Anaplastic Lymphoma Kinase (ALK) Activates Stat3 and Protects Hematopoietic Cells from Cell Death

Anaplastic lymphoma kinase (ALK) activates Stat3 and protects hematopoietic cells from cell death

Alberto Zamo1,2,4, Roberto Chiarle1,3,4, Roberto Piva1, Jennifer Howes1, Yan Fan1, Marco Chilosi2, David E Levy1 and Giorgio Inghirami*,1

1Department of Pathology and Kaplan Comprehensive Cancer Center, New York University School of Medicine, 550 First Avenue, New York, NY 10016, USA; 2Department of Pathology, University of Verona, Borgo Roma, Italy; 3Department of Pathology, University of Torino, Via Santena 7, Torino, Italy

The anaplastic lymphoma kinase (ALK) gene is Introduction characteristically translocated in Anaplastic Large Cell Lymphomas (ALCL) and the juxtaposion of the ALK Non-Hodgkin's lymphomas (NHL) are a heteroge- gene to multiple partners results in its constitutive neous group of malignancies representing the neoplas- protein tyrosine kinase activity. We show here that tic clonal expansion of B or T lymphocytes expression of activated ALK induces the constitutive transformed at dierent stages of dierentiation and phosphorylation of Stat3 in transfected cells as well as in maturation. Among the NHL, the CD30 positive primary human ALCLs. Furthermore, immunohisto- Anaplastic Large Cell Lymphomas (ALCL) represent chemical studies demonstrate that among distinct human a distinct subset of neoplasms with unique morpholo- B and T cell lymphomas, activation of Stat3 nuclear gical, immunophenotypical and clinical features (Harris translocation is uniquely associated with ALK expres- et al., 1994). ALCL include a distinct subset of tumors sion. NPM-ALK also binds and activates Jak3; however, displaying chromosomal translocations involving the Jak3 is not required for Stat3 activation or for cell Anaplastic Lymphoma Kinase (ALK) (2p23;5q35) transformation in vitro. Moreover, src family kinases are gene. Since the ®rst molecular characterization of the not necessary for NPM-ALK-mediated Stat3 activation t(2;5)(2p23;5q35) translocation involving the nucleo- or transformation, suggesting that Stat3 may be phosmin (NPM) and ALK genes (Morris et al., 1994), phosphorylated directly by ALK. To evaluate relevant several groups have identi®ed additional translocations targets of ALK-activated Stat3, we investigated the in which ALK is fused to other partners including the regulation of the anti-apoptotic protein Bcl-xL and its 5-aminoimidazole-4-carboxamide ribonucleotide trans- role in cell survival in NPM-ALK positive cells. NPM- formylase/IMP cyclohydrolase (ATIC) and the clathrin

ALK expression caused enhanced Bcl-xL transcription, heavy chain (CLTCL) genes (Drexler et al., 2000). largely mediated by Stat3. Increased expression of More recently, translocations juxtaposing the Translo-

Bcl-xL provided sucient anti-apoptotic signals to cation Promoter Region (TPR) to ALK were also protect cells from treatment with speci®c inhibitors of found in approximately 50% of in¯ammatory myo®- the Jaks/Stat pathway or the Brc-Abl kinase. These broblastic tumors (Lawrence et al., 2000). studies support a pathogenic mechanism whereby The deregulated expression of ALK chimera allows stimulation of anti-apoptotic signals through activation the transformation of immortalized rodent cells and of Stat3 contributes to the successful outgrowth of ALK the escape of BaF3 cells from IL-3 dependence. All positive tumor cells. the chimeric ALK proteins have domains responsible Oncogene (2002) 21, 1038 ± 1047. DOI: 10.1038/sj/onc/ for homo-dimerization, resulting in constitutive 1205152 activation and phosphorylation of the ALK tyrosine kinase (Bischof et al., 1997). However, the pathogenic Keywords: ALCL; t(2;5); NPM-ALK signaling; Stat3; mechanisms leading to ALK-mediated T-cell trans-

Bcl-xL, apoptosis formation remain unclear. Thus far, several signaling molecules have been shown to interact with NPM- ALK. These include Grb2, Shc, IRS-1, PLC-g and PI3K (Bai et al., 1998, 2000; Bischof et al., 1997; Fujimoto et al., 1996; Slupianek et al., 2001). The pathogenic role of Shc and IRS-1 in ALK-mediated transformation has been excluded by mutagenesis studies (Fujimoto et al., 1996), and the contribution of PLC-g remains controversial. For instance, while *Correspondence: G Inghirami; E-mail: [email protected] mutation of the PLC-g binding site (Y664A) rendered 4These authors contributed equally to this work Received 5 July 2001; revised 25 October 2001; accepted 31 NPM-ALK unable to release transfected BaF3 cells October 2001 from IL-3 dependency, truncated ALK proteins NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1039 lacking residue 664 could eciently transform murine Receptor tyrosine kinase (RTK) or fusion RTK NIH3T3 ®broblasts (Bai et al., 1998). In addition, we proteins can activate multiple signaling pathways, and others have recently shown that forced expres- including Ras, PI3K and Jak-Stat pathways. To sion of ALK leads to the phosphorylation of AKT study whether the constitutive expression of ALK via PI3K. Moreover, and studies in vivo showed that fusion proteins leads to activation of the Jak and NPM-ALK transformation is sensitive to inhibition Stat proteins, we investigated their phosphorylation by DNRas, suggesting that the Ras pathway is status after in vitro transfection and in fresh human required for ALK-mediated transformation of T- ALCL tumors. Our ®ndings show that among the lymphocytes. Jak-Stat molecules, only Jak3 and Stat3 are con-

