MAP Kinase PrMPK9-1 Contributes to the Self-Incompatibility Response1[CC-BY]

A TDY-class MAPK is functionally involved in mediating self-incompatibility in Papaver. Mitogen-activated protein kinases (MAPKs) form important signaling modules for a variety of cellular responses in eukaryotic cells. In plants, MAPKs play key roles in growth and development as well as in immunity/stress responses. Pollen-pistil interactions are critical early events regulating pollination and fertilization and involve many signaling events. Self-incompatibility (SI) is an important mechanism to prevent self-fertilization and inbreeding in higher plants and also is known to utilize signaling to achieve incompatible pollen rejection. Although several pollen-expressed MAPKs exist, very little is known about their function. We previously identified a pollen-expressed MAPK (p56) from Papaver rhoeas that was rapidly activated during SI; several studies implicated its role in signaling to SI-induced programmed cell death involving a DEVDase. However, to date, the identity of the MAPK involved has been unknown. Here, we have identified and cloned a pollen-expressed P. rhoeas threonine-aspartate-tyrosine (TDY) MAPK, PrMPK9-1. Rather few data relating to the function of TDY MAPKs in plants currently exist. We provide evidence that PrMPK9-1 has a cell type-specific function, with a distinct role from AtMPK9. To our knowledge, this is the first study implicating a function for a TDY MAPK in pollen. We show that PrMPK9-1 corresponds to p56 and demonstrate that it is functionally involved in mediating SI in P. rhoeas pollen: PrMPK9-1 is a key regulator for SI in pollen and acts upstream of programmed cell death involving actin and activation of a DEVDase. Our study provides an important advance in elucidating functional roles for this class of MAPKs.

Cellular responses require an integrated signal perception and signal transduction network. Mitogen-activated protein kinases (MAPKs) form highly conserved signaling networks that are common to all eukaryotic cells, including animals, plants, and fungi. They are commonly used to transduce information from sensors to cellular responses. A MAPK cascade minimally consists of three kinases: a MAPKKK, a MAPKK, and a MAPK, which sequentially phosphorylate and activate each other. MAPKs are Ser/ Thr kinases, and upon phosphorylation, which activates them, the MAPK can phosphorylate a range of target proteins, which include other kinases and transcription factors. This results, ultimately, in changes in gene expression, metabolism, cell division, and growth. In plants, they function in diverse signaling networks, including development, but perhaps the best characterized and well known involve responses to biotic stresses stimulated by interactions with pathogens, involving defense responses, and to a range of abiotic stresses, including wounding, drought, cold, heat, UV, touch, osmotic shock, and salt (Rodriguez et al., 2010;Meng and Zhang, 2013;Xu and Zhang, 2015). The involvement of MAPKs in signaling to the expression of defense genes, involving apoptosis and programmed cell death (PCD), also is well documented (Pitzschke et al., 2009;Mase et al., 2012;Bigeard et al., 2015).
Arabidopsis (Arabidopsis thaliana) has 20 MAPK genes, and there are a similar number in the sequenced genomes of other plant species. Plant MAPKs have two types of activation motif: either a Thr-Glu-Tyr (TEY) activation motif or a Thr-Asp-Tyr (TDY) activation motif, which is unique to plant MAPKs (Ichimura et al., 2002;Hamel et al., 2006). Plant MAPKs are highly conserved between species and can be divided into four clades (A-D) based on sequence similarities. The plant TEY MAPKs can be classified into three groups (A-C), while the TDY MAPKs belong to a more distant clade (D; Ichimura et al., 2002;Hamel et al., 2006). Group A MAPKs are well characterized; many are functionally involved in stress responses and include the Arabidopsis MPK3 and MPK6 and their orthologs in other species, such as the stress-activated MAPK from alfalfa (Medicago sativa), salt-induced MAPK, tobacco (Nicotiana tabacum) salicylic acid-induced protein kinase, and wounding-induced protein kinase. The group B MAPKs (Arabidopsis MPK4 and MPK11) also have functions implicated in pathogen defense and stress responses and also play an essential role in cell division (for review, see Meng and Zhang, 2013). The functions of MAPKs in groups C and D are currently less well established (Meng and Zhang, 2013;Zhang et al., 2014).
Plants use PCD for a wide range of functions, including disease resistance and development as well as reproduction (for review, see Van Hautegem et al., 2015;Daneva et al., 2016;Van Durme and Nowack, 2016). Animal cells use a cascade of caspases (Cys proteases), which, when activated, cleave target proteins, resulting in cellular dismantling and, ultimately, death. Although there are no true caspase gene homologs in plants, there is good biochemical evidence for several caspase-like protease activities and their involvement in plant PCD (Bozhkov et al., 2004;Sanmartín et al., 2005). Caspase-like activities have been detected in plants through cleavage of the animal caspase-specific tetrapeptide substrates (e.g. Ac-DEVD-AMC for caspase-3; del Pozo and Lam, 1998;Richael et al., 2001;Danon et al., 2004;Thomas and Franklin-Tong, 2004;Thomas et al., 2006).
Pollination involves many interactions between the pistil and pollen grain/tube as it germinates and grows to achieve fertilization (for review, see Dresselhaus and Franklin-Tong, 2013). Many angiosperms face the possibility of self-fertilization and have evolved selfincompatibility (SI) to avoid this problem. This is an important genetically controlled system that has evolved independently several times; several different mechanisms regulate SI in different plant species (for review, see Takayama and Isogai, 2005;Franklin-Tong, 2008). SI is genetically controlled by a multiallelic S-locus that encodes linked pollen and pistil S-determinants. These allow recognition and rejection, so when male and female S-determinant allelic specificities match, self-(incompatible) pollen is recognized and rejected before fertilization can occur.
The SI system found in Papaver rhoeas (the field poppy) has provided a model system in which to study the components involved in cell-cell recognition and rejection, in particular the signaling network downstream of the interaction between the pistil S-determinant, PrsS, and the pollen S-determinant, PrpS. PrsS encodes a small secreted protein (Foote et al., 1994), and PrpS encodes a small, novel transmembrane protein (Wheeler et al., 2009). The major focus for SI-induced signaling is the initiation of PCD (Thomas and Franklin-Tong, 2004;Bosch and Franklin-Tong, 2007). A long-standing model for SI in P. rhoeas is that PrsS acts as a signaling ligand, initiating a network of signaling events within incompatible pollen that results in the rapid inhibition of pollen tube growth and, ultimately, PCD. A key target for SI signals is the actin cytoskeleton. Within a few minutes of SI initiation, F-actin in incompatible pollen is depolymerized, and later, F-actin reorganizes into punctate foci Poulter et al., 2010). These largescale alterations to the actin polymerization status of incompatible pollen activate a signaling network resulting in PCD . The predominant caspaselike activity during P. rhoeas SI is a caspase-3-like/ DEVDase activity (Thomas and Franklin-Tong, 2004;Bosch and Franklin-Tong, 2007). More recently, it has been demonstrated that reactive oxygen species (ROS) play a key role in signaling to alterations in actin and SI-PCD . How these events integrate to participate in SI-PCD is reviewed by Wilkins et al. (2014).
We previously showed that a MAPK named p56 was activated in SI-induced pollen (Rudd et al., 2003). This implicated MAPK signaling in P. rhoeas SI. Activation of the p56-MAPK peaked ;10 min after SI, so it occurred after the inhibition of pollen tube growth, suggesting a function in later events. A further study showed that the inhibition of SI-induced p56 activation resulted in the alleviation of SI-induced caspase-3-like activity. This provided evidence that the MAPK p56 participates in initiating the PCD signaling cascade (Li et al., 2007). Here, we have identified and cloned a pollen-expressed P. rhoeas MAPK, PrMPK9-1, that encodes the p56-MAPK. We provide evidence that it functions to participate in signaling to SI-mediated inhibition of growth and actin alterations that are key features of the SI response in P. rhoeas and in the activation of a DEVDase, which is required for SI-PCD. The identification of Pr-MPK9-1 provides an important step forward in identifying components involved in the SI signaling network. Moreover, our demonstration that this is a TDY-type MAPK and that it participates in signaling upstream of actin alterations and DEVDase activation in the PCD network provides an important example of the function of a cell-specific TDY MAPK, rather few of which have been characterized to date.

