts-v-Abl system, in particular, requires an active cyclin D1:cdk4 complex to activate c-myc expression (39). Thus, the absence of both G₁ cyclin:cdk proteins and c-myc upon v-Abl inactivation likely prevents posttranslational modification and degradation of p27 in SCID-ts cells (22). Intriguingly, the levels of c-myc, inversely, and Foxo3a, directly, are correlated with p27 protein levels during v-Abl inactivation and reactivation in SCID-ts cells (Fig. 2, B–D). Several reports demonstrate that inhibition of Abl with a specific inhibitor, STI571, leads to elevated Foxo3a and p27 and reduced c-myc levels (36, 40). The elevated Foxo3a seen in SCID-ts cells during v-Abl reactivation may reflect their inability to transmit v-Abl-mediated signals to down-regulate p27, thus propagating a cellular status resembling senescence. Elevated Foxo protein has been implicated in maintaining cellular senescence with a long life span but limited growth (41). In contrast, the constitutive expression of c-myc and absence of Foxo3a in the tumor cells may promote p27 degradation, which in turn permits tumor cells to bypass the senescence for continuous growth.

DNA damage signaling cascades and other cellular stresses may also enhance Foxo3a activity to induce cell cycle arrest or apoptosis (42, 43, 44). Therefore, up-regulation of Foxo3a in SCID-ts cells may reflect an attempt to initiate a cell cycle checkpoint in response to unresolved recombination intermediates. In line with this argument, the low level of Foxo3a in tumor cells may be attributed to few or no recombination intermediates generated. Even though these tumor cells were found to elevate their RAG1/2 mRNA they failed to initiate recombination at the Ig-κ locus, presumably due to their gene inaccessibility, as revealed by the lack of κ₀-germline transcript. Lymphoid-specific genes BSAP, Pax5, and E2A activate RAG1/2 transcription in pre-B cells (45, 46), whereas κ_o-germline transcription relies primarily on NF-κB1/RelA and its downstream components, IRF-4 and Spi-B (6, 34, 47). It is clear that these pathways are differentially regulated during pre-B cell differentiation by v-Abl inactivation because tumor cells still retain the ability to up-regulate RAG1/2 expression in response to v-Abl inactivation, yet are unable to activate NF-κB/Rel to transcribe the κ locus. This suggests that inhibition of NF-κB/Rel is advantageous to cellular survival, presumably by preventing Ig-L gene rearrangement and reducing the chances of defective recombination.

In addition to inducing pre-B cell differentiation and Ig-L-gene rearrangement, NF-κB/Rel transcription factors may contribute to the prolonged cell cycle arrest, because their levels appear to be inversely correlated with cell cycle progression in the SCID-ts cells. This finding is consistent with previous reports that overexpression of RelA and NF-κB1 induce apoptosis or reduce proliferation, respectively, in Ab-MuLV-transformed pre-B cells (42, 48). NF-κB/Rel activation triggered by LPS stimulation and v-Abl inactivation relies on the inhibitory κB factor kinase (IKK) complex (IKKαβγ) to induce degradation of IκBα and release NF-κB/Rel transcription factors to translocate to the nucleus (29, 49, 50). This IKK-dependent signaling pathway may be impaired in S4-derived tumor cells because they contain elevated levels of IκBα at 39°C despite an inactive v-Abl (Fig. 4 A). Taken together, the inability of these tumor cells to respond to both v-Abl inactivation and LPS stimulation argues that the IKK complex or an upstream activator of IKK may be deregulated in tumor cells and may serve as a potential target for malignant transformation.

Activation of NF-κB/Rel can be induced by a wide variety of stimuli, including atypical inducers such as irradiation or chemical agents (51, 52). Over the last several years, studies have demonstrated that certain genotoxic agents activate NF-κB/Rel, resulting in apoptosis and cell cycle arrest rather than the traditional role as a tumor promoter (for review, see Ref. 53).

Incidentally, both ATM and DNA-PKcs have been suggested to participate in the regulation of IKKβ and IκBα in response to DNA-damaging agents to prevent apoptosis and allow time for repair (54, 55, 56). In light of these studies, it is possible that the combination of both elevated DNA damage induced by v-Abl inactivation and DNA-PKcs deficiency in SCID-ts cells would lead to heightened activation of Foxo3a and NF-κB/Rel to maintain survival without growth. Yet unique to these pre-B ts-Ab-MuLV cells is the concurrent activation of Foxo3a and NF-κB/Rel, which have been described (57) in other studies to function in mutually exclusive pathways. It is possible that the v-Abl-signaling cascade involves several down-stream pathways, which may independently regulate Foxo3a and NF-κB/Rel. For example, upon v-Abl inactivation, phosphorylation and subsequent degradation of Foxo3a is inhibited (20), which leads to increased levels of Foxo3a in pre-B ts-Ab-MuLV cells. Although Foxo3a may exert a direct inhibitory effect on NF-κB/Rel activation (57), a separate pathway may activate NF-κB/Rel independently of Foxo3a. Conversely, the presence of genotoxic stress induced in the SCID-ts cell model may be the dominating factor, activating one or both of the Foxo3a and NF-κB/Rel pathways. These multiple signaling pathways induced by v-Abl inactivation at 39°C may be conflicting, leading the cells to a senescent state. Therefore, alterations in regulating Foxo3a, c-myc, p27, and NF-κB/Rel may allow SCID-ts cells to escape from the senescent stage and possibly undergo malignant transformation. Further studies are needed to determine possible upstream activators of these components.

We thank Dr. N. Rosenberg for providing the ts-Ab-MuLV cell lines, 103-B4. We greatly appreciate technical support from Arizona State University, W. M. Keck Laboratory for phosphorimaging analyses, and DNA Laboratory at School of Life Sciences.

The authors have no financial conflict of interest.