SIRT2 and lysine fatty acylation regulate the oncogenic activity of K-Ras4a

Ras proteins play vital roles in numerous biological processes and Ras mutations are found in many human tumors. Understanding how Ras proteins are regulated is important for elucidating cell signaling pathways and identifying new targets for treating human diseases. Here we report that one of the K-Ras splice variants, K-Ras4a, is subject to lysine fatty acylation, a previously under-studied protein post-translational modification. Sirtuin 2 (SIRT2), one of the mammalian nicotinamide adenine dinucleotide (NAD)-dependent lysine deacylases, catalyzes the removal of fatty acylation from K-Ras4a. We further demonstrate that SIRT2-mediated lysine defatty-acylation promotes endomembrane localization of K-Ras4a, enhances its interaction with A-Raf, and thus promotes cellular transformation. Our study identifies lysine fatty acylation as a previously unknown regulatory mechanism for the Ras family of GTPases that is distinct from cysteine fatty acylation. These findings highlight the biological significance of lysine fatty acylation and sirtuin-catalyzed protein lysine defatty-acylation.

, which agrees with our hypothesis. It was likely that K185 could 231 also be fatty acylated for the K182R mutant, because the K182/185R mutant slightly but significantly 232 decreased lysine fatty acylation levels compared with the K182R and K185R single mutants (Fig. 3a). 233 The K185 fatty acylation was not detected by MS most likely because the modified tryptic peptide was 234 too short and hydrophobic (K fatty-acyl C prenyl, oMe ). Overall, these data indicate that K182/184/185 are fatty 235 acylated redundantly and that the 3KR mutation is needed to abolish lysine fatty acylation on the C-236 terminus of K-Ras4a.

246
Quantification of the fluorescent intensity relative to K-Ras4a WT is shown in the bottom panel. Statistical evaluation was by 247 unpaired two-tailed Student's t test. Error bars represent SEM in three biological replicates. *P < 0.05; **P < 0.01; ***P < 248 0.001. Representative images from three independent experiments are shown.

250
K-Ras4a has been shown to be prenylated on cysteine 186 and palmitoylated on cysteine 180 19 . To 251 examine whether cysteine prenylation or palmitoylation play a role in lysine fatty acylation, we 252 generated cysteine-to-serine C180S and C186S mutants. Mutation of the prenylcysteine (C186S) 253 completely abolished the fatty acylation of K-Ras4a (Fig. 3c), which is consistent with the model that 254 prenylation of the cysteine on the CaaX motif of Ras proteins is required for the subsequent fatty 255 acylation 37 . On the other hand, mutation of the palmitoylated cysteine (C180S) led to a substantial, but 256 not a complete, loss of K-Ras4a lysine fatty acylation (Fig. 3c). The fatty acylation on the C180S mutant 257 was NH 2 OH-resistant and was abolished by combining the C180S and 3KR mutations (Fig. 3d), 258 implying that the C180S mutant was fatty acylated on K182/184/185. These data suggest that cysteine 259 palmitoylation might play an important but nonessential role in the occurrence of lysine fatty acylation. 260 It is possible that cysteine palmitoylation facilitates the lysine fatty acyl transfer reaction, or the delivery 261 of K-Ras4a to where lysine fatty acylation occurs. 262 We next assessed whether SIRT2 regulates fatty acylation of K-Ras4a on K182/184/185. SIRT2 263 removed lysine fatty acylation from K-Ras4a WT, the 4KR mutant and the C180S mutant, but not the 264 3KR mutant in vitro ( Fig. S4a & b). SIRT2 KD in HEK293T cells increased lysine fatty acylation of K-265 Ras4a WT and the C180S mutant, but not the 3KR and C180S-3KR mutants (Fig. 3e), indicating that 266 fatty acylation on K182/184/185 is regulated by SIRT2. 