Figure 1 Constitutive expression of Stat3 in NPM-ALK positive cells. (a) Stat3 is constitutively phosphorylated in tyrosine 705 and serine 727 in NPM-ALK positive lymphoid cell lines. Total cells extracts from NPM-ALK negative (Namalwa and Jurkat) and positive (DHL, JB and Karpas 299) lymphoid cell lines were transferred and immunoblotted with the indicated antibodies. (b) NPM-ALK leads to the phosphorylation of Stat3 in cotransfection experiments. NPM-ALK, ATIC-ALK or mutant K210RNPM- ALK and Stat3 expression vectors were cotransfected into h293T cells as indicated. The phosphorylation status of Stat3 was evaluated using a speci®c anti-pY-Stat3 antibody. CDK2 was used for normalization. (c) Stat3 is speci®cally upregulated by NPM- ALK. PC12 cells were transduced with VSG retroviral vectors carrying the wt or trunctated NPM-ALK (T498 and T526) constructs. These two truncated constructs were designed to maintain the catalytic domain and its activity but to lose the Shc and PLCg (T498) or PLCg (T526) binding sites, respectively. The death mutant K210RNPM-ALK was used as negative control. Cells were cultured for 36 h and subsequently lysed and immunoblotted with the indicated antibodies. (d) NPM-ALK leads to the upregulation of Stat3 mRNA transcripts. PC12 cells were transduced with the indicated retroviral vectors as above. Total RNA was extracted using a Triazol system (Sigma). cDNAs were subsequently ampli®ed with speci®c Stat3 oligoprimers for 30 cycles. (e) Stat3 is constitutively phosphorylated in human NPM-ALK positive ALCL. Total extracts derived from cryopreserved tissue blocks from ALK positive and negative ALCLs were immunoblotted with anti-ALK, anti-Stat3 and Stat5 and anti-phospho-Stat3 and Stat5 antibodies. CDK2 protein expression was used to normalize the protein input in each lane. (f) ALK positive human lymphomas display nuclear Stat3. Immunohistochemical stains using monoclonal anti-Stat3 (upper left panel) or rabbit anti-ALK (upper right panel) antibodies were performed on serial paran-embedded tissue sections of human ALCLs. Nuclear Stat3 and primarily cytoplasmic ALK positivity were restricted to tumor cells of NPM-ALK positive ALCL. No detectable stain was documented in normal lymphoid cells of AKL+ or ALK7 ALCL samples (lower panels). Sections were counterstained with Hematoxylin

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1040 stitutively activated in ALK positive cells. Moreover, Table 1 Stat3 and ALK protein expression in human Non-Hodgkin's lymphoma ALK leads to the upregulation of Bcl-xL, which in turn provide anti-apoptotic signals, showing that Diagnosis Total number ALK number Stat3 number ALK not only stimulates tumor cell growth but also ALCL ALK+ 21 21 20 protects cells from apoptosis. These signals may ALCL ALK7 43 0 1 synergistically contribute to the pathogenesis of LyP 9 0 0 ALK positive tumors. MF 10 0 0 PTCL 10 0 1 B-CLL/SLL 10 0 0 MzL 6 0 0 Results FCCL 10 0 0 DLCL 30 0 3

