¹Division of Medical Oncology, Department of Medicine, University of Washington, and Veterans Affairs Medical Center, Seattle, Washington, USA

²Turku Centre for Biotechnology, University of Turku/Åbo Akademi University, Turku, Finland

³Division of Medical Oncology, Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado, USA

Correspondence to: Michael Lilly, Center for Molecular Biology and Gene Therapy, Loma Linda University School of Medicine, Loma Linda, California 92354, USA

We have examined potential mechanisms by which the Pim-1 kinase acts as a hematopoietic cell survival factor. Enforced expression of the wild type 33 kd (FD/hpim33) and 44 kd (FD/mpim44) Pim-1 proteins in murine factor-dependent FDCP1 cells prolonged survival after withdrawal of IL-3, while expression of a dominant negative Pim-1 protein (FD/pimNT81) shortened survival. Following removal of IL-3 FDCP1 cells exhibited loss of mitochondrial transmembrane potential and production of reactive oxygen species, as determined by flow cytometry analysis. The wild type Pim-1 proteins decreased these changes while the dominant negative protein enhanced mitochondrial dysfunction. The antiapoptotic activity of the kinases could not be attributed to modulation of glutathione, catalase, or superoxide dismutase activities. Both the FD/hpim33 and FD/mpim44 cells maintained expression of bcl-2 mRNA following cytokine removal, while a substantial decrease was seen in FD/neo cells. To modulate Bcl-2 protein levels, a bcl-2 antisense RNA construct was coexpressed with the wild type pim-1 cDNAs. FD/hpim33 cells with low cellular Bcl-2 protein levels had shortened cytokine-independent survival compared with FD/hpim33 clones with high Bcl-2 expression. However survival of FD/mpim44 cells after IL-3 withdrawal was substantially independent of cellular Bcl-2 protein levels. The 33 kd protein delayed, and the 44 kd protein completely prevented enhanced cell death associated with enforced expression of human Bax protein however. Our results suggest that the 33 kd Pim-1 kinase may enhance cell survival through cooperation with and regulation of bcl-2. In addition the 44 kd kinase may regulate the expression or activity of other pro- and anti-apoptotic members of the bcl-2 family.

apoptosis; pim-1; bcl-2; mitochondria; IL-3; FDCP1

Introduction

Hematopoietic cells are absolutely dependent on peptide growth regulators for survival and proliferation, but the secondary mediators of such signals are known only in part. The hematopoietic cytokine granulocyte-macrophage colony-stimulating factor (GM-CSF), and the related growth factors interleukin (IL)-3 and IL-5, can effect a spectrum of responses in target cells, including proliferation, differentiation, and induction or prevention of programmed cell death (apoptosis). GM-CSF is known to regulate expression of the antiapoptotic protein Bcl-2 through signal pathways involving ras (Kinoshita et al., 1995a) and Jak2 (Sakai and Kraft, 1997), as part of its program for inhibiting cell death. Whether additional signaling intermediates or pathways are involved in promoting cell survival in GM-CSF-treated cells is unclear.

The central role of mitochondria in the development of apoptosis is now well-established (Zamzami et al., 1996). Loss of mitochondrial transmembrane potential (Zamzami et al., 1995), formation of transition pores (Hirsch et al., 1997), and production of reactive oxygen species (Zamzami et al., 1995) are early events in cells destined to undergo programmed cell death. Subsequently there is release of cytochrome c from mitochondria, which results in activation of caspases and other effectors of DNA fragmentation and cell death (Bossy-Wetzel et al., 1998). Proteins which are members of the bcl-2 family can modulate several of these steps. Bcl-2 protein itself is primarily a mitochondrial protein, and can prevent loss of mitochondrial transmembrane potential and other aspects of mitochondrial dysfunction (Shimizu et al., 1998), act as an antioxidant (Hockenbery et al., 1993), and block release or activity of cytochrome from mitochondria (Kluck et al., 1997).

We have recently reported that the cytoplasmic serine/threonine kinase Pim-1 also acts as a survival factor to inhibit apoptosis in myeloid cells deprived of cytokines (Lilly and Kraft, 1997). pim-1 appears to be a true oncogene in that its enforced expression in transgenic mice leads to an increased incidence of tumors (van Lohuizen et al., 1989; Allen and Berns, 1996; Breuer et al., 1991). The kinase is a labile protein whose expression is especially high in hematopoietic and germ cells. A single Pim-1 protein of 33 kd exists in humans, while mice have both 33 and 44 kd forms. The latter results from an alternate translational start site missing in the human gene (Saris et al., 1991). Approximately equal amounts of the two murine forms are expressed in hematopoietic cells (Saris et al., 1991; Lilly and Kraft, 1997). We have noted that the Pim-1 protein is expressed in primitive hematopoietic cells selectively in response to cytokines whose receptors are members of the hematopoietin receptor superfamily (Lilly et al., 1992). Overexpression of Pim-1 protein is also seen in adult acute leukemias (Amson et al., 1989), and chronic lymphocytic leukemia (Lilly, unpublished observations 1997). The cloning of the highly homologous pim-2 gene (van der Lugt et al., 1995) and kid-1 gene (Feldman et al., 1998) suggests that there likely is a family of pim kinases.