267 268 Lysine fatty acylation regulates subcellular localization of K- Ras4a 269 We next set out to study the effect of lysine fatty acylation on K-Ras4a. A variety of PTMs on Ras 270 proteins, such as cysteine palmitoylation 38,39 , phosphorylation 40,41 and ubiquitination 42 , function to 271 deliver the molecule to the right place within the cell. We hypothesized that lysine fatty acylation may 272 also be critical for the correct subcellular distribution of K-Ras4a. To test this hypothesis, we fused 273 Aequorea coerulescens Green Fluorescent Protein (GFP) to the N-terminus of K-Ras4a WT and the 274 3KR mutant and performed live imaging with confocal microscopy in Ctrl and SIRT2 KD HEK293T 275 cells to visualize K-Ras4a localization. The levels of over-expressed K-Ras4a WT and 3KR were equal 276 in Ctrl and SIRT2 KD cells (Fig. S5a). We also imaged cells with similar GFP intensity under the same 277 settings to avoid potential false positive observations caused by different levels of expression. In Ctrl 278 KD cells, both K-Ras4a WT and the 3KR mutant displayed predominant localization to the plasma 279 membrane (PM). However, the presence of 3KR on intracellular puncta was noticeably more 280 pronounced compared to WT. SIRT2 KD decreased the intracellular punctate-localized K-Ras4a WT 281 compared to Ctrl KD, whereas it had no effect on the punctate localization of the 3KR mutant ( Fig. 4a &  282 b), indicating that the effect of the 3KR mutation on K-Ras4a localization was due to lack of lysine fatty 283 acylation. Similar effects of the 3KR mutation were obtained for K-Ras4a in HCT116 cells and for 284 oncogenic K-Ras4a-G12V, which exhibited a comparable lysine fatty acylation level to K-Ras4a WT 285 S5c & d). On the other hand, SIRT2 KD did not affect the intracellular 286 punctate localization of H-Ras ( Fig. 4a & b), which is consistent with our observation that H-Ras was 287 not regulated by SIRT2 through lysine defatty-acylation. Taken together, these data indicate that lysine 288 fatty acylation inhibits the intracellular punctate localization of K-Ras4a and SIRT2 promotes this 289 localization by defatty-acylation. In addition, the K-Ras4a-C180S mutant that lacks cysteine 290 palmitoylation and the majority of lysine fatty acylation extensively localized to internal membranes, 291 which was distinct from the punctate localization of the 3KR mutant that is deficient in lysine fatty 292 acylation but retains cysteine palmitoylation ( Fig. S5e & f, Movie S1, 2, 3). This implies that cysteine 293 palmitoylation might facilitate the punctate localization of K-Ras4a in the absence of lysine fatty 294 acylation, while lysine fatty acylation inhibits it. 295 It has been shown that Ras proteins associate with and signal from endomembrane compartments, 296 including the endoplasmic reticulum (ER), Golgi, endosomes and lysosome 18,42-48 . Therefore, we next 297 set out to identify the endomembrane compartments where lysine defatty-acylated K-Ras4a is localized. 298 We performed colocalization analyses with a series of membrane compartment markers. Compared with 299 K-Ras4a WT, the 3KR mutant exhibited more pronounced cytoplasmic colocalization with trans-Golgi 300 network (TGN) marker STX6, early endosome marker EEA1, recycling endosome marker Rab11, and 301 lysosome marker LAMP1 (Fig. 4c & d), but not with the ER marker Sec61, trans-Golgi marker GalT 302 and late endosome marker Rab7 (Fig. S5g & h). Time-lapse confocal imaging also revealed that K-303 Ras4a-3KR displayed more internalization from the plasma membrane into punctate structures than did 304 the WT (Supplemental Movie S1 & 2). These results suggest that removal of lysine fatty acylation from 305 K-Ras4a promotes its localization to endomembranes in endocytic pathways, by which it may be routed 306 from early endosome to the lysosome for degradation and to the TGN or recycling endosomes to return 307 to the plasma membrane 49 . 308 309 Pearson's coefficient (n = 11, 11, 11, 11, 17, 17, 10, 10 cells for each sample from left to right, respectively). Statistical 317 evaluation was by two-way ANOVA. Centre line of the box plot represents the mean value, box represents the 95 % 318 confidence interval, and whiskers represent the range of the values. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant.

319
Representative images are shown.

355
Representative images (a, b) from at least three independent experiments are shown.