Stat3 is constitutively phosphorylated in ALK positive Immunohistochemistry was performed on the following categories of tumors and is upregulated by NPM-ALK human lymphomas: ALCL=Anaplastic Large Cell Lymphoma; LyP=Lymphomatoid Papulosis; MF=Mycosis Fungoides; Since tyrosine kinases (TK) are known to activate PTCL=Peripheral T-cell Lymphoma; B-CLL/SLL=B-chronic Lym- Jak/Stat family members, we investigated whether phocytic Leukemia/Small lymphocytic lymphoma; MzL=B-Marginal NPM-ALK stimulated Stat activation in lymphoid Zone Lymphoma; FCCL=Follicular Center Cell Lymphoma; DLCL=Diuse Large Cell Lymphoma cells lines carrying the t(2;5) translocation. NPM-ALK positive cell lines ± DHL, JB, and Karpas 299 ± expressed Stat3 phosphorylated on tyrosine 705 and serine 727 in contrast to Stat5 that was largely in its phosphorylation, the detection of Stat3 within the unphosphorylated form. The NPM-ALK negative nucleus by immunohistochemistry was used to identify lymphoblastoid cell lines Namalwa and Jurkat did activated Stat3. In the majority of ALK positive not show any detectable activation of Stat3 or 5 lymphomas, Stat3 showed a nuclear staining in the (Figure 1a). To prove that Stat3 phosphorylation in tumor cells (Figure 1f), whereas surrounding non- ALCL-derived cell lines was directly related to the neoplastic lymphocytes were negative. In contrast, expression of NPM-ALK, we performed similar ALK negative cases showed primarily cytoplasmic studies in h293T cells transfected with the full length Stat3 with only rare nuclear staining. Moreover, other NPM-ALK or ATIC-ALK (Drexler et al., 2000) groups of ALK-negative NHL only rarely contained a fusion constructs and the kinase dead mutant, few tumor cells expressing nuclear Stat3 (Table 1). K210RNPM-ALK. The forced expression of functional Thus, Stat3 activation largely correlated with ALK ALK constructs, but not of the inactive form of expression in human lymphomas. NPM-ALK, led to the phosphorylation of Stat3 (Figure 1b). Since the Western blot analysis demon- ALK positive tumors display phosphorylated Jak3 and do strated higher levels of endogenous Stat3 in ALK not require Src family members to stimulate Stat3 positive than in ALK negative cell lines (Figure 1a), phosphorylation we investigated whether NPM-ALK could directly upregulate the expression of Stat3. Since ALK To determine whether NPM-ALK activated Jak or receptor is normally expressed only in neuronal cells Tyk2 proteins, we ®rst investigated the phosphoryla- (Iwahara et al., 1997; Morris et al., 1997) we decided tion status of these molecules in human Karpas 299 to use the PC12 cells as a model. After the and DHL (data not shown) ALK positive cell lines. transduction of PC12 cells, we could demonstrate that Only Jak3 was found to be constitutively phosphory- the expression of functionally active NPM-ALK lated (Figure 2a). Furthermore, a series of immuno- resulted in higher protein (Figure 1c) and mRNA precipitation experiments showed that Jak3 bound levels (Figure 1d) of Stat3. Taken together, these data eciently to catalytically active NPM-ALK or suggest that ALK causes, in addition to Stat3 T526NPM-ALK proteins but not to the kinase-dead phosphorylation, enhanced transcription and/or stabi- mutant K210RNPM-ALK (Figure 2b). These ®ndings lization of Stat3 mRNA levels, as well as increased have been recently con®rmed in thymocytes derived levels of Stat3 protein in both PC12 and lymphoid from NPM-ALK transgenic mice (data not shown). To cells. further identify a potential Jak3 docking site within To further explore the biological signi®cance of these NPM-ALK, a series of deletion constructs were ®ndings, we studied the phosphorylation status of Stat3 generated and tested for their ability to bind Jak3 in a series of human ALK positive and negative and/or to transform NIH3T3 cells. Interestingly, lymphomas. Western blotting analysis showed that deletion of residues 118 ± 138 was sucient to abrogate ALK positive, but not ALK negative, tumors displayed Jak3 binding to ALK. Nevertheless, D118 ± 138 NPM- detectable levels of phosphorylated Stat3 (Figure 1e). ALK still eciently induced Stat3 phosphorylation and These observations were con®rmed by immunohisto- transformed NIH3T3 cells (Figure 2c), suggesting that chemical analysis of a large number of human ALCLs Jak3 binding and activation was not essential for and a representative panel of distinct B and T cell biological activity. To further validate this notion in lymphoproliferative disorders as comparison. Since lymphoid cells, we treated DHL and Karpas 299 tumor nuclear translocation of Stat proteins occurs following cells with AG490, a speci®c inhibitor of Jak enzymes.

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1041

Figure 2 Characterization of Jak3 activation and Jak3-NPM-ALK interactions. (a) Jak-3 is constitutively activated in NPM-ALK. Total extracts (500 mg) from NPM-ALK Karpas 299 cells were immunoprecipitated with anti-Jak1, anti-Jak-2, anti-Tyk-2 and anti- Jak-3 Abs. The immunoprecipitates were analysed by immunoblotting with anti-pY Ab. The same membrane was subsequently stripped and blotted with anti-Jaks and anti-Tyk-2 antibodies (lower panels). The single band detected after immunoprecipitation with anti-Jak3 by the anti-pY antibody had the same molecular weight of the band detected by the speci®c anti-Jak3 antibody. (b) Phosphorylated Jak-3 binds to NPM-ALK. h293T cells were co-transfected with NPM-ALK, K210RNPM-ALK, D526NPM-ALK and Jak-3 constructs. After incubation for 36 h, cells were lysated and immunoprecipitated with a rabbit anti-ALK. NPM-ALK immunoprecipitates were subsequently subjected to immunoblotting with anti-pY, anti-Jak-3, and anti-ALK. (c) Schematic representation of trunctated NPM-ALK constructs. The NPM-ALK activation, the phosphorylation status of Stat3 and Jak3, and the transformation of NIH3T3 cells for each individual construct are indicated. The ALK and Jak-3 activation were detected by an immunoprecipitation-Western blotting assay using anti-ALK or anti-FLAG (T1-498, T1-458Y191F/Y445F, T1-436 and T1-363) or anti-Jak-3 antibodies for the immunoprecipitation and Western blotting with anti-pY antibody (Material and methods). Stat3 phosphorylation was documented by using a direct Western blotting and a speci®c anti-pY-Stat3 Ab. Transformation of NIH3T3 cells was evaluated documenting the presence of transformed foci as described in Material and methods. (d) ALK-mediated activation of Stat3 does not require Src, fyn and/or yes. MEF of wt and of SV40 transformed triple knockout Src/fyn/yes were infected with retroviral vectors carrying the mock vector (Pallino) or the NPM-ALK construct. Stat3 phosphorylation was studied by direct Western blotting. ALK and Src expression were evaluated using speci®c Abs. Notably, NPM-ALK expression was sensibly higher in the SYF7/7 cells. (e) the Src-family speci®c inhibitor PP2 does not change the phosphorylation levels of Stat3 in NPM- ALK positive ALCL cell lines. The ecacy of PP2 was ®rst documented studying the inhibition of the Src-mediated activation of Akt in COS-1 cells after treatment with EGF (upper panel). COS-1 cells were serum starved for 24 h and subsequently incubated with dierent concentration of PP2 for 2 h prior the treatment with EGF (10 min, 100 ng/ml, UBI). Phosphorylation levels of Akt were evaluated using an anti-Akt Ab (left upper panel). DHL ALCL cells were also incubated with 10 mg/ml or 50 mg/ml of PP2 for 2 h (right upper panel) and Stat3 phosphorylation levels were evaluated by direct Western. To study where longer exposure to PP2 could modulate the phosphorylation levels of Stat3, Karpas 299, DHL and JB cells were treated with 10 mg/ml of PP2 for 24 h (right bottom panel). Cellular lysated were blotted and Stat3 phosphorylation levels were tested as described. Total protein levels were normalized using an anti-b-tubulin Ab