The mechanism by which pim-1 acts as a survival factor is obscure. We have examined the effects of enforced expression of the kinase on mitochondrial dysfunction accompanying withdrawal of IL-3 from FDCP1 cells, and on the expression and function of the mitochondrial death regulatory genes bcl-2 and bax. Our data suggest that the 33 kd Pim-1 kinase acts to inhibit cell death in part through a bcl-2-dependent mechanism, while the 44 kd kinase may modulate expression or activity of multiple bcl-2 family members.

Results

The pimNT81 mutant is a dominant negative Pim-1 protein

To most clearly identify effects resulting from enforced expression of the Pim-1 kinase we initially sought to develop dominant negative pim-1 constructs. We identified the NT81 mutation (deletion of the N-terminal 80 amino acids of the human Pim-1 protein) as a possible dominant negative protein. Expression of the NT81 mutant in FDCP1 cells via retroviral transduction resulted in the appearance of an anti-Pim-1-reactive protein of the appropriate size (Figure 1a, lane 4). Survival of FDCP1 cells expressing this mutant (FD/pimNT81) was then compared with that of cells expressing the 33 kd (FD/hpim33) or 44 kd (FD/mpim44) Pim-1 proteins, or cells infected with a neo-transducing retrovirus (FD/neo) (Figure 1b). As expected (Lilly and Kraft, 1997) both of the wild type variants markedly prolonged survival of the cells after IL-3 withdrawal. In contrast the FD/pimNT81 cells showed substantially shorter survival than did the FD/neo control cells, an effect opposite to that seen with the normal proteins and consistent with the effects of a dominant negative protein.

To further clarify whether the NT81 mutant is a dominant negative Pim-1 protein, we investigated whether it was able to abolish biological effects of wild type Pim-1 protein. One recently identified substrate for Pim-1 kinase is the nuclear factor of activated T-cells (NFATc) whose transcriptional activity is enhanced by Pim-1 in Jurkat and J-TAg cells (Rainio, Sandholm, Eisenman, Koskinen: manuscript in preparation). By using a luciferase reporter assay we noticed that ectopic expression of the NT81 mutant efficiently suppressed endogenous NFATc activity whether or not the wild type Pim-1 protein was coexpressed, suggesting that the NT81 mutant can inhibit also endogenous Pim-1 activity in a dominant negative fashion (Figure 1c).

Pim-1 expression delays mitochondrial dysfunction associated with IL-3 deprivation

Previous studies have identified mitochondrial dysfunction as an early event in the onset of programmed cell death (Zamzami et al., 1995). Parameters known to be perturbed include maintenance of transmembrane potential and production of ROS. FD/neo cells deprived of IL-3 included a population that was low in staining by the fluorescent probe DiOC₆ (Zamzami et al., 1996), reflecting loss of mitochondrial transmembrane potential (Figure 2a). FD/hpim33 and FD/mpim44 demonstrated a much smaller proportion of DiOC₆(3)-low cells, while FD/pimNT81 had a higher per cent of dim staining cell than did the FD/neo population (Figure 2b). Time course studies showed onset of loss of transmembrane potential between 3 and 8 h after removal of IL-3. Mitochondrial dysfunction appeared to `peak' between 20 and 30 h after IL-3 withdrawal, with lower values thereafter. These curves result from a large fraction of the FDCP1 cells moving simultaneously into apoptosis between 20 and 30 h after cytokine depletion. Only a few surviving FD/neo and FD/pimNT81 cells existed by 40 h. These continued to die, though at a slower rate (lower proportion of cells with mitochondrial dysfunction). Since the flow cytometry data were acquired by `gating' on the intact cell population this gives the artifactual appearance of `recovery'. In fact, all of the FD/neo and FD/pimNT81 cells eventually died after IL-3 removal.

Similarly, FD/neo cells cultured without IL-3 developed a population that stained highly with the probe HE, reflecting intracellular oxidation of the probe to fluorescent ethidium (Figure 3a) as a result of intracellular production of ROS. Again, FD/hpim33 and FD/mpim44 cells showed a lower ability to oxidize intracellular probe, while FD/pimNT81 cells showed an exaggerated production of ROS (Figure 3b). The time course studies showed similar onset of ROS production as was seen with loss of transmembrane potential.

Pim-1 proteins support expression of bcl-2 mRNA without affecting expression of bax message

The ability of Pim-1 proteins to delay mitochondrial dysfunction and production of ROS in IL-3-deprived FDCP1 cells suggested that pim-1 may regulate cellular antioxidant systems. However we found no effect of enforced expression of Pim-1 on glutathione levels, catalase activity, or superoxide dismutase activity (data not shown). Since Pim-1 proteins inhibited production of ROS without modulating classic antioxidant systems, we examined whether pim-1 might modulate expression of the bcl-2 gene, whose product also has antioxidant activity (Hockenbery et al., 1993), inhibits mitochondrial dysfunction (Shimizu et al., 1998), and supports IL-3-independent survival of FDCP1 cells (Nunez et al., 1990). FD/neo, FD/hpim33, and FD/mpim44 cells were removed from IL-3 and cultured for 9 h. Total RNA was isolated and used for analysis of bcl-2 and bax mRNA by RT - PCR (Figure 4a). bcl-2 message was easily detectable in exponentially growing cells (lanes 1 - 3). After removal from IL-3 for 9 h however, bcl-2 mRNA was substantially lower in FD/neo cells (lane 4). In FD/hpim33 and FD/mpim44 cells bcl-2 message persists after IL-3 withdrawal (lanes 5 and 6). In contrast to these results, bax mRNA levels were the same in each cell line after cytokine withdrawal. Amplification of the cDNAs with -actin primers demonstrated that there were equal amounts of intact cDNA in each reaction. Levels of the Bcl-2 protein, measured 12 h after removal of IL-3, also were supported by enforced expression of the Pim-1 kinase, in parallel with the results of the mRNA analysis (Figure 4b). In contrast enforced expression of the bcl-2 gene did not support expression of endogenous Pim-1 proteins (Figure 1a, lane 5).

