Isolation and characterization of dominant and recessive IL-3 ... - Nature

Oncogene (2006) 25, 6595–6603

& 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc

ORIGINAL ARTICLE

Isolation and characterization of dominant and recessive IL-3-independent hematopoietic transformants KF Kiser1, M Colombi1 and C Moroni Institute for Medical Microbiology, Department of Clinical-Biological Sciences, University of Basel, Basel, Switzerland

Retroviral integration mutagenesis and treatment with the frameshift mutagen ICR191 were used to transform v-H-ras expressing PB-3c cells to interleukin-3 (IL-3) independence. Six clones displayed viral integrations into the 30 region of the IL-3 gene thus acting posttranscriptionally by disrupting the AU-rich instability element. Two clones contained reverse orientation integration into the raf-1 gene revealing an enhancer insertion mechanism. Growth by this mechanism was sensitive to the Raf-1 inhibitor BAY 43-9006 and the Mek inhibitor U0126. Following treatment with ICR191, IL-3-independent clones were recovered and studied by cell fusion. With 21/22 clones, IL-3 independence resulted from a recessive mechanism as cellular hybrids with parental cells reverted to IL-3 dependence. Recessive clone D2c displayed increased phospho-Erk1/2 levels and was growth sensitive to U0126, but not to BAY43-9006. The single dominant clone, D5a, showed no signs of mitogenactivated protein kinases pathway activation but displayed constitutive phosphorylation of Stat5. We conclude that PB-3c has several options to acquire IL-3 growth autonomy involving transcriptional or post-transcriptional mechanisms affecting the distal regulators Erk or Stat5. The reported panel of independent dominant and recessive transformants should provide a useful tool for inhibitor profiling. Oncogene (2006) 25, 6595–6603. doi:10.1038/sj.onc.1209673; published online 15 May 2006 Keywords: mutagenesis; Raf-1; ICR191; mast cells; stat

Introduction Malignant transformation of hematopoietic cells can involve mechanisms, which abrogate growth factor dependency, stimulate the cell cycle, antagonize apoptosis or interfere with differentiation resulting in the production of leukemic blast cells. Understanding these Correspondence: Dr C Moroni, Institute for Medical Microbiology, Department of Clinical-Biological Sciences, University of Basel, Petersplatz 10, Basel, Basel-Stadt, 4051, Switzerland. E-mail: [email protected] 1 These two authors contributed equally to this work. Received 27 December 2005; revised 5 April 2006; accepted 7 April 2006; published online 15 May 2006

processes is expected to provide a basis for designing drugs for rational therapy. The molecular alterations underlying the malignant process involve frequently specific chromosomal translocations activating transcription factors, tyrosine kinases or other mutations affecting cell signaling. Cytokine-dependent cell lines provide useful model systems to study aspects of hematopoietic malignancy as growth factor independence, a hallmark of the transformed state, can easily be selected for. Interleukin-3 (IL-3) has been a prototype in the investigation of cytokine-induced signaling mechanisms. This cytokine is produced mainly by T cells (Ihle et al., 1981), but also by mast cells (Wodnar-Filipowicz et al., 1989), and is recognized by a dimeric receptor, whose a- and b- subunits are members of the cytokinereceptor family. IL-3, GM-CSF and IL-5 share the common b-subunit of the receptor. IL-3 binding leads to b-chain autophosphorylation on tyrosine residues and activation of the Jak-Stat, mitogen-activated protein kinases (MAPK) and PI3K-/PKB pathways resulting in proliferation and inhibition of apoptosis (for references see Blalock et al. (1999)). IL-3-independent growth has been observed via activating mutations of the common IL-3R b-subunit (D’Andrea et al., 1994; Hannemann et al., 1995; McCormack and Gonda, 1997) and a variety of mechanisms involving bcr-abl (Jiang et al., 2002), v-erbB (Shounan et al., 1995; McCubrey et al., 2004), v-src (Overell et al., 1987), c-kit (Piao and Bernstein, 1996), activated Stat5 (Onishi et al., 1998) pim-1 (Nosaka and Kitamura, 2002), Flt3 (Hayakawa et al., 2000), mpl (Onishi et al., 1996b) and H-ras (Andrejauskas and Moroni, 1989). For a recent review see Steelman et al. (2004). The PB-3c cell line, used in this work, is a near diploid immortalized bone marrow-derived mast cell line that is strictly dependent on IL-3 with virtually no spontaneously occurring mutations leading to growth autonomy. IL-3-independent growth, however, occurred following conditional expression of the human H-Ras gene at very high levels from a MMTV-long terminal repeat (LTR), which allowed cells to switch from IL-3dependent to inducer (dexamethasone)-dependent growth and back (Andrejauskas and Moroni, 1989). Weaker expression of v-H-ras, driven by a Moloney LTR, led to a reduction in the requirement of exogenous IL-3, but not to autonomous growth. Upon in vivo inoculation, however, v-H-ras expressing PB-3c cells progressed into autocrine IL-3 producing tumor cells