357
A-Raf is involved in the regulation of K-Ras4a by lysine fatty acylation 358 The dynamic regulation of Ras localization is known to be closely coupled to its signaling output 359 43,59,60 . We decided to further explore the molecular mechanism underlying the regulation of K-Ras4a-360 mediated transformation by lysine fatty acylation. We first sought to examine whether lysine fatty 361 acylation affects K-Ras4a activation by a pull-down assay with the Ras-binding domain (RBD) of Raf1, 362 which only binds to the GTP-bound form of Ras 43 . Neither the 3KR mutation nor SIRT2 KD affected 363 EGF-stimulated GTP loading of K-Ras4a ( Fig. S7a & b) or the constitutively GTP-loaded state of K-364 Ras4a-G12V ( Fig. S7c & d). We then determined whether K-Ras4a at endomembranes exists in its 365 GTP-bound state using DsRed-RBD (DsRed fused to the N-terminus of RBD) as a probe. Notably, we 366 observed more colocalization of DsRed-RBD with K-Ras4a on intracellular puncta in cells expressing 367 the 3KR mutant ( Fig. S7e, f, g) than in cells expressing K-Ras4a WT. Furthermore, SIRT2 KD 368 decreased the colocalization of DsRed-RBD with K-Ras4a WT at intracellular puncta, but not with the 369 3KR mutant (Fig. S7e). 370 The results above suggest that SIRT2-dependent lysine defatty-acylation may promote the 371 localization of activated (GTP-loaded) K-Ras4a at endomembranes, which raises the possibility that 372 lysine defatty-acylation may alter the signaling specificity of K-Ras4a by recruiting different effector 373 proteins to endomembranes. We therefore investigated whether lysine defatty-acylation influenced the 374 binding and activation of the three most well characterized Ras effectors: Raf1, PI3K, and RalGDS 61 . 375 Co-immunoprecipitation demonstrated that neither the 3KR mutation nor Sirt2 KD altered the binding 376 of K-Ras4a-G12V with Raf1, PI3K, or RalGDS (Fig. S8a). We also assessed the capacity of K-Ras4a-377 G12V and -G12V-3KR in Ctrl and Sirt2 KD cells to activate Raf1, PI3K, and RalGDS signaling 378 pathways using phosphorylated Erk, phosphorylated Akt, and phosphorylated Jnk as reporters, 379 respectively. K-Ras4a-G12V and -G12V-3KR induced comparable levels of Erk activation, which was 380 not affected by Sirt2 KD. Sirt2 KD resulted in a reduction of Akt and Jnk activation, but the effect was 381 similar for both K-Ras4a-G12V and -G12V-3KR (Fig. S8b), suggesting other Sirt2 targets are important 382 for Akt and Jnk activation. These results suggest that SIRT2 catalyzed lysine defatty-acylation of K-383 Ras4a does not affect the activation of Raf1, PI3K or RalGDS by K-Ras4a. 384 To identify proteins whose binding to K-Ras4a is regulated by lysine fatty acylation, we performed a 385 protein interactome study using stable isotope labeling by amino acids in cell culture (SILAC) (Fig.  386 S9a). We cultured NIH3T3 cells with stable K-Ras4a-G12V and K-Ras4a-G12V-3KR overexpression in 387 light-isotope-and heavy-isotope-labeled medium, respectively. We then performed FLAG IP, mixed the 388 eluted fractions from both IPs, digested with trypsin and analyzed by MS to identify proteins with 389 Heavy/Light (H/L) ratios > 1.3 or < 0.77, which were candidates that would potentially bind to K-390 Ras4a-G12V and K-Ras4a-G12V-3KR differently. The experiment was also repeated after swapping the 391 heavy and light SILAC labels. Additionally, to confirm that the effect of the 3KR mutation on the K-392 Ras4a-G12V interactome was due to the lack of lysine fatty acylation, we also examined the K-Ras4a-393 G12V interactome in Ctrl and Sirt2 KD cells with SILAC, which enabled the identification of proteins 394 (H/L > 1.3 or < 0.77) whose binding to K-Ras4a-G12V was regulated by Sirt2. Integration of the three 395 interactome experiments resulted in 175 interacting proteins with at least two unique peptides and H/L 396 ratio. Among them, nine proteins exhibited increased binding to K-Ras4a-G12V-3KR compared to K-397 Ras4a-G12V, and their interaction with K-Ras4a-G12V was inhibited by Sirt2 KD, suggesting that 398 lysine defatty-acylation enhanced K-Ras4a-G12V interaction with these proteins. On the other hand, one 399 protein showed decreased binding to K-Ras4a-G12V-3KR compared to K-Ras4a-G12V, and its 400 interaction with K-Ras4a-G12V was increased by Sirt2 KD, suggesting that lysine defatty-acylation 401 repressed K-Ras4a-G12V interaction with it ( Fig. S9b). 