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1042 Phosphorylation levels of Stat3 did not change autocrine loop of secreted IL-6 that upregulates anti- following drug treatment (data not shown), con®rming apoptotic signals via a Jak-Stat3 dependent-pathway. that Jak catalytic activity is not required for Stat3 The treatment with AG490 blocks this Jak-Stat3 activation. pathway and makes the U266 cells sensitive to Interestingly, truncated and/or deleted constructs spontaneous and to Fas-mediated apoptosis (Catlett- lacking ¯anking sequences near and/or within the Falcone et al., 1999). Consistent with this model, putative catalytic domain of ALK abolished its kinase treatment of control K210RNPM-ALK U266 cells with activity and the transformation ability of the corre- AG490 led to enhanced cell death. In contrast, NPM- sponding proteins (Figure 2c). These ®ndings suggest ALK positive cells displayed signi®cantly lower the catalytic domain of ALK, determined by the apoptotic rates in the presence of the inhibitor. Both analysis of the homology with the catalytic domains lines were equally resistant to Fas-mediated apoptosis of known kinases (Morris et al., 1994), is insucient in absence of AG490. Nonetheless, only NPM-ALK and it requires additional ¯anking regions possibly for positive cells showed a greater resistance to Fas- the correct tridimensional folding of the protein. Src mediated apoptosis after the treatment with AG490 and src family members play crucial roles in signaling (Figure 3b). These ®ndings strongly suggest that NPM- from RTK and are responsible for Stat activation ALK anti-apoptotic signaling does not require Jak under some circumstances (Wang et al., 2000). To activation. study whether NPM-ALK mediated activation of Stat3 To explain the molecular mechanisms responsible for required src, fyn or yes, the src family proteins the ability of ALK to protect from cell death, we

expressed in ®broblasts, we examined Stat3 phosphor- studied the expression of Bcl-xL protein and other

ylation in mouse embryonal ®broblasts (MEFs) known anti-apoptotic proteins. As expected, Bcl-xL speci®cally deleted for all three src-related genes protein levels substantially decreased in K210RNPM-

(Klinghoer et al., 1999). As shown in Figure 2d, ALK U266 after AG490 treatment. Conversely, Bcl-xL NPM-ALK-mediated Stat3 phosphorylation could be and survivin protein levels remained relatively higher in eciently achieved in the absence of all these kinases. NPM-ALK positive cells. On the other hand, NPM- Finally, to con®rm and extend these results in ALK expression did not signi®cantly modulate the lymphoid cells, we investigated whether the phosphor- protein levels of Bcl-2 and Mcl-1 (Figure 3c). ylation of Stat3 of NPM-ALK positive cells could be To con®rm these ®ndings and to prove that these modulated by PP2, a potent inhibitor of src kinases eects were not uniquely restricted to U266 cells, we (Hanke et al., 1996). The activity of PP2 was ®rst tested the anti-apoptotic role of NPM-ALK in Bcr- demonstrated by showing its ecacy in blocking src- Abl positive human acute leukemia K562 cells treated mediated phosphorylation of Akt of COS-1 cells with the speci®c inhibitor of the Abl kinase, STI571 treated with EGF (Chen et al., 2001). We subsequently (Deininger et al., 1997). Survival of K562 is used a high dose of PP2 (50 mM) for a short period of dependent on the levels of anti-apoptotic signals that time (2 h) or an intermediate dose (10 mM) for a longer are largely mediated by the activity of the deregu- period of time (24 h; 50 mM PP2 were toxic when used lated abl kinase, and may also involve activated Stats for 24 h) to treat ALCL cells. Under both conditions (Nieborowska-Skorska et al., 1999; Gesbert and all ALCL cell lines showed no changes in their Stat3 Grin, 2000; Vigneri and Wang, 2001). K562 cells phosphorylation (Figure 2e, lower panel). These were transduced with K210RNPM-ALK-GFP and ®ndings strongly argue that Stat3 phosphorylation NPM-ALK-GFP viruses and then treated with levels in NPM-ALK positive cells do not require the STI571. Apoptotic cells were identi®ed using Annexin activity of src. V staining. As shown, in Figure 4a, treatment with STI571 enhanced cell death of K210RNPM-ALK-GFP positive cells but not NPM-ALK-GFP cells (Figure The constitutive activation of Stat3 via ALK leads to 4a). Since the survival of K562 is dependent on Bcl-xL upregulation and protection from cell death activation of Bcl-xL through Stat5, which in turn Transcriptional activation of several genes has been depends on the tyrosine kinase activity of the Bcr- implicated as targets for Stat3, including Cyclin D1, Abl fusion protein (Donato et al., 2001), we also

