In planta assessment of the role of thioredoxin h proteins in the regulation of S-locus receptor kinase signaling in transgenic Arabidopsis.

The self-incompatibility (SI) response of the Brassicaceae is mediated by allele-specific interaction between the stigma-localized S-locus receptor kinase (SRK) and its ligand, the pollen coat-localized S-locus cysteine-rich protein (SCR). Based on work in Brassica spp., the thioredoxin h-like proteins THL1 and THL2, which interact with SRK, have been proposed to function as oxidoreductases that negatively regulate SRK catalytic activity. By preventing the spontaneous activation of SRK in the absence of SCR ligand, these thioredoxins are thought to be essential for the success of cross pollinations in self-incompatible plants. However, the in planta role of thioredoxins in the regulation of SI signaling has not been conclusively demonstrated. Here, we addressed this issue using Arabidopsis thaliana plants transformed with the SRKb-SCRb gene pair isolated from self-incompatible Arabidopsis lyrata. These plants express an intense SI response, allowing us to exploit the extensive tools and resources available in A. thaliana for analysis of SI signaling. To test the hypothesis that SRK is redox regulated by thioredoxin h, we expressed a mutant form of SRKb lacking a transmembrane-localized cysteine residue thought to be essential for the SRK-thioredoxin h interaction. We also analyzed transfer DNA insertion mutants in the A. thaliana orthologs of THL1 and THL2. In neither case did we observe an effect on the pollination responses of SRKb-expressing stigmas toward incompatible or compatible pollen. Our results are consistent with the conclusion that, contrary to their proposed role, thioredoxin h proteins are not required to prevent the spontaneous activation of SRK in the A. thaliana stigma.

Many flowering plants possess self-incompatibility (SI), a genetic system that promotes outcrossing by preventing self-fertilization. In the Brassicaceae family, the SI response is controlled by haplotypes of the S locus, each of which contains two genes that encode highly polymorphic proteins, the S-locus receptor kinase (SRK), which is a plasma membrane resident single-pass transmembrane Ser/Thr receptor kinase displayed at the surface of stigma epidermal cells (Stein et al., 1991;Takasaki et al., 2000), and the S-locus Cysrich protein (SCR), which is the pollen coat-localized ligand for SRK (Schopfer et al., 1999;Kachroo et al., 2001;Takayama et al., 2001). SRK and SCR exhibit allele-specific interactions, whereby only SRK and SCR encoded by the same S-locus haplotype interact. In a self-pollination, the binding of this "self" pollen-borne SCR to the extracellular domain of SRK activates the SRK kinase, thereby triggering a cellular response in stigma epidermal cells that causes inhibition of pollen germination and tube penetration into the stigma epidermal cell wall (for review, see Tantikanjana et al., 2010).
Tight regulation of SRK kinase activity and its signaling cascade is critical for productive pollenstigma interactions because constitutive (i.e. SCRindependent) activity of the receptor is expected to result in sterile stigmas that reject both compatible and incompatible pollen. In the classical view of ligandactivated receptor kinases, the receptor occurs as catalytically inactive monomers in the absence of ligand and only becomes activated upon ligand-induced dimerization (for review, see Lemmon and Schlessinger, 2010). However, some receptor kinases in both animals (Chan et al., 2000;Ehrlich et al., 2011) and plants (Giranton et al., 2000;Wang et al., 2005Wang et al., , 2008Shimizu et al., 2010;Bücherl et al., 2013) form catalytically inactive dimers or oligomers in the absence of ligand, with receptor activation presumably resulting from ligand-induced higher order oligomerization or conformational changes (Lemmon and Schlessinger, 2010). Similar to the latter receptors, SRK forms oligomers in unpollinated stigmas, i.e. in the absence of SCR (Giranton et al., 2000), at least partly via ligandindependent dimerization domains located within the SRK extracellular domain (Naithani et al., 2007). It has been proposed that these ligand-independent SRK oligomers are maintained in an inactive state by thioredoxins, the ubiquitous small oxidoreductases that reduce disulfide bridges in proteins (Buchanan and Balmer, 2005). This hypothesis is supported by the following observations: (1) two Brassica napus thioredoxins, the Thioredoxin H-Like proteins THL1 and THL2, were identified as SRK interactors in a yeast (Saccharomyces cerevisiae) two-hybrid screen that used the B. napus SRK 910 kinase domain as bait (Bower et al., 1996); (2) when purified from pistils or insect cells, the Brassica oleracea SRK 3 variant was found to exhibit constitutive autophosphorylation activity in vitro, and this activity was inhibited by Escherichia coli-expressed THL1 proteins and was restored by addition of pollen coat proteins containing self but not of pollen coat proteins containing nonself SCR (Cabrillac et al., 2001); (3) the catalytic activity of THL1 was required for its inhibition of SRK 3 autophosphorylation activity in vitro (Cabrillac et al., 2001); and (4) antisense suppression of THL1/THL2 gene expression in the stigmas of a self-compatible B. napus strain reportedly produced a low-level constitutive incompatibility (Haffani et al., 2004), as might be expected if the THL1/THL2 proteins prevent the spontaneous activation of SRKmediated signaling in stigmas.