402 Among these 10 proteins, the serine/threonine-protein kinase A-Raf and Apoptosis-inducing factor 1 403 (Aif), whose interaction with K-Ras4a-G12V might be increased by lysine defatty-acylation, attracted 404 our attention. A-Raf is a member of the Raf family of serine/threonine-specific protein kinases, acts as a 405 Ras effector and plays an important role in apoptosis 62,63 and tumorigenesis 64-66 . In response to 406 apoptotic stimuli, Aif is released from the mitochondrial intermembrane space into the cytosol and 407 nucleus, where it functions as a proapoptotic factor 67 . Since suppression of apoptosis is linked to  induced transformation 68 , it is plausible that A-Raf and Aif are involved in the regulation of K-Ras4a 409 transformation activity by lysine fatty acylation. To test this hypothesis, we first validated the 410 interactome results by co-IP. Although more interaction of Aif with K-Ras4a-G12V-3KR was observed 411 than with K-Ras4a-G12V, Sirt2 KD did not affect the interaction of Aif with either K-Ras4a-G12V or 412 K-Ras4a-G12V-3KR (Fig. S9c & d) and was not investigated further. However, a greater interaction of 413 A-Raf with K-Ras4a-G12V-3KR was observed than with K-Ras4a-G12V, and Sirt2 KD significantly 414 decreased the interaction of A-Raf with K-Ras4a-G12V but not with K-Ras4a-G12V-3KR ( Fig. 6a & b). 415 Thus, we concluded that A-Raf was an effector protein of K-Ras4a that was regulated by lysine fatty 416 acylation and SIRT2. Unlike the effect of Sirt2 KD on K-Ras4a-G12V-A-Raf interaction, Sirt2 KD did 417 not alter H-Ras-G12V-A-Raf interaction ( Fig. 6a & b). As mentioned earlier, lysine fatty acylation did 418 not affect the binding between C-Raf (Raf1) and K-Ras4a-G12V (Fig. S8a). We also assessed the 419 interaction of K-Ras4a-G12V with another Raf family member, B-Raf. Co-IP indicated that neither 3KR 420 mutation nor Sirt2 KD altered the binding of B-Raf to K-Ras4a-G12V (Fig. S8c). These results 421 collectively demonstrate that removal of lysine fatty acylation from K-Ras4a by SIRT2 results in its 422 preferential association with A-Raf, but not B-Raf or C-Raf. 423 Our results suggest that SIRT2-mediated lysine defatty-acylation does not affect the magnitude of K-424 Ras4a activation but promotes endomembrane localization of active K-Ras4a. It has been reported that 425 the efficient activation of certain effector pathways by Ras is dependent on the entry of Ras to the 426 endosomal compartment 42,69 . Therefore, it is plausible that lysine defatty-acylation may facilitate the 427 endomembrane recruitment of A-Raf by K-Ras4a, thereby increasing K-Ras4a oncogenic activity. Live 428 cell imaging revealed that A-Raf colocalized with K-Ras4a-G12V at both the PM and endomembranes. 429 K-Ras4a-G12V-3KR showed more colocalization with A-Raf on the endomembranes than K-Ras4a-430 G12V did. Sirt2 KD inhibited the endomembrane recruitment of A-Raf by K-Ras4a-G12V but not that 431 by K-Ras4a-G12V-3KR (Fig. 6c & d). These results are in line with our hypothesis. Thus, it is likely 432 that by regulating endomembrane recruitment of A-Raf, K-Ras4a lysine fatty acylation may alter its 433 signaling output through A-Raf, thereby modulating its transforming activity.  Protein lysine fatty acylation was discovered over two decades ago 4-8 . However, very little is known 462 about its functional significance. Our current study furnishes a model where K-Ras4a is fatty acylated on 463 lysine residues at its C-terminal HVR, and the removal of lysine fatty acylation by SIRT2 facilitates its 464 endomembrane localization and interaction with A-Raf, thus enhancing its transforming activity (Fig.  465   S11). These findings demonstrate that a Ras protein is modified and regulated by a previously under-466 appreciated PTM, lysine fatty acylation, which expands not only the regulatory scheme for Ras proteins,467 but also the biological significance of lysine fatty acylation. Moreover, our study reveals the first lysine 468 defatty-acylation substrate for SIRT2 and uncovers the physiological relevance of SIRT2 as a lysine 469 defatty-acylase 12,14,15 . 