RAB27B controls palmitoylation-dependent NRAS trafficking and signaling in myeloid leukemia

RAS mutations are among the most prevalent oncogenic drivers in cancers. RAS proteins propagate signals only when associated with cellular membranes as a consequence of lipid modifications that impact their trafficking. Here, we discovered that RAB27B, a RAB family small GTPase, controlled NRAS palmitoylation and trafficking to the plasma membrane, a localization required for activation. Our proteomic studies revealed RAB27B upregulation in CBL- or JAK2-mutated myeloid malignancies, and its expression correlated with poor prognosis in acute myeloid leukemias (AMLs). RAB27B depletion inhibited the growth of CBL-deficient or NRAS-mutant cell lines. Strikingly, Rab27b deficiency in mice abrogated mutant but not WT NRAS–mediated progenitor cell growth, ERK signaling, and NRAS palmitoylation. Further, Rab27b deficiency significantly reduced myelomonocytic leukemia development in vivo. Mechanistically, RAB27B interacted with ZDHHC9, a palmitoyl acyltransferase that modifies NRAS. By regulating palmitoylation, RAB27B controlled c-RAF/MEK/ERK signaling and affected leukemia development. Importantly, RAB27B depletion in primary human AMLs inhibited oncogenic NRAS signaling and leukemic growth. We further revealed a significant correlation between RAB27B expression and sensitivity to MEK inhibitors in AMLs. Thus, our studies presented a link between RAB proteins and fundamental aspects of RAS posttranslational modification and trafficking, highlighting future therapeutic strategies for RAS-driven cancers.

JMML (12). RAS mutations cause hyperactivation of the RAF/MEK/ ERK signaling pathway. A method to directly target RAS has eluded RAS biologists for decades, owing to the lack of druggable pockets on the surface of RAS proteins and picomolar affinity binding of RAS with GTP. Recent advances have resulted in the break-through development of targeted inhibitors (sotorasib and adagrasib) for one specific mutant form of KRAS (G12C), and these have recently been approved for the treatment of non-small cell lung cancer (13,14). However, G12C mutations are rare in all solid tumors other than those in the lung and they rarely occur in leukemia.
One tenet of RAS biology is that signaling is dependent on the subcellular localization of the GTPase. Nascent RAS proteins are synthesized on free polysomes and encounter farnesyltransferase in the cytosol. After farnesylation of its C-terminal CaaX motif, they gain an affinity for the ER, where they encounter CaaX-processing enzymes. Following CaaX modification, most of the KRAS proteins directly proceed to the plasma membrane (PM). In contrast, NRAS and HRAS proceed to the Golgi apparatus, where they are palmitoylated by the ZDHHC9-GOLGA7 complex (DHHC domain-containing 9 palmitoyl-acyltransferase and Golgin subfamily A member 7, a Golgi complex-associated protein) (15). Palmitoylation increases the affinity of RAS proteins for membranes by up to 100-fold. This increased affinity creates a kinetic trap that enriches NRAS and HRAS at the Golgi membranes, allowing for subsequent trafficking on the cytoplasmic face of exocytic vesicles destined for the PM. At the PM, RAS encounters and can be activated by receptor-Grb2-SOS complexes. Activated RAS proteins recruit RAF to the PM where it becomes active and initiates the MEK/ERK signaling cascade. NRAS and HRAS are discharged from the membrane by depalmitoylation and move by retrograde transport back to the Golgi and/ or ER for another round of palmitoylation. This dynamic cycle is important for RAS activation and function, because inhibition of either RAS palmitoylation or depalmitoylation abrogates RASmediated signaling or cell growth (16)(17)(18). Oncogenic RAS proteins transform cells only when associated with cellular membranes. Therefore, a mechanistic understanding of RAS lipid modification and trafficking may open up new avenues for better and more effective therapies in treating RAS-mutated myeloid malignancies.
RAB27A and RAB27B are small RAB GTPases that have been shown to function in the regulation of exocytic pathways, which involves intracellular vesicle trafficking, docking, and fusion with PM (19). Although RAB27A and RAB27B share similar functions, they also can be involved in different vesicles and different cell types, or can have distinct functions within the same cell type (20,21). In addition, RAB27A and B control different steps of the exosome secretion pathway (20). In murine hematopoietic cells, Rab27b is highly expressed in megakaryocytes and is important for proplatelet formation (22,23); it also influences neutrophil recruitment by regulating vesicle trafficking in neutrophils (24). RAB27B has been shown to be upregulated in solid cancers, which correlates with metastasis and poor survival (25). However, little is known about its role in hematopoietic malignancies or signaling via oncogenic pathways.
In this study, we identified that RAB27B promoted myeloid malignancies by regulating NRAS palmitoylation, trafficking, and signaling. RAB27B was upregulated by oncogenic CBL/JAK2/RAS signaling, implicating a potential link between CBL mutations, mary cells from JAK2 V617F MPN patients (Supplemental Table 2). We found that primary peripheral blood mononuclear cells (PBMCs) from JAK2 V617F MPN patients showed significantly higher RAB27B mRNA and protein levels than healthy controls, but RAB27A expression was not significantly changed ( Figure 2D). These data are consistent with a published genome-wide transcriptome analysis of cells, evidenced by the increased premRNA level ( Figure 2C), suggesting that the CBL-controlled signaling network indirectly regulates RAB27B expression.
