K27-linked ubiquitination of BRAF by ITCH engages cytokine response to maintain MEK-ERK signaling

BRAF plays an indispensable role in activating the MEK/ERK pathway to drive tumorigenesis. Receptor tyrosine kinase and RAS-mediated BRAF activation have been extensively characterized, however, it remains undefined how BRAF function is fine-tuned by stimuli other than growth factors. Here, we report that in response to proinflammatory cytokines, BRAF is subjected to lysine 27-linked poly-ubiquitination in melanoma cells by the ITCH ubiquitin E3 ligase. Lysine 27-linked ubiquitination of BRAF recruits PP2A to antagonize the S365 phosphorylation and disrupts the inhibitory interaction with 14–3–3, leading to sustained BRAF activation and subsequent elevation of the MEK/ERK signaling. Physiologically, proinflammatory cytokines activate ITCH to maintain BRAF activity and to promote proliferation and invasion of melanoma cells, whereas the ubiquitination-deficient BRAF mutant displays compromised kinase activity and reduced tumorigenicity. Collectively, our study reveals a pivotal role for ITCH-mediated BRAF ubiquitination in coordinating the signals between cytokines and the MAPK pathway activation in melanoma cells.


Reviewer #1, Expertise: E3 ligase, ubiquitination, immune regulation (Remarks to the Author):
In this study, the authors show that ITCH-mediated non-degradable ubiquitination of BRAF via K27linked poly-ubiquitin chain in response to pro-inflammatory cytokines from tumor microenvironment in melanoma. K27-linked ubiquitination of BRAF controls ability to bind with inhibitory 14-3-3, leading to activate MEK/ERK signaling and melanoma cell survival. This study provides scientific insights for atypical ubiquitin mediated post translational modification of BRAF by ITCH in the regulation of melanoma survival. Although the data are in general convincing, one major concern is that whether such observations will have biological support, namely, whether it is relevant from the point view of Itch-deficient mice. Although the authors used Itch-/-MEFs, how this would be related to BRAFcontrolled transformation in melanoma remains unclear. Since Itch-promoted K27-chain formation is already known, the significance of findings in this paper is quite limited at the current stage.
Response: We thank the reviewer for recognizing the novelty and the potentially impact of this study. We also agree with the reviewer that it is important to provide further experimental evidence to demonstrate that ITCH-mediated atypical ubiquitination of BRAF is an important mechanism to modulate the MAPK pathway, particularly in the setting of cytokine/TNFα treatment. Enlightened by the constructive comments from the reviewer, we have obtained the following results to substantiate the physiological importance of our findings. 1) In support of a positive role of ITCH in activating the RAF/MEK/ERK signaling cascade, we found that ectopic expression of WT-ITCH or a constitutive active form of ITCH (3D-ITCH) in murine melanocyte melan-a led to elevated MEK and ERK activities (Fig. 4m). The 3D-ITCH mutant (S199D, T222D and S232D) was generated to mimic the three JNK phosphorylation sites that serve to activate ITCH (Chang L et al. (2006). Cell. 124,[601][602][603][604][605][606][607][608][609][610][611][612][613]. 2) More importantly, overexpression of ITCH supported TPA-independent growth of melan-a cells ( Fig. 4n-o). Since proliferation of melan-a cells in vitro require growth factors or mitogens such as TPA (12-O-Tetradecanoylphorbol-13-acetate), TPA-independent growth has been used in multiple studies as an evidence for melanocyte transformation (Garraway LA et al. (2005). Nature. 436, 117-122;Wan L et al. (2017). Cancer Discovery. 7,[424][425][426][427][428][429][430][431][432][433][434][435][436][437][438][439][440][441]. Moreover, although expression of 3D-ITCH alone failed to promote anchorage-independent growth of melan-a cells ( Fig. 4p-q), when Pten was further depleted (Supplementary Fig. 7m-o), 3D-ITCH-expressing melan-a cells formed colonies in the soft agar ( Fig. 4p-q). These results together suggest an oncogenic function of ITCH in transforming melanocytes, which is partly through activating the RAF/MEK/ERK signaling pathway. 3) In addition to MAPK activation, increased MITF expression was also observed in ITCHoverexpressed melanocytes (Fig. 4m). The upregulated MITF level may be due to decreased c-Jun expression as previous reports demonstrated that MITF is negatively regulated by c-Jun in melanoma cells (Riesenberg S et al. (2015). Nature Communications. 6, 8755;Smith MP et al. (2014). Cancer Discovery. 4, 1214-1229. In further support of this notion, depletion of ITCH in melanoma cells resulted in a decrease of MITF ( Fig. 3c-d). MITF has been shown to promote melanocyte and melanoma proliferation (Goding CR (2000). Genes andDevelopment. 14, 1712-1728;Garraway LA et al. (2005). Nature. 436, 117-122), the positive regulation of MITF by ITCH in melanoma cells may function as another mechanism through which ITCH facilitates melanocyte transformation. 4) To further interrogate an oncogenic role for ITCH in promoting melanoma cell survival, we generated a WM1346 melanoma cell line expressing a doxycycline-inducible shITCH construct (backbone EZ-Tet-pLKO-Hygro from Addgene). We found a marked decrease of pMEK and pERK levels as well as cell growth after doxycycline treatment ( Fig. 3e and 3j). Furthermore, doxycycline treatment in vivo significantly suppressed the growth of Teton-shITCH-WM1346 melanoma cell in a xenograft mouse model ( Fig. 3l-n). 5) Different from the Itch a18H/a18H mouse, which is viable (Perry WL et al. (1998). Nature Genetics 18, 143-146;Fang D et al. (2002). Nature Immunology. 3,[281][282][283][284][285][286][287], germline deletion of Braf tended to be embryonic lethal due to vascular defects (Wojnowski L et al. (1997). Nature Genetics 16,[293][294][295][296][297]. We compared the cell viability and growth of melan-a cells with shRNAs targeting Itch or Braf, and found that melan-a cells were quite tolerate to either Itch or Braf deficiency (Supplementary Fig. 5e-g). Moreover, depletion of Braf resulted in a slightly enhanced growth inhibition compared to the Itch knockdown ( Supplementary Fig. 5e-g). On the contrary, in melanoma cells, depletion of either ITCH or BRAF resulted in a massive cell growth arrest and subsequent cell death ( Fig. 3i-j and Supplementary Fig. 4j-p). These findings support the notion that melanoma cells are more addicted to the RAS/RAF/MEK/ERK signaling cascade to support cell survival. The molecular modules including ITCH that fuels MAPK hyperactivation are therefore vital to melanoma cells. However, in normal cells where minimal MAPK signaling strength is required, ITCH-mediated BRAF activation might be relatively dispensable.
We agree with the reviewer that an Itch-deficient melanomagenesis mouse model will be critical to further assess the pathophysiological role of ITCH in melanoma development. However, due to the large volume of data associated with this manuscript, we hope the reviewer could concur with us that these studies would be more suitable for a separate manuscript in the near future.