470 We found that H-Ras and K-Ras4a possess lysine fatty acylation that could be hydrolyzed by sirtuins 471 in vitro or in cells (Fig. 1, 2 & S2a). Although our attempt to detect N-Ras lysine fatty acylation by MS 472 was not successful, the N-Ras-K169/170R (2KR) mutant presented decreased NH 2 OH-resisitant fatty 473 acylation compared with WT, suggesting that N-Ras might be lysine fatty acylated (Fig. 1c & S1b). Therefore, lysine fatty acylation of Ras proteins might have been missed based on the mutagenesis 481 results. Similar to, but slightly different from these previous reports, we found that mutating the 482 palmitoylated cysteine of K-Ras4a decreased lysine fatty acylation by nearly 90% (Fig. 3c) but not 483 completely. A similar effect of the palmitoylated cysteine to serine mutation was also observed for R-484 GTPases contain lysine-rich sequences at their C-termini. It is therefore of great interest to us to 492 investigate whether lysine fatty acylation could act as a general regulatory mechanism for many Ras-493 related small GTPases. 494 The discovery of lysine fatty acylation on K-Ras4a raises the question of the relative abundance of 495 lysine versus cysteine fatty acylation. Semi-quantification of the fluorescence intensity from Alk14 496 labeling results enables us to roughly estimate the stoichiometry of lysine fatty acylation. Based on this, 497 K-Ras4a exhibits nearly 50% of lysine fatty acylation relative to total fatty acylation (Fig. 1c). Therefore, 498 the ratio of cysteine palmitoylation to lysine fatty acylation may be close to 1:1 on K-Ras4a. The 3KR 499 mutation decreased K-Ras4a lysine fatty acylation by about 50% (Fig. 1f & 3e), suggesting that the C-500 terminal lysine fatty acylation regulated by SIRT2 is around 50% of the lysine fatty acylation and 25% 501 of the total fatty acylation. Regarding endogenous K-Ras4a, by quantifying the K-Ras4a western blot 502 signal from the streptavidin beads and supernatant in Fig. 2i, we estimated that about 28% and 50% of 503 the fatty acylated K-Ras4a is lysine fatty acylated in Ctrl KD and SIRT2 KD HCT116 cells, respectively. 504 Unfortunately, precise quantitation of protein fatty acylation still remains a significant unsolved 505 challenge and we could not determine the ratio of fatty acylated versus unmodified K-Ras4a. 506 To study the physiological function of K-Ras4a lysine fatty acylation, we utilized the K-Ras4a-3KR 507 mutant in combination with SIRT2 KD. The lysine-to-arginine mutant maintains the positive charge of 508 the polybasic patch, which makes it a good lysine fatty acylation-deficient mimic. Recently, Zhou et al. 509 reported that lysine and arginine residues are not equivalent in determining the membrane lipid binding 510 specificity of K-Ras4b C-terminus, which raises the possibility that the effect of 3KR mutation might 511 not be solely due to lack of lysine fatty acylation 75 . Likewise, changes in the SIRT2 KD cells could be 512 mediated through other substrates for SIRT2. Therefore, it is critical to employ both the 3KR mutant and 513 SIRT2 KD to rule out these possibilities. SIRT2 KD enhances the lysine fatty acylation of K-Ras4a WT 514 but not the 3KR mutant. Thus, if a biological effect is due to lysine fatty acylation, SIRT2 should have a 515 greater impact on the effect of K-Ras4a WT than that of the 3KR mutant. Of note, H-Ras, which shares 516 similar properties with K-Ras4a, but is not a lysine defatty-acylase target for SIRT2, also serves as a 517 good control for the effect of SIRT2 KD. Indeed, SIRT2 KD repressed the endomembrane localization, 518 transforming activity, and A-Raf binding of K-Ras4a WT more than that of K-Ras4a 3KR or H-Ras, 519 indicating that SIRT2-dependent lysine defatty-acylation facilitates endomembrane localization of K-520 Ras4a, enhances its interaction with A-Raf, and thus promotes cellular transformation. Tsai et al. 19 and we observed that the K-Ras4a-C180S mutant, which possesses no cysteine 525 palmitoylation and little lysine fatty acylation, localizes to ER/Golgi-like internal membranes (Fig. S5e  526 & f). Differently, we found that removal of the lysine fatty acylation by SIRT2, which results in K-527 Ras4a with only cysteine palmitoylation, promotes endomembrane localization of K-Ras4a (Fig. 4a). 