|Home | About | Journals | Submit | Contact Us | Français|
Human T-cell leukemia virus type I (HTLV-I) encodes a Tax oncoprotein that has crucial roles in both virus replication and cell transformation. Our recent studies suggest that the counterbalance between HTLV-I/Tax and PDZ-LIM domain-containing protein PDLIM2 may determine the outcome of HTLV-I infection. Although HTLV-I represses PDLIM2 epigenetically and specifically in transformed cells, PDLIM2 shuttles Tax into the nuclear matrix for ubiquitination-mediated proteasomal degradation, thereby suppressing the transforming ability of HTLV-I. Here, we have further shown that PDLIM2 binds to Tax directly, which was mediated by a putative α-helix motif of PDLIM2 at amino acids 236–254. Consistently, selective disruption of this short-helix crippled PDLIM2 in shutting Tax to the nuclear matrix for ubiquitination-mediated degradation, therefore, PDLIM2 lost the ability in tumor suppression. Although the C-terminal LIM domain of PDLIM2 was not required for Tax binding, it was important for PDLIM2 to interact with the nuclear matrix. Accordingly, the LIM domain was essential for PDLIM2-mediated Tax repression. On the contrary, the N-terminal PDZ domain of PDLIM2 was dispensable for all these events, although the PDZ domain was involved in PDLIM2 binding to cytoskeleton. These studies dissect functional sequences within PDLIM2 and their distinct roles in Tax regulation.
Human T-cell leukemia virus type I (HTLV-I) is an oncogenic retrovirus etiologically associated with adult T-cell leukemia (Gallo, 2005; Takatsuki, 2005; Yoshida, 2005). The virus encodes a 40-kDa regulatory protein named Tax. The Tax protein is not only required for virus replication but also required to immortalize many different cells including human primary T cells (Pozzatti et al., 1990; Tanaka et al., 1990; Grassmann et al., 1992). In addition, Tax-transformed cells induce tumors when introduced into nude or severe combined immunodeficiency mice (Pozzatti et al., 1990; Oka et al., 1992). More importantly, the HTLV-I genome without Tax loses its original transforming ability (Yamaoka et al., 1992), whereas Tax transgenic mice develop various tumors depending on the type of the promoters used to drive Tax expression (Nerenberg et al., 1987; Grossman et al., 1995; Peebles et al., 1995). Notably, Tax-immortalized lymphocytes in vitro and Tax-mediated T-cell lymphoma in animal resemble closely the phenotype of HTLV-I-transformed T cells and HTLV-I-induced adult T-cell leukemia, respectively (Akagi et al., 1995; Kwon et al., 2005; Hasegawa et al., 2006).
The Tax oncoprotein exerts its oncogenic role largely through deregulation of cellular transcription factors that are critical for cell growth and division, such as nuclear factor-κB (Sun and Yamaoka, 2005; Grassmann et al., 2005). In the cytoplasm, Tax recruits the inhibitor of nuclear factor-κB kinase complex into specific perinuclear structures for inhibitor of nuclear factor-κB kinase activation, resulting in degradation of inhibitor of nuclear factor-κB and subsequent nuclear translocation of nuclear factor-κB factors including p65, the prototypic member of nuclear factor-κB (Xiao et al., 2006). In the nucleus, Tax recruits p65 and other cellular transcriptional components into interchromatin granules to form discrete transcriptional hot spots named ‘Tax nuclear bodies’ or ‘Tax nuclear foci’ (Semmes and Jeang, 1996; Bex et al., 1997).
Although the mechanisms by which Tax hijacks cellular signaling for its oncogenic action have been extensively investigated, the molecular studies on how HTLV-I/Tax is regulated by cellular factors are still lacking. Recently, we have identified PDLIM2, a newly identified PDZ–LIM domain-containing protein with ubiquitination-promoting activity, as a potent inhibitor of Tax. In addition to promoting Tax polyubiquitination, PDLIM2 also associates with and shuttles Tax from its functional sites including Tax nuclear foci into the nuclear matrix for proteasomal degradation (Yan et al., 2009a). Accordingly, PDLIM2 coexpression prevents downstream signaling and subsequent tumorigenicity of Tax. Interestingly, PDLIM2 is conversely repressed by HTLV-I, which involves the epigenetic regulation but independent of the Tax oncoprotein (Yan et al., 2009b). More importantly, reconstitution of PDLIM2 into HTLV-I-transformed T cells blocks their tumorigenicity in mice. These findings suggest that the counterbalance between PDLIM2 and HTLV-I/Tax may determine the outcome of viral infection. Thus, it is very interesting and important to address the molecular actions of PDLIM2 on Tax suppression.