Upregulated RAB27B expression is found in JAK2 V617F MPNs and correlates with poor survival in AMLs. Since CBL loss enhances JAK2 activity and signaling (9), we examined RAB27B expression in pri- analysis to examine RAB27B and RAB27A protein levels and halflives (n = 3) in DKO+D compared with Ctrl cells, as shown in TF-1 (B and C) and U937 (D) cells. CHX: cycloheximide that inhibits nascent protein synthesis. Quantification of RAB27B half-lives is shown in the right panels of C and D. (E and F) WB analysis to examine RAB27B and RAB27A protein level and degradation (n = 3) in TF-1 cells after single or double knockout (DKO) of CBL and CBL-B. Quantification of RAB27B half-lives is shown in the right panel of F. (G and H) TF-1 cells stably expressing CBL C381A mutant, CBL WT or empty vector (EV) were established. (G) WB analysis to examine RAB27B protein halflife (n = 3). Quantification of RAB27B half-lives is shown in the right panel. (H) RAB27B protein level in the presence of DMSO, MG132 (MG, 10 μM) or Chloroquine (CL, 5 μM). pSTAT5 is used as a control for proteasomal degradation inhibited by MG132. Data for the halflife studies are represented as mean ± SD, and determined by 2-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001.
al phosphorylation level of JAK2 was the highest in CMK cells, followed by JAK2 V617F+ HEL and SET-2 cells, and the lowest in K562 and TF-1 cells. Interestingly, RAB27B protein and mRNA levels seemed to be positively correlated with the basal JAK2 phosphorylation level in these cell lines (Supplemental Figure 1, B and C). We next sought to perturb JAK2 activation using various approaches to assess if JAK2 inactivation results in changes in RAB27B levels.
CD34 + cells from a large number of MPN patients (28), demonstrating that CD34 + cells from JAK2 V617F MPNs express a significantly higher level of RAB27B, but not RAB27A, compared with healthy controls ( Figure 2E and Supplemental Figure 1A). To further study the correlation between JAK2 signaling and RAB27B expression, we compared multiple leukemia cell lines and found that they exhibit a wide range of RAB27B and RAB27A expression levels. The bas- qRT-PCR to examine RAB27B and RAB27A mRNA levels in TF-1 cells stably expressing CBL C381A mutant or CBL WT compared with empty vector (EV). (C) qRT-PCR to examine RAB27B nascent and mature RNA level in TF-1 DKO cells compared with Ctrl cells. Two different pairs of primers were used to detect premRNA (designated as pre 1 and pre 2, depending on the primer set). Mature messenger RNA is labeled as mRNA. (D) RAB27B protein (left) and mRNA (right) levels in primary human PBMCs from healthy donors (C1-C3, n = 3) and patients with JAK2 V617F+ MPN (n = 4) are shown. (E) RAB27B mRNA levels in BM CD34 + cells from healthy donors (n = 15) and patients with JAK2 V617F+ MPN (n = 43) plotted using the expression data from GSE103176 (28). Each symbol indicates individual subject. (F) RAB27B expression level in patients with AML and healthy controls (GEPIA Cancer Database). (G and H) Kaplan-Meier plot of overall survival for patients with AML with low or high expression of RAB27B. UALCAN (G) top 25% or bottom 75% (low/medium) expression of RAB27B; CTGA database from BloodSpot (H) top 50% or bottom 50% expression of RAB27B. P values determined by log-rank t test are shown. In all relevant panels, data are represented as mean ± SD. 1-way ANOVA was used in panels A, B and D; Student's 2-tailed t tests were used in Figure C and E; *P < 0.05; **P < 0.01; ***P < 0.001. J Clin Invest. 2023;133(12):e165510 https://doi.org/10.1172/JCI165510 lation as well as JAK2 expression were elevated in TF-1 DKO cells compared with Ctrl cells, as previously reported (9) ( Figure 3D and Supplemental Figure 5A). Importantly, we found a significant reduction in phosphorylated-ERK (pERK) upon RAB27B depletion in TF-1 DKO cells, but AKT or STAT5 phosphorylation remained unchanged ( Figure 3D and Supplemental Figure 5B). In contrast, RAB27A depletion did not inhibit cell growth, and it did not affect the activation of AKT, ERK, and STAT5 signaling pathways in TF-1 DKO cells (Supplemental Figure 6, A-C). Of note, RAB27B depletion did not affect cytokine-induced signaling in TF-1 parental cells (Supplemental Figure 3B). We next assessed the upstream signals that activate ERK. Basal c-RAF and MEK phosphorylation were increased in TF-1 DKO cells compared with Ctrl cells ( Figure 3D left panel). Importantly, RAB27B depletion dramatically reduced MEK and c-RAF phosphorylation in TF-1 DKO cells ( Figure 3D right panel), implicating a potential involvement in RAS signaling. We thus performed RAS-GTP pulldowns using GST-RBD-conjugated beads, followed by WB analysis using isoform-specific RAS or pan-RAS antibodies. Total RAS activity, especially NRAS activity, and, to a lesser extent, KRAS activity, was elevated in TF-1 DKO cells compared with Ctrl cells ( Figure 3E left panel). Strikingly, whereas activation of NRAS was almost completely abrogated upon RAB27B depletion in TF-1 DKO cells, GTP-loading of KRAS was preserved ( Figure 3E right panel). We next validated our data in OCI-AML3 cells that harbor an NRAS Q61R mutation, a hotspot mutation that lies within the GTP-binding region of the NRAS protein. Consistently, RAB27B depletion reduced cell growth, ERK signaling, and NRAS activity in OCI-AML3 cells (Figure 3, F-H). Thus, our results revealed what we believe to be a previously unappreciated role for RAB27B in regulating NRAS activity, signaling, and leukemia cell growth.
