N-Arachidonoyl Dopamine Inhibits NRAS Neoplastic Transformation by Suppressing Its Plasma Membrane Translocation

RAS proteins are small GTPases that act as molecular switches, transducing signals from many activated receptors that regulate cell proliferation, survival, and differentiation (1). The mammalian RAS include four highly homologous proteins: H-, N-, KRAS4A, and 4B, the latter two being alternatively spliced forms (2). Oncogenic mutations of RAS genes are common in human cancers (1, 3). However, because the enzymatic activity of RAS is used to turn itself off and its GTP binding affinity is very high, RAS proteins have been difficult to target (4). Despite intensive research over three decades, cancers harboring RAS mutations remain the most difficult to treat and are refractory to current targeted therapies (5). Identification of alternative targets that block RAS signaling is critical to develop therapies for RAS mutation-driven cancer.

The biological activity of RAS proteins relies upon their proper cellular membrane localization (6). To translocate to membranes, RAS need to be modified by prenylation first, which is the addition of an isoprenoid lipid to the cysteine residue of the COOH-terminal CAAX (C, cysteine; A, aliphatic amino acid; X, undetermined amino acid) motif, followed by removal of the -AAX tripeptide and methylation on endoplasmic reticulum (ER; ref. 7). Afterward, the prenylated RAS binds to phosphodiesterase δ subunit (PDEδ), facilitating the traffic and accumulation of prenylated proteins on the cytoplasmic face of ER (8). Though prenylation of RAS by Farnesyl Protein Transferase (FTase) is the obligate first step in the RAS post-translational modifications (PTM) and it has been shown to be essential for RAS membrane association and neoplastic transformation (9), successes of therapies targeting FTase are modest (10). The main reason is that in the presence of FTase inhibitors (FTI), RAS protein can be alternatively prenylated by geranylgeranyltransferase (GGTase) (11). Inhibitors targeting FTase and GGTase in combination have been proved too toxic to be clinically useful (12). Because the strategy of using FTIs to target RAS was not as successful as initially hoped, the interest has been turned to post-prenylation events as possible anticancer targets.

The prenylation of RAS proteins provides the minimal signal for their membrane association. Though RAS signaling on Golgi and ER has been detected, the efficiency is much reduced than the plasma membrane-localized RAS protein (13). The polybasic lysine tract of KRAS4B provides a second signal for its transit directly from the ER to the plasma membrane that bypasses the Golgi (14). The polybasic region of KRAS4B has been shown to be critical for its plasma membrane targeting and functional activity (15). NRAS, HRAS, and KRAS4A are further palmitoylated on the Golgi, increasing their affinity to bind the plasma membrane (16, 17). We have shown previously that interrupting the plasma membrane localization of NRAS and KRAS4A could significantly abrogate their leukemogenic potential (18–20). However, the exact mechanism of RAS proteins translocating from endomembranes to the plasma membrane is still elusive (17). Discovering small molecules interfering RAS plasma membrane localization would help to illustrate this essential process and discover novel therapeutic strategies for RAS-related cancer.

Because the biological activity of RAS proteins relies upon lipid modifications and RAS regulates lipid metabolisms in cancer cells (21), we screened a bioactive lipid library with a RAS-specific cell viability assay for potential therapeutics against RAS-driven cancer. We found that N-arachidonoyl dopamine (NADA) inhibits the NRAS oncogenic transformation by suppressing its plasma membrane translocation.

Retroviral constructs expressing NRAS-G12D, KRAS4A-G12D, KRAS4B-G12D, and NRAS-G12D-C181S as N-terminal GFP-fusion proteins were generated as previously described (19, 22). The retroviruses were produced using Bosc23 cells and were titered as previously described (22).