Bcl-x, p21 and c-myc (Bowman et al., 2000). To study investigated the levels of Bcl-xL in transduced cells in whether the expression of Stat3 regulated genes was the presence of STI571. As expected, STI571 blocked

also enhanced by ALK, we studied the activity of two Stat5 phosphorylation and led to a decrease in Bcl-xL K210R Bcl-xL promoter constructs in the presence of NPM- protein levels in control wt and NPM-ALK cells ALK and K210RNPM-ALK. NPM-ALK, but not the (Figure 4b). In NPM-ALK positive K562 cells, kinase-dead mutant control, enhanced Bcl-x promoter however, the presence of the inhibitor led to a activity, and this eect was completely blocked by similar downregulation of Stat5 phosphorylation but

Stat3b, a dominant negative form of Stat3 (Figure 3a) the cells maintained relatively high levels of Bcl-xL.It (Catlett-Falcone et al., 1999). To validate the biological is likely that NPM-ALK sustained Stat3 phosphor- signi®cance of these ®ndings, we investigated whether ylation, readily detectable only in NPM-ALK expres- NPM-ALK could protect human U266 myeloma cells sing cells (lanes 7 ± 9), was largely responsible for Bcl-

from cell death in the absence of cytokine signaling. xL expression (Bromberg et al., 1999) in the absence The survival of U266 is largely dependent on an of Bcr-Abl catalytic function.

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1043

Figure 3 NPM-ALK protects U266 cells from apoptosis. (a) NPM-ALK leads to the positive transcription of Bcl-xL through Stat3. NIH3T3 cells were transfected with the indicated plasmids (see Material and methods). After 36 h cells were lysed and the luciferase reported assay was performed with the luciferase reported kit (Promega). (b) NPM-ALK protects AG490 treated cells from apoptosis. U266 cells were transduced with VSV-G retroviruses carrying the wt NPM-ALK or K210RNPM-ALK and GFP as a reporter. Transduced cells were treated with AG490 (50 mM) and or with anti-Fas Mab (10 mg/ml) for 3 days and the percentage of apoptotic cells was measured among GFP positive cells by Annexin V staining and ¯ow cytometry; (c) NPM-ALK maintains Bcl-xL and survivin protein levels in U266 cells treated with AG490. U266 transduced cells and treated with AG490 for 72 h (see above) were lysed and their protein expression was evaluated by Western blotting with the indicated Abs. Untreated cells were used a negative control

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1044 Discussion results in constitutive activation of the tyrosine kinase (Bischof et al., 1997; Trinei et al., 2000) which is Cell transformation requires the acquisition of multiple absolutely required for cell transformation (Drexler et genetic defects that ultimately lead to deregulated cell al., 2000). In vitro and in vivo studies have also shown growth. Moreover, in addition to alterations that that activated ALK ®res several signaling pathways, directly promote cell growth, neoplastic cells often including PLCg and PI3K (Bai et al., 1998, 2000). In acquire defects that protect tumor cells from cell death. this study, we have shown that human ALK positive In this study, we have demonstrated that the tyrosine ALCL lymphomas always display constitutively acti- kinase NPM-ALK responsible for ALCL malignancy vated Stat3. Because Stat3 activation is largely causes constitutive activation of Stat3, which in turn restricted to ALK positive ALCLs, its detection oers promotes cell survival by inducing expression of the a new diagnostic tool in the analysis of those tumors in

anti-apoptotic genes, Bcl-xL and survivin. which the identi®cation of ALK remains problematic. Translocations involving the ALK gene occur among Moreover, we can speculate that the use of multiple NHLs exclusively in ALCL. The molecular mechan- markers could allow the identi®cation of new subsets isms responsible for this exquisite speci®city are of ALCLs with possibly unique clinical courses completely unknown and the molecular events leading (Drexler et al., 2000). Nieborowska-Skorska et al. to ALK mediated lymphoid transformation are still (2001) have recently shown that the forced expression largely unclear. Several investigations have demon- of NPM-ALK in BaF3 cells leads to constitutive strated that homodimerization of ALK chimeras activation of Stat5. Interestingly, our data did not document any substantial activation of Stat5 or other Stat members (data not shown) in T/null ALCL tumors. If the in vitro ®ndings obtained with the BaF3 cells are con®rmed using human ALK positive B cell tumors, we may conclude that ALK can lead to the preferential activation of speci®c Stats in cells derived from B or T lymphocytes. The analysis of ALK positive cells has shown that NPM-ALK, in addition to promoting the phosphorylation of Stat3, also leads to increase Stat3 expression, enhancing the levels of Stat3 mRNA as a result of enhanced transcription or stabilization of the mRNA. Akira, (1999) have previously demonstrated that Stat3 levels are upregulated by IL-6 signaling, suggesting that a positive loop may be operational (Takeda et al., 1998). A similar scenario may also occur in ALK positive cells. Many molecules have been shown to have a role in the activation of Stats. Members of the Jak family have been demonstrated to be of crucial importance in many cases (Schlessinger, 2000). In our study, we have found that Jak3 eciently binds to the active forms of NPM-ALK, suggesting that the kinase activity of the fusion protein is required for the docking of Jak3 to ALK. Since the region to which Jak3 binds to ALK does not display any canonical Jak binding sequence (118 ± 138 aa), we cannot exclude the possibility that an unknown adaptor molecule may also be involved in this interaction. Nevertheless, it appears that Jak3 activation is not strictly required for ALK mediated Stat3 activation, Bcl-