The 33 kd Pim-1 protein, but not the 44 kd Pim-1 kinase, requires Bcl-2 for its antiapoptotic effect

While Pim-1 proteins may modulate bcl-2 expression, they may promote cell survival independently of an effect on this gene. We examined directly whether Bcl-2 protein was required for the pim-1 survival effect, by expressing an antisense bcl-2 RNA in FD/hpim33 and FD/mpim44 cells. IL-3-independent survival of FD/hpim33 cells with low Bcl-2 expression was markedly shorter than that of high Bcl-2 expressers, and was similar to that of FD/neo cells (Figure 5a). By contrast, survival of FD/mpim44 cells after IL-3 withdrawal was largely independent of Bcl-2 level (Figure 5b). Only at 72 h after IL-3 removal was survival of FD/mpim44 (Bcl-2 low) cells significantly worse than that of FD/mpim44 (Bcl-2 high) cells (P=0.05 for no difference, by t-test).

Prevention of mitochondrial dysfunction (reflected by transmembrane potential and production of ROS) by pim-1 kinases demonstrated a similar partial dependence on bcl-2 expression (Figure 6). In this analysis cytokine-deprived FD/hpim33 cells with low intracellular Bcl-2 showed similar mitochondrial dysfunction to FD/neo cells. In contrast, FD/hpim33 cells with high Bcl-2 expression preserved mitochondrial function in the absence of IL-3. While down-regulation of Bcl-2 slightly impaired the ability of the 44 kd Pim-1 protein to prevent mitochondrial dysfunction, this effect was small compared to the overall 44 kd kinase effect. Thus modulation of cell survival and mitochondrial function by the pim-1 gene product are both partly dependent on the presence of Bcl-2 protein.

Bcl-2 may inhibit apoptosis in part by complexing with the proapoptotic protein Bax (Korsmeyer et al., 1993) and antagonizing its effects. Excess Bax protein can over-ride the death inhibitory effect of Bcl-2 (Korsmeyer et al., 1993), and Bax has been referred to as a dominant inhibitor of Bcl-2 function (Krajewski et al., 1994). However, Bax protein may also promote cell death independently of any association with Bcl-2 (Knudson and Korsmeyer, 1997). When a FLAG-tagged Bax protein was expressed in FDCP1 cells (Figure 7a), survival after IL-3 withdrawal was shortened (Figure 7b). We then coexpressed Bax with either 33 kd (FD/hpim33/Bax) or 44 kd (FD/mpim44/Bax) Pim-1 proteins. In FD/hpim33 cells bax effect was delayed, but ultimately was not prevented by the 33 kd kinase (Figure 7c). In contrast, enforced expression of the 44 kd kinase substantially prevented the exaggerated cell death resulting from bax overexpression (Figure 7d).

Discussion

IL-3, GM-CSF, and related cytokines are well known to inhibit apoptosis in hematopoietic cells. Their major role appears to be as survival factors, though they likely can act as mitogens, agents which promote transit of the G1/S checkpoint (Fairbairn et al., 1993). These cytokines are known to regulate expression of mRNA for the antiapoptotic gene bcl-2 (Kinoshita et al., 1995b), whose expression (along with similarly acting family members) likely mediates at least part of the survival effect. Signaling pathways utilized by these growth factors to effect such expression are being clarified. A distal domain of the common GM-CSF/IL-3/IL-5 receptor beta_c chain has been implicated in regulation of bcl-2 expression through ras, raf-1 and MAP kinases (Kinoshita et al., 1995a,b; Sato et al., 1993). An additional cascade downstream of ras, sensitive to wortmannin (Kinoshita et al., 1997) has also been described to inhibit cell death. In addition a ras-independent pathway activated by IL-3, and involving Shc tyrosine phosphorylation, has recently been implicated in the cytokine-dependent expression of bcl-2 (Gotoh et al., 1996), as have other regulators such as Jak2 (Sakai and Kraft, 1997) and protein kinase C (Rinaudo et al., 1995).

It is clear that the cytoplasmic serine kinase Pim-1 also acts as a survival factor to inhibit apoptosis in IL-3 deprived hematopoietic cells (Lilly and Kraft, 1997). The current studies utilizing the dominant negative NT81 pim-1 construct demonstrate that Pim-1 protein is involved in the regulation of mitochondrial function as well as overall cell survival after cytokine withdrawal. pim-1 mRNA and protein expression are highly regulated by IL-3, GM-CSF and related cytokines by a poorly defined signal cascade originating from membrane-proximal domains of the alpha and beta_c chains of the receptors (Polotskaya et al., 1993; Sato et al., 1993) and likely involving the Jak2 kinase (Sakai and Kraft, 1997). It appears that the ras-raf-1-MAP kinase pathway is not involved in pim-1 induction however since activated forms of raf-1 (Wingett et al., 1996) or ras (Lilly, unpublished observations, 1997) fail to support constitutive expression of pim-1.