RAB27B is critical for PM localization and palmitoylation of NRAS. It is well established that membrane association is required for RAS activation and signaling. At a steady state, pools of NRAS have been observed on the PM and the Golgi, as well as in the cytosol (32,33). Since RAB27B is reported to be a Golgi-associated RAB GTPase that plays a role in vesicle transport (20,34), we sought to examine if it plays a role in NRAS trafficking. PM association of GFP-NRAS stably expressed in TF-1 DKO cells via lentiviral transduction was diminished upon shRNA silencing of RAB27B ( Figure 4, A-C). Because hematopoietic cells are small and difficult to image, we performed live-cell imaging of GFP-NRAS in SK-MEL-147 cells, a melanoma cell line that harbors an NRAS Q61R mutation, the same NRAS mutation found in OCI-AML3. As we have previously reported (33), GFP-NRAS localized to the PM and Golgi in SK-MEL-147 cells. Silencing RAB27B with siRNA in these cells significantly diminished the PM but not the Golgi expression ( Figure 4, D and E). We next evaluated the subcellular localization of endogenous NRAS with and without depletion of RAB27B in TF-1 DKO cells using detergent-free subcellular fractionation assays (33,35). RhoGDI, TIE2, and LAMIN serve as markers for cytosol, membrane, and nuclear fractions, respectively. Though NRAS was recovered predominantly in the membrane fraction of control shRNA cells, it was distributed among both membrane and cytosolic fractions in RAB27B-deficient TF-1 DKO cells ( Figure 4F and Supplemental Figure 6D), which is consistent with our imaging data. In contrast, RAB27B depletion did not affect the subcellular Treatment for 24 hours, but not 5 hours, with the JAK2/1 inhibitor, ruxolitinib (Ruxo) in TF-1 CBL DKO cells significantly reduced RAB27B but not RAB27A mRNA levels (Supplemental Figure 1D). We stably expressed the myc-tagged WT JAK2 or the V617F mutant in TF-1 cells and found that JAK2 V617F with aberrant JAK2 activation increased RAB27B, but not RAB27A, protein and mRNA levels (Supplemental Figure 1, E and F). As we previously reported, LNK acts as the adaptor protein for CBL-mediated JAK2 degradation, and LNK-deficient cells gain higher JAK2 protein level and enhanced JAK2 signaling (9). Consistently, the protein and mRNA levels of RAB27B, but not RAB27A, were increased in LNK-deficient cells (Supplemental Figure 1, G and H). In addition, we knocked down JAK2 in JAK2 V617F HEL cells to approximately 50% and observed a corresponding reduction in RAB27B protein and mRNA levels (Supplemental Figure 1, I and J). Of note, TF-1 and HEL cells were not able to proliferate with sustained or complete KD of JAK2, as JAK2 is essential to their growth. Together, these data suggest that aberrant CBL signaling and JAK2 activation resulted in the upregulation of RAB27B transcription.
RAB27B regulates NRAS activity, signaling, and leukemia cell growth. The upregulation of RAB27B in JAK2 V617F MPNs and its correlation with AML prognosis prompted us to investigate the potential role of RAB27B in myeloid malignancies. We first examined its function in the growth of TF-1 DKO cells, in which the RAB27B level is markedly elevated. We designed 2 efficient shRNAs targeting human RAB27B confirmed by WB and qRT-PCR ( Figure 3, A and B). As we previously reported, TF-1 DKO cells exhibited cytokine-independent growth and were hypersensitive to cytokines compared with TF-1 parental cells (9). Notably, RAB27B depletion significantly blunted the growth of TF-1 DKO cells ( Figure 3C), but not the growth of parental TF-1 cells (Supplemental Figure 3A). RAB27B is reported to be involved in exosomes and in cytokine secretion (30,31). However, our data suggest that RAB27B depletion does not affect exosome secretion in TF-1-DKO cells (Supplemental Figure 4A). In addition, the growth of TF-1-DKO cells was not affected by the conditioned medium from TF-1-DKO cells and from RAB27B-depleted TF-1-DKO cells (Supplemental Figure 4, B-D), suggesting that RAB27B does not regulate TF-1 DKO cell growth via an autocrine pathway.
To study the mechanisms by which depletion of RAB27B inhibits leukemia cell growth, we evaluated intracellular signaling pathways important for the growth of hematopoietic cells, i.e., JAK-STAT, PI3K-AKT, and ERK. The basal AKT and ERK phosphory-distribution of endogenous KRAS, and RAB27A-depletion did not affect NRAS localization ( Figure 4F and Supplemental Figure 6D).
Palmitoylation of NRAS on cysteine 181 regulates its trafficking between the Golgi and PM, and, therefore, regulates its signaling (36). To detect RAS palmitoylation, we performed an acyl-PEG exchange (APE) assay that allows for the detection of endogenous protein palmitoylation in total cell lysates (37). We first set out to confirm previous findings and validated our APE assays by stably expressing WT and oncogenic NRAS proteins with and without their C181S mutant counterparts. The C181S mutation , and TF-1 DKO cells with or without RAB27B depletion (right), were cultured in media containing serum only or serum and GM-CSF. Cell lysates were subjected to WB analysis with indicated antibodies to examine RAF/MEK/ERK activation. (E) TF-1 cells as described in (D) were cultured in media containing serum only. RAS GTPase activities were measured by RAS GTP pulldowns using RAF-1 RBD agarose beads, followed by WB with indicated antibodies. GTP-bound RAS represents active RAS. Input lysates were subjected to WB analysis with indicated antibodies as controls.
(F-H) RAB27B was stably depleted via lentiviral-shRNA mediated KD in OCI-AML3 cells, with shLuc used as a control. (F) Cells were plated at equal cell numbers and cell growth was determined by counting of live cells. (G) ERK activation was determined by WB. (H) NRAS activity was determined by RAF-1 RBD agarose bead pulldown followed by WB using anti-NRAS antibodies. Input lysates were subjected to WB analysis with the indicated antibodies as controls. In all relevant panels, data are represented as mean ± SD, and 2-way ANOVA was used for comparing cell growth; *P < 0.05; **P < 0.01; ***P < 0.001. of the bone marrow and spleen (Table 1 and Supplemental Figure 8A). Moreover, BM cells from the mice transplanted with Rab27b-deficient LSKs expressing NRAS Q61R showed reduced constitutive pERK activation compared with Rab27b fl/fl LSKs expressing NRAS Q61R ( Figure  5J), consistent with the ex vivo LSK data in Figure 5D.