These observations notwithstanding, the in planta role of thioredoxin h proteins as negative regulators of SRK activity has not been conclusively demonstrated. To date, this proposed function has only been evaluated in a self-compatible strain of B. napus (Haffani et al., 2004). Consequently, it is not known if the proposed inhibitory effect of these thioredoxins on SRK catalytic activity is manifested in self-incompatible stigmas and if it applies to all SRK variants, be they derived from Brassica spp. or other self-incompatible species of the Brassicaceae such as Arabidopsis lyrata.
In this study, we tested the in planta role of thioredoxin h proteins in the regulation of SI signaling using a transgenic self-incompatible Arabidopsis thaliana model that we generated by transforming A. thaliana with the SRKb-SCRb gene pair isolated from the Sb haplotype of self-incompatible A. lyrata (Kusaba et al., 2001;Nasrallah et al., 2002Nasrallah et al., , 2004. We had previously shown that the stigmas of A. thaliana SRKb-SCRb transformants can exhibit an SI response that is as robust as the SI response observed in naturally selfincompatible A. lyrata, demonstrating that A. thaliana, which harbors nonfunctional S-locus haplotypes (Kusaba et al., 2001;Sherman-Broyles et al., 2007;Shimizu et al., 2008;Boggs et al., 2009c), has nevertheless retained all other factors required for SI. In view of the availability in A. thaliana of a highly efficient transformation method and numerous genetic resources, the A. thaliana SRK-SCR transgenic model has enabled the use of experimental approaches that are difficult or impossible to implement in Brassica species and has thus proven to be an invaluable platform for in planta analysis of SRK and SI signaling Boggs et al., 2009aBoggs et al., , 2009bTantikanjana et al., 2009;Tantikanjana and Nasrallah, 2012).
We therefore used this transgenic A. thaliana selfincompatible model to determine if abolishing the proposed SRK-thioredoxin h interaction or eliminating expression of the major thioredoxin h proteins expressed in stigmas would affect the outcome of selfor cross pollination. To this end, we expressed a mutant form of SRKb that lacked the Cys residue previously shown to be required for the interaction of SRK with THLs (Mazzurco et al., 2001), and we analyzed plants carrying knockout insertional mutations in thioredoxin h genes. Our results are inconsistent with the proposed role of thioredoxin h proteins as negative regulators of SRK catalytic activity and SI signaling.

THL1 and THL2 Orthologs in the A. thaliana Stigma
The A. thaliana thioredoxin h family of proteins consists of eight members (Reichheld et al., 2002). Phylogenetic analysis shows that three of these, the Thioredoxin H3 (AtTRX3), H4 (AtTRX4), and H5 (AtTRX5) genes, are the most closely related to Brassica spp. THL1 and THL2 proteins (Fig. 1A). Furthermore, AtTRX3, AtTRX4, and AtTRX5 are unique among all A. thaliana thioredoxin proteins in having the same reduction-oxidation (redox)-active site as Brassica spp. THL1 and THL2 proteins, which consists of the Trp-Cys-Pro-Pro-Cys sequence instead of the canonical sequence Trp-Cys-Gly-Pro-Cys found in other A. thaliana thioredoxin proteins ( Fig. 1B; Gelhaye et al., 2005). Because the amino acid residue immediately after the first Cys within the active site of thioredoxin is thought to play a major role in the protein's activity and specificity (Bréheĺin et al., 2000), the substrate specificities of Brassica spp. THL1 and THL2 and of AtTRX3, AtTRX4, and AtTRX5 are likely to be different from those of other A. thaliana thioredoxin h proteins. This conclusion is supported by the observation that AtTRX3 and AtTRX4 interact with Brassica spp. SRKs, while AtTRX1 and AtTRX2, two thioredoxin h proteins that contain the Trp-Cys-Gly-Pro-Cys active site sequence (Fig. 1B), do not (Mazzurco et al., 2001). Thus, we focused on the AtTRX3, AtTRX4, and AtTRX5 proteins as possible regulators of SRK catalytic activity in the A. thaliana stigma.
