CLIC4 (Chloride intracellular channel 4) is a member of a family of intracellular chloride channel proteins that are ubiquitously expressed in multiple tissue types1
. In addition to chloride channel activity in artificial and biological membranes, CLIC family members participate in cell cycle control, cytoskeletal function, mitosis and differentiation1
, but the pathway(s) through which they function are undefined. Analysis of the molecular structure of CLIC1 and CLIC4 reveals dimorphic proteins that exist in both soluble and membrane-bound configurations at least in part regulated by redox potential2–4
. CLIC4 is essential for apoptosis mediated by p53 and c-Myc, and the CLIC4 promoter is a direct downstream target of these transcription factors 5,6
. Cytoplasmic CLIC4 translocates to the nucleus under conditions of metabolic stress, growth arrest, apoptosis and DNA damage mediated by a functional nuclear localization signal (NLS) on the carboxy terminus of the protein7
. Nuclear CLIC4 residence is an essential component of its pro-apoptotic and growth arrest activity in keratinocytes 8
. In contrast, CLIC4 is excluded from the nucleus in epithelial cancer cells but upregulated in tumour stromal cells associated with myofibroblast conversion9
. CLIC4 was previously linked to myofibroblast conversion in TGF-β treated mammary fibroblasts10
. Thus, while CLIC4 participates in growth control and tissue remodelling, the signalling pathway through which CLIC4 participates is not known.
Yeast two-hybrid assays were carried out using six CLIC4 sequences spanning the entire protein as baits. The CLIC4 bait sequence with amino acids 120–254 interacted with several potential binding proteins, one of which was amino acids 814–1167 of Schnurri-2, a zinc finger protein known to function in the TGF-β superfamily signalling pathway11–13
. The interaction of CLIC4 and Schnurri-2 was validated using co-immunoprecipitation assays in primary cultures of mouse keratinocytes. HA-tagged CLIC4 or empty vector was expressed and Schnurri-2 interacting proteins were co-immunoprecipitated using anti-Schnurri-2 antibody. Both endogenous and exogenous CLIC4 were revealed as Schnurri-2 interacting proteins upon immunoblotting of SDS-PAGE separated immunoprecipitates (, Supplementary Information, Fig. S1a
), confirming the yeast two-hybrid results. CLIC4 and Schnurri-2 do not associate outside of cell context, since in vitro
transcribed and translated full length Schnurri-2 or its interacting domain or the interacting domains of both proteins failed to coimmunoprecipitate with recombinant full-length CLIC4 (data not shown).
TGF-β enhances the expression and association of CLIC4 and Schnurri-2
Deletion constructs of C-terminal V5-His tagged CLIC4 (Supplementary Information Fig. S1b
) in adenoviral vectors were expressed in primary keratinocytes Supplementary Information Fig. S1c
indicates that Schnurri-2 antibody co-precipitates full length CLIC4 (aa 1–253) and the region of CLIC4 minus the NLS (1–197) but not the N-terminal CLIC4 half (1–120) or 1–60 that contains the transmembrane domain. Interaction with the 1–120 or 1–60 domains could not be detected even at the highest exposures. Thus, Schnurri-2 interacts between amino acid 121–197 of CLIC4 consistent with bait region attracting Schnurri-2 in the yeast two-hybrid assay.
Since CLIC4 causes cell cycle arrest and Schnurri-2 participates in TGF-β superfamily (Dpp/BMP) signalling, we asked if TGF-β could influence the interaction of CLIC4 and Schnurri-2. Whole cell protein lysates from keratinocytes that were treated with TGF-β1 for different periods of time show a persistent increase in CLIC4 and Schnurri-2 levels up to 24h of TGF-β1 treatment (). The increase in CLIC4 and Schnurri-2 by TGF-β is transcriptional and is sustained (Supplementary Information, Fig. S2a
). Furthermore, skin biopsies obtained from single and double transgenic mice overexpressing active TGF-β1 in the epidermis in response to doxycycline stimulation 14
(Supplementary Information Fig. S2b
), show elevated transcripts for both Schnurri-2 and CLIC4 by 4–5 days after doxycline. At this time point, transcripts for the TGF-β response gene Smad7 are also increased as previously reported in this model. Thus, CLIC4 and Schnurri-2 respond to TGF- β stimulation similarly in vitro
and in vivo
. TGF- β also stimulates the interaction of CLIC4 and Schnurri-2.. Supplementary Information, Fig. S3
shows by co-immunoprecipitation that TGF-β1 treatment of keratinocytes increases the interaction of both endogenous and exogenous CLIC4 with Schnurri-2 within 30 minutes. The interaction in subcellular fractions appears to occur only in the cytoplasmic compartment of TGF-β1-treated cells and persists for at least 6 hours (). Interaction of these proteins in the nucleus could not be detected. The TGF-β stimulated interaction of CLIC4 and Schnurri-2 is not Smad dependent since siRNA mediated knockdown of Smad2, Smad3 or Smad4 did not interrupt this interaction (Supplementary Information, Fig. S4