The mechanisms through which pim-1 acts to promote cell survival and inhibit programmed cell death have not been characterized previously. Reactive oxygen species (superoxide, hydrogen peroxide, and hydroxyl radicals) are produced during the development of programmed cell death and appear to mediate at least some of the death phenotype, since antioxidants can suppress apoptosis resulting from a variety of stimuli (Oppenheim, 1997; Packham et al., 1996). We found that both Pim-1 proteins can greatly inhibit the intracellular oxidation of dihydroethidium in FDCP1 cells deprived of IL-3. Since we found no evidence for a direct antioxidant effect by the kinases (Lilly, unpublished, 1997) we postulate that one role for pim-1 is to inhibit production of ROS during the development of programmed cell death. Interestingly, a previous study found no evidence for excessive production of ROS in FDCP1 cells following IL-3 withdrawal (Packham et al., 1996). The opposite finding from our studies likely results from a combination of technical factors. Since production of ROS appears to be transient the previous study may have failed to detect these oxidizers by sampling only at early time points. Another possible factor is the much lower concentration of dihydroethidium in our studies, which may give sharper discrimination between baseline and stimulated ROS production. Our studies suggest that the process of apoptosis following hematopoietic cytokine withdrawal resembles that induced by cytotoxic agents, ionizing radiation, and death-inducing cytokines by exhibiting a variety of mitochondrial function deficits including excessive production of ROS (Zamzami et al., 1995).

Many of the phenotypes which we have found to be associated with enforced expression of the Pim-1 proteins can also be seen in FDCP1 cells expressing a human bcl-2 cDNA (data not shown). Since enforced expression of human Bcl-2 protein in FDCP1 cells did not lead to constitutive expression of Pim-1 kinase we examined whether pim-1 might regulate bcl-2 expression. Some transforming proteins or signal transduction mediators which promote cell survival appear to regulate expression of the bcl-2 gene. These include the Bcr-Abl (Sanchez-Garcia and Grutz, 1995) and Aml-1/Eto (Klampfer et al., 1996) chimerical proteins, the v-H-ras gene product (Kinoshita et al., 1995a), activated c-Myb (Salomoni et al., 1997), and activated Akt (Ahmed et al., 1997) and activated Jak2 (Sakai and Kraft, 1997) kinases. Furthermore the p53 protein, which is apoptogenic, acts as a negative regulator of bcl-2 (Miyashita et al., 1994). Since the pim-1 and bcl-2 genes share at least some upstream regulators (Sakai and Kraft, 1997) a linear relationship between the two survival signals seemed possible. We found that both the 33 and 44 kd Pim-1 proteins supported continued expression of bcl-2 mRNA after IL-3 withdrawal, without a corresponding suppression of bax mRNA expression. These observations are consistent with the hypothesis that pim-1 may act in part as a normal upstream regulator of bcl-2 expression. Regulation of c-Myb activity by pim-1 via the p100 transcription factor may be a possible mechanism through which pim-1 supports bcl-2 expression (Leverson et al., 1998). Our observations are clearly relevant to the mechanisms by which pim-1 acts as an oncogene. Whether pim-1 acts similarly in normal hematopoietic cells is less clear, since loss of the kinase may actually increase survival in non-transformed hematopoietic cells (Domen et al., 1993).

While the kinase may support bcl-2 expression, this does not require that Bcl-2 protein be necessary for the survival effects resulting from enforced expression of pim-1. The observation that bcl-2 and pim-1 interact synergistically in transgenic mice to enhance lymphomagenesis suggests that these two survival signals likely do not have entirely redundant effects (Acton et al., 1992). Our studies showed that a lack of Bcl-2 protein completely prevented the enhanced cell survival and preserved mitochondrial function seen with enforced expression of the 33 kd Pim-1 protein. Remarkably the 44 kd kinase inhibited apoptosis in a Bcl-2-independent manner. Downregulation of Bcl-2 impaired the ability of the larger kinase to preserve mitochondrial function, but the Bcl-2 dependent component was only a minor part of the overall effect of 44 kd Pim-1 on mitochondrial function. These data demonstrate that, while Bcl-2 mediates part of the antiapoptotic effects of Pim-1 proteins, other cellular mediators may be involved. Since pim-1 did not appear to regulate bax expression, we examined if the regulation of Bax activity could be implicated in the Bcl-2-independent effects of 44 kd Pim-1. We found that the larger kinase (whose survival effects were largely independent of Bcl-2), but not the 33 kd form, could completely antagonize the proapoptotic effect of overexpressed Bax. Such antagonism might result from a direct effect of the kinase on Bax, though we have not yet found evidence that Pim-1 can directly phosphorylate Bax or Bcl-2. Another way that Pim-1 could regulate Bax activity is through the proapoptotic protein Bad (Yang et al., 1995). Bad appears to displace Bax from complexes with Bcl-2 or Bcl-xL, resulting in apoptosis. Phosphorylation of Bad prevents this displacement and the resultant cell death. Two Bad kinases have been identified thus far - Akt and Raf-1 (Datta et al., 1997; Wang et al., 1996). Pim-1 shares several features with these enzymes. All are serine/threonine kinases which are activated or expressed in response to IL-3 (Cleveland et al., 1994; del Peso et al., 1997; Lilly et al., 1992) and can prevent death in IL-3 deprived hematopoietic cells. Like Pim-1, Akt can up regulate bcl-2 expression (Ahmed et al., 1997) though apparently activated Raf-1 cannot (Cleveland et al., 1994). Our initial studies have shown that Pim-1 kinase can directly phosphorylate recombinant Bad in vitro (Lilly and Kraft, unpublished, 1998). Studies to demonstrate whether similar phosphorylation occurs in intact cells are in progress.