Isolation and characterization of dominant and recessive IL-3 KF Kiser et al

6596

which grew readily in vitro without added IL-3 (Nair et al., 1989). Interestingly, the autocrine state was induced by both transcriptional and post-transcriptional mechanisms. The transcriptional mechanism was dominant, as cell hybrids between tumors and parental PB-3c cells remained autocrine and oncogenic, whereas corresponding hybrids with the other tumor class reverted to IL-3 dependence with loss of oncogenicity upon cell fusion (Hirsch et al., 1993). This indicated that the establishment of the IL-3 autocrine state involved a lossof-function step that can be rescued by cell fusion. Ectopic expression of TTP, a protein promoting decay of certain AU-rich element (ARE) containing mRNAs including IL-3, suppressed tumor formation in vivo when introduced into a line with stabilized IL-3 mRNA (Stoecklin et al., 2003). However, the molecular lesion in the class of tumors with recessive mRNA stability is not known. In this work, we applied retroviral insertion mutagenesis as well as chemical mutagenesis with the frameshift mutagen ICR191 to a v-H-ras expressing clone of IL-3dependent PB-3c cells and selected for IL-3-independent growth. The aim was to isolate dominant and recessive transformants to be characterized for pathway activation and inhibitor sensitivity.

Results Cell transformation by retroviral integration PB-3c-15V4 cells are IL-3-dependent and driven rapidly into apoptosis after cytokine removal. Based on this property, we attempted to identify by retroviral mutagenesis genes mediating IL-3 independence. Cells were infected with a defective retrovirus encoding green fluorescent protein (GFP), which was generated by transfecting the retroviral packaging line platE (Morita et al., 2000). IL-3 was removed 24 h after infection and IL-3-independent clones were isolated within 1–2 weeks, subcloned and characterized. From about 80 unrelated pools consisting each of 3 107 cells, we isolated eight IL-3-independent clones. No IL-3-independent clone was obtained from noninfected cells (>109). The retroviral integration site of the eight independent clones was identified by an inverse polymerase chain reaction (PCR) strategy. Genomic DNA was digested with TaqI, which cuts once within the LTR and, presumably, within a reasonable distance, in the flanking DNA. The resulting fragments were circularized and nested PCR was performed with 2 LTR-specific primer pairs. Submitting the sequenced flanking DNA to a database search revealed two integration hot spots, namely the IL-3 and the raf-1 genes. In six clones, integrations into the IL-3 locus were located within 66 nucleotides upstream or 156 nucleotides downstream of the TAA stop codon (Figure 1a arrows), which is upstream of the six distal AUUUA pentamers of the ARE responsible for rapid mRNA degradation (Stoecklin et al., 1994). These integration events apparently functionally dissociate the ARE from Oncogene

the normally short-lived IL-3 transcript, thus causing mRNA stabilization. Clone GIC was selected from what we called the IL-3 activation group for further characterization. Its retroviral integration is located between the AUUUA-pentamers 2 and 3, 25 nucleotides upstream of the third pentamer (Figure 1a). When tested by actinomycin D chase experiment, IL-3 mRNA of GIC cells was stable over 2 h (Figure 1b), whereas an ARE carrying GFP reporter transcript from control cells was expectedly short-lived. In two of the clones, the integration occurred in the raf-1 locus either in the first intron or in the promoter region about 2 kb upstream of exon 1 (Figure 1c). In both cases, integration was in reverse orientation, identifying enhancer insertion as the mechanism. Clone IIF2G was characterized further. When compared to the precursor line PB-3c-15V4, it displayed strongly elevated Raf-1 protein levels (Figure 1d, left panel) and in addition augmented phospho-Mek1/2 levels (Figure 1d, right panel). In contrast, the IL-3 integration clone GIC used for comparison expressed only moderate levels of Raf-1. To see whether transformation to growth factor independence had occurred by an autocrine mechanism, conditioned medium of GIC and IIF2G cells were assayed on IL-3-dependent parental PB-3c-15V4 cells by 3 H-thymidine incorporation (Figure 1e). At a 1:30 dilution of GIC supernatant maximal 3H-thymidine incorporation was observed, whereas IIF2G showed no bioactivity at any dilution tested. These data suggest that IL-3 independence of PB-3c-15V4 cells can be achieved by either inducing autocrine IL-3 secretion or by upregulation of the MAPK pathway via Raf-1. The Raf-1 clone IIF2G is sensitive to inhibition of Raf-1 and Mek To investigate the role of the MAPK pathway in mediating IL-3 independence, we blocked this pathway at two different levels by a Raf-1 inhibitor, BAY 43-9006, or the Mek inhibitor U0126 and monitored 3 H-thymidine incorporation after 42 h of incubation. The experiments shown in Figure 2a were performed in the absence (left panels) or presence (right panels) of IL-3 to test for possible IL-3-dependent rescue from the MAPK pathway inhibition. The Raf-1 clone IIF2G showed growth inhibition at 3 mM U0126 both with or without added IL-3 (Figure 2a, upper panels), whereas GIC and parental PB-3c-15V4 cells were inhibited by U0126 only at the apparently toxic concentration of 10 mM. The Raf-1 inhibitor BAY 43-9006 inhibited growth of IIF2G at a concentration of 4 mM in a statistically significant fashion, both in the presence or absence of IL-3 (Figure 2a, lower panels). GIC and precursor PB-3c-15V4 cells showed inhibition at higher concentrations, which is probably nonspecific as suggested by the more linear decrease in growth. The effect of the inhibitors on Erk1/2 phosphorylation was monitored by Western blotting (Figure 2b), with total Erk1/2 serving as a loading control. As expected, IIF2G cells showed strong Erk1/2 phosphorylation,


