Isoform-specific cleavage of 14-3-3 proteins in apoptotic JURL-MK1 cells

The proteins of 14-3-3 family are ubiquitously expressed in all eukaryotic cells and involved in the regulation and coordination of many important biological processes [Masters and Fu, 2001; Mackintosh, 2004; Pozuelo Rubio et al., 2004; Aitken, 2006; Gardino et al., 2006; Porter et al., 2006]: among others, they prevent undesirable triggering of apoptotic cell death programs and help to maintain the correct progression of the cell cycle. In humans, the family consists of seven distinct isoforms (β/α, γ, ζ/δ, ε, η, σ, and τ/θ) with tissue-specific expression. The molecular weight (MW) of these acidic proteins is 27–28 kDa, with the exception of the isoform ε, which is somewhat larger (29 kDa). They form homo- or heterodimers which can bind one or two client proteins, mostly through a phosphoserine/phosphothreonine containing motif, although non-phosphorylated binding sites were also identified (e.g., on the proapoptotic protein Bax). The binding site is usually internal, but it can also be located at the C-terminus of the client protein [Coblitz et al., 2006]. Each 14-3-3 monomer comprises nine antiparallel α-helices and two subunits associate through their N-terminal domains to form a flattened U-shaped structure with two amphipathic grooves for ligand binding. While the inner surface of the dimer is composed of highly conserved residues, the C-termini of the monomers are the most divergent regions in sequence and adopt no strictly ordered structure. The affinity of 14-3-3 proteins for their targets is regulated by phosphorylation, both on the target proteins and on 14-3-3's themselves.

Although the high-affinity binding motifs are recognized by all 14-3-3 proteins, the different mammalian 14-3-3 isotypes appear to have specific functions [Kjarland et al., 2006; Jagemann et al., 2008; Niemantsverdriet et al., 2008]. However, despite the large number of known 14-3-3 targets, little data are available as to the specificity in substrate recognition and/or independent regulation of the expression level and activity for the individual 14-3-3 isoforms.

Binding of client proteins to 14-3-3s can have multiple effects including changes in enzymatic activity, subcellular localization, and protein complex formation. Figure 1 shows some of the well-documented interaction points between 14-3-3 proteins and the apoptotic machinery [Bonnefoy-Berard et al., 1995; Bae et al., 2001; Brunet et al., 2002; Nomura et al., 2003; Won et al., 2003; Zhang et al., 2003, 2007; Tsuruta et al., 2004; Pendergast, 2005; Yoshida et al., 2005; Malissein et al., 2006; Qi et al., 2006; Lee et al., 2008; Yan et al., 2008]. Signals arriving from growth factor receptors activate kinases (Akt, PKA, etc.) which phosphorylate number of proapoptotic actors (e.g., Bad, FKHRL1, FOXO1, etc.) and thereby promote their binding to 14-3-3 proteins and inactivation. On the other hand, pro-death signals activate among others the kinase JNK which phosphorylates some of 14-3-3 isoforms and induces the release of proapoptotic client proteins (Bax, c-Abl, etc.) which can thus trigger the apoptosis. In a similar way, the family of 14-3-3 proteins interacts with multiple proteins which are essentially involved in the regulation of the cell-cycle progression, cell adhesion and migration, stress response, and other fundamental processes [Jin et al., 2004; Hermeking and Benzinger, 2006].

In this work, we describe changes in the expression level of five 14-3-3 isoforms induced by the treatment of human leukemic cells with the kinase inhibitor Imatinib mesylate. JURL-MK1 [Di Noto et al., 1997] and K562 cell lines express the aberrant protein Bcr-Abl which is the primary cause of chronic myelogenous leukemia (CML) and whose inhibition by Imatinib mesylate results in apoptosis induction. We show that the lowered expression and cleavage of 14-3-3 proteins occurring during the apoptosis are to a large extent independent of caspase-3 activity and probably involve a non-caspase protease. The relative contributions of caspase-3-dependent and -independent mechanisms vary among the individual 14-3-3 isoforms.