Although our previous studies showed that PDLIM2 is able to associate with Tax in vivo (Yan et al., 2009a), it remains unknown whether the association is direct and independent of other cellular proteins. To address the important issue, we examined whether PDLIM2 and Tax proteins purified from bacteria bind to each other using the in vitro co-immunoprecipitation assay. As shown in Figure 1a, purified Glutathione S-Transferase (GST)-Tax could not be pulled down by anti-Myc antibody. In the presence of Myc-PDLIM2, however, GST-Tax was efficiently pulled down by the anti-Myc antibody. These in vitro binding assays indicated that PDLIM2 can directly interact with Tax.
To define the sequences within PDLIM2 that are responsible for its interaction with Tax, we initially tested the role of the N-terminal PDZ domain of PDLIM2 in Tax binding (Figure 1b), as it is known that Tax contains a PDZ domain-binding motif (PBM) at the C-terminus (Rousset et al., 1998). Surprisingly, mutation of the PBM motif within Tax (Tax ΔPBM) failed to block the Tax–PDLIM2 interaction (Figure 1c). This result suggested that the PDLIM2–Tax interaction is not mediated by Tax PBM and PDLIM2 PDZ. In further support of this, deletion of the PDZ domain of PDLIM2 (ΔPDZ) did not affect its interaction with Tax either (Figure 1d). Similarly, deletion of the C-terminal LIM domain (ΔLIM) had no effect on Tax binding. In fact, simultaneous deletion of both PDZ and LIM domains (ΔPDZ/LIM) did not affect the PDLIM2–Tax interaction much. In sharp contrast, deletion of the sequences between the PDZ and LIM domains (Δ79–278) prevented the PDLIM2–Tax interaction. Thus, the sequences between the PDZ and LIM domains contain Tax-binding motif.
To pinpoint the sequences within PDLIM2 required for Tax binding, we generated various small internal deletion mutants of PDLIM2 (Figure 1b). Interestingly, internal deletion of amino acids 243–253 (Δ243–253) caused complete loss of the PDLIM2–Tax interaction (Figure 1e). The loss of Tax binding seemed not because of overall structure disruption of PDLIM2, because deletions of the sequences ahead of or behind the motif (Δ195–207 and Δ258–278) did not affect the PDLIM2–Tax interaction. Most importantly, deletion of this motif did not affect PDLIM2 binding to p65 (Figure 1f), another target of PDLIM2 (Tanaka et al., 2007). These results indicated that the amino acids 243–253 are required specifically for the Tax binding.
Interestingly, the amino acids 243–253 are located within a putative α-helix motif of PDLIM2 at amino acids 236–254 (http://www.predictprotein.org; Figure 1b). Of note, our previous studies suggested that short helix is the preferential structure for Tax binding (Xiao et al., 2000). To test whether the short α-helix is responsible for the PDLIM2–Tax interaction, we disrupted the helix by smaller internal deletions or point substitutions with prolines. As shown in Figure 1g, disruption of the short helix, by its middle part small deletion (Δ241–245 or Δ246–250) or leucine to proline substitutions (LL241/242PP), completely abolished the PDLIM2–Tax interaction, although mutations of the N- or C-terminal of the short helix (Δ236–240 or Δ251–255, or EE249/250PP) already led to large loss of the interaction. Together, these studies clearly suggested that the putative α-helix motif of PDLIM2 is the Tax-binding motif essential for the PDLIM2–Tax interaction.
To identify the sequences with Tax involved in PDLIM2 binding, we took advantage of the Tax small deletion mutants that systematically cover the whole sequences of Tax (Figure 2a; Fryrear et al., 2009). Interestingly, all those mutants still retained the ability of PDLIM2 binding, same as the wild-type Tax (Figure 2b), suggesting that different sequences within Tax can complementarily interact with PDLIM2. This data are consistent with the fact that Tax is an intrinsically disordered protein, which contains typical disordered regions that exist dynamically for protein interaction (Radivojac et al., 2007; Boxus et al., 2008).