Using absolute quantitative real-time PCR (Wong and Medrano 2005), we found that AtTRX3, AtTRX4, and AtTRX5 are all expressed in A. thaliana stigmas, albeit at various levels ( Fig. 1C). AtTRX3 was expressed at the highest levels, followed by AtTRX4 and AtTRX5, which were expressed at approximately 11-and 19-fold lower levels, respectively. This result is consistent with microarray data available in GENEVESTIGATOR (Supplemental Fig. S1; Zimmermann et al., 2004) and with the results of our previous genome-wide transcriptional analysis of genes expressed specifically in stigma epidermal cells (Tung et al., 2005), both of which show AtTRX3 expression levels to be 6-fold higher than those of other thioredoxin h genes in A. thaliana stigmas.
Two lines of evidence indicate that AtTRX3 and AtTRX4, and not AtTRX5, are the orthologs of Brassica spp. THL1 and THL2, respectively (Supplemental Table S1). Amino acid sequence comparisons show that AtTRX3 is 78% identical to THL1 and AtTRX4 is 85% identical to THL2, while AtTRX5 is only 69% and 60% identical to THL1 and THL2, respectively. Furthermore, the A. thaliana chromosomal regions that encompass the AtTRX3 and AtTRX4 genes exhibit a high degree of synteny with the Brassica rapa chromosomal regions that contain Bra027469 and Bra025762 ( Fig. 1D; Supplemental Table S1), two thioredoxin h genes whose products exhibit, respectively, 98% and 99% amino acid sequence identity to B. napus THL1 and THL2. By contrast, the A. thaliana chromosomal region containing AtTRX5 is syntenous with a B. rapa chromosomal region that contains another thioredoxin h gene, Bra014037, whose protein product is 84% identical to AtTRX5 ( Fig. 1D; Supplemental Table S1), and is thus the likely ortholog of AtTRX5. This observation, together with the very low expression level of AtTRX5 in stigmas, led us to focus on AtTRX3 and AtTRX4 for assessing the potential role of thioredoxin h proteins in the pollination responses of the A. thaliana stigma.
A TM-Localized Cys Essential for the SRK-THL Interaction Is Not Required for the Inhibition of SRK-Mediated Signaling in the Absence of SCR Ligand In their analysis of the interaction of Brassica spp. SRKs with THL1 and THL2, Mazzurco et al. (2001) concluded that the SRK-THL1/THL2 interaction requires a Cys residue located near the C-terminal end of the predicted transmembrane (TM) domain of SRK ( Fig. 2A). This conclusion was predicated on the results of yeast two-hybrid interaction assays showing that the SRK-THL1/THL2 interaction was abolished when the TM-localized Cys was deleted or replaced by Ser, while the ability to interact with THL1/THL2 was acquired when a Cys was inserted at the appropriate position in the TM domain of an SRK-related receptor kinase that lacked this Cys and did not interact with THL1/THL2 (Mazzurco et al., 2001). Furthermore, the fact that this TM-localized Cys residue was conserved in all SRK sequences available at the time but was missing from SRK-related proteins that do not function in SI was interpreted as indicating that interaction with THL was a distinctive feature of SRK and the SI response it determines (Mazzurco et al., 2001).
