JAZ is essential for ligand specificity of the COI1/JAZ co-receptor

Significance

Phytohormones play a key role in plant environmental adaptation and survival and are perceived by specific receptors. Uncovering the determinants of ligand specificity of hormone receptors is a fundamental scientific question essential to understanding how signaling pathways appeared and evolved. By comparing the specific features in the COI1/JAZ co-receptor that determine its recognition of JA-Ile in angiosperms or dn-OPDA in bryophytes, we have discovered a JAZ role in determining ligand specificity of the COI1/JAZ co-receptor. Furthermore, we have identified that, by retaining a co-receptor with features of both bryophytes and angiosperms, lycophytes represent a key transition intermediate in the evolution of the ligand–receptor interaction in the jasmonate pathway.

Abstract

Jasmonates are phytohormones that regulate defense and developmental processes in land plants. Despite the chemical diversity of jasmonate ligands in different plant lineages, they are all perceived by COI1/JAZ co-receptor complexes, in which the hormone acts as a molecular glue between the COI1 F-box and a JAZ repressor. It has been shown that COI1 determines ligand specificity based on the receptor crystal structure and the identification of a single COI1 residue, which is responsible for the evolutionary switch in ligand binding. In this work, we show that JAZ proteins contribute to ligand specificity together with COI1. We propose that specific features of JAZ proteins, which are conserved in bryophytes and lycophytes, enable perception of dn-OPDA ligands regardless the size of the COI1 binding pocket. In vascular plant lineages beyond lycophytes, JAZ evolved to limit binding to JA-Ile, thus impeding dn-OPDA recognition by COI1.

Jasmonates (JAs) are essential hormones for plant survival since they regulate environmental adaptation and developmental processes through extensive transcriptional reprogramming (1, 2). JAs are major players in the activation of plant responses to necrotrophic pathogens and insect feeding (3). They also regulate developmental traits such as growth, flower development and fertility, trichome density, and secondary metabolism.

JAs are fatty acid-derived oxylipins structurally similar to animal prostaglandins (4) and regulate biological processes by activating a signaling pathway conserved in land plants. A co-receptor complex formed by COI1 and a JAZ protein perceives JA ligands (5–8). COI1 is the F-box component of an SCF-type E3-ubiquitin ligase (9, 10). JAZs play a dual role, balancing their function as part of the hormone co-receptor complex with their role in the repression of downstream transcription factors (TFs) such as MYCs and other TFs involved in JA signaling (5, 7, 11–15). In basal conditions, these TFs are repressed by JAZ recruitment of a corepressor complex formed by TOPLESS and Histone Deacetylases through the adaptor protein NINJA (16). JAZs also prevent the interaction of TFs with the general transcriptional machinery through MED25, a component of the MEDIATOR complex (17, 18). Termination of JA signaling by MYC2 is achieved by an autoregulatory negative feedback loop involving MYC2-targeted bHLH1 TFs that impair the formation of the MYC2-MED25 complex (19). Accumulation of the hormone triggers the interaction of the two components of the co-receptor (COI1 and JAZ), which leads to JAZ ubiquitination and degradation by the proteasome. JAZ degradation liberates the TFs from the corepressor complex and allows activation of the JA-induced transcriptional reprogramming (2, 17).

The core signaling components (COI1, JAZ, and MYC) are functionally conserved in all embryophytes, although their number is very different between bryophytes and eudicots. For instance, in the liverwort Marchantia polymorpha there is only one JAZ, whereas Arabidopsis contains thirteen (20–24).

Despite the conservation of the signaling components in embryophytes (20), the ligand activating this pathway is different in bryophytes and eudicots. Thus, whereas the COI1/JAZ ligand in Arabidopsis thaliana is JA-Ile (5–7, 25), the ligand recognized by this co-receptor in the liverwort M. polymorpha is dn-OPDA, a precursor of JA-Ile in angiosperms (8). Indeed, bryophytes, including M. polymorpha, synthesize OPDA and dn-OPDA but do not produce JA-Ile due to a lack of key biosynthetic enzymes (OPR3 and JAR1; (8)) and do not respond to this hormone because their COI1 receptor has a smaller binding pocket unable to accommodate JA-Ile (8). In contrast, the AtCOI1 receptor, having a bigger pocket, is able to bind both JA-Ile and dn-OPDA and to complement the Mpcoi1 mutation (8).