6597

Figure 1 Characterization of clones transformed by retroviral integration. (a) IL-3 mRNA with its 50 cap, the coding region, the AU-rich element (ARE) and the poly-A tail is shown. The ARE consists of eight AUUUA pentamers (filled circles) and one hexamer shown left. Top: Arrows mark retroviral integration sites of six transformed clones. The integration site of the clone GIC is indicated. All these integrations dislocate the functional ARE and thus stabilize the IL-3 mRNA. Bottom: Published mutagenesis data (Stoecklin et al., 1994) is shown for comparison. Open circles correspond to mutated AGGUA pentamers. Shown on the right is the mRNA halflife. Note that the functional ARE, that is, the six terminal pentamers are distal to the retroviral integration sites shown at the top. (b) Actinomycin D chase experiment. RNA from GIC or 15V4wt18 cells containing GFP with the 30 -terminal ARE from IL-3 was hybridized with a IL-3 30 UTR-specific probe. GAPDH served as loading control. (c) Integration into the raf-1 gene. One insertion (IIF7L) is located 2 kb upstream of the first exon, the other is in the first intron. (d) Western blots showing Raf-1 (left panel) and phospho-Mek1/2 levels (right panel). Erk1/2 and Mek1/2 levels serve as loading controls. Clone IIF2G was grown in the presence and absence of IL-3 as indicated. (e) 3H-thymidine incorporation assay. PB-3c-15V4 cells were grown for 42 h in the presence of the indicated supernatants.

independently of added IL-3 (lanes 3 and 5). After 2 h with 1 mM U0126 (Figure 2b, left) the phospho-Erk1/2 levels were significantly reduced. GIC cells, in contrast, showed only moderate levels of phospho-Erk1/2, which were also downregulated by the Mek inhibitor. Pre-

cursor PB-3c-15V4 cells had only baseline activation of the MAPK pathway. The Raf-1 inhibitor BAY 43-9006 showed a strong impact on Erk1/2 phosphorylation of IIF2G after 2 h treatment at a concentration of 4 mM (Figure 2b, right panel). This effect was again indepenOncogene


6598

Figure 2 Sensitivity to MAPK pathway inhibitors. (a) Sensitivity of IIF2G and GIC cells to the Mek inhibitor U0126 (upper panels) and the Raf-1 inhibitor BAY 43-9006 (lower panels) was monitored by 3H-thymidine incorporation. The cells were grown for 42 h in presence of indicated concentrations of inhibitor either in the absence (left panels) or the presence of IL-3 (right panels). The IL-3dependent precursor cell line PB-3c-15V4 was used as a control. Data are average determinations from three independent experiments (standard deviations are shown unless to small to be represented). (b) Phospho-Erk1/2 activation. Cells were grown with or without IL-3 and incubated with 1 mM UO126 or 4 mM BAY 43-9006 as indicated for 2 h, and processed for Western blotting.

dent of added IL-3. Together, these observations indicate that IL-3 independence of IIF2G is mediated by increased activation of the MAPK pathway, whereas the sensitivity of autocrine GIC cells to the inhibitor is in the range of the IL-3-dependent precursor cells. Dominant and recessive IL-3-independent clones can be generated by frameshift mutagen ICR191 We used frameshift mutagen ICR191 as a second transformation tool and used a GFP-expressing PB-3c-15V4 clone as target (15V4wt18). Assuming that both alleles of hypothetical loci have to be knocked out by the mutagen, we anticipated that serial mutagenesis would be required as previously shown for the knockout of the BRF1 gene (Stoecklin et al., 2002). After five rounds, the transformation frequency was about 4.5 105 in v-H-ras expressing 15V4 cells. No transformants were detected after up to five rounds of mutagenesis in PB-3c-20puro cells (data not shown). As ICR191 is a frameshift mutagen and is expected to generate predominantly loss-of-function mutations, the resulting mutants should be mainly recessive. To verify this proposition, isolated clones were fused with IL-3-dependent puromycin resistant PB-3c-20puro cells. As 15V4wt18 cells contain a plasmid coding for Oncogene