The Bosc23, Ba/F3, WEHI-3, and NIH3T3 cell lines were obtained from ATCC in 2009 and cultured as previously described (23). The growth factor dependent Ba/F3 cells were cultured in RPMI 1640 (Gibco), including 10% fetal bovine serum (FBS, Gibco), and supplemented with 15% WEHI-3–conditioned medium as a source of IL3 or with recombinant IL3 (Roche) at the final concentration of 1 ng/mL. The Ba/F3 and NIH3T3 cells were transduced with retroviruses as described (22, 24). The HeLa cell line was bought from ATCC in 2009 and cultured according to the instruction. The HeLa cells expressing Golgi marker were prepared by transfecting TGOLN with C-tRFP tag for Golgi apparatus marking (Origene) using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer's instruction. The leukemia cell lines HL-60, THP-1, and KG-1 were obtained from ATCC in 2011 and cultured according to the instruction. In short, HL-60 cells were cultured in IMDM (Hyclone) with 20% FBS. THP-1 cells were cultured in RPMI 1640 plus 10% FBS and 50 μmol/L 2-mercaptoethanol (Sigma Aldrich). NOMO-1, NB4, and MOLM-13 were from DSMZ in 2011. KG-1 was cultured according to the manufacture's instruction. NB4 and NOMO-1 were cultured in RPMI 1640 with 10% FBS, while MOLM-13 was cultured in RPMI 1640 with 15% FBS. All media were supplemented with penicillin (100 U/mL) and streptomycin (100 mg/mL). All of the cell lines were authenticated by ATCC and DSMZ using short tandem repeat profiling analysis. The cells were expanded and frozen at low passage (<3) after the receipt of the original stock. The frozen cells were thawed and used in this study with the cell passages limiting to 15 for all the experimental procedures. All of the cells were detected periodically to ensure mycoplasma free and maintained in normal condition.

The bioactive lipid library containing 202 compounds was obtained from the National Compound Resource Center (http://www.chemicallibrary.org.cn/11115.html). N-Arachidonoyl dopamine (Cayman), Palmitoyl dopamine (Cayman), Oleoyl dopamine (Cayman), N-arachidonoyl ethanolamine (Cayman), dopamine (Sigma Aldrich), PD0325901 (Selleck), staurosporine (Selleck), and 2-Bromopalmitate (Sigma Aldrich) were prepared and stored according to the manufacturer's instructions.

For the screen assay, Ba/F3-N cells were seeded into 96-well cell culture cluster plates at a density of 5,000 cells per well in 100 μL of culture medium. In primary screen, cells were then treated with compounds from Bioactive Compound Library at a concentration of 3 μmol/L. At the same time, PD0325901 (3 μmol/L) and an equal volume of solvent were included as controls in each plate. After 48 hours of incubation, cell viability was measured using CellTiter Glo (Promega) according to the manufacturer's instructions. The luminescence was detected using an Envision plate reader (PerkinElmer). Z-factor was calculated as described (25), and only data with Z-factor >0.6 were used in following analysis. In secondary screen, the compounds were diluted with 1640 medium in a 9-point, 3-fold dilution series and added according to volume of each well to achieve a 3-fold concentration gradient starting at 10 μmol/L (final solvent concentration, <1/1000). The same detection method and cell proliferation rate calculation equation were used as in the primary screen. The dose–response curve was fitted based on the relative survival cell percentage in nonlinear fitness (curve fit) using GraphPad Prism 6 software (http://www.graphpad.com/scientific-software/prism/). The software build-in analyses “nonlinear regression (curve fit)” and equation “log (inhibitor) vs. response-Variable slope” were used for the data analysis and IC₅₀ calculation.

Total RNAs from Ba/F3-N cells treated with vehicle or 1 μmol/L NADA for 12 hours were prepared, and, after quality control, subjected to the Affymetrix Mouse Genome 430 2.0 Array according to the manufacturer's instructions. Raw array signals were processed using the Robust Multi-Chip Average (RMA) method for background correction, normalization, and summarization. The normalized microarray data were submitted to the Gene Expression Omnibus (GEO) database with accession number GSE86541. All of the genes whose expression significantly changed in at least one comparison of genotype or drought treatment (P ≤ 0.05 and log₂ fold change ≥ 1 or log₂ fold change ≤ −1) were applied for further analysis. GSEA and gene set collection of KEGG pathways were used for enrichment analysis. The significance of the gene set was empirically determined by 1,000 gene set permutations.

Western blot analyses were performed as previously described (22). Primary antibodies used were: anti-RAS, anti-phospho ERK1/2, anti-ERK1/2, anti-phospho MEK, anti-MEK (Cell Signaling Technology); anti-NRAS, anti-KRAS4B (Santa Cruz Biotechnology); anti-β-Actin (Sigma Aldrich) and HRP-linked streptavidin (Sigma Aldrich); secondary antibodies were: Anti-mouse and anti-rabbit IgG secondary antibodies (Cell Signaling Technology). The western blot images were developed with a LAS4000 (Fujifilm). Each western blot shown is a representative of a minimum of 2 experiments.