Public databases now contain 44 SRK sequences containing the TM domain. A survey of these sequences, which are derived from self-incompatible strains of various species, showed that the majority of SRK variants contained a Cys residue at the appropriate location within the region that corresponds to the SRK TM domain as predicted by the TMpred program (Hofmann and Stoffel, 1993). By contrast, this TM-localized Cys residue is replaced by Trp, Phe, Ser, or Arg in five SRK variants from B. oleracea (BoSRK 68 ), A. lyrata (AlSRKa and AlSRK 3 ), Arabidopsis halleri (AhSRK 3 ), and Capsella grandiflora (CgSRK 7 ; Fig. 2A). Moreover, the TM-localized Cys is not required for the negative regulation of SRK in planta. The SRKs that lack this Cys have not been reported to produce a constitutive SI response in their native species or when introduced into A. thaliana. For example, A. thaliana stigmas expressing the CgSRK 7 variant ( Fig. 2A) exhibit a robust SI response toward self SCR 7 -expressing pollen but do not exhibit constitutive rejection of nonself pollen (Boggs et al., 2009b). These observations raise doubt about the notion that THLs, by virtue of their interaction with SRK through this TM-localized Cys residue, function as negative regulators of SRK in the absence of SCR ligand.
The yeast interaction studies of Mazzurco et al. (2001) had also shown that the interaction of Brassica spp. SRKs with the A. thaliana AtTRX3 and AtTRX4 proteins, like the SRK-THL1/THL2 interaction, requires the TM-localized Cys residue of SRK. Therefore, we reasoned that in transgenic A. thaliana stigmas, A. lyrata-derived SRKb, which contains this Cys residue, might interact with AtTRX3 and AtTRX4 and that mutating the TM-localized Cys in SRKb would abolish this interaction. In this manner, we would be able to assess in planta the importance of this Cys residue and of the SRK-thioredoxin h interaction for the regulation of SRK-mediated signaling. It should be noted that it is not feasible to test the Brassica spp. SRK variants used in the yeast interaction studies, such as the B. napus SRK 910 variant (Bower et al., 1996;Mazzurco et al., 2001). Despite repeated attempts, it has not been possible to transfer the SI trait to A. thaliana by expressing Brassica spp. SRKs, including SRK 910 (Bi et al., 2000) or several B. oleracea and B. rapa receptors belonging to highly diverged classes of SRKs (our unpublished results). This result indicates that Brassica spp. SRKs do not function in A. thaliana, possibly because the approximately 15 million years of evolution that separate Figure 2. Analysis of the TM-localized Cys residue in SRK. A, Amino acid sequence alignment of a region encompassing the TM region of SRK protein variants. Bold characters indicate the TM region predicted by TMpred. The box highlights the Cys residue, C463 in AlSRKb, thought to be essential for the interaction of SRK with thioredoxin h proteins. The arrow shows the C463W mutation introduced into SRKb. The underlined amino acids in BnSRK 910 indicate the residues that were included in the SRK kinase domain fragment used as bait in yeast two-hybrid studies (Bower et al., 1996;Mazzurco et al., 2001). B, Pollination assays of stage 13 stigmas from C24 transformants expressing the SRKb(C463W) mutant. Note the intense incompatibility response (manifested by lack of pollen tube growth) produced by C24 SCRb-expressing pollen and the fully compatible response (manifested by profuse pollen tube growth) produced by C24 wild-type (WT) pollen. Bar = 10 mm.
the Brassica spp. and A. thaliana lineages may have produced extensive divergence between the SRK substrate(s) in the two taxa.
To assay for the role of the TM-localized Cys residue, we used the AtS1pr::SRKb chimeric gene in which the SRKb transcriptional unit is placed downstream of the stigma epidermal cell-specific AtS1 promoter (Boggs et al., 2009a). We replaced the TM-localized Cys (C463) of SRKb with a Trp codon to recapitulate the substitution that occurs in two SRK variants ( Fig. 2A), and the resulting AtS1pr::SRKb(C463W) gene was introduced into A. thaliana plants of the C24 accession. As controls for this experiment, we used C24 plants transformed with the wild-type AtS1pr::SRKb transgene, which expresses a robust and SI response that persists throughout stigma development (Nasrallah et al., 2004;Boggs et al., 2009a).
