Identification of BRAF as a novel SPOP interactor
Previously, we and others have identified multiple oncoproteins as degradative or non-degradative substrates of the CRL-SPOP complex using the yeast two-hybrid method or affinity purification coupled with mass spectrometry. Substrate-binding consensus (SBC) motifs (Φ-π-S/T-S/T-S/T, where Φ is a nonpolar residue and π is a polar residue) have been well-characterized in the majority, if not all, of these substrates [9]. To expand the SPOP interactome to a proteome-wide level, we performed an SBC motif search against the human proteome sequence using the webtool ScanProsite (https://prosite.expasy.org/scanprosite/). Among the hundreds of potential candidates identified by this search, we noticed that BRAF harbors a perfectly matched SBC motif (120-VTSSS-124 aa), which is very similar to those present in several known SPOP substrates, including ERG, ATF2, Caprin1, and DAXX (Fig. 1a). Notably, this motif was absent in BRAF paralogs of ARAF and CRAF (Fig. 1b). Given that BRAF is frequently hyperactivated in various human cancers, we explored whether BRAF is an authentic substrate of the CRL3-SPOP complex, and whether BRAF activity is dysregulated in SPOP-mutated cancers. Using reciprocal co-IP assays, we demonstrated that ectopically expressed BRAF interacted with SPOP (Fig. 1c, d). In contrast, ectopically expressed ARAF or CRAF did not interact with SPOP (Fig. 1e). Only SPOP, but none of the other CRL3-based BTB domain-containing adaptors examined, interacted with BRAF (Fig. 1f). Moreover, a specific endogenous interaction between SPOP and BRAF was observed in Ishikawa endometrial cancer cells (Fig. 1g, h). In accordance with a previous study showing that the MATH domain of SPOP is responsible for recruiting substrates [9], deletion of the MATH domain, but not the CUL3-binding BTB domain, completely abolished the interaction between SPOP and BRAF (Fig. 1i, j). Taken together, our data indicate that SPOP interacts with BRAF in cells.
SPOP promotes non-degradative ubiquitination of BRAF
Next, we investigated whether BRAF protein stability was controlled by the ubiquitin–proteasome pathway. Treatment of Ishikawa cells with the proteasome inhibitor MG132 had no obvious effect on BRAF protein levels, indicating that BRAF has a relative long half-life, at least in Ishikawa cells. MLN4924, a small-molecule inhibitor of the NEDD8-activating enzyme required for the activation of the CRL complex, did not alter BRAF protein levels. In contrast, MG132 or MLN4924 treatment caused a marked increase in the protein levels of BRD4 and Caprin1, which are two reported substrates of SPOP (Fig. 2a). Ectopic overexpression of SPOP-WT or its domain deletion mutants did not alter the levels of co-expressed BRAF (Fig. 2b). To characterize the effect of SPOP on endogenous BRAF, we generated Tet-on-inducible Ishikawa cells. Induction of FLAG-SPOP by doxycycline treatment did not affect BRAF protein levels, whereas BRD4 was destabilized in a time-dependent manner (Fig. 2c). Depletion of SPOP by shRNA-mediated knockdown or CRISPR/Cas9-mediated KO in EC cell lines elevated the protein levels of BRD4 and Caprin1, but not BRAF (Fig. 2d, Additional file 1: Fig. S1a–c). Moreover, the half-life of BRAF was comparable between parental and SPOP KO cells (Fig. 2e, f). Although BRAF was not degraded by SPOP, it was robustly polyubiquitinated in response to the co-expression of SPOP-WT but not by the SPOP-ΔBTB or -ΔMATH mutant (Fig. 2g). We further demonstrated that the SPOP–CUL3–RBX1 E3 ubiquitin ligase complex catalyzed BRAF polyubiquitination in vitro (Additional file 1: Fig. S1d). Accordingly, SPOP depletion decreased the ubiquitination levels of endogenous BRAF (Fig. 2h). These results indicate that SPOP induces non-degradative polyubiquitination of BRAF.
Given that SPOP-mediated BRAF ubiquitination is non-degradative, we examined the polyUb chain-linkage specificity of BRAF. We performed in vivo ubiquitination assays using a panel of ubiquitin mutants containing single KR mutations at each of the seven lysine residues. Co-expression of either Ub-KR mutant did not obviously alter SPOP-mediated BRAF ubiquitination. We also used a reciprocal series of Ub-KO mutants that contain only one lysine residue, with the other six lysine residues mutated to arginine. The expression of either Ub-KO mutant nearly abolished SPOP-mediated BRAF ubiquitination (Fig. 2i), indicating that SPOP may catalyze the synthesis of mixed-linkage polyUb chains on BRAF. We also utilized linkage-specific K27/K29/48/63-Ub antibodies to demonstrate that ubiquitinated BRAF contains K27-, K29-, K48-, and K63-Ub linkages (Additional file 1: Fig. S1e). Taken together, our data indicate that SPOP promotes non-degradative ubiquitination of BRAF.