To confirm our studies using mutant NRAS Q61R , we examined the effect of RAB27B in NRAS G12D , the most common mutant form of NRAS in human myeloid malignancies (41)(42)(43) (Figure 6). Consistent with our Q61R results, mice transplanted with Rab27b fl/fl LSKs expressing NRAS G12D mostly developed MML, while those transplanted with Rab27b-deficient LSKs expressing NRAS G12D had reduced incidence of CMML and instead died of T-ALL (Table 2), as evidenced by flow cytometric analysis ( Figure 6B) and the histology of the bone marrow and spleen ( Figure 6C and Supplemental Figure  8B). Consequently, mice transplanted with Rab27b-deficient LSKs expressing NRAS G12D survived moderately but significantly longer than those transplanted with Rab27b fl/fl LSKs expressing NRAS G12D ( Figure 6D). Taken together, these data demonstrate that Rab27b deficiency abrogates oncogenic NRAS-mediated ERK signaling and myeloid leukemia development in vivo.
RAB27B depletion reduces clonogenic growth, ERK activation, and NRAS palmitoylation in primary NRAS mut AMLs. Our data suggest that RAB27B plays a critical role in the growth of NRAS mutant myeloid malignancies. To test the clinical significance of our findings, we examined primary NRAS mut or NRAS WT AML patient samples (Supplemental Table 3). We chose AMLs with high NRAS mut allele frequency in order to detect the full extent of NRAS mut effects in subsequent studies. We infected primary cells from the BM or PB of AML patients with lentiviruses expressing shRNA to RAB27B or luciferase (Luc) as a control. Of note, RAB27B depletion significantly reduced the colony formation ability of primary AML cells from NRAS mut patients, but not those of BRAF mut or KRAS mut patients (NRAS WT ) ( Figure 7A). These data are consistent with the signaling studies showing that RAB27B depletion reduced ERK phosphorylation in AMLs ( Figure 7B). More importantly, RAB27B depletion reduced the palmitoylation level of endogenous NRAS proteins in primary AML cells ( Figure 7C).
Our data suggest that RAB27B promotes AML cell growth by regulating NRAS/MEK/ERK signaling, therefore, we next interrogated data from BeatAML (44) to assess if RAB27B expression correlates with responses to MEK inhibitors (MEKi). Indeed, AML patients with high RAB27B, but not RAB27A, expression were sensitive to 4 different MEK inhibitors, as evidenced by lower AUC values ( Figure 7D and Supplemental Figure 9A). However, no significant correlation was found between RAB27B expression and sensitivity to PI3K inhibitors (Supplemental Figure 9B). Thus, our data implicate a critical role for RAB27B in conferring aberrant NRAS/ERK signaling and AML cell growth.
RAB27B interacts with ZDHHC9 and regulates ZDHHC9-mediated NRAS palmitoylation. To explore the mechanisms by which RAB27B affects NRAS palmitoylation and trafficking, we first assessed the potential interaction between RAB27B, NRAS, and its palmitoyl acyltransferase (PAT) complex ZDHHC9/GOLGA7 by coimmunoprecipitation (coIP) (45). We overexpressed tagged RAB27B and NRAS in 293T cells and found that RAB27B pulled down ZDHHC9 but not NRAS ( Figure 8A). Neither WT nor oncogenic mutant NRAS bound to RAB27B (Supplemental Figure 10A). RAB27B specifically interacted with ZDHHC9 but not GOLGA7 (Supplemental Figure   completely abrogated palmitoylation of WT and oncogenic NRAS, as well as basal activation of ERK in TF-1 cells (Supplemental Figure  7, A and B). Notably, disruption of NRAS palmitoylation and basal ERK phosphorylation abrogated cytokine-independent growth, whereas GM-CSF-mediated cell growth was unaffected (Supplemental Figure 7C). Using the APE assay, we found that the basal pan-RAS and NRAS palmitoylation levels were elevated in TF-1 DKO cell compared with Ctrl cells, while KRAS was not palmitoylated ( Figure 4G left panel). More importantly, RAB27B depletion significantly inhibited pan-RAS, and, in particular, NRAS palmitoylation in TF-1 DKO cells ( Figure 4G right panel). Consistently, RAB27B depletion also reduced NRAS palmitoylation in OCI-AML3 cells ( Figure 4H). HRAS activity was very low in these cell lines, indeed, the effect of RAB27B on this RAS isoform could not be determined (data not shown). These data reveal that RAB27B is critical for NRAS palmitoylation and explain our observation that RAB27B is required for NRAS trafficking to the PM.
Rab27b deficiency in mice inhibits oncogenic NRAS-mediated signaling, HSPC growth, and myeloid leukemia development in vivo. To validate our findings in cell lines, we employed Rab27b-deficient mice. Germline Rab27b KO mice are grossly normal (38). Vav-cre expresses active Cre recombinase resulting in panhematopoietic deletion of the floxed sequences. To determine a cell-intrinsic role of RAB27B in hematopoietic cells, we generated conditional KO mice using Vav-cre (39). Rab27b fl/fl;Vav mice have largely normal hematopoiesis (data not shown). We isolated LSK (Lineage-Kit + S-ca1 + ) cells from Rab27b fl/fl;Vav and Rab27b fl/fl mice and infected them with retrovirus expressing either WT or Q61R mutant NRAS with GFP as a fluorescent marker. Infected cells were purified and subjected to cell-based and biochemical studies ( Figure 5A) (40). Oncogenic NRAS conferred cytokine-independent HSPC cell growth and colony-forming ability. This also occurred in the presence of low concentrations of GM-CSF ( Figure 5, B and C). Strikingly, Rab27b deficiency significantly reduced mutant NRAS Q61R -conferred cell growth and clonogenic ability. In contrast, Rab27b deficiency had no effect on the growth of cells expressing NRAS WT ( Figure 5, B and C). Concordantly, Rab27b deficiency diminished ERK hyperactivation induced by the NRAS Q61R mutation ( Figure 5D) and reduced palmitoylation level of NRAS ( Figure 5E). Hence, Rab27b deficiency in mice inhibits mutant NRAS-mediated signaling and cell growth.