The crystal structure of A. thaliana COI1-JAZ complex showed that JA-Ile or its mimic COR bind the COI1 binding pocket and JAZ interacts with COI1 and the hormone, trapping the ligand in the COI1 pocket (7). Amino acid substitutions in the COI1 pocket result in changes in ligand specificity (8, 26), whereas a mutation between the F-box and the Leucine-rich repeats (LRRs) restores the JA sensitivity of deleterious mutations in COI1 (27). From an evolutionary perspective, we previously demonstrated that a single amino acid change in COI1 was responsible for the switch in ligand specificity from dn-OPDA in bryophytes to JA-Ile in angiosperms (8). Thus far, the contribution of JAZ to ligand specificity has not been studied. Even though JAZ1 establishes direct interactions with the ligand in the crystal structure (7), there is no evidence for different JAZ proteins determining recognition of different ligands by the co-receptor complex.

Dn-OPDA is a precursor of JA-Ile biosynthesis, and, therefore, accumulates in vascular plants. To understand whether dn-OPDA could also be a ligand of the AtCOI1-AtJAZ co-receptor in A. thaliana and thus regulate responses so far attributed solely to JA-Ile, we studied the determinants of ligand specificity in the co-receptors. We found that, besides COI1, JAZ also determines ligand specificity. MpJAZ specifies binding of the co-receptor to dn-OPDA, independently of the origin of COI1 protein (MpCOI1 or AtCOI1). Indeed, by combining the different components of the co-receptor, we were able to fully switch ligand specificity from dn-OPDA to JA-Ile in M. polymorpha plants.

Furthermore, we identified a key evolutionary step within the lycophytes as a transition intermediate with a co-receptor combination of bryophyte-type JAZ and angiosperm-type COI1, leading to a broad ligand specificity. Our results demonstrate how evolution shaped the ligand–receptor interaction in the jasmonate pathway and how the evolution of JAZ repressors in angiosperms limited COI1 binding to JA-Ile, impeding the recognition of dn-OPDA as a ligand.

Results

Ectopic Overexpression of AtJAZ Confers OPDA Insensitivity in M. polymorpha.

As described in the introduction, activation of the JA pathway by exogenous treatment, or derepression in jaz mutants, inhibits growth (21). We have previously shown that JAZ function is conserved in land plants (21). The expression of either MpJAZ or AtJAZ3 restores the wild-type phenotype of dwarf Mpjaz-1 plants in control conditions, indicating that JAZ from different species can effectively repress the jasmonate pathway in M. polymorpha (Fig. 1 A and B; (21)). However, in contrast to the complementation by MpJAZ, which almost restores wild-type (WT) responsiveness to the dn-OPDA precursor OPDA, Mpjaz-1 plants overexpressing AtJAZ3 were mostly insensitive to OPDA (Fig. 1 A and B). This suggested that despite acting as a repressor in M. polymorpha, AtJAZ3 escapes the regulation by MpCOI1. To test if this escape is caused by a lack of AtJAZ3 interaction with MpCOI1, we performed pull-down assays between plant-expressed MpCOI1-FLAG and recombinant JAZ (from At or Mp) fused to MBP in the presence of dn-cis-OPDA, which derives from OPDA and is one of the bioactive ligands of MpCOI1/MpJAZ (8). Dn-cis-OPDA induced the interaction between MpCOI1 and MpJAZ but not between MpCOI1 and AtJAZ3 or the truncated version of MpJAZ lacking the Jas domain (Fig. 1C). This result indicates that AtJAZ3 cannot interact with MpCOI1 to form an active receptor complex with dn-cis-OPDA, therefore explaining the insensitivity to the hormone of AtJAZ3/Mpjaz-1 plants.

Fig. 1.

AtJAZ3 expression confers OPDA insensitivity in M. polymorpha. (A) Growth-inhibitory effect of OPDA (10 and 50 µM) on 14-d-old M. polymorpha gemmalings of WT Tak-1, Mpjaz-1 mutant and EF:MpJAZ-Citrine/Mpjaz-1 and EF:AtJAZ3-Citrine/Mpjaz-1 plants. Scale bar, 1 cm. (B) Growth quantification of plants shown in (A). (C) Immunoblot (anti-flag antibody) of recovered MpCOI1-flag (from 35S:MpCOI1-flag Arabidopsis extracts) after pull-down reactions using recombinant MBP, full-length (FL) MpJAZ-MBP, MpJAZΔJas-MBP or AtJAZ3-MBP protein alone (mock; −) or with dinor-cis-OPDA (50 µM). Bottom, Coomassie blue staining of recombinant proteins after cleavage with Factor Xa showing cleaved MBP.

JAZ Determines Ligand Specificity Together with COI1.