hygromycin resistance, hybrid cells could be selected with puromycin and hygromycin. Of 22 hybrid clones tested, all were near-tetraploid in fluorescence-activated cell sorting (FACS) analysis (data not shown), and 21 had reverted to IL-3 dependence. Interestingly, one clone (D5a) was dominant and grew without IL-3. The data are shown in Figure 3, which also includes a control fusion between PB-3c-20puro and 15V4wt18 cells and shows data from fusions with three recessive and the single dominant clone. We focused on the recessive clone D2c and the dominant clone D5a for further characterization. D2c cells are sensitive to Mek inhibitor U0126 and can be rescued by IL-3 It is reported that conditionally-active Mek1 can relieve IL-3 dependence of TF-1, FDCP1 and FL5.12. cells (Blalock et al., 2000), but not of Ba/F3 cells (Perkins et al., 1996). To investigate a possible role for Mek1/2 or the MAPK pathway in mediating IL-3 independence of D2c and D5a cells, we treated cells with U0126 and BAY 43-9006 and measured 3H-thymidine incorporation after 42 h cultivation (Figure 4a). The recessive clone D2c was strongly inhibited at 1 mM U0126 or higher concentrations (Figure 4a, upper left panel).


6599

Figure 3 IL-3 independence of cell hybrids. Four IL-3-independent cells obtained after treatment with ICR191 were fused to IL-3dependent PB-3c-20puro in the presence of IL-3. To assess growth factor independence, hybrids were grown for 4 days in the presence or absence (open symbols) of IL-3. For control, IL-3-dependent 15V4wt18 were also fused to PB-3c-20puro cells.

Interestingly, this growth inhibition could be rescued by added IL-3 (Figure 4a, upper right panel). The dominant clone D5a was not inhibited by the Mek inhibitor, similar to the parental PB-3c-15V4 cells. Inhibition at 10 mM U0126 most likely reflects toxicity. When the extent of Erk1/2 phosphorylation and its sensitivity to the inhibition was analysed (Figure 4b), we observed in D2c cells a strong phospho-Erk signal, which was sensitive to U0126 (left panel). BAY 43-9006 was not inhibitory in the presence of IL-3, and only slightly in its absence (right panel). Interestingly, neither the protein levels nor the phosphorylation status of Mek1/2, the upstream regulator of Erk1/2, were elevated (Figure 4c), pointing to a Mek1/2-independent mechanism. The data of the recessive clone D2c are in agreement with the results of the proliferation experiments and indicate that Erk1/2 had become critically activated, suggesting that growth rescue from UO126 by added IL-3 might have occurred by ‘pathway switching’. In the dominant clone D5a the phospho-Erk1/2 levels were weaker than in D2c and insensitive to BAY 43-9006, but sensitive to UO126 (Figure 4b). Taken together with the proliferation data from Figure 4a it appeared that the MAPK pathway is not responsible for IL-3-independent growth of D5a. Stat5 is constitutively active in D5a cells IL-3 is known to activate the Jak/Stat pathway whereby receptor-activated Jak2 phosphorylates Stat5, which

dimerizes and translocates to the nucleus where it activates transcription of its target genes (Mui et al., 1995). To monitor possible activation of this pathway in our mutants, electrophoretic mobility shift assays (EMSA) were performed using the Stat5 binding sequence as a specific probe and nuclear extract of the indicated cell lines. PB-3c-15V4 cells grown in IL-3 showed a weak band (Figure 5a upper panel, lane 2), which disappeared after starvation for 6 h in IL-3-free medium containing 0.2% fetal calf serum (FCS) (lane 3). If, however, the starved cells were pulsed for 10 min with IL-3 containing medium, strong Stat5 activation could be observed (lane 4). To prove the specificity of the band, we used increasing amounts of unlabeled Stat3 binding sequence as unspecific, and unlabeled Stat5 as specific competitors. Even at a 1000-fold excess of the Stat3 binding site, the bandshift remained while a 10-fold excess of unlabeled Stat5 sequence led to a loss of signal. Next, we tested transformed cells for activation of Stat5 (Figure 5a lower panels). As expected, the autocrine GIC clone displayed Stat5 activation mediated by IL-3 (lane 11). The recessive clone D2c (lanes 5 and 6) and the Raf-1 clone IIF2G (lanes 9 and 10) showed no Stat5 activation either in absence or in presence of IL-3. The dominant clone D5a, in contrast, showed constitutive Stat5 activation independently of IL-3 (lanes 7 and 8). To confirm these findings, we performed Western blotting using a phospho-Stat5 specific antibody (Figure 5b). A phosphorylation signal could be induced by growth factor withdrawal and readdition (lanes 1-3), and was present in PB-3c-15 and PB-3c-20 cells in IL-3. The transformed clone D5a showed a strong signal consistent with the bandshift data. D2c cells grown in the absence of IL-3 were negative, while addition of IL-3 evoked a weak signal. This is consistent with the IL-3 rescue effect from UO126 inhibition observed in Figure 4.