Response:
We apologize for this mistake; the correction has been made in the revised manuscript.

ITCH activity and ITCH-mediated K27 ubiquitination of BRAF should be compared in normal skin cells and melanoma cells to validate physiological importance of ITCH-mediated K27 ubiquitination of BRAF for survival and invasion of melanoma.
Response: We agree with the reviewer that it is critical to compare the ITCH-mediated K27 ubiquitination of BRAF in both normal melanocytes and melanoma cells. As kindly instructed, in the Supplementary Fig. 6e, we found that TNFα stimulated K27-linked ubiquitination of BRAF in both melan-a and WM1346 cells. Compared to melan-a, TNFα−stimulated BRAF ubiquitination appeared slightly stronger in WM1346 cells. Moreover, depleting ITCH in both cells led to a marked reduction of BRAF ubiquitination upon TNFα treatment (Supplementary Fig. 6i-j), suggesting that in both normal melanocytes and melanoma cells, ITCH promotes BRAF ubiquitination in response to cytokine stimulation. As we have addressed above, compared to melanoma cells ( Fig. 3i-j), depletion of Itch in melanocytes only moderately suppressed melan-a cells proliferation (Supplementary Fig. 5e-g). These results indicate that although ITCH-mediated BRAF K27-ubiquitination occurs in both cell types, due to the oncogenic addiction of melanoma cells to the MAPK signaling, the loss of ITCH, which significantly abolishes MEK and ERK activity in melanoma cells, led to a much stronger inhibition of cell proliferation in melanoma cells ( Fig. 3i-n). These results thus suggest a therapeutic window for targeting ITCH in melanoma patients, a future direction warrants in-depth investigation. Fig 3b,