Twelve independent transformants were analyzed by pollination assays and microscopic monitoring of pollen tube growth (Supplemental Table S2; "Materials and Methods"). As shown in Figure 2B, when transgenic AtS1pr::SRKb(C463W) stigmas were pollinated with "self" SCRb-expressing pollen from a C24[SRKb-SCRb] plant, a robust SI response was observed that was indistinguishable from that observed in the stigmas of C24 plants expressing the wild-type AtS1pr::SRKb transgene. These results demonstrate that the C463W mutation neither abolished nor weakened SRKb function and that all 12 AtS1pr::SRKb(C463W) transformants expressed the SRKb(C463W) protein at levels sufficient to confer SI. By contrast, when the same AtS1pr::SRKb (C463W) stigmas were pollinated with wild-type untransformed pollen (i.e. pollen lacking SCRb; Fig. 2B; Supplemental Table S2), we observed a compatible response that was identical, both in terms of pollen tube number and rate of tube growth, to that observed when the stigmas of C24 plants expressing the wild-type AtS1pr::SRKb transgene were pollinated with wildtype untransformed pollen. This observation indicates that, in the absence of SCRb, the SRKb(C463W) mutant is catalytically inactive or is active at levels that are too low to trigger SI signaling and inhibition of pollen. This result is inconsistent with the previously proposed notion that disrupting the SRK-thioredoxin h interaction in stigmas expressing a functional SRK would result in a constitutive incompatibility response that inhibits both self-and cross pollination (Cabrillac et al., 2001;Haffani et al., 2004).

Loss-of-Function Mutations in the AtTRX3 and AtTRX4 Genes Do Not Affect Pollination Responses in A. thaliana SRKb-Expressing Stigmas
In a second test of the potential role of AtTRX3 and AtTRX4 in pollination responses, we assessed the consequences of eliminating expression of these proteins in the self-incompatible stigmas of A. thaliana SRKb-SCRb plants carrying transfer DNA (T-DNA) insertions in these genes. Because T-DNA insertion lines are available in the Columbia (Col-0) accession but not in the C24 accession, we used Col-0 plants for this experiment. We had previously shown that Col-0 [SRKb-SCRb] transformants express transient SI, whereby the stigmas of young floral buds, stage 13 and early stage 14 according to Smyth et al. (1990), express as intense an SI response as observed in the naturally self-incompatible A. lyrata, with breakdown of SI at later stages of flower development (Nasrallah et al., 2002;Liu et al., 2007). Therefore, all pollination assays described below were performed using stage 13 bud stigmas.
We obtained two strains, hereafter designated trx3-1 and trx4-1, which carry T-DNA insertions in AtTRX3 and AtTRX4, respectively. We determined that the T-DNA in trx3-1 was inserted 28 bp before the second exon of the AtTRX3 gene (Fig. 3A) and that AtTRX3 transcripts were undetectable in the stigmas of trx3-1 homozygous plants (Fig. 3B). These results demonstrate that trx3-1 is a null mutation, as previously reported by Park et al. (2009), who failed to detect TRX3 protein in trx3-1 homozygotes. Similarly, we found that the stigmas of trx4-1 homozygotes, in which the T-DNA was inserted within the first intron of the AtTRX4 gene (Fig.  3A), lacked AtTRX4 transcripts (Fig. 3B), demonstrating that trx4-1 is also a null mutation.
Our observation that null mutations in either AtTRX3 or AtTRX4 have no effect on the pollination responses of Col-0[SRKb-SCRb] stigmas might be due to functional redundancy of the two genes. Therefore, we crossed trx3-1[SRKb-SCRb] and trx4-1[SRKb-SCRb] to generate SRKb-SCRb plants that were homozygous for both trx3-1 and trx4-1. Reverse transcription (RT)-PCR demonstrated that the stigmas of these trx3-1 trx4-1 [SRKb-SCRb] plants lacked AtTRX3 or AtTRX4 transcripts (Fig. 3B), and pollination assays showed that these stigmas exhibited an intense SI response toward pollen from Col-0[SRKb-SCRb] and a fully compatible response toward pollen from untransformed wild-type Col-0 (Fig. 3C). Because the pollination responses of trx3-1 trx4-1[SRKb-SCRb] were indistinguishable from those of wild-type Col-0[SRKb-SCRb] plants, we conclude that loss of AtTRX3 and AtTRX4 expression in A. thaliana stigmas has no effect on the catalytic activity of SRKb, either in the presence or absence of SCRb.