We have previously demonstrated that COI1 contributes to ligand specificity in land plants (8). Thus, MpCOI1 binds dn-OPDA but not JA-Ile or COR, and, therefore, M. polymorpha plants are insensitive to JA-Ile/COR. In contrast, AtCOI1 can accommodate all three molecules (8). Expression of AtCOI1 in M. polymorpha complements Mpcoi1 insensitivity to OPDA and confers additional responsiveness to both JA-Ile and COR (8). As dn-cis-OPDA is an endogenous molecule in A. thaliana, and AtCOI1 mediates dn-cis-OPDA perception when expressed in M. polymorpha (8), we questioned whether AtCOI1 can form a co-receptor with any AtJAZ proteins to perceive this molecule in A. thaliana. Our pull-down experiments with the 12 canonical AtJAZ showed that none of them could form a complex with AtCOI1 and dn-cis-OPDA (Fig. 2A), which suggests that dn-cis-OPDA is not a hormone perceived by COI1/JAZ in A. thaliana. The co-receptor complex AtCOI1/MpJAZ was the only one formed in the presence of dn-cis-OPDA. These results suggest that not only COI1, but also JAZ determine ligand specificity.

Fig. 2.

JAZ determines ligand specificity together with COI1. (A) Immunoblot (anti-flag antibody) of recovered AtCOI1-flag (from 35S:AtCOI1-flag Arabidopsis extracts) after pull-down reactions using recombinant MBP, AtJAZ1-12-MBP, or MpJAZ-MBP protein alone (mock; −) or with dinor-cis-OPDA (50 µM). Bottom, Coomassie blue staining of recombinant proteins after cleavage with Factor Xa. (B) Growth-inhibitory effect of OPDA (50 µM), (+)-JA-Ile (50 µM), or COR (0.5 µM) on 14-d-old M. polymorpha gemmalings of WT Tak-1, EF:AtCOI1/Mpcoi1-1/EF:AtJAZ3-Citrine/Mpjaz-1, EF:AtJAZ3-Citrine/Mpjaz-1, EF:AtCOI1/Mpcoi1-1 and Mpcoi1-1 mutant plants. Scale bar, 1 cm. (C) Growth quantification of plants shown in (B).

To further explore whether JAZ modulates ligand specificity of the co-receptor complex in planta, we replaced the M. polymorpha co-receptor MpCOI1/MpJAZ with AtCOI1/AtJAZ3 by crossing AtCOI1/Mpcoi1-1 female plants and AtJAZ3/Mpjaz-1 male plants. The resulting AtCOI1/AtJAZ3/Mpcoi1-1/Mpjaz-1 plants were fully responsive to JA-Ile and COR, but completely insensitive to OPDA (Fig. 2 B and C), indicating that the AtCOI1/AtJAZ3 complex cannot perceive the OPDA-derived hormone dn-OPDA in M. polymorpha. Therefore, the combination MpCOI1/MpJAZ (in WT Tak-1) specifies the dn-OPDA ligand (Fig. 2 B and C; (8)), AtCOI1/AtJAZ3 (in AtCOI1/AtJAZ3/Mpcoi1-1/Mpjaz-1 plants) specifies JA-Ile and COR, AtCOI1/MpJAZ (in AtCOI1/Mpcoi1) recognizes all three dn-OPDA, JA-Ile, and COR, and finally, MpCOI1/AtJAZ3 (in AtJAZ3/Mpjaz-1) does not recognize any of the ligands. These results indicate that in replacing MpJAZ with AtJAZ, the perception of JA ligands becomes restricted to JA-Ile and COR, as the recognition of the natural Marchantia ligand, dn-OPDA, is lost. Importantly, these results demonstrate that JAZ proteins participate, together with COI1, in the specification of the JA ligand, (Fig. 2 B and C). In other words, only the replacement of both COI1 and JAZ can cause switch of the specificity of the Marchantia co-receptor, which perceives dn-OPDA but not JA-Ile and COR, to the Arabidopsis co-receptor, which binds JA-Ile and COR, but not dn-OPDA. Thus, regardless of the COI1 protein in play (i.e., either that from bryophytes or angiosperms), it is MpJAZ that enables dn-OPDA binding to the co-receptor complex.

The Combination of EQ Amino Acids in the JAZ Degron Specifies Binding to dn-OPDA.

To identify the determinants of ligand specificity by JAZ, we compared the Jas domain, responsible for the interaction with COI1, between Marchantia and Arabidopsis JAZ (Fig. 3A). Alignment of the 27 amino acids of the Jas domain of MpJAZ and AtJAZs did not show major differences between MpJAZ and AtJAZ degrons (Fig. 3A). Nevertheless, we observed that MpJAZ has the positively charged basic residue (R) in the C terminus of the Jas domain, before the well-conserved PY, that was not present in any AtJAZ (Fig. 3A). Instead of R, all AtJAZ showed polar uncharged (Q/T) or nonpolar (V/I) amino acids in this position (Fig. 3A). To test the impact of this R from MpJAZ in ligand specificity, we introduced this R and its neighboring residues (GKA) in the equivalent positions of AtJAZ3, resulting in the chimeras AtJAZ3^R and AtJAZ3^RGKA. Pull-down experiments using these recombinant chimeras, MpCOI1-FLAG expressed in planta and two dn-OPDA isomers, showed that the introduction of these residues did not result in MpCOI1/dn-cis-OPDA binding, thus indicating that these residues are not involved in ligand specification (Fig. 3 A and B).