Response:
We thank the reviewer for pointing out that in the original Fig. 3b, 20 min after TNFα treatment, Itch protein level was reduced. We also agree with the reviewer that it is important to examine if TNFα treatment affects ITCH stability/activity in both MEFs and melanoma cells. As kindly suggested, in the revised manuscript, we have obtained the following experimental evidence to address this concern: 1) ITCH stability is mainly controlled by its auto-ubiquitination and subsequent degradation. We found that compared to WT-ITCH, which exhibited an approximate four-hour half-life ( Supplementary Fig. 5a-b), the catalytically inactive CS-ITCH was completed stabilized (Supplementary Fig. 5b). In support of this notion, c-Jun was significantly stabilized in CS-ITCH-expressing cells as evidenced by an extended half-life (Supplementary Fig. 5b). 2) TNFα treatment didn't significantly destabilize ITCH as we found in melanoma cells, 240 min after TNFα or IL-1β treatment, the ITCH protein level remained largely unchanged ( Fig. 4a-b and Supplementary Fig. 6a-d). We also rerun the lysates of original Fig. 3b, and found that the previous reduction of ITCH at 20 and 30 min was likely due to the reblotting issue. In the revised Fig. 3b, the ITCH protein level remains unchanged at the 20 and 30 min time points. 3) Since in our experimental conditions such as in Fig. 3b and Fig. 4a-b, newly synthesized ITCH may complement the loss of ITCH caused by proteolysis. Following the reviewer's insightful comment, we compared the half-lives of the ITCH protein in melanoma cells upon TNFα treatment. As shown in Supplementary Fig. 5c, TNFα appears to destabilize ITCH likely through accelerating its auto-ubiquitination since the CS-ITCH remained stable even in the presence of TNFα. 4) Furthermore, in support of a non-proteolytic function of ITCH-mediated BRAF ubiquitination, BRAF stability was not affected in the experimental conditions we examined ( Supplementary  Fig. 5a-c). Taken together, our newly obtained results suggest that TNFα activates ITCH and promotes its autoubiquitination and degradation, while at the same time promotes ITCH-mediated ubiquitination of substrates including BRAF and c-Jun ( Supplementary Fig. 7c), which in turn activates BRAF and destabilizes c-Jun, respectively. The simultaneous activating an ubiquitin E3 ligase and destabilizing it via auto-ubiquitination has been observed for a number of E3 ligases including ITCH and other NEDD4 family E3 ligases (Gallagher E et al. (2006). Proc Natl Acad Sci U S A. 103, 1717-1722Chen Z et al. (2017). Mol Cell. 66, 342-345).

In Fig 4a and 4b, the authors showed the effect of TNFa on MEK/ERK pathway activation accompanied with the activation of JNK. The protein expression of ITCH and direct evidence of ITCH activity (phosphorylated ITCH, c-FLIPâ€¦) should be tested to support the coordination between BRAF/MEK/ERK and JNK/ITCH signaling in response to TNFa.
Response: We thank the reviewer for this brilliant suggestion to examine ITCH phosphorylation and subsequent activation after TNFα treatment. As kindly instructed, we found that TNFα treatment led to a marked increase of ITCH serine/threonine phosphorylation using the p-(S/T)P antibody against immunopurified ITCH protein (Supplementary Fig. 7a). Furthermore, as described in the response to comment #3, we found that both ITCH and its ubiquitin substrate c-Jun were destabilized after TNFα treatment as evidenced by shortened half-lives, whereas in cells expressing catalytic inactive CS-ITCH, both ITCH and c-Jun were stabilized (Supplementary Fig. 5b-c). Our newly obtained results thus support previous reports that JNK phosphorylation relieves the auto-inhibitory conformation of the ITCH protein (Gallagher E et al. (2006). Proc Natl Acad Sci U S A. 103, 1717-1722), which in turn promotes the poly-ubiquitination of its downstream targets including proteolytic substrates c-Jun, JunB and c-FLIP as well as non-proteolytic substrates like BRAF.