MATERIALS AND METHODS

RESULTS

We have previously shown that Imatinib mesylate treatment induces apoptosis in JURL-MK1 cells [Kuželová et al., 2005]. Figure 2 shows the kinetics of Imatinib-induced caspase-3 activation measured using the fluorogenic caspase-3 substrate (panel A) and apoptotic DNA fragmentation detected by TUNEL method (panel B). We subsequently performed proteomic analysis of Imatinib effects on JURL-MK1 cell line by means of 2D electrophoresis and MALDI–TOF MS. Among multiple proteins which were altered by Imatinib treatment, some spots were identified as members of 14-3-3 protein family (Fig. 3A). The treatment resulted in a marked decrease of the spot intensities corresponding to 14-3-3 ε, τ, β + ζ (overlapping spots of β and ζ isoforms) and γ + ζ. Additionally, as a result of Imatinib treatment new spots identified as 14-3-3 ε, τ, β, and γ appeared at lower MWs suggesting possible fragmentation of the intact 14-3-3 proteins. Figure 3B shows the comparison of sequence coverage obtained using MALDI peptide mass mapping of two spots which were identified as 14-3-3 τ. The MALDI data of the spot corresponding to the intact form of 14-3-3 τ cover both N- and C-terminus of the protein while the C-terminal peptide was missing from the presumed fragment.

Changes in the expression levels of 14-3-3 proteins were further studied by immunoblotting using specific antibodies against the isoforms ε, β/α, γ, τ, and ζ/δ. The Western blots contained no unspecific bands and the attribution of 28–29 kDa bands to 14-3-3 isoforms could be made unambiguously. As it is shown in Figure 4, the amount of all these 14-3-3 isoforms was lowered during the apoptosis induced by Imatinib mesylate. In parallel, we detected a progressive appearance of fragments of 14-3-3 β, γ, ε, and τ at about 27 kDa. These fragments were not detected when we used antibodies specific for C-terminal sequences of 14-3-3 ε and β (Fig. 4, right part). In the case of ζ isoform, the amount of fragment formed as a result of Imatinib treatment was low compared to the full-length protein and in some experiments was not detected at all.

As some 14-3-3 isoforms had been shown to be substrates for caspase-3 [Nomura et al., 2003; Won et al., 2003], we analyzed the role of caspases in the observed 14-3-3 processing using irreversible caspase inhibitors: zDEVD-fmk as a relatively specific inhibitor of caspase-3 and the general caspase inhibitor zVAD-fmk. These compounds are able to penetrate the cell membrane and they bind covalently to the active site of caspases providing thereby the irreversible inhibition of caspase activity. We have previously shown by fluorometric monitoring of caspase-3 activity and by anti-caspase-3 immunoblotting that both these inhibitors are fully efficient at concentrations lower than 5 µM [Kuželová et al., 2007]. We verified that in the presence of 5 µM zDEVD-fmk or 5 µM zVAD-fmk, the DEVDase activity was maintained well under the control value (untreated JURL-MK1 cells) for up to 72 h following sample treatment with Imatinib mesylate (Fig. 2 A, open symbols). We also tested the effects of these inhibitors on the rate of cleavage of the fluorogenic peptide substrates Ac-LEHD-AFC and Ac-VDVAD-AFC, which are efficiently cleaved by caspase-9 and -2, respectively. As shown in Figure 5, Imatinib mesylate induced about 20-, 2-, and 4-fold increase in the cleavage rate of Ac-DEVD-AFC, Ac-LEHD-AFC, and Ac-VDVAD-AFC, respectively, and this increase was completely inhibited by simultaneous addition of 0.5 µM zDEVD-fmk. On the other hand, the inhibition of caspases under these conditions had only partial effect on the changes in 14-3-3 protein levels induced by Imatinib mesylate (Fig. 6A, lanes 3 and 5). The caspase inhibitors fully prevented 14-3-3 processing only in the case of γ isoform. The isoform 14-3-3 ε was responsive to caspase-3 inhibition, but we observed a large variability in the extent of 14-3-3 ε cleavage inhibition by zDEVD-fmk in individual experiments. The intensity of the fragment band was reduced to 3–77% of the original value (mean reduction to 39% from eight experiments) using 5 µM zDEVD-fmk. In the case of 14-3-3 β, the cleavage was reduced by zDEVD-fmk but significant amount of 14-3-3 fragment was still present in all experiments even at 50 µM inhibitor. Finally, the cleavage of 14-3-3 τ was almost unaffected by zDEVD-fmk in repeated experiments, even at 50 µM concentration (Fig. 6A, lane 4). The broad-range caspase inhibitor zVAD-fmk at 50 µM (but not 5 µM) concentration was able to prevent the changes in the expression level of all the studied 14-3-3 isoforms (Fig. 6A, lane 6) except for 14-3-3 τ.