xL upregulation and/or transformation. In fact, NPM- ALK can eciently transform ®broblast cells which do not express detectable Jak3 (i.e. NIH3T3 cells), and NPM-ALK mutants lacking the Jak3 binding site are still capable of inducing cell transformation as eciently Figure 4 NPM-ALK protects K562 cells from STI571-mediated cell death. (a) K562 infected cells with wt NPM-ALK-GFP or as wild type NPM-ALK. Finally, the treatment of K210RNPM-ALK-GFP viruses were incubated with STI571 human ALCL cells with the Jak3 inhibitor AG490 did (10 mM) for 72 h. The percentage of apoptotic cells calculated not produce any change in the phosphorylation status of among GFP positive cells; (b) NPM-ALK expression upregulates Stat3. Nevertheless, we cannot exclude the possibility Bcl-x (32 kDa) protein levels after STI571 treatment. K562 L that Jak3 has other roles in lymphoid or non-lymphoid transduced cells were incubated with STI571 (10 mM) or AG490 (50 mM) for 72 h. Protein expression was evaluated using speci®c ALK positive tumors, perhaps independent of Stat3 antibodies by Western blotting (Lamant et al., 2000). Similarly, src family kinases were

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1045 not required for ALK-mediated Stat3 activation in Material and methods ®broblasts. These data strongly suggest that ALK may activate Stat3 by direct phosphorylation. Nonetheless, Pathological samples and antibodies the possibility that intermediate proteins might be involved cannot be totally excluded. A panel of well-characterized Non-Hodgkin's lymphoma were retrieved from both the archives of the Hematopathol- In the last few years, several genes have been ogy Laboratory of the New York University Medical Center demonstrated to be directly regulated by Stat3, among and the Surgical Laboratory of the University of Torino and them dierent proteins that control cell cycle progression Verona, Italy. The lymphomas were classi®ed according to and cell survival (Bowman et al., 2000). In this study we the international lymphoma study group, based on hematox- focused on the role of NPM-ALK activated Stat3 targets ylin-eosin and immunoperoxidase stains, as previously that could give tumor cells a biological advantage and described (Chiarle et al., 2000). ALCL were characterized contribute to their overall viability and ability to grow. using a large panel of antibodies speci®cally recognizing T- Because a well established in vitro model enabling us to cell and B-cell restricted antigens and ALK proteins (rabbit investigate the role of ALK and its relationship with the polyclonal, Zymed). activation of Stat3 in T cells is not available, we have focused on two alternative systems in which cell survival Immunohistochemical staining largely depends on the activity of Stats. In the case of Immunohistochemical stainings were performed on formalin U266 cells, survival depends on an autocrine loop, IL6/ ®xed paran embedded tissues using the avidin biotin IL6R, leading to activation of Stat3 and upregulation of peroxidase complex method and a semi-automated immu-

Bcl-xL (Catlett-Falcone et al., 1999). Using this model we nostainer (DAKO or Ventana System) as described (Chiarle demonstrated that NPM-ALK can bypass the block et al., 2000). Neoplasms were considered positive for Stat3 imposed by the AG490 Jak kinase inhibitor, leading to when a nuclear staining was detected in at least 25% of the maintenance of substantial levels of phosphorylated tumor cells. Stat3 which in turn upregulates the expression of

Bcl-xL and survivin preventing cell death. Interestingly, Cell lines and cell cultures NIH3T3 cells transformed by a constitutively active h293T, COS-1, PC12, U266, K562, Jurkat, Namalwa and form of Stat3 display high levels of survivin (Shen et al., Src, Yes and Fyn triple de®cient (SYF7/7) MEF cell lines 2001). were obtained from the American Tissue Culture Collection The validity of these ®ndings was also con®rmed (ATCC). DHL, Karpas 299, and JB, were kindly provided by using a dierent and alternative model taking advantage Dr Lorenzana. Transfections were performed with Eectene of K562 cells. Notably, these cells require Stat5 for their Reagent following the manufacturers' instructions (Qiagen). At least 1 mg of DNA was transfected in each transfection. survival mediated by upregulation of Bcl-xL (Horita et al., 2000). However, the forced expression of NPM- After transfection (36 h), cells were washed and lysed. ALK was able to compensate for the loss of Bcr-Abl Transduction experiments were performed using a VSV-G kinase activity due to the speci®c inhibitor of the Bcr- retroviral system (Clontech). Retroviruses were produced by co-transfecting 293GP cells (Clontech) with the GFP-Pallino- Abl kinase, STI571, maintaining cell survival, likely based (Grignani et al., 1998) retroviral vector (1 mg) and a through activated Stat3. Similar and complementing VSV-G plasmid (1 mg) (Clontech). data have been recently reported by two independent Transformation assays were performed using NIH3T3 cells groups, demonstrating that NPM-ALK can upregulate transfected with the indicated GFP-NPM-ALK plasmids phosphorylated Akt via PI3K, leading to survival of (0.5 m of each NPM-ALK vector and 0.5 mg of carrier BaF3 cells (Bai et al., 2000; Slupianek et al., 2001). DNA pcDNA [Invitrogene]). The morphology and the However, we were unable to document any signi®cant number of transformed foci were evaluated after 2 weeks of change in the phosphorylated levels of Akt in our cell culture, as previously described (Bai et al., 1998). Src models. This discrepancy may be due to the overall inhibitor PP2 was purchased from Calbiochem. levels of ALK in dierent cell types or to the fact that unique signaling pathways are operational in dierent Flow cytometry cells. This notion is supported by the fact that in both Flow cytometry was performed as described (Chiarle et al., U266 and K562, representing mature B cells and 1999). Brie¯y, cells were washed once with PBS and myeloid cells, Stat3 upregulation via ALK leads to the resuspended in staining buer (10 mM HEPES pH 7.4, positive upregulation of Bcl-xL and survivin but not 140 mM NaCl, 2.5 mM CaCl) with 5 ml of AnnexinV-PE Mcl-1, a gene known to be positively regulated by Stat3. (AnnexinV kit, Roche). After 15 min of incubation, in the The mechanisms responsible for these cell type-speci®c dark, cells were directly analysed in the FACScan. (Becton- phenomena are unclear and will require further Dickinson). Cells were evaluated using a CellQuest Program investigation. Finally, the contribution(s) of other (Becton-Dickinson). Data were plotted using a histogram pathways ± including Erk1/2 ± in ALK-mediated algorithm, in which the intensities of Annexin V positive cells were represented in a log 10 scale. survival and/or transformation are still largely unde- termined. More studies are required to conclusively demonstrate that ALK elicits anti-apoptotic signals not Constructs only in B- and myeloid cells, but also in T cells, which Somatically mutated NPM-ALK constructs were generated represent the most common target cells of ALK using speci®c oligonucleotide primers by using a site- translocations. directed mutagenesis kit (Statagene). Truncated constructs