The SBC motif in BRAF is required for SPOP-mediated BRAF ubiquitination
To examine whether the potential SBC motif is required for the SPOP-BRAF interaction, we generated a BRAF mutant in which the SBC motif was deleted. Co-IP assay results showed that SPOP only interacted with wild-type BRAF but did not bind to the BRAF-ΔSBC mutant (Fig. 3a, Additional file 1: Fig. S2a). The deletion of the SBC motif in BRAF abrogated SPOP-mediated BRAF ubiquitination (Fig. 3b). Mutations in amino acids in the SBC motif considerably reduced the SPOP-BRAF interaction (Fig. 3c) and SPOP-mediated BRAF ubiquitination (Fig. 3d).
Given that BRAF interacts with SPOP in an SBC motif-dependent manner, we further explored whether cancer-associated BRAF mutations that occur in the SBC motif would attenuate the SPOP-BRAF interaction and permit BRAF to evade SPOP-mediated ubiquitination. To this end, we examined the cancer sequencing data deposited in cancer gene mutation databases (cBioPortal and COSMIC). We identified a BRAF-T121I mutation in glioblastoma and BRAF-S122Y mutation in osteosarcoma (Fig. 1b, Additional file 1: Table. S3). We found that the SPOP-binding capacity of these BRAF mutants was markedly reduced (Fig. 3e, Additional file 1: Fig. S2b), and SPOP-mediated ubiquitination of these mutants was also markedly attenuated (Fig. 3f). Taken together, our data indicate that cancer-derived BRAF mutations occurring in the SBC motif allow BRAF to evade SPOP-mediated ubiquitination.
EC- and PCa-associated SPOP mutants are defective in promoting BRAF ubiquitination
To date, the vast majority of SPOP mutations associated with EC or PCa occur within the MATH domain, which is responsible for selective substrate binding (Fig. 4a, b) [10, 11]. We postulated that EC- or PCa-associated SPOP mutants may be defective in mediating BRAF ubiquitination. The BRAF-binding ability of EC-associated SPOP mutants was moderately or severely impaired as compared to that of SPOP-WT (Fig. 4c). SPOP-mediated BRAF ubiquitination was moderately or severely attenuated in these mutants (Fig. 4d). Similar effects were observed when PCa-associated SPOP mutants were tested (Fig. 4e, f). In accordance with a previous study showing that cancer-associated SPOP mutants function as dominant-negative variants to deregulate their substrates [12], we found that co-expression of the SPOP mutant markedly reduced the interaction between SPOP-WT and BRAF (Fig. 4g), resulting in the suppression of wild-type SPOP-mediated BRAF ubiquitination (Fig. 4h). Taken together, our data indicate that EC or PCa-associated SPOP mutants exert dominant-negative effects by downregulating BRAF ubiquitination due to impaired BRAF binding capacity.
Cytoplasmic but not nuclear SPOP promotes BRAF ubiquitination
BRAF is a cytoplasmic protein kinase. Our previous study showed that SPOP is localized exclusively in the nucleus as speckles or in both the cytoplasm and nucleus [19]. We investigated the subcellular localization of the SPOP-BRAF interaction. BRAF was diffusely localized in the cytoplasm. When SPOP was localized in both the cytoplasm and nucleus, BRAF was recruited into SPOP speckles in the cytoplasm. When SPOP was localized exclusively in the nucleus, it did not co-localize with cytoplasmic BRAF. We found that SPOP lacking the NLS sequence (SPOP-ΔNLS) accumulated exclusively in the cytoplasm in punctate patterns and colocalized perfectly with cytoplasmic BRAF. Moreover, the deletion of the SBC motif in BRAF did not alter its diffuse cytoplasmic localization, and the SPOP-induced speckle pattern of BRAF was not observed (Fig. 5a).
