Wounding and insect feeding trigger two independent MAPK pathways with distinct regulation and kinetics

Wounding is caused by abiotic and biotic factors and triggers complex short- and long-term responses at the local and systemic level. These responses are under the control of complex signaling pathways, which are still poorly understood. Here, we show that the rapid activation of MKK4/5-MPK3/6 by wounding is independent of jasmonic acid (JA) signaling and that, contrary to what happens in tobacco, this fast module does not control wound-triggered JA accumulation in Arabidopsis. We also demonstrate that a second MAPK module, constituted by MKK3 and the clade-C MAPKs MPK1/2/7, is activated by wounding in an independent manner. We provide evidence that the activation of this MKK3-MPK1/2/7 module occurs mainly through wound-induced JA production via the transcriptional regulation of upstream clade-III MAP3Ks and particularly MAP3K14. We show that mkk3 mutant plants are more susceptible to the larvae of the generalist lepidopteran herbivore Spodoptera littoralis, indicating that the MKK3-MPK1/2/7 module is involved in counteracting insect feeding. One sentence summary Wounding induces the parallel activation of a rapid signaling module (MKK4/5-MPK3/6) and a JA-dependent slow one (MAP3K14-MKK3-MPK1/2/7/14) to restrict insect feeding.


23
Wounding and insect feeding trigger two independent MAPK pathways with distinct 24 regulation and kinetics 25

Short title 26
Independent MAPK modules in wound signaling 27 Material distribution footnote 28 The author(s) responsible for distribution of materials integral to the findings presented in this 29 article in accordance with the policy described in the Instructions for Authors 30 (www.plantcell.org) is Jean Colcombet (jean.colcombet@inra.fr). 31

49
Wounding is a common stress for plants that can be caused by abiotic factors such as 50 wind, heavy rain, hail and snow or during biotic interactions, mostly with herbivorous 51 organisms such as insects. Injury may cause harsh damages to plant tissues and facilitate the 52 entry of pathogens (Savatin et al., 2014). Plants respond to these challenges by activating 53 several mechanisms to rapidly heal tissues and restrict potential pathogen entry. The cuticle 54 and trichomes are important constitutive structures involved in the prevention of wounding. 55 Once wounding occurred, intracellular molecules released from dead cells or damaged cell 56 wall components act as signaling molecules named DAMP (for Damage Associated 57 Molecular Pattern) (Maffei et al., 2012). Additionally, in the vicinity of wound sites, living 58 cells sense the mechanical disturbance through the activation of mechanosensitive channels 59 which trigger intracellular signaling pathways and local responses (Farmer et al., 2014). Such 60 responses are mediated by efficient and complex intracellular signaling mechanisms involving 61 9 activated by JA and the wound-induced activation of MPK3/6 was not compromised in the 243 JA-sensing-deficient coi1-34 mutant (fig. S14). Overall, these data show that two MAPK 244 modules are activated by wounding with different kinetics, a rapid MKK4/5-MPK3/6 module 245 that is independent of JA, and a slower MKK3-MPK2 module that depends on JA signaling. 246

MKK3-MPK2 and MKK4/5-MPK3/6 modules are activated by wounding independently of 247 each other 248
The fact that the MKK4/5-MPK3/6 module is activated rapidly might suggest that it 249 could function upstream of MKK3-MPK2. This idea is supported by the finding that, in 250 tobacco, SIPK and WIPK, which are homologues of Arabidopsis MPK3 and MPK6, were 251 proposed to play a role in wound-induced JA synthesis (Seo et al., 1999;Heinrich et al., 2018;252 Seo et al., 1995252 Seo et al., , 2007  We report in this study the activation of two independent MAPK modules by wounding 282 in Arabidopsis plants ( fig. 8). The first module is defined by MKK4/5-MPK3/6. The second 283 module, defined by MKK3-clade-C MAPKs, was suspected from previous preliminary 284 information (Ortiz-Masia et al., 2007). These two modules are activated independently of 285 each other with very distinct kinetics. We also demonstrated that jasmonic acid, which is 286 produced in response to wounding and herbivores, is an important mediator of the activation 287 of the MKK3 module. Overall, this work provides insight into wound signaling of 288 Arabidopsis. 289