DISCUSSION
Reversible thiol-based redox modifications have emerged as a major mechanism for regulating the activity of proteins involved in many biological processes, ranging from transcription to signal transduction. In particular, studies in mammalian systems have shown that receptor kinase-mediated signaling can be either stimulated or inhibited by the disulfide bond-reducing activity of thioredoxins (Truong and Carroll, 2013). Consequently, the proposal that SRK catalytic activity is inhibited by thioredoxin h proteins in the absence of its cognate SCR ligand and is stimulated when binding of cognate SCR disrupts the SRK-thioredoxin h interaction (Cabrillac et al., 2001) provides an attractive mechanism for explaining the tight control of SRKmediated signaling in the stigmas of self-incompatible crucifers.
However, our results do not support the hypothesis that SRK is redox regulated by thioredoxin h proteins. Certainly, any regulation of SRK by these thioredoxins, if it occurs at all, does not apply to all SRKs. We found no evidence that AtTRX3/AtTRX4 and their proposed interaction with SRKb contribute to the outcome of pollinations, be they incompatible or compatible. Neither mutating the TM-localized Cys residue that was reportedly required for the interaction of Brassica spp. SRKs with thioredoxin h proteins nor introducing null mutations in the AtTRX3 and AtTRX4 genes produced qualitative or quantitative changes in the pollination responses of SRKb-expressing A. thaliana stigmas toward either SCRb-expressing pollen or wild-type pollen. Despite the total lack of AtTRX3 and AtTRX4 transcripts in the stigmas of trx3-1 trx4-1[SRKb-SCRb] plants, we were unable to reproduce the constitutive incompatibility reported for the stigmas of B. napus antisense THL1/THL2 transformants, in which THL1 and THL2 transcripts were reduced but not eliminated (Haffani et al., 2004). The images illustrate pollination responses toward untransformed wild-type Col-0 pollen (top row) and Col-0 SCRb-expressing pollen (bottom row). The genotype of stigmas used for pollination is indicated in each section. Note that, irrespective of their genotype at the AtTRX3 and AtTRX4 loci, all stigmas pollinated with wild-type pollen produced profuse pollen tube growth and all stigmas expressing SRKb exhibited an intense incompatibility response toward SCRb-expressing pollen. Bar = 10 mm.
We conclude that the AtTRX3 and AtTRX4 thioredoxin h proteins do not regulate SRKb or its downstream effectors through their disulfide-reducing activity, nor do they regulate SRKb signaling independent of their catalytic activity, as described for the tomato (Solanum lycopersicum) signaling pathway that is mediated by the Cladosporium fulvum resistance-9 (Cf-9) protein and confers resistance to pathogen strains expressing the Avirulence gene9 (Avr9) elicitor. In this pathway, the Cf-9-interacting thioredoxin CITRX is thought to function as a negative regulator, not via its reducing activity, but by acting as an adaptor protein that recruits the cytosolic kinase Avr9/Cf9 induced kinase1 to the Cf-9 cytoplasmic domain (Rowland et al., 2005;Nekrasov et al., 2006). Importantly, our results are not due to some peculiar property of the SRKb variant or SRKs from A. thaliana species, because the TM-localized Cys residue thought to be essential for the SRK-THL and SRK-AtTRX3/ AtTRX4 interactions is missing in SRK variants from Brassica spp. and C. grandiflora as well as A. lyrata and A. halleri. This observation underscores the conclusion that this Cys residue is not required for proper regulation of the activity of all SRKs.
How then might our results be reconciled with published reports on the role of thioredoxin h proteins in redox regulation of SRK? Regarding the observed interaction in yeast between SRK and THLs via the TM-localized Cys residue of SRK, one possibility is that this interaction is an artifact of the yeast twohybrid assays used to reveal and subsequently analyze the SRK-THL interaction. Thioredoxins reduce proteins by Cys thiol-disulfide exchange, a process that involves the formation of a disulfide bond between a Cys in the thioredoxin active site and a Cys in the target protein. Thus, the notion that a cytosolic thioredoxin can interact with a Cys residue that is buried in the TM domain of SRK is inherently problematic. Interestingly, the SRK fragments that were used as bait for yeast two-hybrid screening contained the kinase domain and an N-terminal extension consisting of six amino acids derived from the TM region ( Fig. 2A; Bower et al., 1996;Mazzurco et al., 2001). In these fragments, the TM-localized Cys residue would not be embedded in a membrane and would be available for disulfide bridge formation with thioredoxin in the yeast nucleus, a situation that does not occur in planta.