To further study the requirement for RAB27B in mutant NRAS-mediated oncogenesis in vivo, we transplanted LSK cells expressing NRAS Q61R into lethally irradiated recipient mice ( Figure 5, F-J). Rab27b fl/fl and Rab27b-deficient cells had similarly high infection rates at the time of transplantation, as evidenced by the percentage of GFP + cells ( Figure 5G). At 6 and 10 weeks, mice transplanted with Rab27b fl/fl LSK cells expressing NRAS Q61R exhibited high white blood cell counts, particularly neutrophils and monocytes ( Figure 5H). In contrast, mice transplanted with Rab27b-deficient LSKs expressing NRAS Q61R had significantly reduced white blood cell counts and lower proportions of GFP + cells and GFP + myeloid cells in the peripheral blood (PB) than those reconstituted with Rab27b fl/fl LSK cells expressing NRAS Q61R ( Figure 5, H and I). Importantly, mice transplanted with Rab27b fl/fl LSKs expressing NRAS Q61R mostly developed MML, while those transplanted with Rab27b-deficient LSKs expressing NRAS Q61R had reduced incidence of MML, instead the moribund mice died of T cell-acute lymphoblastic leukemia (T-ALL), as evidenced by analysis dant with the palmitoylation data, dual expression of ZDHHC9 and GOLGA7 partially rescued the growth of RAB27B-depleted cells, but expression of ZDHHC9 or GOLGA7 alone did not ( Figure 8F).

Discussion
In this study, we uncovered a previously unappreciated role for RAB27B in regulating NRAS palmitoylation, subcellular trafficking, and signaling in myeloid malignancies, in part via interacting with the palmitoyl acyltransferase ZDHHC9. Furthermore, we identified a signaling axis, where oncogenic CBL/JAK2 signaling upregulates RAB27B to enhance NRAS activity and ERK phosphorylation, thereby implicating a potential link between CBL and JAK2 V617F mutations and RAS/ERK signaling. Notably, RAB27B is important for the growth of leukemia cells with CBL or NRAS mutations but does not affect normal hematopoiesis. Hence, this work reveals what we believe to be new signaling dynamics that enhance our understanding of compartmentalized RAS signaling, suggesting RAB27B as a therapeutic target to abrogate oncogenic CBL/JAK2 and RAS-driven myeloid malignancies.
High RAB27B expression is shown to be an unfavorable prognostic factor in many solid cancers such as non-small cell lung carcinoma, colorectal cancer, and ovarian cancer (25); however, its role in leukemia was poorly established. In this study, we found that high RAB27B expression correlated with poor survival of patients with AML. The recent 2016 WHO classification of myeloid malignancies has called for the recognition of proliferative CMML (pCMML) and dysplastic (dCMML) subtypes, with the former having higher white blood cell counts (46). RAS pathway mutations, including NRAS and CBL, define the pCMML phenotype. This phenotype is aggressive, linked to dismal outcomes, and associated with increased transformation to AML, as demonstrated by the exome sequencing of primary human samples and the Nras G12D mouse model (47). Moreover, RAS/CBL mutations predict resistance to JAK inhibitors in myelofibrosis and are associated with poor prognostic features (48). The findings in this work provide a potential mechanistic link between CBL/JAK2-mutated myeloid malignancies and activation of RAS signaling via upregulation of RAB27B. Interestingly, the CBL family of E3 ubiquitin ligases and oncogenic JAK2 signaling does not directly control RAB27B protein stability; rather, they upregulate RAB27B gene transcription. Thus, our work suggests that the downstream gene network triggers a positive feedback loop to amplify oncogenic signaling and contribute to disease progression.
We showed in this study that RAB27B depletion dramatically reduced NRAS palmitoylation, activity, and downstream RAF/ MEK/ERK signaling in both leukemia cell lines and primary human AMLs. RAB27B depletion impaired the PM localization of NRAS 10B). In light of the recent discovery of ABHD17 as an acyl thioesterase that depalmitoylates NRAS (16), using coIP we examined whether RAB27B interacts with ABHD17. The results showed that RAB27B interacted with ZDHHC9 but not ABHD17A (Supplemental Figure 10, C and D). Importantly, we established TF-1 DKO cells stably expressing HA-tag ZDHHC9 and found that HA-ZDHHC9 pulled down endogenous RAB27B, but not RAB27A ( Figure 8B). As expected, ZDHHC9 interacted with NRAS and GOLGA7 in 293T and TF-1 cells. Next, we overexpressed 23 HA-tagged mammalian ZDHHCs along with Myc-tagged RAB27B and found that ZDHHC9, 14, 18, and 23 bound to RAB27B. Among these ZDHHCs, ZDH-HC9 and ZDHHC18 demonstrate the strongest interaction with RAB27B (Supplemental Figure 10E). However, ZDHHC18 depletion did not affect NRAS palmitoylation in TF-1 DKO cells (Supplemental Figure 11), consistent with previous reports that identified ZDHHC9 as a PAT that modifies NRAS (45).