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1046 were derived through a recombinant PCR approach and HRPO-conjugated goat anti-rabbit (1 : 2000, Amesham, 1 h speci®c overlapping primers. The truncated constructs T1- at RT). Detection of immunocomplexes was performed with 498, T1-458, T1-458Y191FY445F, T1-436 and T1-363 were chemiluminescence (ECL, Amersham). For immunoprecipita- tagged using a Flag sequence at the COOH terminus. tion assays lysates were ®rst precleared with protein G- Tagging was necessary because the truncated proteins lost agarose beads and then incubated with the primary antibody the ALK peptide which is speci®cally recognized by the (1 h 48C) followed by protein G beads (1 h 48C). rabbit anti-ALK antibody. Automated DNA sequencing Immunocomplexes were then precipitated, washed, boiled was performed on all constructs to con®rm the presence of and electrophoresed. Detection of the blotted proteins was the mutations. performed as described above.

Western blot analysis and immunoprecipitation Luciferase assay Frozen tissue samples and cell from cell lines were lysed NIH3T3 cells were co-transfected with the NPM-ALK, (20 mM Tris-HCl pH 8, 150 mM NaCl, 1% Triton X-100, K210RNPM-ALK, v-src and Stat3b (1 mg of plasmid each)

5mM EDTA, 1 mM Na3VO4 and 1 mM PMSF) and spun. At in association with the Bcl-xL promoter constructs (Bcl-xL 1.2 least 10 mg of total protein lysates were electrophoresed in and 0.6, 0.5 mg, each plasmid, Catlett-Falcone et al., 1999) SDS ± PAGE gel and transferred onto nitrocellulose mem- driving the expression of the luciferase gene. After transfec- branes. The ®lters were ®rst blocked (5% low fat milk in PBS tion (36 h) cells were lysed and the luciferase assay was with 0.1% Tween 20; 1 h at RT) and then incubated with performed using a luciferase detection kit (Promega). The anti-Stat3 (1 : 1000, Zymed), anti-phospho-Stat3 (1 : 1000, eciency of transfection was normalized by GFP expression. New England Biolabs), anti-Stat5 (1 : 1000, Zymed), anti- phospho-Stat5 (1 : 1000, New England Biolabs), anti-ALK (1 : 4000, Zymed), anti-Cyclin D1 (1 : 10, from Dr M Pagano), anti-Cdk2 (1 : 500, rabbit polyclonal from Dr M Pagano), Acknowledgments anti-phospho-Akt (1 : 1000, Calbiochem), anti-b-tubulin We thank Drs R Jove and D Altieri for providing the

(1 : 2000, Sigma), anti-Bcl-xL (1 : 250, Santa Cruz Biotechnol- Stat3b and the Bcl-xL promoter constructs and the anti- ogy), anti Mcl-1 (1 : 500, Dako) and anti-survivin (5 ml/ml, survivin Ab. A special thank to Drs I Guasparri and A from Dr D Altieri). Following three washes, ®lters were Pesci for their technical assistance. This work was reacted with HRPO-conjugated goat anti-mouse or with supported by NIH grant AI28900 and an AIRC grant.