We also examined the effects of an alternative apoptosis inducer, 17-AAG, an inhibitor of the heat shock protein HSP90, on the expression level of 14-3-3 proteins. The treatment of JURL-MK1 cells with 1 µM 17-AAG resulted in caspase-3 activation closely similar to that induced by Imatinib mesylate (data not shown) as well as in 14-3-3 cleavage (Fig. 6, lane 7). The MW of these fragments was apparently the same as in the case of Imatinib mesylate. The effect of caspase inhibition by 5 µM zDEVD-fmk during 17-AAG treatment was also isoform-dependent: while the inhibitor prevented or reduced the cleavage of 14-3-3 γ, ε, β, and ζ, it was not able to inhibit the cleavage of 14-3-3 τ (Fig. 6, lane 8).

The panels B and C of Figure 6 document caspase-3 inhibition by zDEVD-fmk and zVAD-fmk in the samples analyzed in the panel A of the figure. They show the rate of in vitro cleavage of the fluorogenic caspase-3 substrate and anti-caspase-3 immunoblots, respectively.

Despite the complete inhibition of caspase-3 activity, the effect of caspase inhibitors on the apoptosis progression was also very limited. The apoptotic DNA fragmentation induced by Imatinib was considerably reduced only in the presence of 50 µM zVAD-fmk (Fig. 7).

We performed similar analysis of the effects of caspase inhibitors on 14-3-3 expression and on the apoptosis progression using the cell line K562 which is also derived from a patient with CML. Imatinib mesylate treatment of K562 cells led to similar changes in anti-14-3-3 immunoblots as in the case of JURL-MK1 cells (i.e., lowered expression and fragmentation of individual 14-3-3 isoforms). While the inhibitor zDEVD-fmk appeared to be to some extent more efficient in K562 cells (compared to JURL-MK1) in inhibiting both 14-3-3 protein processing and DNA fragmentation, we still observed considerable discrepancy between the concentrations required for caspase inhibition (0.5 µM) on one hand and for significant effects on these processes (50 µM) on the other hand (data not shown).

DISCUSSION

Altogether, these results clearly show that while caspase-3 probably contributes to 14-3-3 cleavage (especially in the case of 14-3-3 γ and ε), other mechanism has to be involved in this process. Although the selectivity of short peptide-based caspase substrates with regard to the individual caspases is limited [McStay et al., 2008], the data presented in Figure 5 indicate that caspases 9 and 2 are not active in the presence of 5 µM zDEVD-fmk. It is important to realize that the specificity of zVAD-fmk for caspases is compromised when the inhibitor is used at high concentrations (such as 50 µM or more) [Schotte et al., 1999; Rozman-Pungercar et al., 2003], so that the inhibitory effect of zVAD-fmk on both 14-3-3 processing and the apoptotic DNA fragmentation (Figs. 6 and 7) is not necessarily mediated by its effects on the caspase family. With regard to the known in vitro affinity of the used caspase inhibitors for different caspases [Ekert et al., 1999] we believe that none of them can be responsible for the observed 14-3-3 τ processing. We have recently shown that the fluorescently labeled caspase inhibitor FAM-DEVD-fmk binds not only to caspase-3 but also to another target in apoptotic JURL-MK1 cells [Kuželová et al., 2007]. The corresponding strong fluorescent signal can be inhibited by zVAD-fmk at high concentration (50 µM) but not by zDEVD-fmk. Thus, a protease with substrate specificity similar to that of caspase-3 may operate in apoptotic cells and be responsible for 14-3-3 cleavage. Possible candidates are cathepsins, non-cysteine proteases cleaving after Asp residue. However, we failed to detect any effect of pepstatin A (up to 7.5 µM, data not shown), which inhibits cathepsin activity, on the cleavage of 14-3-3 proteins.