In Fig 4c, to clarify whether TNFa induced ubiquitination of BRAF is ITCH-dependent, ubiquitination assay should be conducted in ITCH +/+ and ITCH -/-cells with or without TNFa treatment.
Response: As kindly suggested by the reviewer, in Fig. 4e, we demonstrated that there was a higher BRAF poly-ubiquitination in Itch +/+ MEFs compared to Itch -/-MEFs. Further, TNFα treatment promoted BRAF poly-ubiquitination only in Itch +/+ MEFs but not its null counterparts. Further, TNFα was not capable of promoting BRAF ubiquitination in ITCH-depleted melan-a cells or WM1346 melanoma cells (Supplementary Fig. 6i-j). In concert with the results from MEFs (Fig. 3b), we found that MEK/ERK activation is largely refractory to TNFα treatment in melanoma cells depleted of ITCH ( Fig. 4d and Supplementary Fig. 6p). These results again support the correlation between TNFαmediated JNK/ITCH activation and subsequent BRAF/MEK/ERK activation. Fig 3 to 6, although melanoma cell lines were treated with TNFa or co-cultured with M2 differentiated THP1, these experiments are not enough to use terms of â€˜cytokine responseâ€™. The authors need to identify cytokines from M2 THP1 macrophages or tumor microenvironments and to provide additional data using identified cytokines as well as TNFa.

Response:
We agree with the reviewer that it is not accurate to use the term "cytokine response" in our original manuscript given our results only used TNFα to stimulate the JNK/ITCH pathway in our experimental settings. Previous studies from the Wellbrock group (Smith MP et al. (2014). Cancer Discovery. 4, 1214-1229 demonstrated that TNFα from melanoma tumor-associated macrophages facilitates tumor growth and contributes to BRAF inhibition resistance. The authors revealed that TNFα is highly enriched in M1-and M2-THP1-conditioned media, and that TNFα treatment stimulated the proliferation of melanoma cell line WM266.4. To demonstrate that TNFα is also a key factor to promote melanoma cell growth in our experimental settings, we adopted the approach used in the Smith et al. paper by using TNFα-blocking antibody in our M2 THP1-melanoma cell co-culture experiment. As shown in Supplementary Fig. 7k-l, we found that the antibody against TNFα suppressed THP1stimulated WM3918 melanoma cell growth while had no effect on the proliferation of WM3918 cells in the absence of M2 THP1. This finding also suggests that TNFα is the major factor from M2-THP1 to facilitate melanoma cell growth.
The major conclusion of our manuscript is that the JNK/ITCH signaling pathway facilitates MEK/ERK activation via atypical ubiquitination of BRAF. We fully agree with the reviewer that in addition to TNFα, there might be other cytokines that could also activate the JNK/ITCH pathway to fuel the MEK/ERK signaling in melanoma cells. For example, CAFs (cancer-associated fibroblasts) exhibit tumor-promoting function through secreting pro-inflammatory cytokines (Erez N et al. (2010). Cancer Cell. 17, 135-147). Among CAFs-derived cytokines, IL-1β promotes melanoma invasion and angiogenesis (Voronov E et al. (2003). Proc Natl Acad Sci U S A. 100, 2645-2650). Enlightened from the reviewer's constructive comment, we found that analogous to TNFα, IL-1β activated JNK and stimulated pMEK and pERK in melanoma cells (Supplementary Fig. 6c-d). Furthermore, IL-1β promoted the growth of melanoma cells in clonogenic survival experiments (Supplementary Fig. 7i-j). These results coherently support the notion that the JNK/ITCH signaling pathway engages upstream stimulations including TNFα and IL-1β to promote BRAF poly-ubiquitination and subsequent MEK/ERK activation (Supplementary Fig. 11m).

Page11, line 246 â€oeCytokine Recruitsâ €¦..â€ should be â€oeTNFa recruitâ€¦.â€.
Response: We fully agree with reviewer that it is not accurate to use "Cytokine" in the original manuscript given that only TNFα was used in our experiments. In the revised manuscript, following the constructive comments from the reviewer, experimental results using IL-1β were included to demonstrate that similar to TNFα, IL-1β activated the JNK/ITCH module to accelerate BRAF polyubiquitination (Supplementary Fig. 6h) and to stimulate BRAF/MEK/ERK in melanoma cells (Supplementary Fig. 6c-d), which in turn facilitated proliferation of melanoma cells ( Supplementary  Fig. 7i-j). Therefore, we hope the reviewer would concur with us that it is now appropriate to use the term "Cytokine" in the subtitle for Fig. 4. 8. In Fig6, the data are not enough to insist the notion that PPP2R2A has ability to dephosphorylate not p-S729 but p-S365. Additional experiments to measure dephosphorylate kinetics of PPP2R2A are needed.