Lastly, and importantly, we reasoned that if the reduced NRAS palmitoylation and the compromised cell growth observed in RAB27B-depleted cells was owing to disrupted ZDHHC9 function, overexpression of ZDHHC9 would be able to rescue this phenotype. To test this hypothesis, we first confirmed the role for ZDHHC9 in NRAS palmitoylation via shRNAs targeting ZDHHC9 or GOLGA7 (Supplemental Figure 12). In TF-1 CBL DKO cells, ZDHHC9 depletion, and, to a lesser extent, GOLGA7 depletion, reduced NRAS palmitoylation (Supplemental Figure 12C). Interestingly, RAB27B depletion exerted a stronger suppression of NRAS palmitoylation than depletion of ZDHHC9 did (Supplemental Figure 12C). Importantly, combined expression of ZDHHC9 and GOLGA7 partially rescued NRAS palmitoylation in RAB27B-depleted TF-1 DKO cells, while ZDHHC9 or GOLGA7 alone did not (Figure 8, C-E). Concor- Table 1  abnormalities, Rab27b deficiency significantly abrogated myeloid leukemia development conferred by oncogenic NRAS. Of note, RAB27B depletion impacted the palmitoylation and PM localization of both WT and oncogenic NRAS, but more effectively suppressed oncogenic NRAS-mediated leukemia cell growth and ERK signaling, suggesting that oncogenic NRAS-conferred cell growth is more sensitive to inhibition of palmitoylation or ERK signaling. This is consistent with prior studies showing that MEK inhibition is effective in abrogating murine Nras-mutant AML while Nras is dispensable for the normal function of HSCs (49). This can be attributed to the fact that KRAS plays an essential role in cytokine-mediated normal hematopoiesis (50). Furthermore, our mouse models utilizing retroviral-mediated overexpression of oncogenic NRAS suggest that RAB27B only impacts MML but not T-ALL development in mice. This might be due to the restricted expression pattern of Rab27b. Of note, this overexpression model infects both myeloid and lymphoid progenitors in addition to HSCs (51,52), which limits its ability to dissect the potential distinct role of RAB27B in different progenitors. Nonetheless, our data are consistent with the notion that RAB27B plays a critical role in leukemia cell growth when NRAS is requiredsuch as in AMLs with oncogenic NRAS or CBL in both NRAS-mutated leukemia and melanoma cells. RAB27B interacted with ZDHHC9, a known PAT for NRAS, and regulated NRAS palmitoylation, thereby modulating NRAS trafficking to the PM. Importantly, overexpression of ZDHHC9 and GOLGA7 complex partially restored NRAS palmitoylation, signaling, and cell growth impacted by RAB27B depletion. The partial rescue suggests that RAB27B may affect the function of additional PATs for NRAS. This is corroborated by our data showing that RAB27B depletion exhibited a more pronounced effect on NRAS palmitoylation when compared with ZDHHC9 depletion or GOLGA7 depletion. We show that RAB27B interacted with ZDHHC9 but not NRAS or ABHD17, suggesting that RAB27B affected ZDHHC9 function. However, our present study does not address if and how RAB27B regulates ZDHHC9 activity or access to the substrate. How RAB27B functions and modulates NRAS palmitoylation cycle merits further future investigation. Interestingly, whereas RAB27B is not required for the growth of parental TF-1 cells cultured in cytokines such as GM-CSF, it is critical for the growth of transformed cells cultured in basal conditions, specifically CBL/CBL-B-deficient TF-1 cells or NRAS mutated OCI-AML3 cells. In agreement, while Rab27b -/mice display no obvious mutationswhile it is dispensable for the growth of AMLs with other dominant mutations or in normal HSPCs.
Considering that palmitoylation is a critical PTM for NRAS activation and signaling, significant efforts have been devoted to identifying specific PATs, depalmitoylases, and inhibitors to indirectly target RAS. It has been established that palmitoylation regulates the trafficking of NRAS between the Golgi and the PM (53), and palmitoylation of oncogenic NRAS is essential for leukemia progression (17). The dynamic palmitoylation/depalmitoylation cycle is important for RAS activation and function because inhibition of either RAS palmitoylation or depalmitoylation abrogates RAS-mediated signaling or cell growth (16)(17)(18). ZDHHC9 has been shown to palmitoylate H-and N-RAS in conjunction with GOLGA7 (45). However, previous studies and our data showed that depletion of ZDHHC9 only partially reduced NRAS palmitoylation and cannot completely abrogate leukemogenic potential of oncogenic NRAS (18,54), thus suggesting that additional PATs can act on NRAS. Moreover, palmitoylation inhibitors cause cell damage due to their pleiotropic effects on lipid metabolism (55). Notably, depalmitoylation inhibitors disrupt the RAS palmitoylation cycle and its cellular localization, thus suppressing the growth of murine oncogenic NRAS mut AML blasts (36,56). Recently, the ABHD17 family was found to be the relevant thioesterases that depalmitoylate NRAS (16). ABD957, an inhibitor of ABHD17, has been shown to be more selective than other ABHD inhibitors, though it only partially dampens oncogenic NRAS depalmitoylation and activity (16). Although we showed that RAB27B does not interact with ABHD17 directly in cell lines, we cannot exclude the possibility that RAB27B affects NRAS depalmitoylation.
Our findings indicate that RAB27B is a safe and promising therapeutic target for CBL or NRAS mutant malignancies. The RAB family are challenging targets for inhibition by small molecules, owing to the flat topology of RAB family effector interface and high affinity for GTP binding. Nexinhib-20 has been reported as a RAB27A inhibitor through inhibiting RAB27A-JFC1 binding (57). Cocrystal structures show that several effectors of RAB27A interact with the RAB27A SF4 pocket (WF-binding pocket) via a conserved tryptophan-phenylalanine (WF) dipeptide motif (58). A recent study took advantage of 2 cysteine residues, C123 and C188, that flank the WF pocket and are unique to RAB27A and RAB27B, which belong to a family of more than 60 RAB proteins, and identified the first covalent ligands for native RAB27A (57). This WF motif is present in RAB27B. Thus, this work provides a platform for identifying suitable lead fragments for future development of competitive inhibitors of the RAB27A and RAB27B-effector interaction interface. Alternatively, the WF pocket may allow the development of RAB27B degraders (59). Future investigations are warranted , and 13 C-Lysine (K8) and 13 C-Arginine (R10) (Cambridge Isotope Laboratories) along with standard supplements for 2 weeks. Cells were washed twice in cold PBS, and cell pellets were collected by centrifugation at 16,000g for 15 seconds at 4°C, then snap frozen at -80°C. Three biological replicates were subjected to quantitative proteomics analysis at the Children's Hospital of Philadelphia Proteomics Core Facility.