Fig. 3.

The Jas degron is likely responsible for differences in hormone binding in bryophytes and angiosperms. (A) Multiple sequence alignment of the Jas degron from MpJAZ and the 12 canonical AtJAZ. Asterisks indicate the position of amino acids mutated in the JAZ3 chimeras. (B) Immunoblot (anti-flag antibody) of recovered MpCOI1-flag (from 35S:MpCOI1-flag Arabidopsis extracts) after pull-down reactions using recombinant MBP, WT, and chimeric versions of AtJAZ3-MBP, and MpJAZ-MBP alone (mock; −) or with dinor-OPDA isomers (cis and iso; 50 µM). Bottom, Coomassie blue staining of recombinant proteins after cleavage with Factor Xa. (C) Comparison between AtCOI1/AtJAZ3/JA-Ile and MpCOI1/MpJAZ/dn-cis-OPDA models. Alignment of both complexes. Details of the AtCOI1/AtJAZ3/JA-Ile (D) and MpCOI1/MpJAZ–dn-cis-OPDA (E) complexes showing the COI1 residues involved in interactions with the seven N-terminal residues of JAZ degrons (shown as sticks). Schemes comparing the interactions for AtCOI1/AtJAZ3/JA-Ile (F) and MpCOI1/MpJAZ–dn-cis-OPDA (G) complexes by representing the sequence fragments that contain the COI1’s residues hydrogen bonded to JAZ degrons. All the structures are shown with the same colors in all the panels: AtCOI1 is in gray, MpCOI1 is in yellow, AtJAZ3 is in cyan, MpJAZ is in purple, JA-Ile and dn-cis-OPDA are in green. Dashed lines represent hydrogen bond interactions. (H) Growth-inhibitory effect of OPDA (10 and 50 µM) on 14-d-old M. polymorpha gemmalings of WT Tak-1, Mpjaz-1, EF:MpJAZ-Citrine/Mpjaz-1, EF:AtJAZ3-Citrine/Mpjaz-1 and EF:AtJAZ3^EQ/Mpjaz-1 plants. Scale bar, 1 cm. (I) Growth quantification of plants shown in (H).

Since the JAZ degron has been shown to be sufficient for AtCOI1/ligand binding (7), we then reassessed our analysis of sequence conservation within the JAZ degron, which is composed by an N-terminal loop and a C-terminal α-helix (Fig. 3A). We observed that each amino acid of the MpJAZ degron was present in at least two AtJAZs (Fig. 3A). Nevertheless, the amino acid combination of both E in position 1 and Q in position 4 of the loop region of MpJAZ degron (ELPQ) is unique and does not appear in any AtJAZ (Fig. 3A). Molecular modeling revealed that the MpJAZ degron (ELPQARK) adopts a different extended loop conformation compared with the conformation of AtJAZ3 in the AtCOI1/AtJAZ3/JA-Ile model, which is similar to the AtJAZ1 conformation in the AtCOI1/AtJAZ1 crystal structure with JA-Ile or COR (Fig. 3 C–G; (7)). Our model predicted MpJAZ intermolecular hydrogen bond interactions between E200 and Q203 from MpJAZ and R439 and S407 from MpCOI1, respectively (Fig. 3 C, E, and G). Moreover, the MpCOI1/MpJAZ/dn-cis-OPDA model showed that the interactions corresponding to AtCOI1 R351 and Y472 with AtJAZ3 L203 and L201 were not maintained since these two AtCOI1 residues were not conserved in MpCOI1 (Fig. 3 C–G). Consistent with our previous results on COI1 ligand specificity depending on AtCOI1^A384/MpCOI1^V377 (8), our models showed that the binding pocket of AtCOI1 is larger than that of MpCOI1. The smaller binding pocket of MpCOI1 can therefore accommodate smaller ligands like dn-cis-OPDA or dn-iso-OPDA. Notably, the AtCOI1/MpJAZ/dn-cis-OPDA model supported the possibility of MpJAZ adopting a similar conformation as in the model with MpCOI1 when bound to a less bulky ligand as dn-cis-OPDA. In such conformation, our model shows that MpJAZ E200 and Q203 would interact with AtCOI1 Q447 and D414 (SI Appendix, Fig. S1A). Focusing on the alternative conformation of MpJAZ with either MpCOI1 or AtCOI1 and the dn-OPDA isomers, we generated models with Glide docking and calculated their Glide energy values. We obtained similar values for all the models including either MpCOI1 or AtCOI1 (SI Appendix, Table S1). The alternative conformation of JAZ containing the N-terminal ELPQ degron might explain why MpJAZ and not AtJAZs form a co-receptor with AtCOI1 to perceive dn-cis-OPDA. To test whether the ELPQ loop was sufficient to enable AtJAZ3 binding to dn-cis-OPDA, we generated additional chimeras where the EQ residues were introduced in AtJAZ3, AtJAZ3^R, and AtJAZ3^RGKA: AtJAZ3^EQ, AtJAZ3^EQR, and AtJAZ3^EQRGKA. Our pull-down experiments showed that the EQ mutations within the JAZ degron are sufficient to induce the interaction between MpCOI1 and AtJAZ3 in the presence of dn-cis-OPDA (Fig. 3B) albeit with a lower affinity than MpJAZ. In contrast, additional RGKA mutations outside the JAZ degron did not seem to influence the interaction between MpCOI1 and AtJAZ3. We confirmed that the mutated version of AtJAZ3 (AtJAZ3^EQ) still interacted with AtCOI1 and COR, as expected from the interaction between MpJAZ and AtCOI1 with COR (SI Appendix, Fig. S1B).