Discussion In this study, we have generated dominant and recessive hematopoietic transformants using retroviral insertion mutagenesis in parallel to treatment with the frameshift mutagen ICR191. Insertion mutagenesis would be expected to yield primarily dominant transformants, whereas ICR191 treatment should generate predominantly recessive loss-of-function mutants. Using a single cellular target, we document here that conversion from growth factor dependent to autonomous growth, a critical step of oncogenesis, can be achieved via both procedures involving different pathways and mechanisms. Using retroviral insertion mutagenesis, we observed that activation of the IL-3 locus was a preferred mechanism of oncogenic conversion. Interestingly, integration occurred in the coding region just before the stop codon (two clones), or in the 30 untranslated region (30 UTR) upstream of the functional domain of the ARE (4 clones). The ARE of IL-3 consists of eight AUUUA pentamers and one upstream AUUUUA Oncogene


6600

Figure 4 Sensitivity of dominant clone D5a and recessive clone D2c to MAPK pathway inhibitors. (a) Cells were grown as indicated in the absence (left panels) of presence (right panels) of IL-3, with indicated concentrations of U0126 (upper panels) or BAY 43-9006 (lower panels). 3H-thymidine incorporation was determined after 42 h. Data are average determinations from three independent experiments (standard deviations are shown unless to small to be represented). (b) Western blotting was performed with D2c, D5a and PB-3c-15V4 cells after treatment for 2 h with 1 mM U0126 (right) or 4 mM BAY 43-9006 (left) either in the presence or absence of IL-3. (c) Levels of phospho-Mek1/2 and Mek1/2 monitored by Western blotting.

hexamer (shown as circles in Figure 1a). A previous systematic mutational analysis of the relative contribution of these motifs to mRNA decay had revealed, that the six downstream, but not the three upstream motifs, are responsible for mRNA decay promoting activity (Stoecklin et al., 1994). Correspondingly, all insertions observed dislocated the IL-3 mRNA from the functionally active downstream motifs. Particularly revealing are two mutants, including GIC, in which the retrovirus integrated between motif 3 and 4 and thus maintained the presence of the three upstream motifs within the mRNA. As these motifs were insufficient to promote rapid mRNA decay, the altered mRNA was stable leading to IL-3 overexpression and autocrine growth (Figure 1). It is surprising that no integration event in the promoter region has been observed, particularly as we have previously observed in PB-3c-derived tumor lines the spontaneous translocation of endogenous retroviral elements into the promoter (Hirsch et al., 1993). Retroviral translocation into the 30 UTR of IL-3 has been observed also by others (Algate and McCubrey, 1993). Oncogene

The raf-1 gene represented a second integration locus in our study. Two clones showed retroviral integrations in the promoter region or the first intron, both in reverse orientation with respect to raf-1, suggesting enhancer insertion as the putative mechanism. These clones overexpress Raf-1 with corresponding strong activation of the MAPK pathway (Figure 1c and unpublished data), and showed sensitivity to both the Mek inhibitor UO126 and the Raf-1 inhibitor BAY 43-9006, whereas the autocrine GIC clone was insensitive to these drugs (Figure 2a). McCubrey and collaborators have reported that wild-type Raf-1 overexpression did not readily relieve IL-3 dependence of FDCP1 cells (Hoyle et al., 2000) nor transform FL5.12 (Shelton et al., 2003) or TF-1 cells (McCubrey et al., 1998), whereas Raf-1 activated by truncation abrogated the cytokine dependence in all three cases. That we isolated transformed Raf-1 clones from PB-3c-15V4 cells which express v-H-ras is surprising as one would not expect to require a second mutation in the MAPK pathway. The greatly increased levels of Raf-1 (Figure 1d, left panel), p-Mek1/ 2 (Figure 1d, right panel) and p-Erk1/2 (Figure 2b, right


6601

Figure 5 Constitutive Stat5 activation in dominant clone D5a. (a) Activation of the Jak/Stat pathway was assessed by electrophoretic mobility shift assay with radiolabeled Stat5 binding sequence. The specificity of Stat5 binding was tested by competition with unlabeled Stat3 or unlabeled Stat5 binding sequence. Lysates of PB-3c-15V4 cells were starved for 6 h in 0.2% FCS and subsequently stimulated with IL-3 (upper panel). The lower panel shows Stat5 activation of indicated cell lines after culture in the presence or absence of IL-3. (b) Stat5 activation revealed by Western blotting using an anti-p-Stat5 antibody. Total Stat5 served as loading control.

panel) indicate that Raf-1 overexpression strongly boosts the pathway. When 15V4wt18 cells were treated with a frameshift mutagen, 21 out of 22 clonal isolates had reverted to factor dependency following cellfusion to parental IL-3dependent cells, revealing that the mutation affecting growth was recessive. Interestingly, one clone (D5a) conferred factor independence to the parental cell hybrid, arguing for a dominant mutation. The underlying mechanism could be the generation of a truncated protein following frameshift mutation, which may exert dominant-negative inhibition on a growth-inhibiting protein, or a truncation in the IL-3 receptor protein, as observed to occur in acute leukemia (Testa et al., 2004). Cloning of the postulated transforming gene should reveal the mechanism. When tested by growth and biochemical criteria, dominant clone D5a differed from recessive clone D2c. A hallmark of the former was phosphorylation of Stat5, which is known to correlate with growth factor independence of megakaryocytic cell lines (Liu et al., 1999), and occurs after Flt3 mutation (Hayakawa et al., 2000) and bcr-abl expression (Jiang et al., 2002). The mechanism of Stat5 activation is not known, but is not due to protein truncation/activation induced by the frameshift mutagen, as its molecular weight was not altered (Figure 5b, lanes 8 and 9). Stat5

