The nuclear receptor NR4A1 is regulated by SUMO modification to induce autophagic cell death

NR4A is a nuclear receptor protein family whose members act as sensors of cellular environment and regulate multiple processes such as metabolism, proliferation, migration, apoptosis, and autophagy. Since the ligand binding domains of these receptors have no cavity for ligand interaction, their function is most likely regulated by protein abundance and post-translational modifications. In particular, NR4A1 is regulated by protein abundance, phosphorylation, and subcellular distribution (nuclear-cytoplasmic translocation), and acts both as a transcription factor and as a regulator of other interacting proteins. SUMOylation is a post-translational modification that can affect protein stability, transcriptional activity, alter protein-protein interactions and modify intracellular localization of target proteins. In the present study we evaluated the role of SUMOylation as a posttranslational modification that can regulate the activity of NR4A1 to induce autophagy-dependent cell death. We focused on a model potentially relevant for neuronal cell death and demonstrated that NR4A1 needs to be SUMOylated to induce autophagic cell death. We observed that a triple mutant in SUMOylation sites has reduced SUMOylation, increased transcriptional activity, altered intracellular distribution, and more importantly, its ability to induce autophagic cell death is impaired.

251 SUMO peptides are conjugated to a Lys residue, frequently within a consensus motif 252 KXE, where  represents any hydrophobic residue and X any amino acid(30). It has 253 been shown that NR4A2 is SUMOylated in two motifs conserved among the family 254 members that lead to SUMO ligation to lysines K91 (32) or K558 (33), depending on the 255 cell context. In order to find additional potential SUMOylation sites also conserved in the 256 three members of the family, we aligned NR4A protein sequences and searched for the 257 SUMO consensus motif KXE. We identified K102 and K577 within a SUMO motifs, which 258 are indeed SUMOylated (26), and observed that K558 is conserved in NR4A1 and NR4A2, 259 and hence it is also a potential site for SUMOylation in NR4A1 (positions numbers refer to 260 the human canonic sequence UniProtKB P22736) ( Figure 1A).

261
To analyze NR4A1 SUMOylation during SP/NK 1 R-induced autosis, we 262 immunoprecipitated NR4A1 at different time points after SP exposure and looked for 263 SUMO1 or SUMO2/3 conjugation by Western blot. Starting at 3 hr after SP induction, a 264 fraction of NR4A1 was SUMOylated by SUMO1 and the modification lasted up to 12 hr, a 265 time point at which SUMO2/3 ligation was also observed ( Figure 1B). We verified that 266 SUMO peptide conjugation of NR4A1 occurs during SP/NK 1 R-induced autosis by over-267 expressing Myc-tagged SUMO1 ( Figure 1C) or HA-tagged SUMO2 ( Figure 1D).  Figure 2A). Nevertheless, a weak signal was still observed, indicative of an additional 286 phosphorylated threonine. We found two additional ERK consensus motifs in the NR4A1 287 sequence ( Figure 1A), although additional phosphorylation sites could be targeted by 288 other kinases.

289
To analyze whether SUMO conjugation influences phosphorylation, we studied the 290 level of phosphorylation when SUMOylation is reduced. As shown in Figure 2A, the level 291 of phosphorylation on threonine residues was not affected in NR4A1_TriMut. This result 292 indicates that phosphorylation is not affected by SUMOylation in this case. On the other 293 hand, SUMOylation resulted dependent on previous phosphorylation in T143, as 294 NR4A1_T143A was barely SUMOylated ( Figure 2B). Therefore, NR4A1 SUMOylation, 295 subsequent to phosphorylation, could alter NR4A1 stability and/or intracellular distribution 296 and, consequently, the ability to induce autosis.