modules? 291
Our data suggest that the MKK3-MPK2 module is activated through the JA-and 292 wound-induced transcription of several clade-III MAP3Ks. Among them, MAP3K14 shows 293 the highest and earliest expression upon wounding. Coherently, we observed a reduction of 294 MPK2 activation by wounding specifically at early time points in the loss-of-function 295 map3k14 CRISPR lines. We also observed a stronger wound-induced MPK2 activation in 296 map3k14-1, which expresses a truncated and more stable form of MAP3K14. In addition, 297 other members of the clade-III MAP3Ks, such as MAP3K18, MAP3K17, MAP3K19 and 298 MAP3K20, which are induced by wounding at a later time point, may maintain MPK2 activity 299 after MAP3K14-triggered initial activation. Among the 8 clade-III MAP3K proteins, only 300 MAP3K13 and MAP3K14 have a carboxyl-terminal extension which could correspond to a 301 transmembrane domain, possibly suggesting a particular role at the membrane, a location 302 thought to play a role in cellular wound sensing (Farmer et al., 2014). Unfortunately, we did 303 not succeed to see any fluorescent signal in lines expressing MAP3K14::YFP after wounding, 304 suggesting a very low expression level of MAP3K14 or alternatively that the YFP fusion at 305 the C-terminus may reduce MAP3K14 stability (data not shown). As expected in a process 306 requiring de novo protein production, the general inhibitor of protein synthesis, 307 cycloheximide, fully abolished the wound-induced MPK2 activation. In the past, de novo 308 synthesis of MAP3Ks has been well documented in the case of the MKK3-MPK7 activation 309 by drought which requires ABA-dependent MAP3K18 protein accumulation (Danquah et al., 310 2015;Boudsocq et al., 2015). Here we describe another stress triggering MKK3 module 311 activation by de novo protein synthesis of MAP3Ks via transcriptional regulation. This new 312 result comforts this model, suggesting that the late responsive signaling MAPK pathways are 12 more general than first thought. Therefore the module could broadly regulate a second layer 314 of events in response to environmental constraints (Colcombet et al., 2016). 315 MKK3 modules have been by far less characterized than the iconic stress-activated 316 module MKK4/5-MPK3/6. An obvious explanation lies in its rather atypical slow activation, 317 which was unexpected; textbooks often teach that MAPK modules are early/rapid signaling 318 actors. Additionally, clade-C MAPKs have never been detected so far by in-gel kinase assays. 319 Nevertheless, phosphorylated activation motifs of clade-C MAPKs have been found in 320 phosphoproteomic approaches suggesting that they are often activated. For example, MPK1/2 321 phosphorylation was reported in response to ABA (Umezawa et al., 2013) and DNA damage-322 inducing irradiations (Roitinger et al., 2015). Interestingly, in map3k14-1 that over-activates 323 MPK2 in response to wounding, the anti-pT-E-pY antibody detected a new band appearing 324 below MPK3 and MPK6 with a kinetics fitting clade-C MAPK activation (Fig. S13E). This 325 suggests that clade-C MAPKs are tightly controlled to avoid full activation. More 326 exhaustively, transcriptional regulation of clade-III MAP3Ks by stresses seems to occur very 327 often based on expression databases (Winter et al., 2007;Zimmermann et al., 2004), and 328 suggests important roles of MKK3 modules in environmental perception. Notably, 329 MAP3K13, MAP3K18 and MAP3K20 are induced by a large number of environmental 330 signals including osmotic, salt and drought stresses. Whereas MKK3 is a hub of the module, 331 the related MAP3Ks and MAPKs are encoded by multigenic families. The input signal 332 specificity is conferred by the transcriptional regulation of the MAP3K genes but we cannot 333 exclude that MAP3K activity is also directly modulated by input stresses, as suggested in 334 ABA signalling (Matsuoka et al., 2015;Mitula et al., 2015). The situation is less clear for 335 MAPKs: MPK1/2/7/14 seem to be activated identically by a stress, based on protoplast 336 experiments and immunoprecipitation from organs (Danquah et al., 2015). Our hypothesis to 337 explain this functional redundancy is that they may be expressed in different 338 cells/compartments or target specific substrates. This should be an important point to address 339 in the future. 340 Last, MKK3 has also been proposed to function upstream of MPK6 and MPK8 in 341 various contexts such as blue light, dark-light transition and ROS homeostasis (Sethi et al., 342 2014;Takahashi et al., 2007Takahashi et al., , 2011Lee, 2015). We previously reported that these functional 343 connections were not found when combining these kinases in protoplast expression systems 344 (Danquah et al., 2015). Additionally, in the present work, we did not see any  functional connection in the context of wounding. Nonetheless, it is possible that such 346 13 connections exist in other physiological contexts in specific cell types or organs (Colcombet 347 et al., 2016). 348