We then expressed the AtJAZ3^EQ chimera in Mpjaz-1 and found that it complemented the Mpjaz-1 dwarf phenotype similar to AtJAZ3 or MpJAZ (Fig. 3 H and I, and SI Appendix, Fig. S1C). However, when grown on OPDA, the expression of AtJAZ3^EQ in AtJAZ3^EQ/Mpjaz-1 plants conferred only a mild increase in OPDA sensitivity compared with AtJAZ3/Mpjaz-1 (Fig. 3 H and I). While these results support that the EQ residues are important to specify the ligand and the co-receptor complex formation, they also indicate that the lower affinity for COI1/dn-OPDA conferred by EQ to JAZ3 compared with MpJAZ may not be sufficient for full complementation of OPDA sensitivity in planta. Therefore, additional residues outside the JAZ degron might also influence the affinity of JAZ for MpCOI1/dn-OPDA in vitro and in planta.

Lycophytes Combine Angiosperm-Like COI1 with Bryophyte-Like JAZ and Respond to dn-OPDA.

Our previous work showed that COI1 from bryophytes differed from tracheophyte COI1 in one key amino acid, and that this particular valine-to-alanine change was responsible for the switch in ligand specificity from dn-OPDA in M. polymorpha to JA-Ile/COR in A. thaliana (8). To understand the evolutionary significance of the EQ combination in the JAZ degron and whether this combination of amino acids also correlates with the switch in hormone identity between bryophytes (dn-OPDA) and tracheophytes (JA-Ile), we searched for the presence of the EQ combination in JAZ from tracheophytes. Using the OneKP database (28, 29), we identified several JAZ proteins containing the ELPQ motif in lycophytes but not in other tracheophyte lineages (SI Appendix, Fig. S2). In addition to the JAZs containing this motif, we identified additional lycophyte JAZ showing nonconserved residues in other positions of the Jas domain (i.e., H in position 5 or I in position 8 of the Jas domain). Lycophytes constitute the sister lineage of euphyllophytes (monilophytes and seed plants). Extant lycophytes are classified in three different lineages: Isoëtales, Selaginellales and Lycopodiales. Notably, we could identify the ELPQ motif in members of all three lineages (SI Appendix, Fig. S2). However, the Selaginella sequences containing the ELPQ motif showed a divergent TIFY domain and were not considered bona fide JAZ (SI Appendix, Fig. S2).

We confirmed that all the species containing the ELPQ JAZ motif showed at least one COI1 sequence with an A in the equivalent position to AtCOI1³⁸⁴ (SI Appendix, Fig. S3A), which in principle indicates that these species could perceive JA-Ile or COR. The presence of this A is consistent with its conservation in other tracheophytes (8). We also identified other lycophyte COI1 sequences, which exhibit different residues in that position. Similar to certain bryophyte COI1s, several Isoëtes and Selaginella sequences showed H, T, or I at this position ((8); SI Appendix, Fig. S3A). Interestingly, three Huperzia selago COI1s showed a V in the same position as MpCOI1 in addition to HsCOI1^A. Due to the similarity to M. polymorpha co-receptor sequences (JAZ^EQ and COI1^V) and the presence of angiosperm-like COI1^A, we decided to focus on H. selago. First, we obtained H. selago plants from natural populations in Spain, France, and Russia. Sequencing of the putative HsCOI1^V gene in the three populations revealed that this gene encodes an A instead of a V indicating an error in the sequence from the OneKP database. The concurrent presence of COI1^A and JAZ^EQ in the same plant may indicate that lycopodiales represent the evolutionary step in which COI1 had already undergone the valine-to-alanine change (expanding ligand recognition to include JA-Ile/COR), whereas JAZ remained similar to bryophyte JAZ^EQ (specifying recognition of dn-OPDA).