NIH3T3 cells expressing GFP-NRAS-G12D, GFP-KRAS4A-G12D, and GFP-KRAS4B-G12D were treated with NADA or vehicle for 6 hours. After being fixed, stained with Hoechst dye (Sigma Aldrich), and mounted, the cells were visualized and photographed by a laser scanning confocal microscopy TCS SP8 (Leica). Ba/F3 cells expressing GFP-NRAS-G12D, GFP-KRAS4A-G12D, and GFP-KRAS4B-G12D and HeLa cells simultaneously expressing RFP-TGOLN and GFP-NRAS-G12D were seeded in cell culture plates with glass bottom (WHB company) and treated with 2 μmol/L NADA or vehicle control for 2 hours. The live cells were directly visualized and photographed by the confocal microscopy TCS SP8 (Leica).

The membrane proteins and cytoplasmic proteins were isolated using the Mem-PER Plus Membrane Protein Extraction Kit per the manufacturer's instruction (Thermo Scientific). Briefly, 1 × 10⁷ cells were harvested by centrifugation at 300×g for 5 minutes and washed 3 times with Cell Wash Solution. Cells were then suspended in 0.75 mL Permeabilization Buffer for 10 minutes at 4°C with constant mixing, followed by centrifugation at 16,000× g for 15 minutes. The supernatant containing cytoplasmic proteins were then isolated. The pellet was suspended in 0.5 mL Solubilization Buffer, and incubate for 30 minutes at 4°C. The supernatant containing membrane proteins were obtained after centrifugation at 16,000× g for 15 minutes.

GraphPad Prism 6 software and Student t test were used for statistical data analysis. Statistical significance threshold was set at P = 0.05.

To identify compounds that could selectively inhibit the proliferation of RAS-dependent cell lines, we established a differential cytotoxicity assay suitable for high-throughput screening. Ba/F3, a murine bone marrow–derived cell line depending on IL3 for viability and proliferation (26), has been successfully used in high-throughput assays for kinase drug discoveries (27). We first tested this system with oncogenic RAS and found that oncogenic NRAS (NRAS-G12D), which was the predominant oncogenic RAS isoform found in hematological malignancies (1, 3), could transform Ba/F3 cells into IL3 independence, while PD0325901, a selective inhibitor for the RAS downstream effector MEK, can inhibit the proliferation of NRAS-G12D–transformed Ba/F3 (Ba/F3-N) cells (Fig. 1A and B). This RAS transformation–specific cell viability assay is an ideal system to screen compounds blocking the RAS oncogenic signaling. Using this system, we screened a bioactive lipid library for inhibitors of RAS transformation, with PD0325901 and the solvent DMSO as positive and negative controls, respectively. NADA, palmitoyl dopamine (PDA) and oleoyl dopamine (ODA) were identified as the top hits that can effectively reduce proliferation and/or viability of Ba/F3-N cells (Fig. 1C). Interestingly, all of the three compounds belong to a family of lipid-soluble dopamine compounds, N-acyl-dopamines, condensated of long-chain fatty acids and dopamine at its NH₂-terminus (28). They are all naturally produced molecules sharing not only common structures but also similar biological characteristics, including function as neurotransmitters and as ligands of cannabinoid and vanilloid receptors (29, 30).

To further validate the antiproliferation potency and target specificity of these three compounds, half maximal inhibitory concentration (IC₅₀) of NADA, PDA, and ODA were profiled on Ba/F3-N with or without IL3 and the parental Ba/F3 cells in the presence of IL3. The IC₅₀ of NADA in inhibiting proliferation of Ba/F3-N cells in the absence of IL3 is approximately 5-fold lower than that of Ba/F3-N and Ba/F3 in the presence of IL3 (Fig. 1D). The IC₅₀ values of PDA and ODA in inhibiting proliferation of Ba/F3-N cells in the absence of IL3 are also lower than those of Ba/F3-N and Ba/F3 in the presence of IL3 (Supplementary Fig. S1A and S1B). We also measured the effect of NADA treatment by counting the cell numbers and confirmed that NADA indeed inhibits the cell proliferation of Ba/F3-N cells, rather than just affecting cellular metabolism (Supplementary Fig. S2A and S2B). These results indicate that NRAS-transformed cells are more susceptible for the growth-inhibitory effect of NADA, PDA, and ODA than control cells. Due to structural similarities and its relatively stronger activity in inhibiting NRAS transformation, we focus our further studies on NADA.