We have consistently seen in multiple assays (cytokine-independent survival, mitochondrial function, modulation of Bax activity, Bcl-2 independence) that the 44 kd Pim-1 protein exerts a greater biologic effect than the 33 kd species. Whether this represents an intrinsic difference in the kinases is unknown. The cDNA for the shorter kinase was of human origin. Since the human and mouse 33 kd proteins show 94% identity in amino acid sequence it seems unlikely that species differences would account for the greater activity of the 44 kd protein. The larger protein is known to have a longer half life (30 - 60 min) than the smaller kinase (less than 10 min; Saris et al., 1991). Furthermore it exists as a high molecular weight intracellular complex whereas the 33 kd enzyme does not (Saris et al., 1991). We have consistently seen higher intracellular levels of the 44 kd protein than the 33 kd kinase, even when expressed from the same promoter. It seems likely that much of the difference between the two forms results from this difference in expression. Identification of physiologic substrates, and direct comparison of recombinant forms of the two kinases in phosphotransferase assays with those substrates, will be helpful in clarifying these issues.

Materials and methods

Cells and culture

All growth experiments used the IL-3 dependent murine hematopoietic cell line FDCP1 (obtained from Dr Scott Boswell, Indiana University). Cells were routinely maintained in RPMI-1640 medium with 10% iron-supplement calf serum (Sigma) and 10% WEHI-3B conditioned medium. Reporter gene studies utilized J-TAg cells, a subline of Jurkat human T leukemia cells (Northrop et al., 1993). These were maintained in RPMI-1640 medium supplemented with 10% FCS, 2 mM L-glutamine, 100 U/ml penicillin and 100 g/ml streptomycin. All cells were maintained at 37°C in 5% CO₂. For enumeration of surviving cells, Trypan Blue-negative cells in 9 mm³ of cell suspension were counted in triplicate with a hemocytometer.

Pim-1 plasmids and cDNA clones

Construction of the retroviral expression plasmids for wild type human 33 kd and murine 44 kd Pim-1 proteins (pLXSN/hpim and pLXSN/pim44, respectively) has been described previously (Lilly and Kraft, 1997). The dominant-negative pimNT81 mutant was produced by amplification using Pfu polymerase, the human cDNA clone pC1 (Meeker et al., 1987) as a template, and the following primers: sense 5'-GTAGAATTCGCCACCATGCCTAATGGCACTCGACTG-3', and antisense 5'-GTACTATTTGCTGGGCCCCGGCGAC-3'. The amplified product was ligated into the plasmid pLXSN (Miller and Rosman, 1989) for production of retroviruses, and packaged in the amphotrophic retroviral packaging line PA317 as described (Lilly and Kraft, 1997). A human FLAG-tagged bax cDNA in pCDNA3, and a murine bcl-2 cDNA in pBSII were obtained from Dr David Hockenbery (Fred Hutchinson Cancer Research Center). The murine bcl-2 cDNA was amplified with Pfu polymerase and ligated in antisense orientation into pLXSN.

Construction of cell lines

We prepared FDCP1-derived cell lines expressing the cDNA for the human 33 kd Pim-1 protein (FD/hpim33), a cDNA which exclusively expresses the full-length murine 44 kd Pim-1 protein (FD/mpim44), or a cDNA encoding a N-terminal truncation of the human pim-1 cDNA (FD/pimNT81) via retroviral transduction. The NT81 mutant pim-1 cDNA encoded a protein which included amino acids 81 - 313 of the human sequence (Meeker et al., 1987). The preparation of the FD/neo, FD/hpim33, and FD/mpim44 cells lines has been described previously (Lilly and Kraft, 1997). The FDpimNT81 cell line was constructed similarly.

bax and antisense bcl-2 constructs were introduced into FDCP1, FD/hpim33, and FD/mpim44 cells via electroporation (GenePulser, BioRad). A puromycin resistance plasmid (pPGK/puro, courtesy of Dr Glenn Begley, WEHI) was cotransfected with these two plasmids to permit selection of transfected clones by antibiotic resistance. Control cell lines for these experiments consisted of FDCP1, FD/hpim33, and FD/mpim44 cells transfected with the pPGK/puro alone.

Mitochondrial studies

Mitochondrial transmembrane potential was assessed by staining cells with the lipophilic probe DiOC₆(3) (Zamzami et al., 1995). Cells were treated with 40 nM probe for 10 min at 37°C, then analysed for fluorescence on the FL1 channel of a FACScan instrument. Production of reactive oxygen species (ROS) was determined by incubating cells with dihydroethidium (HE, 8 M) for 10 min at 37°C (Zamzami et al., 1995), then analysing intracellular oxidized, fluorescent probe on the FL3 channel of a FACScan instrument. Both probes were obtained from Molecular Probes (Eugene, OR, USA).

Enzyme assays

Catalase was measured by the method of Ou and Wolff (1996), using commercially available reagents (PeroXOquant, Pierce). Superoxide dismutase activity was determined with a commercially available kit (SOD-525, Oxymetrics). Glutathione was quantified by the method of Hedley and Chow (1994), using reagents purchased from Sigma. Total protein on the cell lysates was measured by the BCA method (Pierce).