As for the in vitro experiments showing that THL1 inhibits SRK activity and that this inhibition is relieved by addition of pollen coat proteins from "self" pollen (Cabrillac et al., 2001), another in vitro study of SRK activation (Takayama et al., 2001) had questioned the involvement of thioredoxins. While both studies concluded that "self" SCR caused activation of SRK as measured by an increase in SRK autophosphorylation activity, this activation was observed in the presence of THL1 in the first study (Cabrillac et al., 2001) but did not require the presence of thioredoxin in the second (Takayama et al., 2001). We suggest that THL1, which is phosphorylated by SRK in vitro (Bower et al., 1996), acts as an SRK pseudosubstrate that competes with SRK for autophosphorylation, leading to the apparent inhibition of SRK autophosphorylation activity upon addition of THL1.
Finally, with regards to the low-level constitutive incompatibility response observed by antisense suppression of THL1/THL2 (Haffani et al., 2004), several factors complicate interpretation of this phenotype. The antisense construct used in this study contained the full-length THL1 or THL2 sequence, raising the possibility that the antisense silencing effect might not have been specific for these THL genes or even other thioredoxin genes. Importantly, the study did not use a strain whose stigmas express SI. Rather, it was performed using a self-fertile strain of B. napus in which the basis of self-fertility was not known and the functionality of the endogenous SRK was not established. Furthermore, no changes in SRK kinase activity or complex formation were observed in the THL1/THL2 antisense plants relative to wild-type plants. As a result, the authors acknowledged that the pollination phenotype observed in antisense THL1/THL2 stigmas does not provide conclusive evidence for the negative regulation of SRK by THL1/THL2 (Haffani et al., 2004).
Our results leave open the question of how the ligand-independent SRK oligomers observed in unpollinated stigmas are maintained in an inactive state. Although AtTRX3, and to a lesser extent AtTRX4, are the major thioredoxins in the A. thaliana stigma, we cannot rule out the possibility that another thioredoxin might function as a negative regulator of SRK catalytic activity by reducing one of several disulfide bridges that link Cys residues located outside the TM domain. However, it is possible that the ligand-independent SRK oligomers are inactive, not only due to the effect of some inhibitory factor, but also because they assume an autoinhibitory conformation in the absence of the SCR ligand. Ligand-activated receptor kinases are typically regulated by multiple mechanisms (Lemmon and Schlessinger, 2010;Endres et al., 2011). A prime example is the A. thaliana BRASSINOSTEROID-INSENSITIVE1 (BRI1) receptor, which is subject to two modes of inhibition, cis-inhibition by its C-terminal domain and trans-inhibition by the BRI1 Kinase Inhibitor1 protein, both of which are relieved by binding of the brassinolide ligand to the BRI1 extracellular domain (Wang et al., 2005;Wang and Chory, 2006;Jaillais et al., 2011aJaillais et al., , 2011b. Further studies of SRK, including resolution of crystal structures for the receptor in its ligand-unbound and ligand-bound forms, are required to understand how tight regulation of SRK activity is achieved to prevent the spontaneous activation of SRK-mediated signaling and the illegitimate inhibition of compatible pollen.

Plant Materials
The Arabidopsis thaliana C24 and Col-0 plants used in this study were grown at 22°C under continuous light. Col-0[SRKb-SCRb] transformants containing the Arabidopsis lyrata SRKb and SCRb genes with their native 59 and 39 regulatory regions were previously described (Nasrallah et al., 2002(Nasrallah et al., , 2004. The T-DNA insertion mutants (Alonso et al., 2003) in AtTRX3 (SALK_111160) and AtTRX4 (SALK_091998) were obtained from the Arabidopsis Biological Resource Center (Ohio State University). Plant genotypes were determined by PCR using the primers listed in Supplemental Table S3.