The changes in the expression levels of all 14-3-3 isoforms can be largely prevented under the experimental conditions which lead to the inhibition of DNA fragmentation (Fig. 7). It thus appears that the decrease of 14-3-3 expression and cleavage of some 14-3-3 isoforms are related to the apoptosis progression and may represent an integral part of the programmed cell death or at least of some apoptotic pathways.

The MW of 14-3-3 fragments which are produced during the apoptosis is only slightly lower in comparison with the full-length proteins and the cleavage thus corresponds to the removal of several amino acids only. If the truncation took place at the N-terminus of the protein, it would probably negatively affect the ability of 14-3-3's to form dimers and thereby the affinity for client proteins. However, we judge from indirect indications that 14-3-3 fragments rather lack C-terminal amino acids: the fragment of 14-3-3 β is recognized by the antibody specific for the N-terminus of the protein (sc-1657) but not by C-terminal anti-14-3-3 β antibody (sc-25276) (Fig. 4). Similarly, the antibody against the C-terminus of 14-3-3 ε (AB 9732) detects solely the full-length protein. The series of peptides identified by MALDI from the spot presumably corresponding to the cleaved 14-3-3 τ contains peptides from the N-terminus while it lacks the C-terminal peptide which is present in the intact protein (Fig. 3B). Moreover, 14-3-3 ε possesses additional amino acids at the C-terminus in comparison with the other isoforms (Fig. 4, top) and this would explain the fact that the lost sequence of 14-3-3 ε is larger than that of 14-3-3 β, γ, ζ, or τ (the shift towards lower MW is larger in the case of 14-3-3 ε). The isoform ε displays also the highest shift in pI on 2D gels (Fig. 3A). It follows from the sequence alignment (top of Fig. 4) that the loss of 6–7 amino acids from the C-terminal tail of 14-3-3 β or τ compared to 17 amino acids from that of 14-3-3 ε would result in loss of 2 compared to 6 acidic charges. In the case of γ isoform, the shift in pI is larger than that for β and τ (Fig. 3A) suggesting that three acidic charges (eight amino acids from the C-terminus) might be removed.

Nomura et al. 2003 have shown using immunoprecipitation that the removal of seven C-terminal amino acids from 14-3-3 τ results in the loss of the affinity for Bax. Similarly, C-terminal cleavage of 14-3-3 ε has been shown to impair the ability of the protein to bind Bad and the expression of the truncated 14-3-3 ε in 293T cells enhanced Bad-mediated apoptosis [Won et al., 2003]. It is thus reasonable to suppose that 14-3-3 cleavage leads to the release of the proapoptotic client proteins and forms a positive feedback allowing for amplification of proapoptotic signals. It is also possible that the truncated forms of 14-3-3 proteins have an additional role in the apoptosis progression. The C-terminal tail of 14-3-3 γ was recently shown to be required for a full and specific interaction with the scaffold protein KSR1 which regulates signal transduction through Raf/MEK/ERK pathway [Jagemann et al., 2008]. It is interesting to note in this regard that the truncation of 14-3-3 ζ has been reported to enhance its affinity for Raf-1 and Bad [Truong et al., 2002]. A study of this isoform has indicated that the C-terminal tail of the protein in ligand-free form occupies the ligand-binding groove and inhibits, to some extent, inappropriate associations [Silhan et al., 2004]. However, we observed only weak cleavage of 14-3-3 ζ isoform during Imatinib-induced apoptosis of JURL-MK1 cells and no cleavage of this particular isoform was found by Nomura et al. after etoposide treatment of HeLa cells.

In conclusion, we believe that the decrease of 14-3-3 protein expression level and the removal of the C-terminus belong to common processes contributing to the apoptosis regulation, regardless of the cell type and of the inducing stimulus. Our results show that the cleavage of 14-3-3 proteins is executed by at least two different mechanisms probably involving caspase-3 and still unknown protease beyond the caspase family, in an isoform-specific manner.