The E3 ubiquitin ligase TRIM21 causes degradation of HuR in response to heat shock
Moderate heat shock, at 43°C for 2 h, to MCF7 cells in 10% serum-containing medium caused a strong time-dependent reduction of HuR protein to ~25% of initial level, with no change in mRNA level (Fig. 1A). Heat shock at 43°C to MCF7 cells in serum-free medium also caused a similar reduction in HuR protein level, as reported earlier in HeLa cells26, suggesting that the HuR response to heat shock is independent of serum or cell type (Suppl. Fig. 1). Exogenously expressed EGFP-tagged HuR showed a similar reduction in response to heat shock (Fig. 1B). Both nuclear and cytoplasmic HuR showed similar degradation in heat-stressed cells, indicating cytoplasmic translocation as not a prerequisite for HuR degradation (Suppl. Fig. 2). Protein synthesis inhibition by cycloheximide (CHX) rapidly reduced HuR in heat-stressed cells compared to cells at 37 °C, demonstrating a significantly higher degradation rate of HuR under heat shock. However, treatment with the proteasomal inhibitor MG132, in presence of CHX, maintained the cellular level of HuR at 43°C (Fig. 1C). The half-life of HuR in heat-stressed cells treated with CHX goes down to ~30 min, whereas it is restored to more than 2 h on MG132 treatment (Fig 1D). This demonstrates that HuR is proteasomally degraded under heat shock.
As the K182 residue of HuR is reported to be polyubiquitinated under heat shock, we expressed 6X-His-tagged wild type (WT) and K182R mutant HuR proteins in cells and subjected them to heat shock. WT HuR was efficiently degraded, whereas K182R mutant HuR was protected from degradation under heat shock (Fig 1E). WT HuR underwent polyubiquitination in cells exposed to heat shock which was significantly reduced in case of the K182R mutant (Fig. 1F).
The K182 residue is polyubiquitinated by the ubiquitin ligase TRIM21 in response to UVC irradiation27. We therefore investigated whether TRIM21 was the E3-ligase responsible for degrading HuR under heat shock. HuR and TRIM21 showed interaction in cells at normal temperature and enhanced interaction in cells exposed to heat shock at 43°C for 1 h and treated with MG132 (Fig. 1G). siRNA-mediated depletion of TRIM21 prevented the degradation whereas overexpression of TRIM21 enhanced the degradation of HuR under heat shock (Fig. 1H and 1I).
To determine the involvement of K182 in the degradation of HuR by TRIM21 under heat shock, TRIM21 was co-expressed in cells expressing either myc-tagged WT or K182R mutant HuR. Upon heat shock, WT-HuR was degraded but the HuR K182R mutant was protected from degradation (Fig. 1J). K182R mutant HuR also showed significantly reduced interaction with TRIM21 upon heat shock (Fig. 1K). Together, these observations demonstrated TRIM21 as the E3 ubiquitin ligase causing the degradation of HuR under heat shock via the ubiquitination of K182.
TRIM21-mediated degradation of HuR protects cells from heat shock-induced cell death
TRIM21 enhances cell proliferation and reverses the DNA damage-induced cell death in UV-irradiated cells28. We therefore investigated the effect of TRIM21 on cell viability and death in response to heat shock. TRIM21 overexpression enhanced cell viability at 37°C and also significantly rescued the viability of heat-stressed cells (Fig. 2A, left panel). Conversely, TRIM21 depletion reduced cell viability and further accentuated the reduction of viability of heat-stressed cells (Fig. 2A, right panel). TRIM21 overexpression reduced caspase 3/7 activity in cells at 37°C and also significantly reduced the enhanced caspase activity in heat-stressed cells (Fig. 2B, left panel). The opposite effect was seen on TRIM21 knockdown (Fig. 2B, right panel). Annexin V/PI staining of cells overexpressing TRIM21 showed significant reduction in the enhanced apoptosis of cells exposed to heat shock (Fig. 2C). The converse was observed in cells in which TRIM21 was knocked down (Fig. 2D). This demonstrated that TRIM21 enhanced the viability and reduced apoptosis of heat- stressed cells.