The major function of PDLIM2 in Tax repression is to promote Tax polyubiquitination and subsequently proteasomal degradation (Yan et al., 2009a). Thus, we next examined the role of the different function sequences of PDLIM2 in Tax polyubiquitination and degradation. Consistent with previous studies (Yan et al., 2009a), PDLIM2 strongly promoted Tax polyubiquitination (Figure 3a). In correspondence with the ubiquitin E3 liagase function of the LIM domain of PDLIM2 (Tanaka et al., 2007), the ΔLIM or ΔPDZ/LIM mutant of PDLIM2 failed to promote Tax polyubiquitination. Similarly, all PDLIM2 mutants defective in Tax binding such as the Δ79–278, Δ243–253, LL241/242PP and EE249/250PP mutants lost their ability in promoting Tax polyubiquitination. On the other hand, the PDLIM2 ΔPDZ and Δ195–207 mutants, which were capable of binding to Tax, still retained the ability in promoting Tax polyubiquitination. These results suggested that through its middle Tax-binding motif, PDLIM2 physically interacts with Tax, which in turn allows the C-terminal LIM domain to promote polyubiquitination of Tax.
Highly consistent with their dispensable role in promoting Tax ubiquitination, the PDZ domain and amino acids 195–207 as well as amino acids 258–278 are not involved in PDLIM2-stimulated Tax proteasomal degradation, because PDLIM2 mutants deleting these sequences, like the wild type PDLIM2, were able to promote Tax degradation in our pulse-chase assays (Figure 3b). In sharp contrast, disruption of the LIM domain (ΔLIM) or the Tax-binding motif (Δ243–253, LL241/242/PP, EE249/250PP) resulted in the disability of PDLIM2 in promoting Tax proteasomal degradation. It should be noted that the functional differences of the PDLIM2 mutants were not because of their different expressions, as their expression levels were comparable (Figures 3a and b, lower panels). Thus, both Tax-binding motif and LIM domain of PDLIM2 are crucial for promoting Tax polyubiquitination and subsequent proteasome-mediated degradation, although they have distinct roles in these events.
As PDLIM2 ΔLIM mutant binds to Tax, but does not induce Tax degradation, it will be interesting to examine whether this mutant can function as a dominant-negative form of PDLIM2. To address this important issue, we generated C8166 cells stably expressing an empty vector, PDLIM2 wild type or PDLIM2 ΔLIM, as this HTLV-I-transformed T cell line expresses a modest level of endogenous PDLIM2 (Yan et al., 2009a). Consistently, Tax protein underwent a modest proteasomal degradation, which was significantly accelerated by stable expression of exogenous PDLIM2 (Figure 3c). More importantly, the Tax degradation was completely blocked by expression of PDLIM2 ΔLIM. These results indicated that PDLIM2 ΔLIM functions as a dominant-negative version of PDLIM2, further suggesting the important roles of the LIM domain and Tax-binding motif in PDLIM2-mediated degradation of Tax.
In addition to promoting Tax polyubiquitination, PDLIM2 shuttles Tax from its functional sites such as perinuclear locations and Tax nuclear foci into the nuclear matrix for proteasomal degradation (Yan et al., 2009a). Our previous studies indicated that the Tax-shuttling function of PDLIM2 can be visualized by the indirect immunofluorescence staining to detect diminishment or disappearance of Tax perinuclear aggregates and nuclear foci (Yan et al., 2009a). In line with previous studies (Semmes and Jeang, 1996; Bex et al., 1997; Yan et al., 2009a), Tax strongly localized to the perinuclear aggregates and discrete nuclear bodies (Figure 4a, panel 1 from left) when Tax was expressed in Hela cells, which lack expression of endogenous PDLIM2 (Yan et al., 2009a). Consistent with our previous studies (Yan et al., 2009a), coexpression of PDLIM2 resulted in significantly less staining of Tax proteins and almost complete disappearance of Tax perinuclear aggregates and nuclear bodies (panel 2). When the proteasome was inhibited, the disappearance of Tax staining in the perinuclear aggregates and nuclear bodies was associated with increased Tax amount in the nuclear matrix (Yan et al., 2009a). Although it lost the cytoskeleton-binding ability, the PDZ deletion mutant could still shuttle Tax (panel 3), similar to the PDLIM2 wild type. In contrast, the LIM deletion mutant largely lost the ability in Tax shuttling, as evidenced by normal expression of Tax in perinuclear aggregates and nuclear bodies (panel 4). In high agreement with their abilities in Tax binding, the amino acids 195–207 or 258–278 deletion mutant behaved like wild-type PDLIM2 (panels 5 and 7) whereas the amino acids 243–253 deletion mutant as well as the LL241/242PP and EE249/250PP mutants of PDLIM2 lost the Tax-shuttling ability (panels 6, 8 and 9). Of note, all the mutations except the PDZ deletion did not affect expression pattern of PDLIM2, particularly the cytoskeleton binding under the immunofluorescence staining, suggesting no overall structure alterations of the PDLIM2 mutants. These results suggested that the Tax-shuttling function of PDLIM2 depends on the Tax-binding motif and LIM domain.