Response:
We completely agree with the reviewer that additional experiments to support a role of PPP2R2A in recruiting the PP2A complex to the ubiquitination BRAF are required. To this end, in the revised manuscript, we have obtained the following experimental results to further demonstrate that PPP2R2A facilitates PP2A-mediated dephosphorylation of BRAF p-S365: 1) Depletion of the regulatory subunit of PP2A, PPP2R2A, led to an increase of pS365-BRAF, but not pS729-BRAF ( Fig. 6l and Supplementary Fig. 9m). Both PPP2CA and PPP2R2A knockdown led to reduced pMEK and pERK levels ( Fig. 6k-l and Supplementary Fig. 9m).
6 3) As suggested by the reviewer, we found that compared to the PPP2CA immunoprecipitates, the PP2A complex containing PPP2CA, PPP2R2A and PPP2R1A exhibited a higher activity in dephosphorylating pS365-BRAF but not pS729-BRAF (Fig. 6h-i and Supplementary Fig. 9l). 4) Further kinetic analyses revealed that the Michaelis-Menten kinetics of dephosphorylating BRAF proteins for PPP2CA immunoprecipitates was V max =4.21±0.63 pmol/min/μg; K m =0.185±0.14 μM, while for the PP2A complex containing PPP2CA, PPP2R2A and PPP2R1A was V max =10.33±1.31 pmol/min/μg; K m =0.74±0.39 μM (Fig. 6i). This result demonstrates that compared to the PPP2CA catalytic subunit alone, the PP2A complex containing the PPP2R2A subunit exhibited a higher phosphatase activity in dephosphorylating BRAF. 5) More importantly, compared to PPP2CA immunoprecipitates, ubiquitinated BRAF appeared to be a better substrate for the PP2A complex containing PPP2CA, PPP2R2A and PPP2R1A (Fig.  6j), indicating that PPP2R2A binds to the K27-linked poly-ubiquitin chain and recruits the PP2A complex to accelerate pS365-BRAF dephosphorylation (Fig. 6e). 6) In addition, a recent proteomic study revealed PPP2R2A as a BRAF interacting protein in cells (Diedrich B et al. (2017). The EMBO Journal. 36,[646][647][648][649][650][651][652][653][654][655][656][657][658][659][660][661][662][663]. Taken together, our newly obtained results further demonstrate a crucial role for the WD40 repeats domain containing subunit PPP2R2A in promoting pS365-BRAF dephosphorylation by the PP2A complex, a mechanism that leads to BRAF activation upon its K27-linked ubiquitination by ITCH. It will be important to further assess if other B55 subunits, which also contain the WD40 repeats domain, function similarly as PPP2R2A, and whether B56 and other regulatory subunits could bind to ubiquitinated BRAF and dephosphorylate pS365-BRAF. However, we hope the reviewer could concur with us that these studies are outside of the main scope of current manuscript and will be more suitable for a separate manuscript in the near future.

Because there are important differences in phenotype and physiological responses between immortalized tumor cell line and primary tumor cells from the patient, the authors need to conduct the key experiments using primary melanoma cells.
Response: We thank the reviewer for this constructive suggestion to examine if ITCH-mediated BRAF ubiquitination promotes MEK/ERK activation and melanoma proliferation. However, due to technical difficulty of propagating primary melanoma cells, especially those harboring WT-BRAF, for efficient in vitro shRNA knockdown experiments, we chose to use early passage (<20 passages) melanoma cell lines M245 and IPC-298 derived from melanoma patients obtained from Dr. Keiran Smalley lab for our experiments. These two cell lines both express WT-BRAF and mutant NRAS (Supplementary Fig. 4de). As shown in Supplementary Fig. 4d-e, depletion of ITCH in both M245 and IPC-298 cells led to reduced pMEK/pERK as we observed in other BRAF WT melanoma cell lines. Notably, these cells displayed poor survival after ITCH knockdown as evidenced by clonogenic survival assays (Supplementary Fig. 4n). Further, TNFα treatment facilitates BRAF ubiquitination, MEK/ERK activation and cell proliferation in both cells (Supplementary Fig. 6a-b, 6g and 7g-h). These results, together with our previous results using WM1346, WM3918 melanoma cell lines, normal melanocytes and MEFs coherently support our conclusion that ITCH-mediated BRAF ubiquitination activates BRAF activity to promote melanoma cell proliferation and invasion.