We then investigated the effect of TRIM21 on the morphological changes of heat-stressed cells. Heat shock caused a visible change in cellular morphology in comparison to cells at 37°C (Suppl. Fig. 3 and 4). The reduction in cell area and perimeter upon heat shock were significantly restored on TRIM21 overexpression (Fig. 2E). Conversely, TRIM21 knockdown further reduced cell area and perimeter of heat stressed cells (Fig. 2F). TRIM21 also enhanced the long-term viability and proliferation of heat-stressed cells. TRIM21 overexpression significantly enhanced the proliferation of cells at 37°C over a 72 h period. While cells exposed to heat shock showed significantly reduced proliferation during this period, TRIM21 overexpression significantly enhanced the proliferation of heat-stressed cells and restored it nearly to the level of cells unexposed to heat shock (Fig. 2G). Depletion of TRIM21 showed the opposite effect (Fig. 2H). Heat-stressed cells formed significantly less number of colonies compared to cells not exposed to heat shock, whereas colony formation by heat-stressed cells overexpressing TRIM21 was nearly similar (Fig. 2I). Together, these observations demonstrated a profound role of TRIM21 in ensuring short term survival and long term viability of cells exposed to heat shock.
We thereafter investigated whether the survival advantage provided by TRIM21 under heat shock was mediated by its degradation of HuR. Overexpression of both WT HuR and HuR K182R, which is refractory to TRIM21-mediated degradation, significantly reduced the enhanced viability of cells observed on TRIM21 overexpression. WT HuR reduced the enhanced viability of heat-stressed cells overexpressing TRIM21 which was further reduced on overexpression of HuR K182R mutant (Fig. 2J, left panel). The opposite effect was observed in case of cellular caspase activity (Fig. 2J, right panel). This demonstrated that WT HuR, and more so the non-degradable K182R mutant HuR, could reverse the increase in cell viability and decrease in apoptosis induced by TRIM21 overexpression in heat-stressed cells. WT HuR and K182R mutant HuR also reduced the enhancement in the proliferation rate caused by TRIM21 overexpression in cells exposed to heat shock (Fig. 2K). Together, these findings demonstrated that TRIM21’s ability to protect cells from heat shock-induced death is, at least partly, mediated by its degradation of HuR.
S100 and E101 residues of HuR are determinants of recognition by TRIM21
Heat shock induced the interaction between HuR and TRIM21 suggesting the presence of specific determinants of TRIM21 binding in HuR which might be post-translationally modified in response to heat shock. We adopted an unbiased approach to delineate the determinants of HuR interaction with TRIM21 under heat shock. His-tagged versions of various deletion mutants, consisting of different domains of the HuR protein, were constructed (Fig. 3A). When expressed in cells exposed to heat shock, only WT HuR and RRM1-2 showed degradation, indicating that the region of HuR interacting with TRIM21 was localized in RRM1-2 (Fig. 3B). Interestingly, RRM2-3, which still contained the K182 residue, did not show degradation under heat shock, suggesting that residues necessary for substrate recognition by TRIM2 were present in RRM1 or in the linker region. Also, only WT HuR and HuR RRM1-2 interacted with TRIM21, further indicating that RRM1-2 contained the residues crucial for recognition of HuR by TRIM21 (Fig. 3C).