Evidence Suggests That p56-MAPK Is Encoded by
PrMPK9-1, an Ortholog of AtMPK9 The strategy adopted to identify the gene encoding p56-MAPK was based on combining extant data on plant MAPKs, properties of the protein, and proteomic analysis. MAPKs comprise a large gene family with 20 members identified in the Arabidopsis genome. Although MAPKs are highly conserved, they fall into four groups: the TEY groups of MAPKs (A-C) are generally smaller (;42-45 kD) than the TDY group of MAPKs (group D), which are ;55 to 79 kD (Supplemental Fig.  S1). This suggested that the poppy p56-MAPK was possibly a TDY group MAPK. Expression data from Genevestigator indicated that four Arabidopsis TDY MAPKs, AtMPK8, AtMPK9, AtMPK17, and AtMPK19, are expressed at high levels in pollen (Supplemental Table S1). Fourier transform-ion cyclotron resonance (FT-ICR) mass spectrometry of p56 isolated by SDS-PAGE fractionation of pollen proteins (see "Materials and Methods") identified five peptides present in AtMPK9, comprising three peptides that were unique to AtMPK9, one peptide unique to AtMPK8, and two peptide sequences that were present in AtMPK8, AtMPK9, and AtMPK15 (Supplemental Table S2). Several other peptides corresponding to TDY MAPKs from Brassica napus, rice (Oryza sativa), and Populus trichocarpa also were identified (Supplemental Table  S2). Taken together, these data suggested that the poppy p56-MAPK was likely encoded by an AtMPK8/9 ortholog.
We next designed primers based on sequences from Arabidopsis TDY MAPKs with high expression in pollen (AtMPK8, AtMPK9, and AtMPK17). An ;950-bp fragment was cloned from poppy pollen, and sequencing of this revealed that it shared highest nucleotide identity with AtMPK8 and AtMPK9 (Supplemental Table S2). To obtain the full-length gene sequence, 59 RACE and 39 RACE were carried out on pollen cDNA (see "Materials and Methods"). An ;1.6-kb mRNA product was amplified, cloned, and sequenced. The pollen open reading frame was 1,479 bp long, predicted to encode a protein of 493 amino acids, with a molecular mass of 56.3 kD and a pI of 6.49. The full-length sequence showed highest homology to AtMPK8 and AtMPK9, sharing 79.4% identity with AtMPK9 and 65.5% identity to the AtMPK8 gene at the amino acid level. Therefore, we designated the poppy gene PrMPK9-1. To determine if the PrMPK9-1 sequence was the gene encoding the SI-activated p56-MAPK, we reanalyzed the FT-ICR mass spectrometry data using the predicted protein sequences obtained. The predicted PrMPK9-1 protein had two peptide matches that were highly significant (KGSYGVVASAVDTHTGEKV and KVDPLALNLLERL; Supplemental Table S2).
We also cloned another pollen open reading frame, designated PrMPK9-2 (1,818 bp, encoding a predicted protein of 606 amino acids), that also showed high homology to AtMPK8 and AtMPK9 and homology with PrMPK9-1 (Supplemental Fig. S2). However, its predicted molecular mass of 68.95 kD was too large to be a good candidate for the gene encoding the SI-activated p56-MAPK. Comparing the PrMPK9-2 predicted protein sequences with the p56 pollen extract, FT-ICR peptide data revealed no significant matches (Supplemental Table S2). This discounted PrMPK9-2 as a candidate for the poppy p56-MAPK and confirmed that PrMPK9-1 was the best candidate.
The expression of PrMPK9-1 was examined using semiquantitative reverse transcription (RT)-PCR. Transcript was detected in mature pollen, with weak expression during late anther development; there was negligible expression in root, leaf, or stigma (Fig. 1A). This expression pattern is consistent with the known function of the p56-MAPK involvement in the SI response in mature pollen. We examined whether PrMPK9-1 was up-regulated after SI induction, as several MAPKs are up-regulated after stimulation, particularly stress treatments (Zhang et al., 2014). RT-PCR expression analysis of PrMPK9-1 revealed increased expression 10 min after SI induction in incompatible pollen specifically; compatible pollen challenged with the same PrsS did not show increased expression, nor did other control pollen samples (Fig. 1B). Thus, PrMPK9-1 also exhibits increased expression after SI induction.