The manuscript by Yin et al presents data implicating K27-linked ubiquitination of wild-type BRAF as a mechanism that modulates BRAF (and presumably also CRAF and ARAF )-mediated ERK pathway activation. The authors use various biochemical and cell based assays to support a model by which ubiquitination of BRAF is mediated by the ITCH E3 ligase via K27, leading to release of inhibitory interaction with 14-3-3 and activation of MEK/ERK signaling in the context of cytokine stimulation. The study presents interesting findings with potentially important implications, but it is somewhat uneven.
Part of the manuscript shows convincing evidence of BRAF ubiquitination, the role of ITCH and the effect of ubiquitination on BRAF's ability to activate ERK in cells. However, the molecular details of the proposed mechanism of BRAF activation by ubiquitination are less well supported by the data, and certain conclusions appear somewhat self-contradictory.

Response:
We thank the reviewer for recognizing the novelty of this study. We also agree with the reviewer that it is crucial to provide further experimental evidence to support the molecular mechanisms by which the kinase activity of poly-ubiquitinated BRAF is promoted. We hope after reading our pointby-point responses below as well as our revised manuscript, the reviewer would concur with us that we have obtained convincing results to demonstrate that our newly discovered molecular circuit serves to activate the RAF/MEK/ERK signaling cascade and contributed to melanoma cell survival.

. In ex vivo assays for ubiquitination (co-IPs), the authors use proteasome inhibitor MG132 for 12 hours. Since, as they mention, K27-linked ubiquitination is non-degradative would the results be similar without proteasome inhibition?
Response: We thank the reviewer for raising the concern regarding the use of MG132 in our ubiquitination assays. The reason that we use MG132 across the panels in Fig. 1 and Fig. 2 is to keep the experimental condition consistent for different ubiquitin mutants. As the reviewer kindly pointed out, wild type (WT) ubiquitin and the ubiquitin mutants that have K48 and K11 available could be utilized to assemble K48-or K11-linked poly-ubiquitin chains, which could be further processed by the 26S proteasome for substrate protein degradation. In our figure panels using WT-ubiquitin and ubiquitin mutants with K48 or K11 available, we added MG132 to the cells to ensure no K48/K11 linkages were disposed of from our ubiquitination assay results.
As kindly suggested by the reviewer, we performed the ubiquitination assay in Fig. 1g without MG132, as the reviewer could find out in the Supplementary Fig. 1f, the results are largely identical to the results we obtained from the condition with MG132.