528 This evidence supports the model that cysteine palmitoylation enables K-Ras4a to undergo vesicular 529 transport, whereas lysine fatty acylation blocks K-Ras4a translocation from the PM to endomembranes 530 (Fig. S11). Furthermore, the C180S mutant suppressed K-Ras4a-G12V-mediated anchorage-531 independent growth and activation of MAPK signaling 19,20 . In contrast, the 3KR mutant increased K-532 Ras4a-G12V-mediated anchorage-independent growth (Fig. 5b & c), exhibited no effect on MAPK 533 signaling (Fig. S8), but activated A-Raf instead (Fig. 6a). Considering the possibility that lysine fatty 534 acylation largely relies on cysteine palmitoylation to occur, it is likely that the reversible lysine fatty 535 acylation adds a layer of regulation for K-Ras4a above that of cysteine palmitoylation. Raf interactions. Based on these previous studies, it is likely that the C-terminal PTM of K-Ras4a 545 regulates its interaction with Raf isozymes and that lysine fatty acylation may inhibit the binding of K-546 Ras4a to A-Raf but not to B-Raf and C-Raf. 547 Mutations that activate Ras are found in about 30% of all human tumors screened. KRAS mutations, 548 which affects both K-Ras4a and K-Ras4b, occur most frequently, accounting for 86% of RAS-driven 549 cancers 82 . Though K-Ras4a is homologous to the transforming cDNA identified in Kirsten rat sarcoma 550 virus 83 , its function and regulation is less characterized compared to K-Ras4b. Recent studies showed 551 that K-Ras4a is widely expressed in human cancers, suggesting that K-Ras4a plays a significant role in 552 KRAS-driven tumors 19,20 . Our findings reveal that K-Ras4a is regulated by SIRT2-dependent lysine 553 defatty-acylation. Depletion of SIRT2 increased lysine fatty acylation and diminished oncogenic 554 transforming activity of K-Ras4a, suggesting that interference with K-Ras4a lysine fatty acylation could 555 be an approach to anti-K-Ras therapy. Promega. 32 P-NAD + was purchased from PerkinElmer. Saponin (S0019-25G) was from TCI America. 614 Sep-Pak C18 cartridge was purchased from Waters. 615 The pLKO.1-puro lentiviral shRNAs constructs for luciferase and human SIRT1, SIRT2, and mouse 616 For SILAC experiments, 'light' NIH3T3 cells were maintained in DMEM media for SILAC (88420, 649 Thermo Fisher Scientific) supplemented with 100 mg/L [ 12 C 6 , 14 N 2 ]-L-lysine, 100 mg/L [ 12 C 6 , 14 N 4 ]-L-650 arginine, non-essential amino acids, and 15% dialyzed FBS (26400036, Thermo Fisher Scientific); 651 'heavy' NIH3T3 cells were cultured in DMEM media for SILAC supplemented with 100 mg/L [ 13 C 6 , 652 15 N 2 ]-L-lysine, 100 mg/L [ 13 C 6 , 15 N 4 ]-L-arginine, non-essential amino acids, and 15% dialyzed FBS. 653 Cells were cultured in SILAC media for at least six doubling times to achieve maximum incorporation 654 of 'labeled' amino acids into proteins before the interactome study was performed. stopped by washing the affinity gel using 1 mL of IP washing buffer for 3 times. On-bead click 695 chemistry and in-gel fluorescence were carried out as described above. 696 High-performance liquid chromatography (HPLC)-based SIRT2 activity assay. SIRT2 or 697 SIRT2-H187Y (1 μM) was incubated in 60 μL of reaction buffer (20 mM Tris, pH 8.0, 1 mM DTT, 1 698 mM NAD) with 32 μM acetyl H3K9, myristoyl H3K9, or myristoyl K-Ras4a-C180 peptides (Fig. S12), 699 respectively, at 37°C for 10 min (deacetylation) or 20 min (demyristoylation). Reactions were quenched 700 with 60 μL ice-cold acetonitrile and spun down at 18,000 g for 10 min to remove the precipitated protein. The supernatant was then analyzed by HPLC on a Kinetex XB-C18 column (100 A, 75 mm × 4.6 mm, 702 2.6 μm, Phenomenex). The peak areas were integrated and the conversion rate was calculated