phosphorylation was absent in IL-3-free D2c and in the raf-1 integration clone IIF2G. In contrast, D2c but not D5a displayed elevated p-Erk levels, and growth sensitivity to the Mek inhibitor U0126. This might indicate that in D2c cells, a loss-of-function event may have activated the MAPK pathway. However, in contrast to the raf-1 integration clone IIF2G, where phosphorylation and growth was sensitive to both the Raf-1 and Mek inhibitors, D2c was sensitive to UO126 but not the Raf-1 inhibitor BAY43-9006. The difference of the latter drug on the two cells is quite striking (Figure 2b and 4b). A possible mechanism may be the loss of a phosphatase in D2c cells, which negatively regulates Erk phosphorylation. Interestingly, growth inhibition mediated by UO126 was rescued by addition of IL-3 in D2c (Figure 4a), but not in IIF2G cells (Figure 2a), further arguing for different underlying mechanisms. This work documents that multiple mechanisms can transform a single target cell to autocrine and nonautocrine states via genetically dominant and recessive mutations and different signaling pathways. This not only highlights the need for developing pathway- or mechanism-specific inhibitors, but also to eventually apply therapeutic drugs in conjunction with corresponding molecular tests to verify the underlying mechanism. Oncogene


6602 Table 1 Transformation Cell

Agent

Mechanism

PB-3c GIC

Retrovirus

IIF2G D2c D5a

Retrovirus Frameshift mutagen Frameshift mutagen

–

– IL-3, autocrine, post-transcriptional Raf-1 Unknown Unknown

Activation status Geneticsa

p-Stat5

p-Erk

Growth inhibition by U0126

BAY 43-9006 Inhibition rescued by IL-3

Dominant

+ +

+ +

Dominant Recessive Dominant

+++

+++ ++ +

+ +

+

No Yes

a

Determined by somatic cell fusion.

A well-know example is Gleevec, with exquisit specificity for the bcr-abl and c-kit tyrosine kinases (Druker et al., 1996). In fact, bcr-abl transformed PB-3c cells to IL-3 independence, with sensitivity to Gleevec and growth rescue by IL-3 (our unpublished data), while the clones described here were insensitive to the drug. Thus, a panel of cells as described here and summarized in Table 1, may be useful for profiling anticancer drugs to reveal pathway specificity. Candidate drugs acting against one but not on other transformants should provide interesting leads for further study and development. Materials and methods Reagents Chemicals were purchased from: actinomycin D (Alexis, Lausen, Switzerland), U0126 (Alexis), BAY 43-9006 (BAYER), phosphatase inhibitor cocktail I and II (Sigma, St Louis, MO, USA), Complete protease inhibitor cocktail (Roche, Rotkreuz, Switzerland), puromycin and hygromycin (Calbiochem, Darmstadt, Germany). Mouse anti-phosphop44/42, rabbit anti-phospho-MEK1/2, rabbit anti-MEK1/2, rabbit anti-p44/42, mouse anti-phospho-Stat5 and rabbit antiStat5 antibodies were obtained from Cell Signaling, Beverly, MA, USA, mouse anti-Raf-1 antibody from BD Biosciences, San Jose, CA, USA. Secondary antibodies horseradishperoxidase-coupled goat anti-mouse IgG (DAKO), horseradish-peroxidase-coupled goat anti-rabbit IgG and alkaline phosphatase-coupled goat anti-rabbit antibody from Southern Biotechnology Associates, Birmingham, AL, USA. Cell culture Culture conditions and PB-3c mast cell subclones 15, 20 and 15V4, the latter expressing v-H-ras and neomycin resistance, have been described previously (Nair et al., 1989). 15V4wt18 cells are PB-3c-15V4 cells transfected with a plasmid carrying GFP with a 30 UTR derived from IL-3 and hygromycin resistance. PB-3c-20puro cells carry a puromycin resistance gene. Conditioned medium from 15-IL3MXh-6a cells (Stoecklin et al., 1994) served as source of IL-3. Retroviral integration mutagenesis PlatE cells (Morita et al., 2000) were transfected with 120 mg of retroviral pMxEGFP plasmid (Onishi et al., 1996a) using lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturers protocol. At 24 h after transfection, virus-containing supernatant was harvested and filtered using a 0.45 mm pore filter. PB-3c-15V4 cells (3 107) were infected Oncogene