PrMPK9-1 Encodes a TDY MAPK
Analysis of the predicted PrMPK9-1 protein sequence reveals the expected features of a MAPK, including 11 characteristic kinase domains, the ATPbinding site, the substrate-binding site, and the kinase interaction motif. The presence of a TDY motif, rather than a TEY motif, is characteristic of a group D MAPK (for review, see Rodriguez et al., 2010). Consistent with this, PrMPK9-1 possesses a long C-terminal extension of ;140 amino acids that is typical of group D MAPKs (Jonak et al., 2002). In addition, the protein lacks the C-terminal common-docking domain that is found in groups A, B, and C MAPKs but not in group D proteins. In addition, a phylogenetic tree (Supplemental Fig. S1) based on ClustalX analysis of the predicted PrMPK9-1 and PrMPK9-2 protein sequences with MAPK sequences from Arabidopsis, rice, P. trichocarpa, alfalfa, Zea mays, and Homo sapiens also placed the proteins in the large Figure 1. Expression analysis of PrMPK9-1. A, Semiquantitative RT-PCR expression analysis of PrMPK9-1. PrMPK9-1 shows increasing expression during pollen development and low expression in other tissues. I, Immature anthers; 23/22/21/+1, stages of anther development (number of days from anthesis); P, hydrated pollen; R, root; L, leaf; S, stigma. GAPD shows control gene expression. B, RT-PCR expression analysis of PrMPK9-1 shows increased expression 10 min after SI induction in incompatible (SI) and compatible (COM) pollen or heatdenatured PrsS (DSI). CK, Control using water as a template; GM, control pollen growing in growth medium (GM). GAPD shows control gene expression.
clade of group D MAPKs. The poppy proteins exhibited highest homology with AtMPK9. Gene duplication is a prominent feature within both the eudicot and monocot MPK gene families (Hamel et al., 2006;Rodriguez et al., 2010). For example, in P. trichocarpa, two paralogs of AtMPK16 have been identified (PtMPK16-1 and PtMPK16-2). The high similarity between PrMPK9-1 and PrMPK9-2 suggests this to be the case for the poppy MPK9. Thus, PrMPK9-1 and PrMPK9-2 appear to be paralogs of AtMPK9, although the latter is present only as a single copy in the Arabidopsis genome.
Evidence of a Direct Link between PrMPK9-1 and p56-MAPK Although evidence indicated that p56-MAPK is not an ortholog of AtMPK3, we had shown previously that p56-MAPK cross-reacts with an anti-AtMPK3 antibody (Li et al., 2007). Therefore, we checked if the product of PrMPK9-1 was detected by the anti-AtMPK3 antiserum. We expressed the full-length His-tagged PrMPK9-1 protein in Escherichia coli. Western blotting confirmed that this recombinant protein could be detected with anti-His tag antiserum ( Fig. 2A) and anti-AtMPK3 (Fig. 2B). Given this observation, we sought to obtain direct lines of evidence to demonstrate that PrMPK9-1 encodes the p56-MAPK.
Initially, we used western blotting to detect PrMPK9-1. As MAPKs share many conserved domains, we made an antibody raised against the divergent C-terminal region (103 amino acids) of PrMPK9-1, as the full-length protein was unlikely to be sufficiently specific. This C-terminal His-tagged PrMPK9-1C truncated recombinant protein (;16 kD) was detected with the anti-His tag antiserum (Fig. 2C). The PrMPK9-1C antiserum cross-reacted with the truncated PrMPK9-1C recombinant protein (Fig. 2D) and with full-length PrMPK9-1 recombinant protein (Fig.  2E). We also ascertained that it detected a protein migrating at 56 kD in crude pollen extracts from P. rhoeas (Fig. 2F). Preimmune serum did not cross-react with the PrMPK9-1C recombinant protein (Fig. 2G).
We then examined pollen extracts that had been SI induced (Fig. 3A) to see if we could detect an ;56-kD MAPK protein that was activated by SI. Western blotting using anti-pTXpY to detect activated MAPKs showed minimal levels of phosphorylated p56-MAPK activity in untreated pollen and increased activity in SIinduced samples (Fig. 3B); several activated MAPKs also were detected, but their activity was approximately the same in both samples. When the membrane was reprobed with anti-PrMPK9-1C ( Fig. 3C), this antibody detected a protein at the same molecular mass as the phosphorylated p56-MAPK, but, as expected, it did not distinguish between phosphorylated and unphosphorylated samples. This provides good evidence that PrMPK9-1 corresponds with the p56-MAPK phosphoprotein. The top band at ;70 kD is a cross-reacting contaminant from the recombinant PrsS used for the SI induction (Supplemental Fig. S3).
We previously used calyculin A (a Ser/Thr protein phosphatase inhibitor) to obtain increased levels of phosphorylated MAPKs in pollen extracts and used a pTXpY antibody to detect the activated form of p56 (Rudd et al., 2003;Li et al., 2007). To determine that PrMPK9-1 corresponded to this phosphoprotein, we treated pollen with calyculin A. Equal amounts of untreated-and calyculin A-treated pollen extracts (Fig.  3D) were immunoprecipitated using biotinylated PrMPK9-1C antiserum and probed with anti-pTXpY antibody (Fig. 3, E and F). This revealed an activated ;56-kD protein immunoprecipitated by anti-PrMPK9-1C in the calyculin-treated sample ( Fig. 3E) but not in the untreated sample (Fig. 3F). Corresponding immunoprecipitates of untreated and SI-induced samples probed with anti-pTXpY antiserum (Fig. 3G) also gave an ;56-kD protein, present in the SI sample and not in the untreated sample. We believe the bottom band to be a degradation product. These data provide good evidence that PrMPK9-1 corresponds with a pollen phosphoprotein that has the properties of the SI-activated p56-MAPK identified previously (Rudd et al., 2003;Li et al., 2007).
To unequivocally determine the identity of the protein in the calyculin A-treated pollen sample, a 56-kD band was excised from the sample immunoprecipitated by PrMPK9-1C antiserum (Fig. 3E) and analyzed using FT-ICR mass spectrometry. Importantly, this revealed that all the peptides present in this 56-kD band had highly significant matches with PrMPK9-1 (Supplemental Table S2). Together, these data provide compelling evidence that PrMPK9-1 encodes the p56-MAPK.  1,000). B, Anti-AtMPK3 antiserum (1:1,000) cross-reacted with recombinant full-length PrMPK9-1. C, C-terminal His-tagged PrMPK9-1C truncated recombinant protein (;16 kD) was detected with the anti-His tag antiserum; PrMPK9-1C antiserum was raised against this protein. D, PrMPK9-1C truncated recombinant protein cross-reacted with the PrMPK9-1C antiserum. E, PrMPK9-1C antiserum cross-reacted with the full-length PrMPK9-1 recombinant protein. The smear at the bottom is a likely degradation product. F, PrMPK9-1C antiserum crossreacted with a protein from P. rhoeas pollen extracts, migrating at 56 kD. G, Preimmune serum did not cross-react with the PrMPK9-1C recombinant protein. The arrows indicate the positions to which full-length PrMPK9-1 and C-terminal truncated PrMPK9-1C migrate.