Once we clarified that H. selago contains angiosperm-like COI1 and bryophyte-like JAZ, we decided to characterize the role of Huperzia JAZ (SI Appendix, Fig. S3B) in determining ligand specificity. We used the sequence of Huperzia myrsinites JAZ (HmJAZ) instead of H. selago JAZ (HsJAZ) because HmJAZ represents the full-length sequence and both TIFY and Jas domains are conserved between HsJAZ and HmJAZ. Due to the similarity between the Jas domain of MpJAZ and HmJAZ, we hypothesized that HmJAZ could establish the same interactions as MpJAZ with MpCOI1 and the two endogenous ligands of COI1 in M. polymorpha, dn-cis-OPDA/dn-iso-OPDA. Pull-down assays showed that HmJAZ can indeed form a co-receptor with MpCOI1 and either dn-cis-OPDA or dn-iso-OPDA, similar to MpJAZ (Fig. 4A). We then confirmed that JAZ is functionally conserved between Huperzia and M. polymorpha, since HmJAZ complemented Mpjaz-1 dwarf phenotype (Fig. 4 B and C) and is therefore able to repress the pathway hyperactivated in Mpjaz-1. Unlike the plants expressing AtJAZ3 or AtJAZ3^EQ, which are insensitive to OPDA 10 µM, HmJAZ/Mpjaz-1 plants responded to OPDA treatment similar to MpJAZ/Mpjaz-1 (Fig. 4 B and C). These experiments indicate that HmJAZ can regulate OPDA responses in M. polymorpha and, therefore, that JAZ function is conserved between bryophytes and lycophytes.

Fig. 4.

Lycophyte and bryophyte JAZ are biochemically and genetically similar. (A) Immunoblot (anti-flag antibody) of recovered MpCOI1-flag (from 35S:MpCOI1-flag Arabidopsis extracts) after pull-down reactions using recombinant MBP, HmJAZ-MBP, and MpJAZ-MBP alone (mock; −) or with dinor-OPDA isomers (cis and iso; 50 µM). Bottom, Coomassie blue staining of recombinant proteins after cleavage with Factor Xa. (B) Growth-inhibitory effect of OPDA (10 and 50 µM) on 14-d-old M. polymorpha gemmalings of WT Tak-1, Mpjaz-1, EF:MpJAZ-Citrine/Mpjaz-1 and EF:HmJAZ/Mpjaz-1 plants. Scale bar, 1 cm. (C) Growth quantification of plants shown in (B).

Because MpJAZ can form a co-receptor with AtCOI1 to perceive not only JA-Ile and COR but also dn-OPDA, and lycophyte COI1 sequences are similar to AtCOI1 in terms of A 384 conservation, we tested whether HmJAZ could also determine dn-OPDA binding when forming a co-receptor with AtCOI1, similar to MpJAZ. In pull-down assays, we observed that HmJAZ formed a co-receptor with AtCOI1 and JA-Ile, COR and dn-cis-OPDA (Fig. 5A). We confirmed that HmJAZ and MpJAZ mediate identical interactions between COI1 and the different ligands in a heterologous system by yeast two-hybrid assays (Fig. 5B and SI Appendix, Fig. S4A). We observed that similar to MpJAZ, HmJAZ enables AtCOI1 binding to dn-iso-OPDA in yeast (Fig. 5B). HmJAZ can therefore form a co-receptor with MpCOI1 to perceive dn-OPDA and with AtCOI1 to perceive not only JA-Ile and COR, but also dn-OPDA. These results indicate that HmJAZ expanded the previously described ligand-specificity of AtCOI1 for JA-Ile/COR to dn-OPDA isomers, similar to MpJAZ.

Fig. 5.

Lycophyte JAZ enables JA-Ile and COR perception. (A) Immunoblot (anti-flag antibody) of recovered AtCOI1-flag (from 35S:AtCOI1-flag Arabidopsis extracts) after pull-down reactions using recombinant MBP, HmJAZ-MBP and MpJAZ-MBP alone (mock; −) or with (+)-7-iso-JA-Ile (50 µM), COR (0.5 µM) and dinor-OPDA isomers (cis and iso; 50 µM). Bottom, Coomassie blue staining of recombinant proteins after cleavage with Factor Xa. (B) Yeast two-hybrid interaction assays between COI1 and JAZ in the absence or presence of dn-cis-OPDA (50 µM), JA-Ile (50 µM), or COR (3 µM). MpASK1 was coexpressed using pTFT vector to favor COI1 stability. AtJAZ9–AtJAZ9 and AtJAZ3-AtJAZ3 interaction were used as a positive control. BD, binding domain; AD, activation domain. Cotransformed yeasts were plated on media lacking either two amino acids (−2; L and W), three (−3; L, W, and adenine), or four (−4; L, W, adenine, and H) to confirm the presence of two (−2) or three (−3) plasmids or assess the interaction (−4). (C) Time-course accumulation of JAs in H. selago plants in control conditions or 15, 30, and 60 min after wounding. (D) Expression of one HsJAZ gene by qPCR in control and wounded (1 h) H. selago plants. (E) Expression of jasmonate markers genes (HsAOS for biosynthesis and two HsJAZ for signaling) by qPCR in control and treated (dn-cis-OPDA 50 µM or COR 0.5 µM; 1 h) H. selago plants.