Is wound-induced JA production connected to MAPK modules? 349
The phytohormone JA plays a critical role in the wound response (Wasternack, 2018). It 350 is massively produced in a few minutes after wounding by chloroplast-and perixisome-351 localized biosynthetic enzymes and is involved both in local and long-distance responses 352 (Wasternack and Hause, 2013;Koo et al., 2009). Plants impaired in JA synthesis and 353 signaling have weakened responses to wounding and are much more susceptible to chewing 354 insects (Howe and Jander, 2008). The JA core signaling module is well described (Thines et 355 al., 2007;Chini et al., 2007): JASMONATE ZIM DOMAIN (JAZ) proteins sequester 356 transcription factors, notably basic helix-loop-helix (bHLH) factors such as MYC2 to prevent 357 them from activating JA-responsive genes. The binding of JA-Ile to COI1 receptor, which is 358 an F-Box E3 ligase, increases COI1 affinity for JAZ proteins, triggering their ubiquitination 359 and consequent proteasome-dependent degradation. The released transcription factors become 360 free to modulate gene expression (Thines et al., 2007). In this article, we report a reduction of 361 wound-triggered MPK2 activation in opr3 and coi1 mutants, which are impaired in JA 362 synthesis and signaling, respectively. Therefore we can conclude that the activation of the 363 MKK3-MPK1/2/7/14 module is downstream of JA signaling and synthesis. The remaining 364 activation in these mutants could be explained either by the existence of JA-independent 365 pathways controlling MAP3K transcriptional regulation in response to wounding or by the 366 fact that they are not total loss-of-function mutants. opr3 mutant was reported to still produce 367 some JA-Ile (Chini et al., 2018) and coi1-34 is not a full loss-of-function allele (Acosta et al., 368 2013). Surprisingly, the wound-triggered up-regulation of MAP3K transcripts is only mildly 369 impaired in the JA mutants whereas the MAPK activation is strongly reduced. It is possible 370 that the active MAP3K amount is limiting, titrated by competition. Alternatively, beside 371 transcriptional induction, MAP3Ks may also require a JA-dependent modification (i.e. 372 phosphorylation) for their activation, and the two impaired regulations would lead to a strong 373 reduction of MPK2 activation. This last hypothesis is supported by the fact that MAP3K18 374 intrinsic activity increases upon ABA treatment (Matsuoka et al., 2015;Mitula et al., 2015). open. MKK3-related modules had been shown to regulate ROS homeostasis in response to 378 wounding, but was then proposed to work upstream of the clade-D MAPK MPK8 (Takahashi 379 14 et al., 2011). Recently, the MAP2K inhibitor PD98059 was shown to reduce JA-triggered 380 activation of enzymes involved in ascorbate and glutathione metabolism in maize leaves 381 (Shan and Sun, 2018). Overall, these results point to redox homeostasis as a putative target of 382 MKK3 module in the general response to wounding and JA. 383 The signaling pathway from wound signaling to JA production is largely unknown. One 384 known event is the transient increase of cytosolic Ca 2+ (Maffei et al., 2004;Kiep et al., 2015) 385 which must be decoded by Ca 2+ -sensing proteins. Two calmodulin-like-proteins, CML37 and 386 CML42, have been identified as positive and negative regulators, respectively, of JA-387 mediated defense in Arabidopsis plants after herbivore attack (Vadassery et al., 2012;Scholz 388 et al., 2014). While CML42 very likely affects the binding of JA-Ile to the receptor, CML37 389 regulates the expression of JAR1, the enzyme catalyzing the JA-Ile formation. Our work 390 shows that JA synthesis is not under the control of MPK3/6. This is surprising as the 391 homologs of MPK3 and MPK6, WIPK and SIPK, respectively, were shown to modulate JA 392 levels in response to wounding in tobacco (Seo et al., 1999(Seo et al., , 1995Heinrich et al., 393 2018). This apparent discrepancy might be explained by different approaches used in the 394 studies. Notably, we took advantage of a unique mkk4mkk5 mutant issued from a tilling 395 screen (Zhao et al., 2014), in which the wound-induced activation of MPK3/6 is strongly 396 reduced but not fully abolished. It is possible that the remaining weak activation of MPK3/6 is 397 sufficient to trigger downstream events such as the hormonal production. Nevertheless, this 398 hypothesis is unlikely as this mutant is clearly impaired in wound-induced ethylene 399 production (Li et al., 2017). Alternatively, and more interestingly, these results suggest that 400  (Bigeard et al., 2015;Rayapuram et al., 2018;Bigeard and Hirt, 2018). 406 However, very little is known about their functions in response to other stresses and notably 407 in response to wounding. The recent discovery that the MKK4/5-MPK3/6 module is involved 408 in wound-induced ethylene production through the transcriptional regulation of ACS genes 409 was the first report of their role in wound signaling in Arabidopsis (Li et al., 2017). 410 Interestingly, the same module is involved in ethylene production in response to flg22 and 411 Botrytis (Li et al., 2012;Liu and Zhang, 2004). This regulation occurs through both a direct 412 phosphorylation-dependent stabilization of the ACS6 enzyme and the up-regulation of ACS 413 genes by the phosphorylation-dependent activation of the WRKY33 transcription factor. In 414 rice, OsMPK1, the closest homolog of AtMPK6, was also shown to interact with and 415 phosphorylate WRKYs in vitro (Yoo et al., 2014). Interestingly, cycloheximide, which is a 416 potent blocker of MKK3 module activation ( fig. 3), is a strong activator of MPK3/6 (fig. S16) 417 and triggers the transcriptional up-regulation of a large set of MAMP-regulated genes 418 (Navarro et al., 2004). Overall, these data suggest that the MKK4/5-MPK3/6 module may be 419 involved in the control of gene expression in response to wounding as for other stresses (Frei 420 dit Frey et al., 2014). 421