NFAT transcriptional activity

J-TAg cells were transiently transfected via electroporation (GenePulser, BioRad) with combinations of the following plasmids: pNFAT-LUC (Northrop et al., 1993), pLTR-poly or pLTR-pim-1 (control and wild type pim-1 expression plasmids; Rainio et al., manuscript in preparation), and pLXSN or pLXSN-pimNT81 (control and dominant-negative pim-1 expression plasmids). Cells were pulsed (10⁷ cells/400 l) in Cytomix (van den Hoff et al., 1992) at 300 V and 960 F. Forty-two hours post-transfection, cells were stimulated for 6 h with 1 M ionomycin and 15 ng/ml of the phorbol ester TPA and then collected for reporter gene assay. The luciferase activity was measured using a Labsysems Luminoskan. Values were normalized against the protein concentration which was measured by the Bradford method.

Detection of mRNA

We used an RT - PCR assay to measure changes in bcl-2 and bax mRNA. RT - PCR utilized as a template single strand, oligo-DT primed cDNA prepared with a standard kit (Superscript, GIBCO) from total RNA. Primers for amplification of murine bcl-2 and bax were purchased from Continental Lab Products, while -actin primers were from Stratagene. Amplification was continued for 25 cycles.

Acknowledgements

This work was supported in part by NIH grants CA45672 (M Lilly) and DK44741 (A Kraft), as well as by funds from the Academy of Finland (PJ Koskinen).

Acton, D, Domen J, Jacobs H, Vlaar M, Korsmeyer S and Berns A. (1992). Curr. Top. Microbiol. Immunol. 182, 293-298. MEDLINE

Ahmed NN, Grimes HL, Bellacosa A, Chan TO and Tsichlis PN. (1997). Proc. Natl. Acad. Sci. USA 94, 3627-3632. Article MEDLINE

Allen JD and Berns A. (1996). Cancer Biol. 7, 299-306.

Amson R, Sigaux F, Przedborski S, Flandrin G, Givol D and Telerman A. (1989). Proc. Natl. Acad. Sci. USA 86, 8857-8861. MEDLINE

Bossy-Wetzel E, Newmeyer DD and Green DR. (1998). EMBO J. 17, 37-49. Article MEDLINE

Breuer M, Wientjens E, Verbeek S, Slebos R and Berns A. (1991). Cancer Res. 51, 958-963. MEDLINE

Cleveland JL, Troppmair J, Packham G, Askew DS, Lloyd P, Gonzalez-Garcia M, Nunez G, Ihle JN and Rapp UR. (. (1994). Oncogene 9, 2217-2226. MEDLINE

Datta SR, Dudek H, Tao X, Masters S, Fu H, Gotoh Y and Greenberg ME. (1997). Cell 91, 231-241. MEDLINE

del Peso L, Gonzalez-Garcia M, Page C, Herrera R and Nunez G. (1997). Science 278, 687-689. MEDLINE

Domen J, van der Lugt NMT, Laird PW, Saris CJM, Clarke AR, Hooper ML and Berns A. (1993). Blood 82, 1445-1452. MEDLINE

Fairbairn LJ, Cowling GJ, Reipert BM and Dexter TM. (1993). Cell 74, 823-832. MEDLINE

Feldman JD, Vician L, Crispino M, Tocco G, Marcheselli VL, Bazan NG, Baudry M and Herschman HR. (1998). J. Biol. Chem. 273, 16535-16543. MEDLINE

Gotoh N, Tojo A and Shibuya M. (1996). EMBO J. 15, 6197-6204. MEDLINE

Hedley D and Chow S. (1994). Methods Cell Biol. 42, 31-44. MEDLINE

Hirsch T, Marzo I and Kroemer G. (1997). Biosci. Rep. 17, 67-76. MEDLINE

Hockenbery DM, Oltvai ZN, Yin XM, Milliman CL and Korsmeyer SJ. (1993). Cell 75, 241-251. MEDLINE

Kinoshita T, Yokota T, Arai K and Miyajima A. (1995a). Oncogene 10, 2207-2212. MEDLINE

Kinoshita T, Yokota T, Arai K and Miyajima A. (1995b). EMBO J. 14, 266-275. MEDLINE

Kinoshita T, Shirouzu M, Kamiya A, Hashimoto K, Yokoyama S and Miyajima A. (1997). Oncogene 15, 619-627. MEDLINE

Klampfer L, Zhang J, Zelenetz AO, Uchida H and Nimer SD. (1996). Proc. Natl. Acad. Sci. USA 93, 14059-14064. Article MEDLINE

Kluck RM, Bossy-Wetzel E, Green DR and Newmeyer DD. (1997). Science 275, 1132-1136. Article MEDLINE

Knudson CM and Korsmeyer SJ. (1997). Nature Genetics 16, 358-363. MEDLINE

Korsmeyer SJ, Shutter JR, Veis DJ, Merry DE and Oltvai ZN. (1993). Semin. Cancer Biol. 4, 327-332. MEDLINE

Krajewski S, Krajewska M, Shakaik A, Miyashita T, Wang HG and Reed JC. (1994). Am. J. Pathol. 145, 1323-1336. MEDLINE