PrMPK9-1 Is Cytosolic and Does Not Alter Its Localization after SI Induction
As some MAPKs translocate to the nucleus after activation (Cheong et al., 2003), we investigated the intracellular localization of PrMPK9-1 in untreated pollen tubes and after SI induction using immunolocalization. Fixed pollen tubes were probed with the purified anti-PrMPK9-1C antiserum and visualized using anti-rabbit IgG-fluorescein isothiocyanate (FITC) secondary antibody (Fig. 4). This revealed that PrMPK9-1C appeared to have a largely cytosolic localization throughout untreated pollen tubes (Fig. 4, A-C). Examining localization after SI induction revealed that PrMPK9-1C retained its mostly cytosolic localization (Fig. 4, D-F). Even several hours after SI induction, PrMPK9-1C localization remained similar to the observed localization in normally growing pollen tubes. We conclude that, unlike some other MAPKs that have been shown to undergo rapid translocation to the nucleus (e.g. PcMPK3a/b and PcMPK6 during the innate immunity response in parsley [Petroselinum crispum] cells; Lee et al., 2004), PrMPK9-1 does not alter its localization after activation.
Antisense Oligonucleotides for PrMPK9-1 Block the SI Response in Pollen Tube Growth Having obtained compelling evidence that PrMPK9-1 is the poppy p56-MAPK, we wished to establish that PrMPK9-1 was involved in the SI response. Although we had speculated earlier that p56 was unlikely to be involved directly in pollen inhibition, as its peak activation was 10 min after SI induction, this did not rule out a possible role in decision-making after the rapid tip inhibition. In order to establish whether the p56-MAPK was functionally involved in the early steps of the SI response, we used in vitro SI bioassays using antisense oligonucleotides. This approach has been used previously to demonstrate a functional role for a pollen-soluble inorganic pyrophosphatase  and for the PrpS pollen S-determinant (Wheeler et al., 2009) in SI, as they block the inhibition response. We hypothesized that if the PrMPK9-1 MAPK is functionally involved in signaling to targets/ mechanisms involved in SI, knockdown of its expression should result in the alleviation of pollen tube inhibition in an S-specific manner. Therefore, we induced SI in vitro in the presence of either antisense (as-ODNs) or sense (s-ODNs) oligonucleotides for PrMPK9-1 to test this hypothesis.
SI induction in pollen grown in vitro, stimulated by the addition of recombinant PrsS proteins, resulted in highly significantly inhibited pollen tube growth (a reduction of 75.25% in length; P = 2.55 3 10 245 , n = 150). Compatible pollen in combination with the same PrsS proteins did not result in any significant difference in lengths compared with compatible untreated pollen tubes (P = 0.062, not significant). When SI was induced in pollen pretreated with antisense MPK9-1 oligonucleotides (as-ODNs), we observed a significant alleviation of the SI-specific inhibition (Fig. 5). Addition of the as-ODNs resulted in a highly significant recovery of SIinduced tubes (333.4% longer pollen tube length compared with SI-induced tubes; P = 2.4 3 10 244 , n = 150). The sense version of this oligonucleotide (s-ODN) did not alleviate SI-induced inhibition, and SI-induced pollen tubes in the presence of s-ODNs were significantly shorter than untreated pollen tubes (P = 2.3 3 10 236 , n = 150). The as-ODN and s-ODNs alone had no effect on normal pollen tube growth (P = 0.856, not significant, and P = 0.181, not significant, respectively; Fig. 5). Together, these data demonstrate that PrMPK9-1 has a crucial role in the signaling network resulting in the SI-induced S-specific irreversible pollen tube inhibition.
Antisense Oligonucleotides for PrMPK9-1 Block the Formation of SI-Induced Actin Punctate Foci As we had shown previously that alterations in actin polymerization status can trigger PCD in pollen  and that the formation of punctate F-actin foci is a key feature of SI induction (Geitmann et al., 2000), we wished to establish whether PrMPK9-1 was functionally involved in mediating the formation of these distinctive actin foci formed specifically in incompatible pollen. To examine this, we took a similar approach, inducing SI in vitro in the presence of either PrMPK9-1 antisense (as-ODNs) or sense (s-ODNs) oligonucleotides and imaging F-actin in pollen tubes after 2 h. SI triggered the formation of the F-actin foci (Fig.  6A), while untreated pollen tubes had normal F-actin configurations (Fig. 6B); as-ODNs in combination with SI induction rescued the pollen F-actin, which showed a normal configuration (Fig. 6C), while s-ODNs in combination with SI induction had no effect (Fig. 6D); control treatment with as-ODNs (Fig. 6E) or s-ODNs alone had no effect. Quantification of these data (Fig.   6F) revealed that pollen tubes pretreated with as-ODNs prior to SI induction had significantly fewer actin foci compared with SI alone (P = 1.14 3 10 210 , n = 200), while s-ODN pretreatment prior to SI was significantly  different from the as-ODN pretreatment with SI (P = 0.0005). The as-ODN and s-ODNs alone had no effect (P = 0.21, not significant, and P = 0.2814, not significant, respectively). Together, these data demonstrate that PrMPK9-1 is upstream of SI signaling to the formation of punctate actin foci.