For protein half-life assays, the cells were treated with 50 μg/ml cycloheximide (CHX) for the indicated time. The cell pellets were collected as described above, lysed in SDS lysis buffer and then boiled for 10 minutes at 95 o C. For WB assay, 20 μg of each protein sample was loaded in an SDS-PAGE gel and then transferred to a nitrocellulose membrane (Bio-Rad) using an Amersham TE 70 ECL Semi-Dry Transfer Unit. The membranes were blocked for 30 minutes using 5% BSA (for phosphoprotein antibodies) or 5% nonfat milk (for total protein antibodies) in TBS + Tween20 (TBST) and incubated with primary antibodies using the same buffer at 4°C overnight, followed by incubation with HRP-conjugated secondary antibodies for 45 minutes at room temperature (RT). The membranes were subsequently washed in TBST and developed with ECL using a KwikQuant imager (Kindle Biosciences, LLC).
Cell fractionation assay. 6 × 10 7 TF-1 cells were resuspended in 2 mL hypotonic homogenization buffer (10 mM HEPES at pH 7.4, 10 mM KCl and 1.5 mM MgCl 2 ), and kept on ice for 10 minutes. The swelled cells were dounced approximately 20 times using a dounce homogenizer with a tightly fitting pestle and cell lysis was confirmed under microscopy. and PMSF (Sigma-Aldrich) in deionized H 2 O) for 30 minutes at 4°C. The supernatant was collected by centrifugation at 16,000g for 5 minutes at 4°C. In some experiments, cell extracts were cleared by a 200,000g ultracentrifugation. After preclearing with protein A/G beads for 1 hour, cell lysates were incubated with HA-EZ Agarose Beads (Sigma-Aldrich) for 3 hours or anti-Myc primary antibody for 3 hours followed by incubation with protein A/G beads for 1 hour at 4°C with gentle agitation. The beads were then extensively washed with IP buffer and boiled for 10 minutes at 95 o C in SDS loading buffer for WB analysis.
RAS GTPase assay. RAS activity was detected using the Ras Activation Assay kits from Cell Biolabs. Briefly, cell pellets were lysed with 1 × Assay Buffer supplemented with PMSF and proteinase inhibitor 293T cells were transfected with constructs to express tagged RAB27B, NRAS and ZDHHC9 as indicated. Cells were then subjected to IP followed by WB using the indicated antibodies. (B) TF-1 DKO cells stably expressing HA-ZDHHC9 were subjected to coIP with anti-HA antibodies followed by WB analysis using the indicated antibodies to examine its interaction with endogenous proteins. (C) Schematic illustration of the hypothesis depicting how RAB27B regulates NRAS signaling (blue arrows) and how to restore NRAS signaling disrupted due to RAB27B loss (red arrows). (D-F) RAB27B-depleted TF-1 DKO cells stably expressing HA-ZDHHC9 or Myc-GOLGA7 were subjected to examination of palmitoylation status and cell growth. (D) WB to examine the efficiency of RAB27B KD or ZDHHC9 and GOLGA7 overexpression in TF-1 Ctrl and DKO cells. (E) APE assay to examine palmitoylation of endogenous RAS proteins. OE, overexpression; HAM, hydroxylamine; Palm, palmitoylated. (F) Cells were cultured in triplicate in different concentrations of human GM-CSF and cell growth after 3 days in culture was determined by MTT absorbance. Data are represented as mean ± SD. **P < 0.01; ***P < 0.001 compared with the shRAB27B group, determined by 2-way ANOVA.
for 10 minutes and resuspended in PBS with DAPI for imaging using a Zeiss LSM 710 confocal microscope equipped with a 100 × /1.4 numerical aperture oil-immersion objective. SK-MEL-147 cells were grown on 35 mm dishes with no. 0 coverslips (MatTek) and transfected with 25 nM ON-TARGETplus Human RAB27B siRNA SMARTPool (Dharmacon, Life Technologies) or nontargeting Control Pool siRNA (Dharmacon, Life Technologies) using DharmaFECT1 (Dharmacon, Life Technologies). Two days after siRNA transfection, the cells were transfected with 1 μg/ dish pEGFP-NRAS using Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. The next day, EGFP fluorescence in live cells was imaged using a Zeiss LSM 800 inverted confocal microscope equipped with a 63 × /1.4 oil objective. Images were analyzed in ImageJ software.
Lentiviral infection of primary human AML cells. As described previously (62), non-TC-treated plates were precoated with RetroNectin (Takara) at 4°C overnight, followed by blocking with 2% BSA in PBS for 30 minutes at RT and washing with HBSS (Gibco). The lentiviral supernatant expressing shRNA targeting Luc or RAB27B was loaded to the plates and centrifuged at 1,800g for 1 hour at 10°C. After the plate was warmed to RT, viral supernatant was removed and human primary cells at 500,000 cells/mL with 2 μL/mL LentiBlast Premium (OZ Biosciences) were added to the plate, then centrifuged at 450g for 90 minutes at 37°C.