297
In previous work we observed that Nr4a1 expression is induced by SP signaling 298 (12), but NR4A1 protein abundance cannot be explained by transcriptional regulation 299 alone. We noticed that NR4A1 produced by expression of its cDNA (lacking both 5' and 3' 300 UTR) from a viral promoter (CMV) that lacks endogenous regulatory regions, also 301 accumulates in response to SP/NK 1 R (for example, look at Figure 2A, Input). We 302 rationalized then that NR4A1 stability should be post-translationally regulated. MAPK 303 signaling activated by SP/NK 1 R does not affect NR4A1 abundance, as it still accumulates 304 when ERK2 signaling is inhibited (12). SUMO modification commonly enhances protein 305 stability by binding to lysine residues that otherwise would be ubiquitinated, targeting the 306 protein for proteasome degradation; nevertheless, it has also been observed that some E3 307 ubiquitin ligases, such as RNF4, which have both SIM (SUMO interacting motif) and RING 308 domains, attach ubiquitin to SUMO-modified proteins (30), and this mechanism has been 309 described to occur to NR4A1(26). Therefore, we asked whether NR4A1 SUMOylation 310 regulated NR4A1 stability during SP/NK 1 R-induced autosis. First, we determined the 311 endogenous NR4A1 half-life. NR4A1 reached a maximum level of expression 3 hr after SP 312 exposure and by 9 hr it was still detected but clearly reduced ( Figure 2C). We then 313 inhibited new protein synthesis (with cycloheximide) after 3 hr of induction with SP (to start 314 with the highest amount of NR4A1), and observed that NR4A1 was degraded very rapidly, 315 as 3 hr after cycloheximide addition it was no longer detected ( Figure 2C).

316
To compare the half-life of wild type NR4A1 with NR4A1_TriMut with reduced 317 SUMOylation, we tagged it with FLAG, and compared FLAG-NR4A1 wild-type half-life with 318 FLAG-NR4A1_TriMut (which allowed to distinguish NR4A1_TriMut from endogenously 319 induced NR4A1). Since these constructs are expressed from a strong constitutive 320 promoter (CMV), FLAG-NR4A1 started in these experiments with a higher amount of 321 protein than the endogenously-induced NR4A1, therefore its expression could be 322 observed it for a longer period of time. We inhibited new protein synthesis 24 hr after 14 323 transfection. Notably, the stability imposed by SP signaling still occurred in FLAG-tagged 324 NR4A1 ( Figure 2D, upper panel). Under basal expression SUMOylation seemed to indeed 325 contribute to NR4A1 stability, as the half-life of TriMut (which was SUMOylated to a lesser 326 degree) was clearly reduced compared to wild type. Surprisingly, however, TriMut 327 abundance still increased in response to SP ( Figure 2D, bottom panel) reducing its 328 degradation rate. Other reports describe that NR4A1 stability can be modulated by 329 acetylation, which also occurs at Lys residues. We looked for acetylation during SP/NK 1 R-330 induced autosis, but no difference was found in the level of acetyl-Lysine content in 331 immunoprecipitated NR4A1 from non-treated cells compared with SP treated cells (data 332 not shown). We do not know at this point which mechanism mediates the increase in 333 NR4A1 stability in response to SP signaling but, possibly, the remaining SUMOylation 334 sites observed in TriMut ( Figure 1D) could be involved. 350 WT in response to SP. As can be seen in Figure 3A, NR4A1_TriMut showed enhanced 351 transcriptional activity in response to SP signaling. Therefore, SUMOylation appears to be 352 a relevant negative regulator of NR4A1 transcriptional activity.

369 NR4A1 SUMOylation is necessary for SP-induced autosis
370 We hypothesized that the signaling pathway activated upon NK 1 R activation by SP binding 371 triggers SUMOylation of NR4A1, conferring upon it the ability to induce autosis instead of 372 apoptosis or proliferation, among other processes. To inhibit NR4A1 SUMOylation we 373 expressed GAM1, a viral protein that inhibits SUMOylation by promoting SAE1/SAE2 and 374 UBC9 degradation (39). As shown in Figure 4A, in the presence of GAM1, but not of an 16 375 inactive GAM1 mutant, the amount of SUMOylated NR4A1 was reduced. Supporting the 376 notion that SUMOylation increases its basal stability, the total amount of NR4A1 was also 377 reduced in the presence of GAM1. More significantly, Gam1 expression completely 378 prevented SP/NK 1 R-induced autosis ( Figure 4B).

379
Finally, to confirm whether SUMOylation of NR4A1 is indeed necessary to mediate 380 SP-induced autosis, we tested the ability of TriMut to induce autosis in response to SP. To 381 eliminate endogenous expression of NR4A1 that could mask the mutant phenotype, we 382 silenced it by targeting small interfering RNAs to the 3' untranslated region of RN4A1 383 mRNA, which is absent in the NR4A1 expression vector. As expected, silencing the 384 expression of endogenous NR4A1 reduced cell death in response to SP, which was 385 restored by the expression of NR4A1 WT but not by the expression of TriMut ( Figure 4C).

425
Finally, we show that the ability of NR4A1 to induce autosis is impaired when 426 SUMOylation is reduced. Autosis could be triggered by specific molecular interactions of