Fig. 3. The authors' suggestion that the effect of ubiquitination on BRAF's ability to activate ERK is through the JNK/ITCH axis is based mostly on indirect evidence. Only TNF? is used to stimulate ERK activity in MEFs. Have they tried to stimulate ERK with PDGF, a known activator of ERK signaling in MEFs? Would the effect of PDGF on ERK also be hampered in Itch-/-MEFs? In such case, alternative mechanisms to JNK should be considered. For example, is the interaction of BRAF with MEK in cells affected by ubiquitination?
Response: We thank the reviewer for the brilliant suggestion to further determine if in addition to TNFα, other cytokines or growth factors also activate BRAF via the JNK/ITCH module. In the original Supplementary Fig. 3a (revised Supplementary Fig. 4a), EGF was used to stimulate Itch +/+ and Itch -/-MEFs and we found that EGF activated MEK and ERK regardless of Itch genetic status. However, compared to Itch -/-MEFs, Itch +/+ MEFs displayed a stronger and sustainable response to EGF (Supplementary Fig. 4a), indicating that although not exclusively, ITCH also participated in EGFtriggered MEK/ERK activation. Moreover, as kindly instructed by the reviewer, we compared the pMEK/pERK response of these MEFs to PDGF. As shown in Supplementary Fig. 4b, quite similar to EGF treatment, PDGF activated MEK and ERK in both Itch +/+ and Itch -/-MEFs. On the other hand, for TNFα treatment, the MEK/ERK activation was almost completely abolished in Itch -/-MEFs (Fig. 3b), suggesting an indispensable role of ITCH in mediating this process. From our results, we found that growth factors triggered a robust but transient activation of MEK/ERK, whereas cytokine-induced MEK/ERK was rather moderate but more sustained. Furthermore, depletion of ITCH in 293T cells (Supplementary Fig. 1h), melanocytes melan-a (Supplementary Fig. 6i) and WM1346 melanoma cells (Supplementary Fig. 6j) abrogated endogenous BRAF ubiquitination, again support a key role of ITCH in mediating TNFα-induced BRAF activation.
Together, our results indicate that growth factors and TNFα utilize different signaling cascades to activate the RAF/MEK/ERK pathway. Our findings in this manuscript advocate a role for JNK/ITCH pathway to stimulate RAF proteins via ITCH-mediated poly-ubiquitination of RAF proteins. However, compared to the canonical growth factor/receptor tyrosine kinase (RTK)/RAS/RAF/MEK/ERK signaling cascade, the MEK/ERK activation is quite moderate after cytokine stimulation, which indicates that the JNK/ITCH axis is not the predominant signal to activate MEK/ERK in normal culturing conditions. However, for melanoma cells that expose to proinflammatory environment, a scenario that could be found in most solid tumors (Lippitz BE (2013). The Lancet Oncology. 14, e218-e228), the cytokine-mediated activation of MEK/ERK may provide a protection against cytokineinduced cytotoxicity. Previous studies have suggested a number of mechanisms to explain cytokinetriggered MEK/ERK activation (Dumitru CD et al. (2000). Cell. 103, 1071Cell. 103, -1083Marques-Fernandez F et al. (2013). Cell Death and Disease. 4, e493;Shao Y et al. (2015). Journal of Investigative Dermatology. 135, 1839-1848. Although the increase of c-FLIP transcription upon NF-κB activation has been shown to regulate CRAF function in neuronal cells, the mechanism that leads to the immediate activation of ERK upon cytokine stimulation remains unclear, which was one the reasons directed us to our current findings.
Enlightened by the comments from both reviewers, in addition to TNFα, we also treated melanoma cells with IL-1β, another proinflammatory cytokine found in tumor microenvironment. As shown in the revised manuscript, IL-1β treatment also led to the increase BRAF ubiquitination and activation of MEK/ERK (Supplementary Fig. 6c-d and 6h). Furthermore, similar to TNFα treatment, cells with the addition of IL-1β displayed enhanced proliferation (Supplementary Fig. 7i-j).
We agree with reviewer that it is critical to examine if BRAF K27-linked poly-ubiquitination also influence its interaction with MEK, its direct downstream target. As shown in Supplementary Fig.  9a, in vitro ubiquitination of BRAF didn't affect its binding to MEK1. Moreover, TNFα treatment, which promoted BRAF ubiquitination in cells, failed to alter its interaction with MEK ( Fig. 6c and Supplementary Fig. 9g). We have tested the binding of ubiquitinated BRAF with a number of its known upstream and downstream interacting proteins. Among all the proteins examined, only 14-3-3 but not MEK1, NRAS, BRAF, CRAF or KSR1, displayed reduced binding with ubiquitinated BRAF both in vitro ( Fig. 6a and Supplementary Fig. 9a-e) and in cells ( Fig. 6c and Supplementary Fig. 9g).
3. Fig 4. As in previous comment, only TNF? is used in RAS-mutant cell lines. What would be the effect of growth fatcors, such as EGF for example, on BRAF ubiquitination?
Response: As kindly suggested, in the revised manuscript, we examined how EGF might affect BRAF ubiquitination and downstream signals in melanoma cells. As shown in Supplementary Fig. 6l, similar to TNFα, EGF treatment promoted BRAF ubiquitination in WM1346 cells. Notably, in melanoma cells, EGF treatment also activated JNK (Supplementary Fig. 6k), which is likely through the EGFR/FAK/MEKK1/JNK signaling circuit (Huang C et al. (2004). Journal of Cell Science. 117,[4619][4620][4621][4622][4623][4624][4625][4626][4627][4628]. To further interrogate whether EGF-induced BRAF ubiquitination is through ITCH, we compared BRAF ubiquitination in WT and Itch -/-MEFs upon EGF treatment, as shown in Supplementary Fig. 6m, BRAF ubiquitination is largely hampered in Itch-deficient cells. These results together suggest that although EGF promotes BRAF ubiquitination through the JNK/ITCH axis, given the direct route from EGFR/RAS/RAF to ERK activation, EGF-dependent BRAF ubiquitination may only play a minor role in tuning BRAF function.
Moreover, enlightened by the constructive comment from the reviewer, we compared TNFα and EGF-stimulated BRAF ubiquitination in WT-BRAF and ubiquitination deficient 5KR-BRAF-expressing melanoma cells. As shown in the revised Supplementary Fig. 10d-e, we found that analogous to ITCH knockdown, 5KR-BRAF-epxressing cells exhibited a diminished response to TNFα− and EGF-mediated BRAF ubiquitination. These results coherently support our conclusion that upstream signals that activate the JNK/ITCH signaling axis promote BRAF ubiquitination and subsequent MEK/ERK activation (Supplementary Fig. 11m).