Leverson JD, Koskinen PJ, Orrico FC, Rainio EM, Jalkanen KJ, Dash AB, Eisenman RN and Ness SA. (1998). Mol. Cell. 2, 417-425. MEDLINE

Lilly M, Le T, Holland P and Hendrickson S. (1992). Oncogene 7, 727-732. MEDLINE

Lilly M and Kraft A. (1997). Cancer Res. 57, 5348-5355. MEDLINE

Meeker TC, Nagarijan L, ar-Rushdi A, Rovera G, Huebner K and Croce C. (1987). Oncogene Res. 1, 87-101. MEDLINE

Miller AD and Rosman G. (1989). Biotechniques 7, 980-990. MEDLINE

Miyashita T, Krajewski S, Krajewska M, Wang HG, Lin HK, Liebermann Da, Hoffman B and Reed JC. (1994). Oncogene 9, 1799-1805. MEDLINE

Northrop JP, Ullman KS and Crabtree GR. (1993). J. Biol. Chem. 268, 2917-2923. MEDLINE

Nunez G, London L, Hockenberry D, Alexander M, McKearn JP and Korsmeyer SJ. (1990). J. Immunol. 144, 3602-3610. MEDLINE

Oppenheim RW. (1997). Adv. Neurol. 72, 69-78. MEDLINE

Ou P and Wolff SP. (1996). J. Biochem. Biophys. Methods 31, 59-67. MEDLINE

Packham G, Ashmun RA and Cleveland JL. (1996). J. Immunol. 156, 2792-2800. MEDLINE

Polotskaya A, Zhao Y, Lilly M and Kraft A. (1993). Cell Growth Differ. 4, 523-531. MEDLINE

Rinaudo MS, Su K, Falk LA, Halder S and Mufson RA. (1995). Blood 86, 80-88. MEDLINE

Sakai I and Kraft AS. (1997). J. Biol. Chem. 272, 12350-12358. MEDLINE

Salomoni P, Perrotti D, Martinez R, Franceschi C and Calabretta B. (1997). Proc. Natl. Acad. Sci. USA 94, 3296-3301. MEDLINE

Sanchez-Garcia I and Grutz G. (1995). Proc. Natl. Acad. Sci. USA 92, 5287-5291. MEDLINE

Saris CJM, Domen J and Berns A. (1991). EMBO J. 10, 655-664. MEDLINE

Sato N, Sakamaki K, Terada N, Arai K and Miyajima A. (1993). EMBO J. 12, 4181-4189. MEDLINE

Shimizu S, Eguchi Y, Kamiike W, Funahashi Y, Mignon A, Lacronique V, Matsada H and Tsujimoto Y. (1998). Proc. Natl. Acad. Sci. USA 95, 1455-1459. Article MEDLINE

van den Hoff MJ, Moorman AF and Lamers WH. (1992). Nucleic. Acids Res. 20, 2902. MEDLINE

van der Lugt NMT, Domen J, Verhoeven E, Linders K, van der Gulden H, Allen J and Berns A. (1995). EMBO J. 11, 2536-2544.

van Lohuizen M, Verbeek S, Krimpenfort P, Domen J, Saris C, Radaszkiewicz T and Berns A. (1989). Cell 56, 673-682. MEDLINE

Wang HG, Rapp UR and Reed JC. (1996). Cell 87, 629-638. MEDLINE

Wingett D, Long A, Kelleher D and Magnuson NS. (1996). J. Immunol. 156, 549-557. MEDLINE

Yang E, Zha J, Jockel J, Boise LH, Thompson CB and Korsmeyher SJ. (1995). Cell 80, 285-291. MEDLINE

Zamzami N, Marchetti P, Castedo M, Decaudin D, Macho A, Hirsch T, Susin SA, Petit PX, Mignotte B and Kroemer G. (1995). J. Exp. Med. 182, 367-377. MEDLINE

Zamzami N, Susin SA, Marchetti P, Hirsch T, Gomez-Monterrey I, Castedo M and Kroemer G. (1996). J. Exp. Med. 183, 1293-1295. MEDLINE

Figure 1 (a) Immunoblot of Pim-1 proteins. Cells were deprived of IL-3´6 h (FD/pimNT81) or 12 h (FD/neo, FD/hpim33, FD/mpim44, FD/hbcl-2) then subjected to a combined immunoprecipitation/immunoblotting procedure using a rabbit polyclonal anti-Pim-1 antibody (Lilly et al., 1992). (b) Survival curve of FD/neo, FD/hpim33, FD/mpim44, and FD/pimNT81 cells. Cells were washed ´3 to remove IL-3, then resuspended in fresh medium at 1.5 - 2.0´10⁵/ml. Trypan Blue negative cells were counted serially. Each point is the mean±s.d. of triplicate determinations from one of two similar experiments. Relative number of cells at start of experiment has assigned value=100. (c) NFATc transcriptional activity is inhibited by pimNT81 in a dominant negative fashion. J-TAg cells were transfected with the indicated plasmids together with the pNFAT-LUC reporter construct, and the NFATc-dependent luciferase activities were measured 2 days later. Amounts of plasmids used: pLXSN (control) and pLXSN-pimNT81 6 g per transfection; pLTR-poly (control) and pLTR-Pim-1 3 g per transfection; pNFAT-LUC 1 g per transfection. Where no plasmid is indicated, equal amounts of empty control plasmid were included. Each bar is the mean±s.d. of pooled observations (n=6) from two independent experiments, each performed in triplicate