Transgenes and Plant Transformation
The pCAMBIA derivative containing the AtS1pr::SRKb chimeric gene, in which the A. lyrata SRKb transcriptional unit is placed downstream of the stigma epidermal cell-specific AtS1 promoter, was previously described (Boggs et al., 2009a). To generate the SRKb(C463W) mutant transgene, the AtS1pr::SRKb-SCRb plasmid served as template for recombinant PCR using the SRKb(C463W) forward (59-GTT CCA TCA TGT TCT GGG TTT GGA GAA GGA-39) and SRKb(C463W) reverse (59-TCC TTC TCC AAA CCC AGA ACA TGA TGG AAC-39) primers. All plasmids were sequenced at the Cornell University Life Sciences Core Laboratories Center to exclude the presence of PCR-generated errors. The plasmids were introduced into C24 wild-type plants by the floral dip method (Clough and Bent, 1998), and transformants were selected on Murashige and Skoog medium (Sigma-Aldrich) containing 50 mg mL -1 hygromycin.

Pollination Assays
One day before performing the assays, stage 12 flower buds were emasculated (for staging of flower development, see Smyth et al., 1990). The stigmas of these buds, now at stage 13 of development, were manually pollinated with pollen grains from mature postanthesis flowers under a stereomicroscope. Two hours after pollination, the stigmas were fixed, stained with decolorized aniline blue, and examined by epifluorescence microscopy as previously described (Kho and Bear, 1968). Each pollination assay was performed in triplicate. In these assays, an incompatible response is manifested by the growth of fewer than 10 pollen tubes per pollinated stigma, a partially incompatible response by growth of 10 to 29 pollen tubes per pollinated stigma, and a compatible response by growth of more than 30 pollen tubes per pollinated stigma.

Expression Analysis
Total RNA was extracted from 25 stage 12 stigmas using the TRIzol reagent (Invitrogen). The RNA was treated with DNaseI (Invitrogen), and 0.5 mg of this RNA was used to synthesize first-strand complementary DNAs using an oligo (dT) primer and the First-Strand cDNA Synthesis Kit for Real-Time PCR (USB). RT-PCR was performed using the gene-specific primers listed in Supplemental Table S1 under the following amplification parameters: 94°C for 30 s, 55°C for 45 s, and 72°C for 2 min.
Absolute quantitative real-time PCR was performed according to Wong and Medrano (2005) in an Applied Biosystems ViiA 7 Real-Time PCR System using the HotStart-IT SYBR Green qPCR Master Mix (USB) and gene-specific primers (Supplemental Table S3). PCR amplification was performed under the following conditions: 95°C for 2 min, followed by 40 cycles of 95°C for 15 s, 55°C for 30 s, and 72°C for 30 s. The amounts of transcripts were determined using the standard curve method, with three replicates of each sample. Standard double-stranded DNAs were prepared from the synthesized firststrand cDNAs by PCR using gene-specific primers (Supplemental Table S3), and the DNA concentration was determined by UV A 260 .

Sequence Analysis
The TMpred Server was used to predict the TM region of SRK protein variants (Hofmann and Stoffel, 1993). Sequences were aligned using ClustalW (Larkin et al., 2007). The thioredoxin h phylogenetic tree was constructed with the neighbor-joining method using the MEGA5.1 program (Tamura et al., 2011). The plant genome duplication database (Lee et al., 2013) was used for analysis of synteny between genomic regions of A. thaliana and Brassica rapa.
The T-DNA left border region and flanking genomic region were amplified by PCR with the 111160RP/LBa1 primer pair for trx3-1 and the 091998RP/ LBa1 primer pair for trx4-1, and the amplified DNA fragments were sequenced at the Cornell University Life Science Core Laboratories Center.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. AtTRX gene expression patterns based on expression data available in public databases.
Supplemental Table S1. Comparison of A. thaliana AtTRX3, AtTRX4, and AtTRX5 and their closest B. rapa relatives.
Supplemental Table S2. Pollination phenotypes of stigmas from C24 plants expressing the SRKb(C463W) mutant or the wild-type SRKb control.
Supplemental Table S3. Primers used for genotyping, real-time PCR, and RT-PCR.