We then adopted molecular dynamics (MD) simulation to identify the putative contact points between HuR and TRIM21. The MD simulation was performed with HuR RRM1-2 (PDB: 4EGL) and the PRYSPRY domain of TRIM21 (PDB: 2IWG), that has been shown to be responsible for substrate binding29. The simulation showed a rapid reduction and subsequent stabilization of the solvent accessible surface area (SASA) of the system indicating interaction between the two molecules (Suppl. Fig. 5A). The contact area was around 30 nm2 and remained nearly constant throughout the length of the simulation post binding. The interaction was also energetically stable, being of the order of 103 kJ/mol and was mostly electrostatic (Suppl. Fig. 5B). The putative contact surface between HuR and TRIM21 showed two regions of HuR, between 98-106 amino acids and between 111-119 amino acids. Among these residues, the highest probability of contact was obtained for two amino acids, S100 and E101, located in the linker region between RRMs 1 and 2 (Fig. 3D). Another two amino acids, R115 and T116, also showed interaction with TRIM21 residues but with lower probability. The molecular distance between S100 and K182, the site of TRIM21-mediated ubiquitination, was estimated to be 24.19 Å, whereas the distance between R115 and K182 was 19.22 Å (Fig. 3E). However, K182 and S100 are located on the same face of HuR, whereas R115 is located on the opposite face. The MD simulation results allowed the molecular docking of HuR RRM1-2 and TRIM21 PRYSPRY domains, showing a close contact between S100 residue of HuR and TRIM21 (Fig. 3F). The MD simulation and molecular docking results suggested S100 and E101 residues as the possible primary determinants of interaction between HuR and TRIM21.
S100 and E101 residues of HuR are required for TRIM21 binding and degradation of HuR under heat shock
The S100 and E101 residues of HuR were mutated to generate HuR S100A and E101A mutants and a S100A/E101A double mutant (Fig. 4A). All the three mutant HuR proteins were found to be refractory to heat shock-induced degradation while WT HuR was degraded efficiently (Fig. 4B). TRIM21 failed to interact with either of the single mutants or the double mutant, while it interacted efficiently with WT HuR (Fig. 4C). The S100A/E101A double mutant HuR also showed reduced ubiquitination compared to WT HuR under heat shock, (Fig. 4D). This demonstrated S100 and E101 residues as necessary for the heat shock-induced ubiquitination and degradation of HuR by TRIM21.
We checked the effect of the HuR S100A/E101A mutant on cell viability and apoptosis after exposure to heat shock. Overexpression of the HuR S100A/E101A double mutant significantly reversed the enhancement of cell viability and reduction in caspase activity observed upon TRIM21 overexpression both in cells exposed to 37°C and 43°C (Fig. 4E, left top and bottom panels). Similarly, overexpression of the HuR S100A/E101A mutant reversed the reduction in apoptosis of heat-stressed cells upon TRIM21 overexpression (Fig. 4E, right top and bottom panels and Suppl. Fig 6). The HuR double mutant also significantly reduced the enhanced proliferation of cells overexpressing TRIM21 during the 72 h period post exposure to heat shock (Fig. 4F). Therefore, the degradation-resistant S100A/E101A double mutant HuR could counteract TRIM21’s ability to ensure cell survival and proliferation under heat shock.
We also tested whether the R115 and T116 residues were required for degradation of HuR and interaction with TRIM21. A R115A/T116A double mutant HuR showed degradation similar to WT HuR upon exposure to heat shock (Suppl. Fig. 7A). Also the R115A/T116A double mutant showed interaction with TRIM21 in cells exposed to heat shock (Suppl. Fig. 7B). Therefore, the R115 and T116A residues did not determine the interaction between HuR and TRIM21 and did not contribute to HuR degradation under heat shock.
Finally, we determined the effect of the S100A/E101A mutant on the mRNA levels of some of the post-transcriptional targets of HuR which are related to cell proliferation, apoptosis and stress responses. qRT-PCR for specific HuR target mRNAs showed that these mRNA levels were strongly reduced in WT HuR-overexpressing cells exposed to 43 °C, whereas overexpression of HuR S100A/E101A double mutant restored these mRNAs to nearly normal levels (Fig. 4G). Also, HuR S100A/E100A double mutant retained the ability to interact with these mRNAs (Suppl. Fig. 8). This indicated that the degradation resistant HuR double mutant protein could bind with and stabilize the HuR target mRNAs which are otherwise destabilized in heat-stressed cells.