Roles of the MAPKs in the interaction with herbivores 422
MAPKs also play an important role in plant-herbivore interactions (Hettenhausen et al., 423 2017). Early studies in Nicotiana tabaccum showed that WIPK is activated rapidly by 424 wounding and that WIPK-silenced plants have reduced levels in defense-related genes and JA 425 (Seo et al., 1995). In Nicotiana attenuata, mechanical wounding and oral secretion of the 426 herbivore Manduca sexta activate SIPK and WIPK within 5 minutes (Wu et al., 2007). In 427 Arabidopsis, grasshopper oral secretion increases the wound-induced activation of MPK3 and 428 MPK6 (Schäfer et al., 2011). In Solanum lycopersicum, M. sexta feeding also activates WIPK 429 and SIPK homologs SlMPK3 and SlMPK1/2, respectively, and silencing SlMPK1/2 reduces 430 herbivory-induced JA levels resulting in enhanced larval growth (Kandoth et al., 2007). 431 However, other studies in N. tabaccum and N. attenuata reveals more complexity in the JA-432 MAPKs relationship depending on herbivores. In response to wounding, tobacco MPK4 is 433 activated in minutes and silencing NtMPK4 compromises JA-responsive gene induction 434 (Gomi et al., 2005). By contrast, silencing NaMPK4 does not affect JA production nor 435 resistance to the generalist herbivore Spodoptera littoralis, but increases resistance to M. sexta 436 (Hettenhausen et al., 2017). This indicates that some functions of MPK4 are specific to 437 particular herbivore interactions and that further studies will be necessary to clarify the role of 438 different MAPK modules in wound-and herbivory-induced JA signaling. 439 440 16