Figure 2 Measurement of mitochondrial transmembrane potential by cellular fluorescence with the probe DiOC₆(3). (a) FD/neo cells deprived of IL-3´21 h (left panel), or maintained in IL-3 (right panel). Left panel shows 17% of cells with low DiOC₆(3) fluorescence. (b) Time course of appearance of DiOC₆(3)-low cells after IL-3 withdrawal

Figure 3 Measurement of ROS production by intracellular oxidation of HE to ethidium. (a) FD/neo cells deprived of IL-3´21 h (left panel), or maintained in IL-3 (right panel). Left panel shows 21% of cells with HE-high fluorescence. (b) Time course of appearance of HE-high cells after IL-3 withdrawal

Figure 4 Regulation of bcl-2 expression by pim-1. (a) RT - PCR measurement of bcl-2, bax, -actin mRNA in FDCP1 cells deprived of IL-3´9 h. Lanes 1, 4: FD/neo cells; Lanes 2, 5: FD/hpim33 cells; Lanes 3, 6: FD/mpim44 cells. (b) Immunoblot of Bcl-2 protein in FD/hpim33, FD/mpim44, FD/neo cells deprived of IL-3´12 h. One hundred micrograms of total cell lysate protein was loaded per lane of the gel

Figure 5 (a) Effect of Bcl-2 protein on cell survival in FD/hpim33 cells. FD/neo, FD/hpim33 (Bcl-2 low), FD/hpim33 (Bcl-2 high) cells were washed ´3 to remove IL-3, then resuspended in fresh medium at 1.5 - 2.0´10⁵/ml. Relative number of cells at time 0=100. Each point is mean±s.d. of triplicate determinations, from one of two similar experiments. Insert: Immunoblot of Bcl-2 protein in puromycin-resistant clones of FD/hpim33 cells transfected with antisense bcl-2 RNA construct and pPGK/puro. One hundred micrograms of total cell lysate protein was loaded per lane of the immunoblot. Equal numbers of three low Bcl-2 expressing clones (5, 7, 9), and equal numbers of three clones expressing the normal, high Bcl-2 level (1, 2, 3) were pooled to give FD/hpim33 (Bcl-2 low) and FD/hpim33 (Bcl-2 high) pools, respectively. (b) Effect of Bcl-2 protein on cell survival in FD/mpim44 cells. FD/neo, FD/mpim44 (Bcl-2 high) and FD/mpim44 (Bcl-2 ion) cells were washed ´3 to remove IL-3, then resuspended in fresh medium at 1.5 - 2.0´10⁵ ml. Relative number of cells at time 0=100. Each point is the mean±s.d. of triplicate determinations from one of two similar experiments. Insert: Immunoblot of Bcl-2 protein in puromycin-resistant clones of FD/mpim44 cells transfected with an antisense bcl-2 RNA construct and pPGK/puro. One hundred micrograms of total cell lysate protein was loaded per lane of the immunoblot. Equal numbers of three low Bcl-2 expressing clones (6, 7, 8), and equal numbers of three clones expressing the normal, high Bcl-2 level (9, 4, 5) were pooled to give FD/mpim44 (Bcl-2 low) and FD/mpim44 (Bcl-2 high) pools, respectively

Figure 6 Effect of Bcl-2 levels on mitochondrial dysfunction in FD/hpim33 and FD/mpim44 cells. Cells were washed ´3 to remove IL-3, then cultured for 13 h in cytokine-free medium. Cells were incubated with DiOC₆(3) or HE, then analysed by FACS for FL1 or FL3 channel fluorescence, respectively. Loss of mitochondrial transmembrane potential is associated with a cell population with low DiOC₆(3) staining. Production of ROS is associated with a cell population with high HE staining. Bar values are the mean±s.d. of triplicate determinations from one of two similar experiments. Derivation of FD/hpim33 (Bcl-2 low), FD/hpim33 (Bcl-2 high), FD/mpim44 (Bcl-2 low), FD/mpim44 (Bcl-2 high) cell pools is as in Figure 5a and b

Figure 7 (a) Immunoblot of FLAG-tagged human Bax protein in FDCP1 cells. Blot was probed with M2 monoclonal anti-FLAG, then an anti-mouse IgG-horseradish peroxidase conjugate, then visualized with Enhanced Chemiluminescence reagents. (b). Cell survival curve for FD/neo (five independent clones), FD/bax cells (six independent clones). Cells were washed ´3 to remove IL-3, then resuspended in fresh medium at 1.5 - 2.0´10⁵ ml. Relative number of cells at time 0=100. Each point is the mean±s.e.m. of means from individual survival curves. (c) Cell survival curve. FD/neo, FD/hpim33 cells were transfected with human FLAG-Bax expression construct and/or pPGK/puro plasmid, and FLAG-Bax positive clones were selected from puromycin-resistant colonies by immunoblotting. FD/neo/Bax, FD/hpim33/Bax, FD/hpim33/puro are pools consisting of equal numbers of cells from each of three independent clones. Method as in panel b. (d) Cell survival curve. FD/neo, FD/mpim44 cells were transfected with human FLAG-Bax expression construct and/or pPGR/puro plasmid, and FLAG-Bax-positive clones were selected from puromycin-resistant colonies by immunoblotting. FD/neo/Bax, FD/mpim44/Bax, and FD/mpim44/puro are pools consisting of equal numbers of cells from each of three independent clones. Method as in panel b