To date, dn-iso-OPDA has only been detected in bryophytes, whereas dn-cis-OPDA accumulate in vascular plants including Arabidopsis (8, 30). To evaluate whether the HsJAZ-mediated interaction with the dn-OPDA isomers could be relevant in H. selago, we performed oxylipin profiling in these plants before and after wounding. We detected OPDA, iso-OPDA, dn-OPDA, and dn-iso-OPDA in control conditions and observed that these molecules were transiently accumulated between 15 and 30 min after wounding (Fig. 5C). Strikingly, we could not detect any JA or JA-Ile in H. selago (Fig. 5C), contrary to the previously proposed evolutionary emergence of JA-Ile biosynthesis in lycophytes (Selaginella moellendorffii; (31)). Based on our metabolomics results, we hypothesized that in the absence of JA-Ile, the dn-OPDA isomers are likely the bioactive ligands in Huperzia. The presence of a JAZ containing the ELPQ motif would be therefore required to confer COI1^A the specificity toward dn-OPDA.

To further explore the oxylipin signaling pathway in H. selago, we analyzed gene expression after wounding or oxylipin treatment in these plants. We used young plants (ca. 5 cm long) grown on their original soil prior to treatment for gene expression analysis. Similar to the marker genes induced by wounding or JA treatment in other plant species, wounding resulted in the upregulation of JA signaling components (HsJAZ) and downstream marker genes (HsChiB and HsERF) in H. selago (Fig. 5D and SI Appendix, Fig. S4B). Moreover, exogenous treatment with dn-cis-OPDA or the JA-Ile mimetic COR induced the upregulation of jasmonate marker genes such as the two HsJAZ and the jasmonate biosynthetic gene HsAOS (Fig. 5E). These results indicate that H. selago COI1/JAZ co-receptor perceives the endogenous bioactive oxylipin dn-OPDA and is already adapted to perceive JA-Ile/COR even though JA-Ile is not synthesized in this plant. Our results highlight the previously unrecognized role of JAZ in determining ligand specificity and identify another key evolutionary step in jasmonate signaling; that is, COI1 evolution to accommodate JA-Ile/COR likely preceded that of the genes essential to the production of JA-Ile while retaining ligand specificity toward dn-OPDA and JA-Ile through the features of a bryophyte-like JAZ. Furthermore, JAZ evolution restricted COI1 binding to JA-Ile/COR in euphyllophytes, meaning that dn-OPDA can no longer act as a COI1/JAZ ligand in these plants.

Discussion

Defining the determinants of ligand specificity of hormone receptors is essential to understanding how signaling pathways appeared and evolved. In this work, we have discovered a JAZ role in determining ligand specificity of the COI1/JAZ co-receptor. Furthermore, we have identified that, by retaining a co-receptor with features of both bryophytes and angiosperms, lycophytes represent a key transition intermediate in the evolution of the ligand–receptor interaction in the jasmonate pathway (Fig. 6).

Fig. 6.

Coevolution of JAs biosynthesis and COI1/JAZ signaling in Streptophytes. Green, blue, and yellow rectangles indicate the presence of conserved sequences (JAs signaling components, JAZ and COI1) or detection of JA-Ile or dn-OPDA in the corresponding plant lineages. Dark blue rectangle for dn-OPDA indicates the lineages (bryophytes and lycophytes) where dn-OPDA acts as COI1/JAZ^EQ ligand. In all lineages (light blue and dark blue rectangles) dn-OPDA has a COI1-independent function in thermotolerance and stress responses (31). Dashed lines in JA-Ile rectangle indicate that the precise origin of JA-Ile biosynthesis within lycophytes has not been elucidated yet. The presence of dn-OPDA in charophytes exclusively refers to Klebsormidium nitens (depicted in the figure). Except for K. nitens (charophyte) and Anthoceros agrestis (hornwort), all the other plant drawings were created with BioRender.com: Chlamydomonas (chlorophyte), M. polymorpha (liverwort), Physcomitrium patens (moss), S. moellendorffii (lycophyte), fern, coniferous tree (gymnosperm), rice (monocot) and Antirrhinum (dicot).