To explore the specificity of NADA on tumor cells harboring NRAS mutations, we evaluated the effects of NADA in six human myeloid leukemia cell lines with or without NRAS mutation (Fig. 1E). NRAS-mutated (NRAS-mt) cell lines THP-1 (NRAS-G12D) and HL-60 (NRAS-Q61L) are, in general, more susceptible to NADA's antitumor activity than that in KRAS-mutated (KRAS-mt) cell lines NB4 (KRAS-A18D), NOMO-1 (KRAS-G13D), as well as RAS wild-type (RAS-wt) cell lines KG-1 and MOLM-13. These results indicate that NRAS-driven leukemia cells are more sensitive to the NADA treatment.

NADA is the amide of dopamine (DA) and arachidonic acid (AA). It can be separated into the N-arachidonoyl ethanolamine (AEA) and DA parts with an ethyl amino overlap (Fig. 2A). In order to check the biological functional group of NADA, AEA and DA were tested for the antiproliferation activity in Ba/F3-N cells. We found that AEA, DA, or AEA and DA in combination did not suppress the growth of Ba/F3-N cells (Fig. 2B). These results indicate that the anti-RAS activity of NADA requires both the dopamine and arachidonoyl moieties in the same molecule.

Both NADA and AEA are endogenous ligands that activate the cannabinoid receptors (CB) and the transient receptor potential vanilloid type 1 (TRPV1) channel (31–33). Cannabinoids and CBs, including CB1 and CB2, have been studied extensively and have multiple functions. The activity of NADA and AEA in binding and activating the CB1 receptor is about 30 times more potent than that on the CB2 receptor (34). AEA does not inhibit NRAS transformation suggests that the anti-NRAS activity of NADA involves molecular events beyond engaging the CB1 or TRPV1 signaling. To test whether the CB2 receptor activation could inhibit the NRAS oncogenic pathway, we examined the anti-NRAS activity of WIN 55,212-2, a nonspecific CB receptor agonist (35), finding no growth suppress of Ba/F3-N cells (Fig. 2C). Together, these results suggest that the inhibition of NADA on NRAS transformation is independent of its ability to engage cannabinoid or vanilloid receptors.

To investigate the antiproliferation mechanism of NADA on Ba/F3-N cells, we checked the cell-cycle status and the viability of Ba/F3-N cells with or without the treatment of NADA. We found that NADA significantly reduces the number of Ba/F3-N cells in the S phase (Supplementary Fig. S3A and S3B). To examine whether NADA induces cell death, NADA-treated Ba/F3-N cells were stained with 7AAD and Annexin V-PE at different time points after the treatment. No apparent increase of Annexin V⁺ cells was detected within 6 hours after NADA treatment (data not shown). If NADA treatment was prolonged to 8 hours, more than 30% of Ba/F3-N cells turned into Annexin V⁺/7AAD⁺, and nearly 80% of the treated Ba/F3-N cells became Annexin V⁺/7AAD⁺ after 12 hours of NADA treatment (Fig. 3A). These results demonstrate that NADA does induce death of Ba/F3-N cells. However, unlike typical apoptosis, in which phosphatidylserine exposure (indicated by Annexin V staining) is an early event and appears ahead of the loss of plasma membrane integrity (indicated by 7AAD staining; ref. 36), most Annexin V⁺ signal co-appears with 7AAD staining in NADA-treated Ba/F3-N cells (Fig. 3A).

We then checked the morphology of NADA-treated Ba/F3-N cells. After 8-hour treatment of NADA, the cell size was significantly reduced, and at 12 hours, most of the cells appeared as small cell debris with patchy chromatin condensation (Fig. 3B). As a comparison, Ba/F3-N cells treated with staurosporine (STSP), a typical cell apoptosis inducer, displayed a different morphology (Fig. 3B). The latter data suggest that NADA induces cell death through a nonapoptotic pathway.