Antisense Oligonucleotides for PrMPK9-1 Block SI-Induced PCD in Incompatible Pollen
We previously obtained data implicating a MAPK as being functionally involved in mediating PCD in incompatible pollen (Li et al., 2007). However, as this involved a pharmacological approach, which would inhibit all pollen MAPKs, and as the identity of p56 at that time was unknown, we could not link p56 to being functionally involved in mediating SI-induced PCD. Here, having cloned p56, we used the fluorogenic caspase-3/7 live-cell substrate, Image iT LIVE Green Caspase-3 and Green Caspase-7 Detection Kit (see "Materials and Methods"), in combination with the antisense PrMPK9-1 oligonucleotides to examine whether PrMPK9-1 was involved in mediating SIinduced PCD in incompatible pollen. We induced SI in vitro in pollen tubes that had been pretreated with either antisense (as-ODNs) or sense (s-ODNs) oligonucleotides and imaged caspase-3-like activity in pollen tubes after 4 to 5 h, when peak activity was expected (Bosch and Franklin-Tong 2007), to test this hypothesis. Untreated pollen tubes showed no FITC fluorescence indicating caspase-3-like activity (Fig. 7, A and B), while SI-treated pollen tubes had increased levels of fluorescence (Fig. 7, C and D); pollen tubes treated with as-ODNs prior to SI had background levels of fluorescence similar to untreated controls (Fig. 7, E and F), while the control treatment with s-ODNs prior to SI had no effect, with DEVDase activity still detected (Fig. 7, G and H). Quantification of these data (Fig. 7I) revealed that the as-ODNs resulted in a highly significant recovery of SI-induced tubes compared with SI-induced treatment alone (P = 8.790 3 10 210 , n = 250), and s-ODN pretreatment prior to SI had a significantly different effect compared with that of the as-ODNs (3.60 3 10 28 ). The as-ODN and s-ODNs alone had no effect (P = 0.250, not significant, and P = 0.409, not significant, respectively). Together, these data demonstrate that PrMPK9-1 plays a key role in the signaling network resulting in caspase-3-like activation, which will induce PCD in incompatible pollen tubes.