Next, we investigated the effect of cancer-associated SPOP mutations on BRAF localization. We focused on three hotspot mutants (E47K, S80R, and W131G). Three cancer-associated SPOP mutants lacking the NLS sequence (ΔNLS-E47K, -S80R, and -W131G) accumulated exclusively in the cytoplasm in punctate patterns similar to those of SPOP-ΔNLS, but these mutants did not colocalize with BRAF. Similarly, the cancer-associated BRAF-T121I mutant did not co-localize with SPOP (Fig. 5b). Our previous studies showed that the SPOP-ΔNLS mutant was more effective in promoting the ubiquitination of its cytoplasmic substrates (INF2, MyD88, Caprin1, and HIPK2) than SPOP-WT, but it was unable to ubiquitinate nuclear substrates [19,20,21,22]. Indeed, we found that SPOP-ΔNLS immunoprecipitated more BRAF than SPOP-WT (Fig. 5c) and SPOP-ΔNLS was more effective in promoting BRAF ubiquitination than SPOP-WT (Fig. 5d). Taken together, our data indicate that SPOP can recruit BRAF into cytoplasmic speckles and that this activity strictly depends on its cytoplasmic localization.
SPOP suppresses MAPK/ERK pathway activation
BRAF plays a critical role in the regulation of MAPK/ERK pathway activation [3]. We examined MAPK/ERK activity in SPOP KO and SPOP overexpressed cells. The induction of FLAG-SPOP by doxycycline treatment led to a marked decrease in the phosphorylation levels of MEK1/2, ERK1/2, and MSK1, indicating that MAPK/ERK activation was attenuated (Fig. 6a). Conversely, stronger MAPK/ERK activation was observed when SPOP was ablated (Fig. 6b). Moreover, SPOP-ΔNLS overexpression attenuated the MAPK/ERK pathway activation in EGF-stimulated cells (Fig. 6c, d). Conversely, SPOP KO enhanced MAPK/ERK pathway activation in EGF-stimulated cells (Fig. 6e, f). Stronger MAPK/ERK activation was observed in SPOP KO Ishikawa cells reconstituted with the SPOP-S80R mutant than in those reconstituted with the SPOP-Y87C mutant, which retains a similar binding capacity to BRAF as SPOP-WT (Additional file 1: Fig. S3a, b).
To determine whether the escape of SPOP-mediated-BRAF ubiquitination has any impact on MAPK/ERK pathway activation, we reconstituted BRAF-KO Ishikawa cells (Additional file 1: Fig. S3c–e) with BRAF-WT or the -T121I mutant, which is incapable of binding to SPOP. As shown in Fig. 6g, h, EGF-induced MAPK/ERK activation was stronger in BRAF-KO Ishikawa cells reconstituted with the BRAF-T121 mutant than in those reconstituted with BRAF-WT. We found that stable overexpression of SPOP-WT or ΔNLS mutant reduced the interaction between BRAF and its upstream regulator, H-RAS, and downstream effectors MEK1/2 and ERK1/2 (Fig. 6i). Conversely, stable overexpression of SPOP-S80R or -F125V, increased the interactions between BRAF and MAPK/ERK pathway components. Surprisingly, stable overexpression of the SPOP-Y87C mutant even moderately reduced the interactions between BRAF and MAPK/ERK pathway components (Fig. 6j). Moreover, the interactions between BRAF and MAPK/ERK pathway components were moderately increased in BRAF-KO Ishikawa cells reconstituted with the BRAF-T121 mutant as compared to those reconstituted with BRAF-WT (Fig. 6k). Taken together, our data indicate that SPOP suppresses MAPK/ERK pathway activation.
BRAF-T121I mutant is more effective to promote cell growth, migration, and invasion than BRAF-WT
We found that stable overexpression of SPOP-WT reduced the growth of Ishikawa, KLE, and SPEC-2 cells as compared to parental cells. In contrast, stable overexpression of EC-associated SPOP mutants increased the growth of KLE, SPEC-2, and Ishikawa cells as compared to parental cells (Additional file 1: Fig. S4a–d). Consistent with the substantial roles of BRAF in cellular malignancy [5], BRAF KO Ishikawa cells grew at a markedly reduced growth rate, as demonstrated by colony formation and CCK-8 assays. Reconstitution with BRAF-WT restored BRAF-KO cell growth in vitro, and more importantly, BRAF-KO cells reconstituted with the BRAF-T121I mutant showed a higher growth rate than those reconstituted with BRAF-WT (Fig. 7a, b). Similarly, Transwell assay results showed that BRAF-T121I mutant expression resulted in increased cell migration and invasion compared to BRAF-WT (Fig. 7c, d). Moreover, the 3D sphere formation assay results showed that the number and size of spheres in BRAF KO cells reconstituted with the BRAF-T121I mutant were markedly increased compared to those reconstituted with BRAF-WT, suggesting that the cancer-derived BRAF mutant increased anchorage-independent cell growth (Fig. 7e, f). Taken together, our data indicate that loss of SPOP-mediated BRAF ubiquitination promotes cell malignancy.