Purification and retroviral infection of mouse LSK cells. LSK cells were purified as described previously (61). In brief, BM cells were isolated from femora and tibiae of mice in PBS containing 0.5% BSA and 2 mM EDTA buffer. Lineage negative cells (Lin-) were isolated using the Lineage Cell Depletion Kit (Miltenyi Biotec). Lin-cells were stained with APC-c-Kit and PE-Sca1 antibodies, and LSK cells were purified with a BD FACSAria Fusion cytometer and cultured in SFEM medium plus 10% FBS supplemented with 20 ng/mL Flt3L, 20 ng/mL IL-6, 100 ng/ mL SCF, 20 ng/mL TPO, and 0.1 mM β-ME. After 48 hours of culture, LSK cells were spin-infected with retroviruses expressing WT or Q61R mutant NRAS preloaded on RetroNectin-coated plates, with 10 μg/ml polybrene (Sigma-Aldrich). Cells were either transplanted 24 hours after infection or sorted for GFP positivity 48 hours after infection for cell proliferation assay, colony assay, or biochemical experiments.
Signaling studies. For cytokine sensitivity, LSK cells were starved in RPMI-1640 media plus 0.5%BSA for 1-2 hours, then stimulated with a graded dose of GM-CSF for 10 minutes. Cell pellets were immediately collected by centrifugation at 16,000g for 15 seconds and snapfrozen. For TF-1 cells, cells were either collected in GM-CSF-containing culture media or cytokine-free serum-containing media for 6-12 hours.
Cell proliferation assay. We seeded 10,000 cells of various cell lines or primary mouse HSPCs in triplicate in 96-well plates in a graded concentration of cytokines at 100 μL media per well. After 3 days of culture, 3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyl tetrazolium bromide (MTT; Invitrogen) was added to a final concentration of 0.5 mg/mL and incubated for 4 hours at 37°C. 100 μL stopping buffer (5% SDS, 2.5% acetic acid, 50% dimethylformamide) was then added to terminate the reaction. The absorbance was measured on a SpectraMax 190 Microplate Reader at 570 nm.
Colony-forming cell assay. For primary human AML cells, 50,000 cells in 1 mL methylcellulose (StemCell) supplemented with 5 U/mL EPO (EPOGEN), 10 ng/mL IL-3, 5 ng/mL SCF and 5ng/mL GM-CSF were thoroughly vortexed, plated in triplicates in 35 mm small petri dishes (Olympus), and kept in an incubator at 37°C and 95% humidity in an atmosphere of 5% CO 2 . The colonies were counted after 10-14 days.
The nuclei were collected by centrifugation at 500g for 10 minutes at 4 o C and sonicated in 3mL 1 × SDS loading buffer, then boiled for 10 minutes at 95 o C. Postnuclear supernatant was transferred to a polycarbonate tube and subjected to ultracentrifugation at 350,000g for 1 hour at 4°C. Supernatant was collected as the cytosolic fraction, and the pellet was lysed in 2 mL 1% NP-40 lysis buffer for 30 minutes at 4°C as the membrane fraction. 1 mL 3 × SDS loading buffer was added to the cytosolic and membrane fractions, which were then boiled for 10 minutes at 95°C. All fractions were then subjected to WB analysis. APE assay. Protein palmitoylation was examined using the APE assay (37). In brief, cells were lysed in 1 × TEA lysis buffer (50 mM triethanolamine pH 7.3, 150 mM NaCl) supplemented with 5 mM EDTA, 4% SDS, and protease inhibitors cocktail in deionized H 2 O and directly sonicated at RT. The cell lysate was reacted with Tris (2-carboxyethyl) phosphine (TCEP; Pierce) for 30 minutes at RT with nutation and subsequently with N-ethylmaleimide (NEM; Sigma-Aldrich) for 2 hours to reduce and block free cysteine residues. Prechilled methanol, chloroform, and deionized H 2 O (4:1.5:3) were sequentially added to remove NEM and precipitate proteins, and this step was performed 3 times. Proteins were dissolved in 1 × TEA lysis buffer supplemented with 5 mM EDTA and 4% SDS and split into 2 tubes treated with or without 1 M hydroxylamine (HAM; Sigma-Aldrich) in 0.2% Triton X-100/TEA buffer for 1 hour at RT with nutation to remove S-fatty acid groups, followed by another 3 repeats of methanol-chloroform-H 2 O precipitation. The exposed cysteines were incubated with 1.33 mM 10 kDa mPEG-Mal (Sigma-Aldrich) for 2 hours at RT. Subsequently, proteins were subjected to methanol-chloroform-H 2 O precipitation once and boiled in SDS loading buffer for WB analysis.
Real-time quantitative PCR. Total RNA was isolated using the RNeasy Plus Mini Kit (#74134, QIAGEN). Reverse-transcription reactions were performed to synthesize cDNA using qScript cDNA Supermix (#95047, Quanta Biosciences). Real-time quantitative PCR (qRT-PCR) were performed using SYBR Green Master Mix (Applied Biosystems) in a ViiA 7 real-time PCR system (Applied Biosystems). The qRT-PCR primers used are as follows: RAB27B-preRNA-F: no. 1-CTGGAATAAGAGCAGT-CATTTGACATC, no. Purification and quantification of exosomes. 2 × 10 6 cells were cultured in 6 mL RPMI plus 10% exosome-depleted serum (Gibco). 1 mL of the media was collected after 1 hour, 3 hours, and 5 hours of culture. For purification of exosomes, the collected media was differentially centrifuged at 2,000g for 10 minutes then at 16,000g for 1 hour at 4°C to remove cells, debris, and bigger vesicles. Subsequently, the supernatant was loaded with a 1 mL syringe into a NanoSight NS300 (Malvern Instruments) for quantification of exosomes.
Immunofluorescence and live-cell imaging. TF-1 DKO cells were infected with pCL20-GFP-NRAS along with retroviruses expressing shLuc or shRAB27B. After 2 days, cells were incubated with Alexa Fluor 647-conjugated wheat germ agglutinin (WGA) for 10 minutes at 37 o C to stain the PM. Subsequently, the cells were washed with cold PBS, fixed in 4% PFA