Expression of exogenous CLIC4 in mouse keratinocytes modestly enhanced the activity of the TGF-β dependent reporter plasmid p3TP-lux () in response to low doses (50 pg/ml) of TGF-β1. CLIC4 increased the reporter activity in a TGF-β1 dose dependent manner (Supplementary Information, Fig. S5a
). This enhancement required the presence of Schnurri-2 as Schnurri-2 siRNA prevented CLIC4 from increasing TGF-β reporter activity (, Supplementary Information, Fig. S5a
). Growth inhibition is a major physiological response of keratinocytes to TGF-β. shows that exogenous CLIC4 alone inhibited DNA synthesis and together with TGF-β 1 inhibited keratinocyte DNA synthesis to a greater extent than TGF-β 1 alone, and this required the presence of Schnurri-2. Keratinocyte growth inhibition by CLIC4 and TGF-β also followed a TGF-β 1 dose dependent course (Supplementary Information, Fig. S5b). Schnurri-2 is a Smad binding partner in Dpp and BMP signalling in Drosophila and mice respectively and alters downstream signalling12,13
. We confirmed that Schnurri-2 also modifies TGF-β 1 signalling in a mammalian system by introducing a Schnurri-2 expression plasmid pAct-FLAG-hShn2 into HEK293, a readily transfectable epithelial cell line. By itself Schnurri-2 caused CLIC4 to translocate to the nucleus and in conjunction activated Smad2 and suppressed c-Myc (Supplementary Information, Fig. S6
). With TGF-β1, Schnurri-2 increased TGF-β dependent signal transduction, transcriptional activity and growth inhibition (, Supplementary Information, Fig. S7a
). These results indicate that Schnurri-2 is involved in TGF-β signalling in addition to its known function in BMP signalling.
CLIC4 and Schnurri-2 enhance TGF-β signaling
We next examined the biological significance of CLIC4 in TGF-β responses using RNA interference to knockdown CLIC4 expression in keratinocytes. TGF-β1-induced p3TP-lux activity and inhibition of thymidine incorporation by TGF-β1 in keratinocytes were reduced by expression of specific CLIC4shRNA further confirming participation of CLIC4 in TGF-β signalling (). Expressing exogenous CLIC4 in keratinocytes also enhanced the TGF-β1 dependent downregulation of c-Myc and upregulation of p21 even in the absence of TGF-β1 (). These changes are known to participate in TGF-β mediated growth inhibition15,16
. Furthermore, suppression of CLIC4 by shRNA enhanced c-Myc and reduced p21 expression in the absence of TGF-β1 and reduced c-Myc inhibition and p21 induction in the presence of TGF-β1 ().
TGF-β1 increases total CLIC4 and causes nuclear trafficking of CLIC4 in keratinocytes within 30 minutes (). Nuclear CLIC4 can be detected by confocal microscopy following exposure to TGF-β1 as early as 1 hour (data not shown) and is sustained for at least 24 hours (). Similarly, exogenous Schnurri-2 (pAct-FLAG-hShn2) translocates to the nucleus of HEK293 cells exposed to TGF-β1 in a time course virtually identical to that observed for endogenous CLIC4 in HEK293 cells (). Confocal microscopy for FLAG staining confirmed nuclear trafficking of Schnurri-2 (). Together these results support a mechanism whereby TGF-β1 enhances its activity through stimulating an interaction of CLIC4 and Schnurri-2 and promoting their nuclear translocation.
TGF-β induces nuclear translocation of CLIC4 to enhance TGF-β signalling
To confirm that nuclear translocation of CLIC4 by TGF-β1 has functional significance, we introduced a nuclear targeted CLIC4 expression adenovirus7
into keratinocytes and monitored the TGF-β responses. Both CLIC4 and nuclear CLIC4 enhanced TGF-β1 dependent reporter activity, but nuclear CLIC4 was 10 fold more potent than wild-type CLIC4 overexpression (). TGF-β dependent inhibition of DNA synthesis in keratinocytes is enhanced by both wild-type and nuclear targeted CLIC4 but much more so by the latter. Furthermore, nuclear CLIC4 caused marked inhibition of DNA synthesis in the absence of exogenous TGF-β (). In order to examine whether growth inhibition by nuclear CLIC4 is TGF-β dependent, we expressed exogenous wild-type CLIC4 and nuclear targeted CLIC4 in primary keratinocytes from Smad3 knockout mice and wild-type littermates and treated with TGF-β1 for 16h. indicates that the lack of Smad3 increased thymidine incorporation in keratinocytes consistent with a growth inhibitory function of the TGF-β pathway for this cell type. CLIC4 and especially nuclear CLIC4 enhanced TGF-β mediated growth inhibition in wild-type keratinocytes but were much less effective in Smad3 knockout keratinocytes. While DNA synthesis in the wild-type keratinocytes was inhibited 83 fold by overexpression of nuclear CLIC4, the Smad3 knockout cells were only inhibited by 9 fold, thus confirming that a completely intact TGF-β signalling pathway is essential for CLIC4 to inhibit keratinocyte growth.