with 25 ml viral supernatant containing 10 mg/ml hexadimethrin (Polybrene, Fluka, Buchs, Switzerland) for 5 h at 371C. After 24 h, IL-3 was removed to select for factor independence. Viable cells were enriched by Ficoll-Paque (Amersham Pharmacia, Little Chalfont, UK). Inverse PCR Genomic DNA was digested with TaqI (NEB), and the resulting fragments were circularized with T4 DNA ligase (NEB) and amplified by nested PCR using hot start Taq polymerase (Qiagen, Hilden, Germany). Two LTR-specific primer pairs were used. 11 PCR primers: TV296 ATCCGAC TTGTGGTCTCGC and M2144 CAAAATGGCGTTACTT AAGC. 21 PCR primers: TV297 AGTGATTGACTACCC GTCAGC and M2145 CTTGCCAAACCTACAGGTGG. Products were sequenced and analysed using the CEQ 6000 sequencer (Beckman-Coulter, Fullerton, CA, USA) following standard protocol. Database search of the mouse genome (Blast) identified the region of viral insertion. Northern blotting Cells were treated with 5 mg/ml actinomycin D (Alexis) and RNA was extracted 0, 1 and 2 h after addition following the method of Gough (Gough, 1988). 30 mg RNA were processed and hybridized as described previously (Benjamin et al., 2004). 3 H-thymidine incorporation assays 104 cells per well were plated in 100 ml medium containing indicated concentrations of inhibitor in 96-well microtiter plates. Cells were grown for 24 h and labeled with 0.5 mCi 3 H-thymidine (Amersham Pharmacia, specific activity 70-86 Ci/mmol) per well for 18 h. Incorporation was measured using Matrix 96 direct beta counter. Counts were normalized to inhibitor-free controls.

Western blotting Cells were incubated for 2 h with 1 mM U0126 or 4 mM BAY 43-9006 and extracted with RIPA lysis buffer (120 mM NaCl, 50 mM Tris/HCl pH 8.0, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS) containing 1 mM phenylmethylsulphonyl fluoride (PMSF), Complete protease inhibitor cocktail (Roche) and phosphatase inhibitor cocktail I and II (Sigma). 30 mg total protein lysate was resolved on a Tris-Glycine-Gradient-Gel 420% (Anamed, Darmstadt, Germany) and transferred to Immobilon-P membranes (Millipore, Bedford, MA, USA). Membranes were blocked in TBS-0.1% Tween-20 (TBS-T) containing 2% Blocking Agent (Amersham Pharmacia) and incubated overnight at 41C with the appropriate antibody. The membrane was washed three times for 15 min in TBS-T before applying the secondary antibody for 45 min at room


6603 temperature. After three washing steps, the signal was detected by chemiluminescence with ECL Advancet Western blotting Detection Kit (Amersham Biosciences) or CDP star (Roche) depending on the type of secondary antibody. ICR191 treatment 15V4wt18 cells, 108, were treated with 2 mg/ml ICR191 (Sigma) for 2 h at 371C in IL-3 containing medium, and allowed to recover for 24 h. This procedure was repeated over several rounds. IL-3-independent clones were selected by IL-3 removal and viable cells were enriched by Ficoll-Paque (Amersham Pharmacia). Cell fusion 107 PB-3c-20puro cells and 107 IL-3-independent 15V4wt18 cells were mixed and fused by electroporation in serum-free IMDM at 220 V and 900 mF. Hybrids were selected in IL-3 containing medium supplemented with puromycin (2 mg/ml) and hygromycin (1000 U/ml). Hybrid cells were subsequently cultured in the absence and presence of IL-3 to assess dominance or recessivity of mutations. Nuclear protein extraction and EMSA PB-3c-15V4 cells were starved for 6 h in IL-3-free medium containing 0.2%. FCS followed by stimulation with IL-3 containing medium for 10 min. Cell pellets were resuspended in cold EMSA buffer (20 mM Hepes pH 7.9, 75 mM NaCl, 1 mM

DTT, 2 mM MgCl2, 0.1% Triton X-100, 1 mM PMSF, Phosphatase inhibitor cocktail I and II (Sigma) and Complete Mini (Roche)) and dounced 30 times. Nuclei were pelleted, washed and extracted with EMSA buffer containing 300 mM KCl for 30 min on ice. Nuclear extract was supplemented with 5% glycerol. Stat1/Stat3 binding site (SIE element) of the c-fos promoter was used for competition: 50 GATCTGCTTCC GGAACGT30 sense oligo and 50 ACGTTCCGGAAGCA GATC30 antisense oligo (Nagata and Todokoro, 1996). As Stat5 binding site the bovine b-casein promoter was used: 50 AGATTTCTAGGAATTCAAATC30 sense oligo and 50 GA TTTGAATTCCTAGAAATCT30 antisense oligo (Joung et al., 2003). The Stat5 sense oligo was 50 radiolabeled and annealed to unlabeled Stat5 antisense oligo. Nuclear extract, 10 mg, were incubated with Stat5 probe (5000 c.p.m.) in the presence of 2 mg poly(dIdC) (Amersham Pharmacia) and 0.5 mM DTT for 30 min on ice. For competition assays 1, 10, 100 and 1000 molar excess of unlabeled nonspecific Stat3 or specific Stat5 competitor sequence were mixed with the radiolabeled probe. Bands were separated by 4% native acrylamide gel electrophoresis. The gel was exposed to phospho-imager screen (Bio-Rad). Acknowledgements We thank Dr Don Benjamin and Bernd Rattenbacher for discussion and comments on the manuscript.