To further understand how does NADA induce death of NRAS transformed cells, we recorded morphological changes of NADA-treated cells with time-lapse light confocal microscopy. Figure 3C shows a typical death process of NADA-treated Ba/F3-N cells. Within the first half-hour after the NADA treatment, the active movements of Ba/F3-N cells were ceased and cells were gradually losing the pseudopodia, a feature of RAS transformation (37, 38), and rounding up (Fig. 3C, stage 1). About 7 hours later, cells underwent a series of rapid and dramatically morphological changes. These changes could be divided into 4 stages. First, the cells underwent limited blebbing and cells became smaller (Fig. 3C, stage 2). Then neurite-like structures grew out of the cell body (Fig. 3C, stage 3) and soon retracted after a few minutes (Fig. 3C, stage 4). Finally, a single blister, devoid of organelles, was formed (Fig. 3C, stage 5). Fluid inside of the blister was quickly leaked out, leaving crushed cell organelles and nucleus in the cell debris.

These characteristics of cell death induced by NADA are consistent with typical features of cell death termed oncosis (36, 39, 40). Like other cases of oncosis (41), cleavage of caspase-3 can be detected in NADA-treated Ba/F3-N cells (Supplementary Fig. S4A). Although the activation of caspase-3 was suppressed by pretreatment with the caspase inhibitor z-VAD-FMK for 2 hours (Supplementary Fig. S4B), the NADA-treated Ba/F3-N cells cannot be rescued by z-VAD-FMK. Taking together, these data demonstrate that NADA induces oncosis in Ba/F3-N cells.

Because KRAS mutations are common in human cancers and KRAS4B is the predominant spliced form, we tested the effect of NADA on KRAS4B-G12D–transformed Ba/F3 (Ba/F3-K4B) cells. Unlike Ba/F3-N cells (Fig. 1D), there is no significant difference between the values of IC₅₀ of NADA in Ba/F3- K4B cells in the absence of IL3 compared with that in the control Ba/F3 cells in the presence of IL3 (Supplementary Fig. S5A). Consistently, NADA does not possess more potent inhibitory effect in KRAS-mt leukemia cells NB4 and NOMO-1 compared with RAS-wt leukemia cell lines MOLM-13 and KG-1 (Fig. 1E). Using time-lapse light confocal microscopy, we observed that NADA also induces oncosis in Ba/F3-K4B cells, though it takes higher concentration and longer time than that in Ba/F3-N cells (data not shown). It is also shown that under the same condition, NADA induces significantly more Ba/F3-N cells to undergo oncosis (indicated by greatly reduced cell size) than that in Ba/F3-K4B cells (Supplementary Fig. S5B and S5C). These data suggest that Ba/F3-N cells are more susceptible to NADA treatment than Ba/F3-K4B cells.

To understand the mechanism by which Ba/F3-N cells are more susceptible to NADA treatment, we first analyzed the transcription profile of Ba/F3-N cells with or without NADA treatment. The transcriptome analyses show that the RAS pathway signature is altered dramatically after NADA treatment (Fig. 4A; Supplementary Data Set), suggesting that NADA interferes with the RAS signaling pathway in Ba/F3-N cells. The RAS oncogenic signaling is transduced through many downstream effectors, such as the RAF/MEK/ERK pathway (21, 42). As NADA can specifically inhibit cell transformation by oncogenic NRAS, but not by oncogenic KRAS4B, at certain concentrations (Fig. 1D; Supplementary Fig. S5A), we checked the effect of NADA on the RAF/MEK/ERK pathway in Ba/F3-N, Ba/F3-K4B, and parental Ba/F3 cells, respectively. We found that phosphorylation of both MEK and ERK was dramatically reduced in NADA-treated Ba/F3-N but not in Ba/F3-K4B, nor in Ba/F3 cells in the presence of IL3 (Fig. 4B; Supplementary Fig. S6A). As expected, the phosphorylated ERK and MEK levels are decreased in control Ba/F3 cells after removal of IL3 (Supplementary Fig. S6B). Interestingly, NADA's inhibitory effect starts within 30 minutes (Fig. 4B).