DISCUSSION
MAPKs have been shown to be key players in regulating many signaling networks in plants (Pitzschke et al., 2009;Meng and Zhang, 2013;Xu and Zhang, 2015). Despite a good knowledge of the MPK gene families for some time now, the function of only a handful of MPK genes have been identified. Here, we have identified the pollen-specifically expressed PrMPK9-1, a developmentally expressed TDY MAPK, with high expression in mature pollen, as being involved in the SI response. We took an unusual backward approach, as we had a well-characterized biological phenomenon (SI) that we had shown previously to involve a MAPK activity (Rudd et al., 2003), but we did not know which of the 20 MAPKs was involved. To our knowledge, our studies are the first implicating a function for a MAPK that is expressed in pollen. Moreover, PrMPK9-1 falls in the (so far) rather poorly characterized group of TDY class (group D, family V) MAPKs. Below, we discuss the implications of our findings.

Identity of PrMPK9-1 as a TDY MAPK
The P. rhoeas MPK9-1 gene belongs to the large MPK clade D, which has a TDY motif instead of TEY and an extended C-terminal region relative to the other MAPKs (Ichimura et al., 2002). The group D MAPKs are unique, owing to their TDY motif in the activation loop and long C-terminal common-docking domain (Rodriguez et al., 2010). Although the existence of the TDY-type clade D MAPKs is well established and the fact that all plants have MAPKs in this group indicates their functional importance, relatively little is known about the function of MAPK members of clade D. Moreover, functional redundancy between MAPKs makes it difficult to employ a genetic approach to determine their functions. Thus, rather little is known about the function of TDY MAPKs in plants. A wealth of information relating to TEY MAPKs exists, and most of what we know about MAPK function relates to the TEY MAPKs, with studies largely focused on a relatively small subset of MAPKs, primarily MPK3/4/6, which are all in the A and B subgroups of TEY-type MAPKs. Early research on plant MAPKs focused on their functions in immunity and stress responses, as many of the TEY MAPKs were found to function in these (for review, see Meng and Zhang, 2013). More recent studies have revealed that many TEY MAPKs also play key roles in plant growth and development (for review, see Xu and Zhang, 2015).
Only a few members of subfamily D have been functionally characterized to date, so insights into functions for TDY MAPKs are limited. However, all four of those that have been characterized to date are phylogenetically closely clustered together in a group. The alfalfa MAPK, TDY1, was one of the first of the TDY-clade MAPKs to be characterized. High expression in root nodules and root and shoot apices suggested a possible role in tip growth initiation/ development, and TDY1 promoter studies revealed induction by mechanical and pathogen wounding (Schoenbeck et al., 1999). In rice, the TDY MAPK BWMK1 (blast-and wound-induced MAPK) was shown to be induced by infection by both infection of the rice blast fungus Magnaporthe grisea and mechanical wounding (He et al., 1999). More recently, it was shown that BWMK1 is activated by fungal elicitors, hydrogen peroxide, salicylic acid, and jasmonic acid (Cheong et al., 2003). Other studies in cotton (Gossypium raimondii) showed that MPK9 (together with MPK13 and MPK25) confer resistance to Verticillium dahlia, which causes defoliation (Zhang et al., 2014). These studies provide evidence that these TDY MAPKs play a role in signaling to defense signaling. In contrast, a landmark study in Arabidopsis revealed that MPK9 was preferentially and highly expressed in stomatal guard cells and has a completely separate function. Studies revealed that MPK9 is functionally redundant with MPK12 (a TEY MAPK) and that together they acted as positive regulators of ROS-mediated abscisic acidinduced stomatal guard cell closure (Jammes et al., 2009). Although most of this study focused on MPK12, it clearly showed that MPK9 and MPK12 function together in regulating transpiration. More recently, it has been shown that these two MPKs also function in methyl jasmonate signaling to induce stomatal closure (Khokon et al., 2015). Data indicate that MPK9 and MPK12 function upstream of S-type anion channel activation and downstream of ROS production, cytosolic alkalization, and [Ca 2+ ] cyt elevation in guard cells (Khokon et al., 2015). A further Arabidopsis TDY MAPK has been functionally characterized recently: Arabidopsis MPK8 is activated by mechanical wounding, and it has been shown that the MPK8 pathway negatively regulates ROS accumulation by controlling the expression of Figure 7. Antisense oligonucleotides block the SI-induced DEVDase/ caspase-3-like activity. A, Representative image of an untreated (UT) pollen tube with no DEVDase/caspase 3-like activity. B, Corresponding bright-field image. C, Representative image of an SI-induced pollen tube after 4 h, using live-cell caspase 3/7 activity probe and an FITC filter set, showing fluorescence. D, Corresponding bright-field image. E, Representative image of an SI-induced pollen tube with antisense PrMPK9-1 oligonucleotide pretreatment (SI + as-ODN) after 4 h, using live-cell caspase 3/7 activity probe and an FITC filter set, with no detectable FITC signal. F, Corresponding bright-field image. G, Representative image of an SI-induced pollen tube with sense PrMPK9-1 oligonucleotide pretreatment (SI + s-ODN) after 4 h, using live-cell caspase 3/7 activity probe and an FITC filter set, with fluorescence detected. H, Corresponding bright-field image. Bar = 10 mm. I, Quantification of the effect of oligonucleotide pretreatment of pollen tubes (as-ODN or s-ODN) on DEVDase activity. Five independent experiments scored 50 pollen tubes in total for each treatment, and the percentage of pollen tubes in each category is indicated. Error bars indicate SE.
the RbohD gene (Takahashi et al., 2011). Here, we have shown that PrMPK9-1, which is closely related to these TDY MAPKs, functions in signaling to SI in incompatible pollen. We speculate that different functions have evolved for these MAPKs in different cell types.