References Algate PA, McCubrey JA. (1993). Oncogene 8: 1221–1232. Andrejauskas E, Moroni C. (1989). EMBO J 8: 2575–2581. Benjamin D, Colombi M, Moroni C. (2004). Nucleic acids research 32: e89. Blalock WL, Pearce M, Steelman LS, Franklin RA, McCarthy SA, Cherwinski H et al. (2000). Oncogene 19: 526–536. Blalock WL, Weinstein-Oppenheimer C, Chang F, Hoyle PE, Wang XY, Algate PA et al. (1999). Leukemia 13: 1109–1166. D’Andrea R, Rayner J, Moretti P, Lopez A, Goodall GJ, Gonda TJ et al. (1994). Blood 83: 2802–2808. Druker BJ, Tamura S, Buchdunger E, Ohno S, Segal GM, Fanning S et al. (1996). Nat Med 2: 561–566. Gough NM. (1988). Analytical biochemistry 173: 93–95. Hannemann J, Hara T, Kawai M, Miyajima A, Ostertag W, Stocking C. (1995). Mol Cell Biol 15: 2402–2412. Hayakawa F, Towatari M, Kiyoi H, Tanimoto M, Kitamura T, Saito H et al. (2000). Oncogene 19: 624–631. Hirsch HH, Nair AP, Moroni C. (1993). J Exp Med 178: 403–411. Hoyle PE, Moye PW, Steelman LS, Blalock WL, Franklin RA, Pearce M et al. (2000). Leukemia 14: 642–656. Ihle JN, Lee JC, Rebar L. (1981). J Immunol 127: 2565–2570. Jiang X, Ng E, Yip C, Eisterer W, Chalandon Y, Stuible M et al. (2002). Blood 100: 3731–3740. Joung YH, Park JH, Park T, Lee CS, Kim OH, Ye SK et al. (2003). Exp Mol Med 35: 350–357. Liu CB, Itoh T, Arai K, Watanabe S. (1999). J Biol Chem 274: 6342–6349. McCormack MP, Gonda TJ. (1997). Blood 90: 1471–1481. McCubrey JA, Shelton JG, Steelman LS, Franklin RA, Sreevalsan T, McMahon M.. (2004). Oncogene 23: 7810–7820. McCubrey JA, Steelman LS, Hoyle PE, Blalock WL, Weinstein-Oppenheimer C, Franklin RA et al. (1998). Leukemia 12: 1903–1929.

Morita S, Kojima T, Kitamura T. (2000). Gene Ther 7: 1063–1066. Mui AL, Wakao H, O’Farrell AM, Harada N, Miyajima A. (1995). EMBO J 14: 1166–1175. Nagata Y, Todokoro K. (1996). Biochem Biophys Res Commun 221: 785–789. Nair AP, Diamantis ID, Conscience JF, Kindler V, Hofer P, Moroni C. (1989). Mol Cell Biol 9: 1183–1190. Nosaka T, Kitamura T. (2002). Exp Hematol 30: 697–702. Onishi M, Kinoshita S, Morikawa Y, Shibuya A, Phillips J, Lanier LL et al. (1996a). Exp Hematol 24: 324–329. Onishi M, Mui AL, Morikawa Y, Cho L, Kinoshita S, Nolan GP et al. (1996b). Blood 88: 1399–1406. Onishi M, Nosaka T, Misawa K, Mui AL, Gorman D, McMahon M et al. (1998). Mol Cell Biol 18: 3871–3879. Overell RW, Watson JD, Gallis B, Weisser KE, Cosman D, Widmer MB. (1987). Mol Cell Biol 7: 3394–3401. Perkins GR, Marshall CJ, Collins MK. (1996). Blood 87: 3669–3675. Piao X, Bernstein A. (1996). Blood 87: 3117–3123. Shelton JG, Steelman LS, Lee JT, Knapp SL, Blalock WL, Moye PW et al. (2003). Oncogene 22: 2478–2492. Shounan Y, Miller M, Symonds G. (1995). Exp Hematol 23: 492–499. Steelman LS, Pohnert SC, Shelton JG, Franklin RA, Bertrand FE, McCubrey JA. (2004). Leukemia 18: 189–218. Stoecklin G, Colombi M, Raineri I, Leuenberger S, Mallaun M, Schmidlin M et al. (2002). EMBO J 21: 4709–4718. Stoecklin G, Gross B, Ming XF, Moroni C. (2003). Oncogene 22: 3554–3561. Stoecklin G, Hahn S, Moroni C. (1994). J Biol Chem 269: 28591–28597. Testa U, Riccioni R, Diverio D, Rossini A, Lo Coco F, Peschle C. (2004). Leukemia 18: 219–226. Wodnar-Filipowicz A, Heusser CH, Moroni C. (1989). Nature 339: 150–152. Oncogene