441
Primer sequences for plant genotyping, molecular cloning and expression analysis are 442 provided in the supplemental table 1. 443

Insect experiments 488
Larvae of the generalist lepidopteran species Spodoptera littoralis were reared as 489 previously published (Vadassery et al., 2012). To monitor MPK2 activation or measure 490 oxylipin production in response to insect, 4 th instar larvae were starved overnight prior to 491 plant feeding for 1h and 3h. To monitor MPK3/MPK6 activation, larvae were removed after 492 15 minutes. For long-term feeding assays (8 days), first instar larvae were used according to 493 Vadassery et al. (2012). When different plant lines were used at the same time, they were kept 494 separated from each other to avoid any kind of contact and placed randomly in the 495 experimental setup. 496

Other assays 504
For gene expression analysis, plants were collected at indicated times and frozen in 505 liquid nitrogen. RNA extraction, cDNA synthesis and qRT-PCR analysis were performed as 506 previously described using primers in table S1 (Danquah et al., 2015). Yeast 2-hybrid assays 507 were performed as previously described except that the selective medium was generated with 508 0.

Fig. S3 -Flg22 rapidly and transiently activates MPK3 and MPK6 in leaf punches
Punches from Col-0 leaves were equilibrated over night in water and then treated for the indicated time with 500 nM flg22 or water (MOCK). MPK3 and MPK6 phosphorylation is monitored by western-blot using an antibody raised against the phosphorylated form of ERK2 (anti-pTpY). Equal loading is controlled by Coomassie staining. B. Kinase activity of MPK2 after immunoprecipitation with an anti-MPK2 antibody from WT and map3k14-1 leaves following wounding. C. Kinase activity of HA-immunoprecipitated MPK2 expressed in mkk3-1 mesophyll protoplasts in the presence or absence of MKK3 and MAP3K14wt and MAP3K14-1. Western-blots show protein expression levels.
Col-0  A and B. Western-blot using antibody raised against the phosphorylated form of ERK2 (anti-pTpY) in indicated genetic backgrounds after wounding. Equal loading is controlled by Coomassie staining. Figure  S13A is an un-cropped version of western blot shown in figure 6A.

Wounding
C. Kinase activity of MPK2, MPK3, MPK6 after immunoprecipitation with an appropriate specific antibody from wounded leaves of Col-0 and mkk4mkk5 plants. MPK3/6 activation was monitored by western-blot using antibody raised against the phosphorylated form of ERK2 (anti-pTpY). Equal loading is controlled by Coomassie staining.
D to F. Western-blots using antibody raised against the phosphorylated form of ERK2 (anti-pTpY) in indicated genetic backgrounds. "?" (D) referred to a map3k14-1 specific band which could correspond to the over activation of clade-C MAPKs. Equal loading is controlled by Coomassie staining.   Kinase activity of MPK2 after immunoprecipitation with an anti-MPK2 specific antibody from leaves on which S. littoralis fed during 15 minutes before to be removed (at t=15'). MPK3/6 activation was monitored by western-blot using antibody raised against the phosphorylated form of ERK2 (anti-pTpY). Equal loading is controlled by Coomassie staining.

B and C.
Kinase activity of MPK2 after immunoprecipitation with an anti-MPK2 specific antibody from leaves on which S. littoralis fed for 1 and 3 hours in Col-0 and coi1-34.

D.
Weight of S. littoralis caterpillars after feeding for 8 days on Col-0 and mkk3-1 rosettes. Box plot shows distribution of caterpillar weight (n>120 in 5 biological replicates

Fig. 8 -Working model of MAPK activation by wounding
Wounding and insect feeding activate two MAPK modules: a rapid one composed of MKK4/5-MPK3/6 which regulates notably ethylene production and a slow one composed of clade-III MAP3Ks-MKK3-MPK1/2/7/14 whose activation is under the control of a JA-dependent production of MAP3Ks. Drought also activates MAP3Ks-MKK3-MPK1/2/7/14 through an ABA-dependent step.