PrMPK9-1 Functions in Pollen in the SI Signaling Network
To our knowledge, to date, no known function for AtMPK9 in pollen has been found, although the Arabidopsis Affymetrix chip data indicate that AtMPK9 is expressed in pollen. This suggests that this MAPK has evolved a specific function within the context of SI. PrMPK9-1 exhibits increased expression after SI induction, and the protein is activated by phosphorylation after SI induction. Here, using an antisense oligonucleotide approach, we have demonstrated that PrMPK9-1 has a crucial role in the signaling network leading to the SI-induced S-specific pollen tube inhibition and PCD. Our study identifies a key role for PrMPK9-1 in signaling to characteristic SI-mediated actin cytoskeleton reorganization (Geitmann et al., 2000;Snowman et al., 2002;Poulter et al., 2010) and activation of a DEVDase activity that is required for PCD in incompatible pollen (Thomas and Franklin-Tong, 2004;Bosch and Franklin-Tong, 2007). Moreover, our previous studies on SI have shown that actin alterations are required for PCD . Our data place PrMPK9-1 as a key player upstream of actin alterations and PCD. MAPK signaling to the actin cytoskeleton is well established (Šamaj et al., 2004). Integrating our existing knowledge of the SI signaling network with our new data on PrMPK9-1, we can place PrMPK9-1 activation, which peaks at 10 min after SI induction (Rudd et al., 2003), downstream of the almost instantaneous increases in [Ca 2+ ] cyt (Franklin-Tong et al., 1993, downstream of increases in ROS that peak at 5 min post SI , and upstream of the later formation of punctate actin foci and DEVDase activation. Thus, PrMPK9-1 occupies a critical position within the SI-PCD signaling network.

Close Relatives AtMPK8 and AtMPK9 May Function in Similar Signaling Networks
It is of considerable interest that two of the few Arabidopsis MAPKs belonging to the TDY group that have been functionally characterized so far, MPK8 and MPK9, are the closest relatives to the P. rhoeas PrMPK9-1 and have been shown to be involved in regulating ROS in different signaling networks (Jammes et al., 2009;Takahashi et al., 2011;Khokon et al., 2015). Thus, together, current data suggest that several TDY MAPKs, AtMPK8, AtMPK9, and PrMPK9-1, play a key role in regulating ROS homeostasis. As all three MAPKs are close relatives, this may not be a coincidence. It will be interesting to see if future studies show that further TDY MAPKs in this cluster generally share this phenomenon. Moreover, a study by Menges et al. (2008), using Gene Ontology overrepresentation analysis to analyze biological processes potentially involved in unknown MAPK functions, identified MPK9 as being potentially involved in cytoskeleton organization. Here, we have provided direct experimental evidence that the P. rhoeas PrMPK9-1 is upstream of and required for regulating the highly distinctive SI-induced actin reorganization into punctate foci. Indeed, it appears that the signaling network found in P. rhoeas pollen undergoing SI-PCD is rather similar (at least superficially) to that for MPK8 expressed in various tissues (Khokon et al., 2015). While the methyl jasmonate pathway in guard cells shares several signaling components with the abscisic acid pathway (ROS, [Ca 2+ ] cyt alterations, and cytosolic alkalization), the SI-PCD network also involves ROS and [Ca 2+ ] cyt increases and alterations in cytosolic pH, but a dramatic acidification is triggered (Wilkins et al., 2015). Future studies should explore whether other TDY MAPKs in this cluster generally share these key signaling components.
In summary, here, we show that PrMPK9-1 is a TDY MAPK that is expressed in a distinct cell type (pollen) and provide evidence that it has a cell type-specific function for PrMPK9-1, with a distinct role from AtMPK9. PrMPK9 is a key regulator for SI in pollen and acts upstream of PCD involving alterations in actin organization and activation of a DEVDase/caspase-3like enzyme. Thus, the role of MPK9-1 expressed in pollen plays a role in a very different biological phenomenon to the other TDY MAPKs analyzed at this level to date (ROS, [Ca 2+ ] cyt alterations, and cytosolic alkalization). However, the signaling network in which it is placed appears (at least superficially) similar to other closely related MAPKs that have been characterized to date. It will be interesting in the future to see what functions and signaling networks further cell type-specific MPKs are involved in.