from the 703 ratio of the free H3K9 peptide peak area over the total peak areas of the substrate and product peptides. experiments, total cell lysates (2 mg of total protein for detecting SIRT1/2, 50 μg for A-Raf and c-Raf, 1 743 mg for B-Raf, p110α and RalGDS, determined by Bradford assay) were incubated with 10 μL 744 suspension of anti-FLAG M2 affinity gel for 2 h at 4 °C. The resulting affinity gel was washed three 745 times with 1 mL IP washing buffer and heated in protein loading buffer (2 × final concentration) at 746 95 °C for 10 min. Western blot was then performed to detect levels of the indicated proteins. 747 Detection of lysine fatty acylation on K-Ras4a using the 32 P-NAD assay. The 32 P-NAD assays 748 were carried out as described previously with minor modification 30 . HEK293T cells were transfected 749 with empty pCMV5 vector or pCMV5-K-RAS4A overnight and lysed in 1% NP-40 lysis buffer with 750 protease inhibitor cocktail. For each reaction, cell lysates (3 mg of total protein, determined by Bradford 751 assay) were incubated with 10 μL suspension of anti-FLAG M2 affinity gel for 2 h at 4 °C. The affinity 752 gel was washed three times with 1 mL of IP washing buffer. The resulting anti-FLAG affinity gel or the 753 synthetic acetyl and myristoyl H3K9 peptides 30 (25 M, positive control) were mixed with 10 μL 754 solutions containing 1 μCi 32 P-NAD, 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM DTT. The 755 reactions were incubated with 1 μM BSA (negative control), SIRT2, or SIRT2-H187Y at 37°C for 30 756 min. A total of 2 μL of each reaction were spotted onto silica gel TLC plates and developed with 7:3 757 ethanol:ammonium bicarbonate (1 M aqueous solution). After development, the plates were air-dried 758 and exposed to a PhosphorImaging screen (GE Healthcare). The signal was detected using Typhoon 759 9400 Variable Mode Imager (GE Healthcare). 760 Biotin pull-down of lysine fatty acylated endogenous K-Ras4a. The assay was carried out as 761 previously described with some modifications 31 . Briefly, HCT116 cells were infected with lentivirus 762 carrying luciferase (Ctrl) or SIRT2 shRNA for 3 days and treated without or with Alk14 (50 μM) for 6 h 763 before being collected. Total proteins were then extracted using 1% NP-40 lysis buffer with protease 764 inhibitor cocktail. 10 mg of total protein extract was subjected to click reaction with 100 μM Biotin-N 3 , 765 500 μM TBTA, 1 mM CuSO 4 and 1 mM TCEP in a final volume of 5 mL. The reaction was allowed to 766 proceed at room temperature for 1 h. Proteins were precipitated by adding 4 volumes of ice-cold 767 methanol, 3 volumes of water, and 1.5 volumes of chloroform. Precipitated proteins were pelleted by 768 centrifugation (4,500× g, 20 min, 4 °C), washed twice with 50 mL of ice-cold methanol and air-dried. 769 The protein pellet was suspended in 4% SDS buffer (4% SDS, 50 mM triethanolamine pH 7.4, and 150 770 mM NaCl, 10 mM EDTA). The solubilized protein mixture was diluted to 1% SDS with 1% Brij 97 (in 771 50 mM triethanolamine pH 7.4, and 150 mM NaCl) and incubated with streptavidin agarose (0.2 ml 772 slurry for 1 mg of protein) for 1 h at room temperature. The streptavidin beads were washed three times 773 with 10 mL of 1% SDS in PBS buffer. The streptavidin beads were incubated with 1 M NH 2 OH (pH For live cell imaging, cells were incubated in the Live Cell Imaging Solution (A14291DJ, Thermo 783 Fisher Scientific) and imaged with a Zeiss 880 confocal/multiphoton inverted microscope (Carl Zeiss 784 MicroImaging, Inc., Thornwood, NY) in a humidified metabolic chamber maintained at 37 °C and 5% 785 CO 2 . For time-lapse movies, 60 single section images were recorded at 1 sec intervals for 1 min. 786 For immunofluorescence, cells were rinsed with 1 × PBS twice and fixed with 4% paraformaldehyde 787 (v/v in 1 × PBS) for 15 min. The fixed cells were washed twice with 1 × PBS, permeabilized and 788 blocked with 0.1% Saponin/5% BSA/1 × PBS for 30 min. The cells were then incubated overnight at 4 789 °C in dark with indicated primary antibody at 1/50 -1/100 dilution (in 0.1% Saponin/5% BSA/1 × PBS). 790