References

Akira S. (1999). Stem Cells, 17, 138 ± 146. Gesbert F and Grin JD. (2000). Blood, 96, 2269 ± 2276. BaiRY,DieterP,PeschelC,MorrisSWandDuysterJ. Grignani F, Kinsella T, Mencarelli A, Valtieri M, Riganelli (1998). Mol. Cell Biol., 18, 6951 ± 6961. D, Lanfrancone L, Peschle C, Nolan GP and Pelicci PG. Bai RY, Ouyang T, Miething C, Morris SW, Peschel C and (1998). Cancer Res., 58, 14 ± 19. Duyster J. (2000). Blood, 96, 4319 ± 4327. Hanke JH, Gardner JP, Dow RL, Changelian PS, Brissette Bischof D, Pulford K, Mason DY and Morris SW. (1997). WH, Weringer EJ, Pollok BA and Connelly PA. (1996). J. Mol Cell Biol., 17, 2312 ± 2325. Biol. Chem., 271, 695 ± 701. Bowman T, Garcia R, Turkson J and Jove R. (2000). Harris NL, Jae ES, Stein H, Banks PM, Chan JK, Cleary Oncogene, 15, 2474 ± 2488. ML, Delsol G, De Wolf-Peeters C, Falini B and Gatter Bromberg JF, Wrzeszczynska MH, Devgan G, Zhao Y, KC. (1994). Blood, 84, 1361 ± 1392. Pestell RG, Albanese C and Darnell Jr JE. (1999). Cell, 98, HoritaM,AndreuEJ,BenitoA,ArbonaC,SanzC,BenetI, 295 ± 303. Prosper F and Fernandez-Luna JL. (2000). J. Exp. Med., Catlett-Falcone R, Landowski TH, Oshiro MM, Turkson J, 191, 977 ± 984. Levitzki A, Savino R, Ciliberto G, Moscinski L, IwaharaT,FujimotoJ,WenD,CupplesR,BucayN, Fernandez-Luna JL, Nunez G, Dalton WS and Jove R. Arakawa T, Mori S, Ratzkin B and Yamamoto T. (1997). (1999). Immunity, 10, 105 ± 115. Oncogene, 14, 439 ± 449. ChenR,KimO,YangJ,SatoK,EisenmannKM,McCarthy Klinghoer RA, Sachsenmaier C, Cooper JA and Soriano P. J, Chen H and Qiu Y. (2001). J Biol. Chem., 276, 31858 ± (1999). EMBO J., 18, 2459 ± 2471. 31862. Lamant L, Pulford K, Bischof D, Morris SW, Mason DY, Chiarle R, Budel LM, Skolnik J, Frizzera G, Chilosi M, Delsol G and Mariame B. (2000). Am. J. Pathol., 156, Corato A, Pizzolo G, Magidson J, Montagnoli A, Pagano 1711 ± 1721. M, Maes B, De Wolf-Peeters C and Inghirami G. (2000). Lawrence B, Perez-Atayde A, Hibbard MK, Rubin BP, Dal Blood, 95, 619 ± 626. Cin P, Pinkus JL, Pinkus GS, Xiao S, Yi ES, Fletcher CD Chiarle R, Podda A, Prolla G, Podack ER, Thorbecke GJ and Fletcher JA. (2000). Am. J. Pathol., 157, 377 ± 384. and Inghirami G. (1999). J Immunol, 163, 194 ± 205. Morris SW, Kirstein MN, Valentine MB, Dittmer KG, Deininger MW, Goldman JM, Lydon N and Melo JV. Shapiro DN, Saltman DL and Look AT. (1994). Science, (1997). Blood, 90, 3691 ± 3698. 263, 1281 ± 1284. DonatoNJ,WuJY,ZhangL,KantarjianHandTalpazM. Morris SW, Naeve C, Mathew P, James PL, Kirstein MN, (2001). Blood, 97, 2846 ± 2853. Cui X and Witte DP. (1997). Oncogene, 14, 2175 ± 2188. DrexlerHG,GignacSM,vonWasielewskiR,WernerMand Nieborowska-Skorska M, Wasik MA, Slupianek A, Salo- Dirks WG. (2000). Leukemia, 14, 1533 ± 1559. moni P, Kitamura T, Calabretta B and Skorski T. (1999). Fujimoto J, Shiota M, Iwahara T, Seki N, Satoh H, Mori S J. Exp. Med., 189, 1229 ± 1242. and Yamamoto T. (1996). Proc. Natl. Acad. Sci. USA, 93, 4181 ± 4186.

Oncogene NPM-ALK leads to Stat3 activation and Bcl-xL upregulation A Zamo et al 1047 Nieborowska-Skorska M, Slupianek A, Xue L, Zhang Q, Takeda K, Kaisho T, Yoshida N, Takeda J, Kishimoto T and Raghunath PN, Hoser G, Wasik MA, Morris SW and Akira S. (1998). J. Immunol., 161, 4652 ± 4660. Skorski T. (2001). Cancer Res., 61, 6517 ± 6523. Trinei M, Lanfrancone L, Campo E, Pulford K, Mason DY, Schlessinger J. (2000). Cell, 103, 211 ± 225. Pelicci PG and Falini B. (2000). Cancer Res., 60, 793 ± 798. Shen Y, Devgan G, Darnell Jr JE and Bromberg JF. (2001). Vigneri P and Wang JY. (2001). Nat. Med., 7, 228 ± 234. Proc. Natl. Acad. Sci. USA, 98, 1543 ± 1548. Wang YZ, Wharton W, Garcia R, Kraker A, Jove R and Slupianek A, Nieborowska-Skorska M, Hoser G, Morrione Pledger WJ. (2000). Oncogene, 19, 2075 ± 2085. A, Majewski M, Xue L, Morris SW, Wasik MA and Skorski T. (2001). Cancer Res., 61, 2194 ± 2199.

Oncogene