To understand the mechanism by which Schnurri-2 facilitates CLIC4 enhancement of TGF-β signalling, we analyzed the effect of Schnurri-2 siRNA on TGF-β1 dependent nuclear translocation of CLIC4. Knockdown of Schnurri-2 prevented CLIC4 from translocating to the nucleus in response to TGF-β1 treatment of keratinocytes () indicating that Schnurri-2 is important for CLIC4 nuclear translocation in response to TGF-β. indicates that depletion of Schnurri-2 prevents wild-type CLIC4 from enhancing TGF-β1 transcriptional activation (also ) but does not prevent nuclear CLIC4 from enhancing TGF-β reporter activity. This result was confirmed using immortalized MEFs (mouse embryonal fibroblasts) from Schnurri-2 knockout mice. Supplementary Information, Fig. S7b
shows that wild-type CLIC4 is unable to enhance TGF-β reporter activity in Schnurri-2 knockout MEFs. However, exogenous expression of Schnurri-2 restores the ability of wild-type CLIC4 to enhance TGF-β signalling in these cells. In contrast, nuclear targeted CLIC4 enhances reporter activity in the absence of Schnurri-2. Thus, the primary requirement for Schnurri-2 in this pathway is to facilitate CLIC4 nuclear translocation and is not required for CLIC4 nuclear action.
Schnurri-2 is essential for TGF-β dependent nuclear translocation of CLIC4
shows that nuclear targeting of CLIC4 enhances nuclear phospho-Smad2 and phospho-Smad3 levels even in the absence of exogenous TGF-β stimulation. Expression of CLIC4shRNA decreased nuclear phospho-Smad2 and phospho-Smad3 in both untreated keratinocytes and keratinocytes treated with TGF-β1 for 1 hour (Supplementary Information, Fig. S8a
). Targeting of a non-specific protein to the nucleus (GFP) by adenoviral vector7
did not enhance levels of phospho-Smad2-3 (data not shown) indicating that nuclear CLIC4 has a specific effect. Phospho- Smad2 and 3 levels are also increased when endogenous CLIC4 is targeted to the nucleus by TGF-β treatment, and the levels are sustained for extended periods when exogenous wild-type CLIC4 translocates to the nucleus by TGF-β treatment (Supplementary Information, Fig. S8b
). Smad4 levels are not altered by nuclear CLIC4 (). Phospho-Smad2 and 3 are predominantly nuclear proteins although phosphorylation occurs in the cytoplasm. Nuclear CLIC4 primarily enhances nuclear phospho-Smad2 and 3 even in the absence of exogenous TGF-β (). Nuclear CLIC4 does not enhance translocation of phospho-Smad2 or 3 from the cytoplasm to the nucleus in the absence or presence of exogenous TGF-β since cytoplasmic phospho-Smad2 or 3 are not reduced in its presence (). The increase in nuclear phospho-Smads is not a consequence of enhanced total Smad protein levels since they remained unchanged ().When keratinocytes were treated with TβRI inhibitor SB431542, the overall levels of phospho-Smad2 decreased due to blockade of basal Smad activation. Nevertheless phospho-Smad2 levels were partially sustained by nuclear CLIC4 (Supplementary Information, Fig. S9a
). This result indicated that CLIC4 was not acting through receptor mediated phosphorylation and suggested that spontaneous dephosphorylation of phospho-Smads was delayed by nuclear CLIC4. indicates that CLIC4 directly interacts with phospho-Smad2-3 in keratinocytes. After TGF-β treatment for just a short time, endogenous CLIC4 coimmunoprecipitates from lysates of the nuclear fraction with endogenous phospho-Smad2 and 3 () although the abundance of phospho-Smads (and hence the level of immunoprecipitated phospho-Smads) diminishes over the 3 hour time course consistent with the reported degradation of nuclear phospho-Smads after TGF-β 17
.Co-immunoprecipitation analysis demonstrates that CLIC4 also interacts with Smad4 (Supplementary Information, Fig. S9b
Nuclear CLIC4 associates with and stabilizes phospho-Smad2 and phospho-Smad3
Among the complexities of TGF-β signalling a dynamic interplay of Smad phosphorylation and dephosphorylation regulates nuclear responses. Ligand stimulation results in R-Smad phosphorylation and accumulation of Smad complexes in the nucleus. where they are dephosphorylated and exported from the nucleus thereby terminating the signal 18,19