Next, we examined whether NADA also affects RAS signaling in human leukemia cells. It has been shown that survival and proliferation of NRAS-mutated cell lines THP-1 and HL-60 cells depend on the expression of NRAS (43). We found that phosphorylation levels of ERK and MEK were dramatically decreased in THP-1 and HL-60 cells shortly after NADA treatment. While in the KRAS-mutated (NB4 and NOMO-1) and RAS-wt (KG-1 and MOLM-13) leukemia cells, NADA did not significantly reduce the level of phosphorylated ERK and MEK (Fig. 4C). These data demonstrate that NADA can specifically and acutely inhibit oncogenic NRAS signaling.

One major difference between NRAS and KRAS4B is their membrane-targeting mechanism. To examine the effect of NADA on RAS membrane localization, we generated NIH3T3 cells transformed by GFP-NRAS-G12D and GFP-KRAS4B-G12D. We then checked the effect of NADA on RAS plasma membrane localization and found that GFP-NRAS-G12D is mis-localized to internal membranes after NADA treatment, while GFP-KRAS4B-G12D plasma membrane localization is not significantly affected by NADA (Fig. 5A). A similar difference was also observed in Ba/F3-N and Ba/F3-K4B cells (Fig. 5B). Consistent with the effect on NRAS signaling, the effect of NADA on inhibiting NRAS plasma membrane translocation occurred quickly after the addition of NADA (Supplementary Fig. S7). The effect of NADA to disrupt NRAS plasma membrane translocation is, therefore, not a result of cell death or non-specific toxicity, as the effect appears long before typical oncosis occurs.

KRAS4A also relies on palmitoylation for high-affinity plasma membrane binding (14, 16). We find that NADA inhibits plasma membrane translocation of GFP-KRAS4A-G12D as well (Fig. 5A and B). Consistently, NADA inhibits growth of KRAS4A-G12D-transformed Ba/F3 cells (Fig. 5C). These data demonstrate that NADA could selectively inhibit the plasma translocation of oncogenic NRAS and KRAS4A.

Next, we checked whether NADA caused the relocalization of NRAS proteins to the cytoplasm or endomembranes. We used a membrane protein extraction kit to separate membrane proteins and cytoplasmic proteins from Ba/F3-N cells. Because NADA, as well as ODA and PDA, all have a palmitoyl-like moiety, which may affect RAS palmitoylation, we include the Ba/F3 cell line expressing the palmitoylation-deficient NRAS-G12D-C181S mutant (Ba/F3-N-C181S) as a control. 2-Bromopalmitate (2-BP), a known general palmitoylation inhibitor (44), was also used as a control for NADA treatment. We found that NRAS-G12D was predominately detected in the membrane portion, whereas NRAS-G12D-C181S was mainly found in the cytoplasmic fraction due to the lack of palmitoylation (Fig. 6A and B). Interestingly, cytoplasmic NRAS-G12D dramatically decreased after the treatment of NADA for 30 minutes (Fig. 6A and C). In contrast, the distribution of NRAS-G12D-C181S was barely changed even after 2 hours of NADA treatment (Fig. 6B and C). The 2-BP treatment did not lead to significant redistribution of NRAS-G12D, nor NRAS-G12D-C181S under the same condition (Fig. 6A–C). These data indicate that NADA redistributes NRAS not only from the plasma membrane to endomembranes but also from the cytoplasm to endomembranes, and that it does so in a palmitoylation-dependent manner.

We also tested whether the same effect of NADA could be seen in human leukemia cells with endogenous RAS proteins, using NRAS- and KRAS4B-specific antibodies. We found that after NADA treatment, NRAS, but not KRAS4B, was redistributed to the endomembrane (Fig. 6D–F).

The fact that NADA redistributes NRAS from cytoplasm to endomembranes in a palmitoylation-dependent manner suggests that NADA does not exert its function through interring NRAS palmitoylation. To verify this notion, we evaluated the effect of NADA on NRAS palmitoylation using an acyl biotin exchange chemistry assay. We found that NADA does not significantly affect the levels of either palmitoylated NRAS or total palmitoylated proteins in Ba/F3-N cells (Supplementary Fig. S8). These results indicate that NADA inhibits the NRAS plasma membrane translocation through a post-palmitoylation event.

To further examine the effect of NADA on NRAS localization, we transduced GFP-NRAS-G12D into HeLa cells expressing the RFP-TGOLN protein as a Golgi marker. We found that after NADA treatment, the plasma membrane localized NRAS was gradually reduced with substantial accumulation in perinuclear structures colocalizing with the Golgi apparatus (Fig. 6G; Supplementary Fig. S9). These data indicate that NADA interrupts the trafficking of NRAS traffic from Golgi apparatus to plasma membrane and to the cytoplasm.