Phosphorylation of S100 is required for TRIM21 binding and degradation of HuR under heat shock
As phosphorylation of HuR have been shown to influence its sub-cellular localization and mRNA-binding ability under different stress conditions24,30, we investigated whether S100 was phosphorylated in response to heat shock. WT HuR showed high level of serine phosphorylation in heat stressed cells whereas the S100A mutant showed strongly reduced serine phosphorylation, demonstrating S100 as a target for phosphorylation under heat shock (Fig. 5A). A phosphomimetic S100D mutant was degraded in cells even at 37°C (Fig. 5B). However, knockdown of TRIM21 prevented the degradation of HuR S100D mutant in the absence of heat shock (Fig. 5C). TRIM21 also efficiently interacted with the HuR S100D mutant both in absence and presence of heat shock, whereas it interacted efficiently with WT HuR only upon exposure to heat shock (Fig. 5D). Together, these data indicated that the phosphorylation of S100 was responsible for recognition of HuR by TRIM21 and subsequent degradation. We also investigated the effect of the S100D mutant on cell viability and apoptosis in response to heat shock. Overexpression of WT HuR reduced cell viability and enhanced caspase activity in heat-stressed cells, as observed before, whereas overexpression of the S100D mutant showed the opposite effect (Fig. 5 E and F). Therefore, the rapid degradation of the HuR S100D mutant provided the same survival advantage to the cells caused by TRIM21-mediated degradation of HuR under heat shock.
AKT1 phosphorylates S100 leading to TRIM21-mediated degradation of HuR under heat shock
We then proceeded to identify the kinase phosphorylating S100 of HuR in response to heat shock. S100 is part of a kinase substrate motif RXXS/T (R97PSS100) which was found to be conserved across vertebrates (Fig. 6A). One of the kinases with the substrate motif RXXS/T is AKT1, which plays a central role in mediating critical cellular responses including cell growth and survival31,32. AKT1 is also reported to be induced by heat shock and involved in suppression of heat shock-induced cell death33,34. Therefore we tested whether AKT1 might be the kinase phosphorylating HuR in response to heat shock. Treatment of cells with an Akt inhibitor prevented the degradation of HuR in heat stressed cells (Fig. 6B). siRNA-mediated knockdown of AKT1 also inhibited the degradation of HuR in response to heat shock (Fig. 6C). Exposure to 43°C increased the phosphorylation of AKT1 in a time-dependent manner, without increasing the level of AKT1, indicating the activation of AKT1 by heat shock (Fig. 6D). The Akt inhibitor prevented serine phosphorylation of HuR in response to heat shock (Fig. 6E). HuR also interacted with phosphoAKT1 in response to heat shock, which was abrogated by treatment with Akt inhibitor, suggesting that activated AKT1 interacted with HuR under heat shock (Fig. 6F). WT HuR interacted with phosphoAKT1 while the S100A mutant HuR failed to do so, showing S100 as the target for AKT1-mediated phosphorylation in response to heat shock (Fig. 6G).
As AKT1-mediated phosphorylation of S100 would be dependent on recognition of the substrate motif RXXS/T by AKT1, we disrupted this motif in HuR by mutating R97 to A. The HuR R97A mutant did not get degraded under heat shock, indicating the importance of the RXXS/T motif for substrate recognition by the kinase (Fig. 6H). The R97A mutant also did not undergo serine phosphorylation under heat shock, further demonstrating the requirement of the AKT1 recognition motif for serine phosphorylation of HuR under heat shock (Fig. 6I). Furthermore, the R97A mutant HuR failed to interact with phosphoAKT1 in response to heat shock (Fig. 6J). Together, these findings demonstrated the presence of S100 within a bona fide AKT1 substrate motif in HuR and AKT1 as the kinase phosphorylating S100 under heat shock.
Finally we checked whether AKT1-mediated phosphorylation of HuR was necessary for interaction of HuR with TRIM21 in response to heat shock. Treatment with the Akt inhibitor drastically reduced TRIM21 interaction with HuR in heat-stressed cells (Fig. 6K). siRNA-mediated knockdown of AKT1 also prevented the interaction of HuR with TRIM21 in response to heat shock (Fig. 6L). Finally, the R97A mutant also failed to interact with TRIM21 under heat shock whereas the WT HuR interacted efficiently (Fig. 6M). Together, these observations showed that AKT1-mediated phosphorylation of S100 in HuR allowed the binding of TRIM21 to HuR, leading to polyubiquitination of K182 and subsequent degradation of HuR in response to heat shock.