MATERIALS AND METHODS
Cloning of Papaver rhoeas MPK9-1: Sequence Analyses and Affymetrix Analysis cDNA was synthesized using the Invitrogen SuperScript II Reverse Transcriptase kit using total RNA from P. rhoeas pollen (RNAeasy Plant Mini Kit; Qiagen). Initially, gene-specific primers based on Arabidopsis (Arabidopsis thaliana) MAPKs expressed in pollen (AtMPK8, AtMPK9, and AtMPK17) were used to obtain partial pollen-expressed P. rhoeas MPK cDNAs. An 880-bp cDNA fragment, amplified using primers MPK17-594 and MPK17-395, was cloned and sequenced and shared highest identity with AtMPK8 and AtMPK9. Full-length cDNAs of PrMPK9-1 were obtained using 39 and 59 RACE PCR. 39 RACE amplification from P. rhoeas pollen cDNA used 39 RACE (39RACE 17AP) and 59 gene-specific (PrMPK8-3R1) primers. The product was PCR purified and reamplified with 39 RACE primer (39RACE UAP) and nested gene-specific 59 primer (PrMPK8-3R2). For 59 RACE, cDNA was synthesized using a primer specific to the 39 end of the gene (PrMPK8-5R1) instead of the oligo(dT) primer. cDNA was cleaned using the QIAquick PCR Purification Kit (Qiagen). A poly(C) tail was added at the 59 end. PCR amplification employed a 59 RACE abridged primer (59RACE AAP) and a second genespecific 39 primer (PrMPK8-5R2). This amplification product was reamplified with a 59 RACE unabridged primer (59RACE UAP) and a third nested genespecific 39 primer (PrMPK8-5R3). Full-length sequence was confirmed by sequencing the mRNA product amplified using primers PrMPK91gen59 and PrMPK8gen39 (Table I).
DNA sequences were analyzed using Chromas software (Technelysium). Homology searches were prepared from the National Center for Biotechnology Information Web site utilizing the Entrez and BLAST (Altschul et al., 1997) functions. Alignments were made using ClustalW 1.7 and ClustalX2 (www.ebi. ac.uk/Tools/msa/clustalw2/). Arabidopsis MAPK gene expression was analyzed using data from Affymetrix gene chip experiments available via Genevestigator (https://www.genevestigator.com). Sequence alignments were prepared by ClustalW, and phylogenetic distributions were prepared by ClustalX on PHYLIP version 3.69.

Proteomic and Bioinformatic Analyses
For mass spectrometry analysis of the p56 protein, total pollen proteins were separated by SDS-PAGE and stained with Coomassie Blue. A band corresponding to 56 kD was excised and digested with Trypsin Gold (Promega). Peptides were identified by FT-ICR mass spectrometry searching NCBInr Green Plant database using the SEQUEST algorithm.

In Vitro Kinase Activity Assays
Kinase assays using pollen protein were performed according to Rudd et al. (1996). Briefly, pollen was homogenized on ice in 50 mM Tris-HCl, pH 7.5, and 0.1% Triton X-100 containing protease inhibitor cocktail, cell debris was removed by centrifugation, and supernatants were stored at 220°C until use. Recombinant His-tagged PrMPK9-1 was incubated with crude pollen extract in 50 mM Tris-HCl, pH 7.5, 1 mM ATP, 1 mM MgCl 2 , and 0.25 mM calyculin A for 15 min at 30°C. The His-tagged protein was isolated using an Ni-NTA Spin Column (Qiagen). Incorporation of 32 P into recombinant His-tagged proteins was measured after 10 min. Phosphorylated MAPK proteins were analyzed using SDS-PAGE followed by autoradiography. Phosphoproteins were enriched using the PhosphoProtein Purification Kit (Qiagen).

Pollen Tube Growth and SI Induction
Pollen was grown and SI was induced as described (Wilkins et al., , 2015. Briefly, recombinant PrsS proteins were expressed using nucleotide sequences specifying the mature peptide of the PrsS 1 and PrsS 3 alleles of the PrsS gene (pPRS100 and pPRS300) as described (Foote et al., 1994;Kakeda et al., 1998). SI was induced by adding recombinant PrsS (to 10 mg mL 21 ) to pollen of the appropriate S-haplotype to generate an incompatible combination (e.g. PrsS 1 and PrsS 3 with pollen from plants of haplotype S 1 S 3 ). For calyculin A treatment, growing pollen was treated with 0.25 mM calyculin A (Sigma-Aldrich) for 10 min.

Pollen and Leaf Protein Extracts
Pollen proteins were prepared according to Rudd et al. (1996). Briefly, pollen was homogenized on ice in 50 mM Tris-HCl, pH 7.5, and 0.1% Triton X-100 containing protease inhibitor cocktail, cell debris was removed by centrifugation, and supernatants were stored at 220°C. Leaf proteins were extracted in a similar way, after grinding with liquid N 2 . Protein concentration was estimated using the Bio-Rad assay.