Fig 5. It is not clear how apparent BRAF activity in cells (as measured by ERK phosphorylation) is
reduced in 5KR but the in vitro kinase activity is not affected. If, as the authors suggest, K-27-linked ubiquitination negatively regulates S365 phosphorylation in BRAF, then the in vitro activity of BRAF should be affected too (see Guan KL et al., JBC, 2000).

Response:
We fully agree with the reviewer that it is puzzling that ubiquitination-deficient 5KR-BRAF displayed compromised activity in cells (Fig. 5a) while it exhibited full potential to phosphorylate GST-MEK1 in the in vitro kinase assay (Fig. 5b). To solve this discrepancy, first we compared the pS365 and pS729 levels of both WT-BRAF and 5KR-BRAF immunoprecipitates (IPs) that were used for in vitro kinase assays (Supplementary Fig. 8e). Surprisingly, both WT-and 5KR-BRAF displayed similar levels of pS365 and pS729. We used Flag-BRAF IPs from transiently transfected 293T cells for in vitro kinase assays. The exogenous Flag-BRAF expressed in 293T cells were apparently much more than the endogenous BRAF (3 μg Flag-BRAF was transfected to a 10 cm dish of 293T cells in order to prepare Flag-BRAF IPs). On the other hand, in assays to determine BRAF activity in cells, much less Flag-BRAF was transfected (90 ng BRAF was transfected to a 10 cm 293 cells). In the latter setting, the exogenous Flag-BRAF expressed at a level close to the endogenous BRAF level (Supplementary Fig.  8f). These results suggest that overexpressed Flag-BRAF in 293T cells are no longer modified in a fashion as the endogenous ones likely due to limited upstream kinases/phosphatases. This notion is supported by the observation that although lower doses of 5KR-BRAF (90 ng/dish) were hyperphosphorylated at S365 and thus incapable to activate MEK, high dose of 5KR-BRAF (1000 ng/dish) exhibited a comparable activity to its WT counterpart to promote MEK activation (Supplementary Fig.  8f). The relative band intensities were labelled underneath the blots, however, it is worth noting that the band intensities for the 1000 ng BRAF lanes were out of linear range, which could only be used as a close estimation.
Response: As kindly instructed, in Supplementary Fig. 9p, we found that similar to ubiquitinationdeficient 5KR-BRAF ( Fig. 7b and Supplementary Fig. 10d-e), ITCH binding-deficient BRAF mutant (S675A/P676A+P751A/P754, termed PA1/PA2) also displayed higher S365 phosphorylation in cells, while the S729 phosphorylation remain largely the same. In addition, we found that adding the S365A mutation to PA1/PA2-BRAF restored the activity of PA1/PA2-BRAF to phosphorylate MEK in 293 cells (Supplementary Fig. 9o). Furthermore, analogous to 5KR-BRAF, we found that PA1/PA2-BRAF could not be efficiently ubiquitinated by ITCH in vitro and thus still bound to 14-3-3 even after E1/E2/E3/UB incubation (Supplementary Fig. 9f). Fig 7 would be more relevant if knockdown and overexpression were inducible, so that the authors could study the effect of overexpression of 5KR-BRAF in established tumors.