Oncogenic mutations of NRAS account for 12% of RAS mutations in human cancer (45). It has been shown that KRAS4A plays an essential role in the development of carcinogen-induced lung cancer in mice and that oncogenic KRAS4A is widely expressed in human cancers (46, 47). Suppression of palmitoylated RAS signaling is, therefore, likely to be an effective therapeutic strategy for a broad range of cancers. Here, we show that neural transmitter NADA can selectively inhibit the viability of oncogenic NRAS- and KRAS4A-transformed cells and the plasma membrane translocation of NRAS and KRAS4A. This class of chemicals may serve as a base for developing effective therapies for RAS-related cancers.

NADA is one of the endocannabinoids, which has been implicated in a growing number of physiological functions. N-acyl-dopamines, including PDA, ODA, and NADA, have also been to have anti-inflammatory effects (48). Besides, the antitumor activity of cannabinoids has been documented (29, 49). Here, we show for the first time that NADA has an anti-RAS transformation activity and that this activity relies on the core structure combining the dopamine and arachidonic acid moieties and is independent of its function through dopamine, cannabinoid, and vanilloid receptors (Fig. 2). The new anticancer activity and a preliminary structure–activity relationship analysis of NADA would contribute to effective therapy development for RAS-driven cancers.

Oncotic cell death involves progressive membrane injury (39). It is possible that NADA affects transportation of molecules that regulate cell osmolarity. With extremely high dose, NADA might induce oncosis in all cells. However, the different susceptibility of NRAS- and KRAS4A-transformed cells versus that of KRAS4B-transformed cells suggests that RAS signaling could inhibit/delay the process of oncosis. More importantly, our data reveal that a therapeutic window might exist for modulating NADA target in treating cancers driven by oncogenic NRAS and KRAS4A.

It is interesting that NADA could significantly redistribute the cytosolic and plasma membrane NRAS to the inner membrane system not only in the NRAS-mutated cell lines but also in the KRAS-mutated cell lines (Fig. 6D and E). This should happen because the localization of RAS is not dependent on their GTPase activities. However, because NRAS in the latter cell lines is the wild-type form and survival of these cells relies on mutated KRAS signaling, the NADA treatment does not affect their proliferation and downstream RAS signaling (Figs. 1E and 4C). These data further support the oncogenic NRAS selective therapeutic effect of NADA.

It is generally believed that palmitoylated RAS proteins traffic to the plasma membrane through the classical exocytic pathway across the Golgi apparatus (17, 50). However, exactly how palmitoylation operates as a trafficking signal is unclear (6). Our data show that NADA blocks the plasma membrane translocation of NRAS without inhibiting its palmitoylation. Of particular interest is the fact that NADA also affects the translocation of NRAS between cytoplasm and the Golgi apparatus in a RAS palmitoylation-dependent manner. Together, these data suggest that NADA may target a novel molecular process involved in the regulation of RAS intracellular trafficking. As such, NADA may serve as a molecular probe to illustrate the mechanism by which palmitoylated RAS proteins are translocated to the plasma membrane, as well as the dynamics of translocation of NRAS between the cytoplasm and Golgi apparatus. This effect of NADA on NRAS and KRAS4A trafficking may also be used as a start point to identify novel therapeutics for RAS-driven cancers.

No potential conflicts of interest were disclosed.

Conception and design: M. Wu, J. Zhang, B. Jiao, R. Ren

Development of methodology: M. Wu, B. Jiao, R. Ren

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Wu, R. Ren

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Wu, J. Huang, J. Zhang, B. Jiao

Writing, review, and/or revision of the manuscript: M. Wu, J. Zhang, C. Benes, B. Jiao, R. Ren

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Jiao

Study supervision: R. Ren

The authors thank Xianming Deng, Ping Liu, Huanbin Zhao, Zheng Ruan, and Donghe Li for their technical assistance and project discussion.

This work was supported by the National Natural Science Foundation of China (81230055 to R. Ren; 81300402 to B. Jiao), Shanghai Outstanding Academic Leader Program (12XD1403500 to R. Ren), and the Doctoral Innovation Fund from Shanghai Jiao Tong University School of Medicine to M. Wu.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.