. In order to understand if physical association of CLIC4 with phospho-Smad2 and 3 could protect the two from dephosphorylation, we exposed nuclear lysates of keratinocytes transduced with either the control (Vector), wild-type or nuclear targeted CLIC4 adenovirus to lambda protein phosphatase, a serine threonine phosphatase. shows that phospho- Smad2 and 3 in lysates containing nuclear CLIC4 are relatively more resistant to phosphatase treatment whether isolated from TGF-β treated or untreated keratinocytes. Relative to their respective untreated samples, 2 units of lambda phosphatase enzyme dephosphorylates p-Smad2 by ~75% and ~50%, in vector only and wild-type CLIC4 overexpressing samples respectively but only by 10% in nuclear targeted CLIC4 expressing samples. Similarly, p-Smad3 is dephosphorylated by ~80–95% in empty vector, ~70–80% in wild-type CLIC4 and only 50% in nuclear CLIC4 expressing samples. Thus in this assay nuclear CLIC4 and to a lesser extent wild-type CLIC4 protect phospho-Smad2–3 from the action of phosphatases. With the short TGF-β exposure (1 hour), only small amounts of wild-type CLIC4 are expected in the nucleus.
PPM1a has been identified as a Smad binding protein and a Smad phosphatase18
. When nuclear CLIC4 is overexpressed in TGF-β treated keratinocytes, co-immunoprecipitation of phospho-Smad2 and 3 with PPM1a is inhibited () . Notably, wild-type CLIC4 also inhibits the interaction of phospho-Smad2-3 with PPM1a consistent with the earlier results that CLIC4 translocates to the nucleus after TGF-β treatment. RNAi mediated knockdown of PPM1a increased levels of phospho-Smad2 and 3 in lysates from keratinocytes transduced with empty vector or wild-type CLIC4 but did not further enhance the levels in samples with nuclear targeted CLIC4 (), indicating that at least part of the nuclear CLIC4 effect on p-Smad2/3 stabilization is due to its inhibition of PPM1a activity on phospho-Smads. That the levels of phospho-Smad2 and phospho-Smad3 in vector and wild-type CLIC4 groups did not rise to the same level as in nuclear targeted CLIC4 after PPM1a knockdown might indicate that other, yet to be described, Smad phosphatases are also inhibited by nuclear resident CLIC4 (or perhaps the knockdown of PPM1a is incomplete).
These results depict a novel mechanism of enhancement of TGF-β signalling. TGF-β stimulates the association of Schnurri-2 and CLIC4 in the cytoplasm leading to Schnurri-2 dependent nuclear translocation of the complex where it dissociates. Nuclear CLIC4 stabilizes nuclear phospho-Smad2 and phospho-Smad3 levels by associating with these proteins and protecting them from PPM1a dependent dephosphorylation and possibly from other Smad phosphatases, thus enhancing and sustaining the TGF-β nuclear signalling (Supplementary Information, Fig. S10
). The protection of nuclear phospho-Smads by CLIC4 enhances both early (c-Myc downmodulation and p21 upregulation) and later (growth inhibition) consequences of TGF-β signalling also prolonging the response. A positive feedback loop may also be established since a consequence of TGF-β signalling is to increase the expression of both Schnurri-2 and CLIC4.
Previous data indicated that the CLIC4 NLS was required for nuclear translocation in response to cell stress7
, suggesting that aside from TGF-β stimulation, there are additional mechanisms regulating CLIC4 nuclear translocation independent of Schnurri-2. Nevertheless, TGF-β signalling could be downstream from CLIC4 nuclear translocation in response to stress and cell cycle arrest as well as the participation of CLIC4 in myofibroblast conversion. CLIC4 loss in tumors could contribute to tumor resistance to TGF-β signalling. Schnurri-2 and CLIC4 now join a panoply of regulators of Smad signalling in the nucleus reflecting the need for precise control of TGF-β signalling. In addition to phospho-Smad dephosphorylation by PPM1a there is the Arkadia mediated degradation of the co-repressor SnoN 20
, the direct action of Arkadia on both activation and degradation of phospho-Smads 17
and retention of Smad proteins in the nucleus by TAZ, a transcriptional co-regulator 21
. Together, the combination of Schnurri-2 and CLIC4 present a unique duo of novel players in a critical pathway controlling diverse aspects of cell behaviour and may serve as targets for modifying TGF-β signalling under conditions where this pathway is involved in pathological changes.