As the nuclear matrix is nuclear framework throughout the nucleoplasm, the expression of Tax and PDLIM2 in these two distinct subcellular locations cannot be distinguished by the immunofluorescence staining. To further confirm the role of different sequences within PDLIM2 in shuttling Tax into the nuclear matrix, we also performed the subcellular fraction assay using 293 cells, which express undetectable endogenous PDLIM2 protein (Yan et al., 2009a). As shown in Figure 4b, Tax was expressed in the cytoplasm, soluble nuclear fraction (nucleoplasm) and insoluble nuclear fraction (nuclear matrix) in the absence of PDLIM2. It should be noted that Tax proteins in the nuclear matrix were resistant to the proteasome-mediated degradation in the absence of PDLIM2, similar to those in the cytoplasm and nucleoplasm (Figure 3b; Yan et al., 2009a). In agreement with our previous studies (Figure 3b; Yan et al., 2009a), coexpression of PDLIM2 resulted in significant increase in Tax’s ratio in the nuclear matrix with decrease in Tax total protein (Figures 4b and c). Proteasome inhibition would increase Tax expression only in the nuclear matrix, but not other cellular fractions, suggesting that PDLIM2 shuttles Tax into the nuclear matrix for proteasomal degradation (Yan et al., 2009a). Similar to the PDLIM2 wild type, PDLM2 mutants deleting the PDZ domain (ΔPDZ), amino acids 195–207 (Δ195–207) and 258–278 (Δ258–278) were able to shuttle Tax into the nuclear matrix. In contrast, PDLIM2 mutants deleting the LIM domain (ΔLIM and ΔPDZ/LM) and mutants disrupting the Tax-binding motif (Δ79–278, Δ243–253, LL241/242PP and EE249/250PP) lost this ability. Together with the immunofluoresence studies, these results suggested that the Tax-binding motif and LIM domain, but not the PDZ domain, are essential for PDLIM2-mediated shuttling of Tax into the nuclear matrix.
Recent studies suggest that PDLIM2 is expressed in both cytoplasm and nucleus with the ability to interact with cytoskeleton (Torrado et al., 2004; Loughran et al., 2005; Tanaka et al., 2005; Yan et al., 2009a, b). However, it still remains largely unknown how PDLIM2 is expressed in the distinct subcellular compartments and the role of this different subcellular expression in PDLIM2 functions. To address this issue, we initially examined the effect of treatment of leptomycin (LMB), a specific inhibitor of chromosome region maintenance-1 (CRM1)-dependent nuclear export, on the subcellular expression of PDLIM2. As shown in Supplementary Figure 1, LMB treatment did not affect subcellular expression patterns of PDLIM2 and Tax. Consistently, LMB did not interfere with PDLIM2-mediated Tax nuclear degradation. The failure of LMB was not owing to its inefficiency in CRM1 inhibition, because inhibitor of nuclear factor-κBα nuclear export was blocked in the same LMB-treated cells. These data suggested that the nuclear exports of both PDLIM2 and Tax are CRM1-inpendent. To define the mechanism of PDLIM2 shuttling between the cytoplasm and the nucleus, we then examined the role of distinct sequences within PDLIM2 in its subcellular expression. In association with its inability to bind to cytoskeleton (Torrado et al., 2004; Loughran et al., 2005; also see Figure 4a), the PDZ domain deletion mutant showed an increased expression in the cytoplasm (Figure 4d). Interestingly, the PDZ domain deletion mutant also showed a decreased expression in the nucleoplasm, but no obvious effect in the nuclear matrix expression, suggesting an additional role of the PDZ domain in PDLIM2 nucleoplasmic expression. Although the LIM domain was not required for PDLIM2 binding to cytoskeleton (Loughran et al., 2005; Figure 4a), it was involved in PDLIM2 expression in the nuclear matrix, as deletion of this important domain led to significant decrease in the nuclear matrix, but increase in the cytoplasm and nucleoplasm of PDLIM2 protein (Figure 4d). This finding explained why the LIM domain is required for PDLIM2-mediated Tax shuttling into the nuclear matrix, although it is not required for PDLIM2 binding to Tax. Consistent with these results, deletion of both PDZ and LIM domains led to almost exclusive expression of PDLIM2 in the cytoplasm, whereas deletion of the entire region between the PDZ and LIM domains led to dramatic increase in the nuclear matrix with significant decrease in the nucleoplasm of PDLIM2 protein. On the other hand, disruptions of the Tax-binding motif or small internal deletions of the middle region of PDLIM2 had no effect on its subcellular distribution. These studies suggested that different domains of PDLIM2 have distinct roles in PDLIM2 subcellular expression and Tax shuttling.
To investigate the significance of these important domains in PDLIM-mediated suppression of Tax tumorigenicity, we generated Rat-1 fibroblasts stably expressing Tax alone or together with different PDLIM2 mutants (Figure 5a). Although expression of Tax alone or together with PDLIM2 or PDLIM2 mutants had no effect on the growth of Rat-1 cells in normal culture condition (Figure 5b); Tax could promote colony formation of Rat-1 cells in soft agar (Figure 5c). Consistent with our previous studies (Yan et al., 2009a), PDLIM2 was able to prevent the Tax-mediated anchorage-independent growth of Rat-1 cells. Highly consistent with their roles in Tax shuttling and proteasomal degradation, deletion of the LIM domain (ΔLIM) or disruption of Tax-binding motif (Δ243–253, LL241/242PP and EE249/250PP) resulted in loss of this suppression function of PDLIM2 on Tax-mediated tumorigenesis. On the contrary, deletion of the PDZ domain (ΔPDZ) or amino acids 258–278 (Δ258–278) had no statistically significant effect on the suppression of Tax-mediated transformation. These in vitro results were further substantiated by our in vivo studies. As shown in Figure 5d, the LIM domain and Tax-binding motif, but not the PDZ domain or amino acids 258–278, were required to suppress Tax-mediated tumorigenesis in severe combined immunodeficiency mice.
In conclusion, we have dissected different functional sequences within PDLIM2. We have shown that a putative short α-helix of PDLIM2 at amino acids 236–254 functioned as the selective Tax-binding motif, therefore required for Tax subcellular redistribution into the nuclear matrix, polyubiquitination, degradation and eventual suppression of Tax tumorigenicity. Interestingly, the C-terminal LIM domain and the N-terminal PDZ domain were respectively involved in PDLIM2 binding to the nuclear matrix and cytoskeleton, although both domains were dispensable for Tax binding. In further support of the fact that PDLIM2 targeted Tax into the nuclear matrix for proteasomal degradation, the LIM domain, but not the PDZ domain, was required for PDLIM2-mediated Tax suppression. These studies thus provide important insights into the molecular actions of PDLIM2 on Tax regulation. These studies also have the general significance in cancer biology and treatment, given our recent findings linking PDLIM2 epigenetic repression to pathogenesis of different cancers such as breast cancer and colon cancer (Qu et al., 2010a, b).
We thank N Raab-Traub for Rat-1 cells, OJ Semmes for Tax-GFP mutants, and NIH AIDS Research & Reference Reagent Program for various reagents. This study was supported by NIH/NCI Grant R01 CA116616 and ACS Grant RSG-06-066-01-MGO to G Xiao.
Conflict of interest
The authors declare no conflict of interest.