Studies in angiosperms, mainly in Arabidopsis, have identified COI1 as the JA-Ile receptor and COI1 has been so far considered the only determinant of ligand specificity (5–7, 25). Supporting this role of COI1, studies in M. polymorpha demonstrated that a single amino acid substitution in the hormone-binding pocket of COI1 (V to A) explains the switch in ligand specificity from the bryophyte hormone (dn-OPDA) to the angiosperm ligand (JA-Ile; (8), Fig. 6). Thus, COI1^V determines dn-OPDA binding, whereas COI1^A determines recognition of JA-Ile and COR. Besides COI1, a role in ligand specificity for JAZ has not yet been considered. Indeed, structural studies of the COI1/JAZ co-receptor together with its ligands have placed JAZ as a mere lid that stabilizes the interaction between COI1 and the ligand (JA-Ile/COR; (7)). The JAZ degron is the minimal JAZ sequence that interacts with COI1 to form a co-receptor perceiving JA-Ile or COR (7). Within the JAZ degron, the N-terminal loop region has been shown to mediate the interaction with the hormone and COI1, and differences in the loop region have been proposed to underlie different affinities of JAZ and COI1/JA-Ile, but not specificity for the ligand (7, 32). However, our results show that MpJAZ determines the receptor binding to dn-OPDA independently of the type of COI1 protein (COI1^V or COI1^A), whereas none of the Arabidopsis JAZ are able to interact with dn-OPDA and any type of COI1. Comparison of the degron sequence in JAZ proteins pinpointed a specific combination of two residues (EQ) within the loop region of MpJAZ, which adopts a different conformation than AtJAZ in our molecular modeling complexes, establishing novel intermolecular interactions with COI1 in the presence of dn-OPDA. Indeed, the introduction of these two residues in the AtJAZ3^EQ mutant facilitated the recognition of dn-OPDA by the MpCOI1/AtJAZ3^EQ co-receptor, whereas this molecule was not recognized by MpCOI1/AtJAZ3. A further proof of the importance of JAZ in ligand specificity is that we were able to switch the ligand specificity in M. polymorpha through the combination of different COI1 and JAZ proteins. Thus, the WT bryophyte combination COI1^V/JAZ^EQ determines exclusive responsiveness to dn-OPDA in M. polymorpha, whereas the angiosperm-type combination COI1^A/JAZ determines exclusive responsiveness to JA-Ile/COR (Fig. 6). Mixed combinations of angiosperm COI1 and bryophyte JAZ (COI1^A/ JAZ^EQ) determine responsiveness to both dn-OPDA and JA-Ile/COR, and the combination of bryophyte COI1 and angiosperm JAZ (COI1^V/JAZ) leads to the lack of recognition of any of these ligands. Therefore, JAZ determines specificity for the ligand together with COI1, and the EQ combination in the JAZ degron is essential for dn-OPDA recognition.

Despite this essential role of the EQ combination in determining ligand specificity, additional residues outside the loop region may also play a role in vivo in the formation of the co-receptor complex perceiving dn-OPDA since plants expressing AtJAZ3^EQ were only slightly more sensitive to OPDA than AtJAZ3/Mpjaz-1. One of the current limitations of the AtCOI1/AtJAZ1 crystal structure is the scarce structural data on the JAZ protein outside the JAZ degron, which is difficult to crystalize due to large disordered regions in JAZ (7, 33). Ideally, crystal structures of COI1/JAZ complexes from different species, including full-length JAZ, should reveal the molecular interactions mediated by JAZ and COI1 residues and the different ligands. Alternatively, the transgene expression level and protein accumulation, rather than residues outside the degron, may influence the sensitivity in vivo, since expression of MpJAZ in M. polymorpha plants also reduces sensitivity to the hormone, and ectopic expression does not fully complement the endogenous MpJAZ function.

To understand the evolutionary implications of these co-receptor/ligand features we searched for the EQ combination in different plant lineages. This search revealed the conservation of both EQ residues in almost all bryophyte JAZ (the exceptions are the liverworts Monoclea gottschei and Frullania sp., and the hornwort Phaeoceros carolinianus), and more strikingly, also in some lycophytes, such as H. selago. This lycophyte represents a key evolutionary step, combining angiosperm-like features of COI1 (COI1^A) with bryophyte-like JAZ (JAZ^EQ; Fig. 6). This combination results in a broadened ligand specificity of the COI1/JAZ co-receptor, which could thus bind dn-OPDA and also JA-Ile and COR. H. selago synthesizes dn-OPDA but not JA-Ile (Fig. 6) and, therefore, although its COI1/JAZ co-receptor uses dn-OPDA as the endogenous ligand, it is already adapted to perceive JA-Ile. Indeed, exogenous treatments with the JA-Ile mimic COR confirmed this hypothesis by showing that COR can activate gene expression and the dn-OPDA pathway in H. selago similarly to dn-OPDA. Being prepared for JA-Ile perception might have an evolutionary advantage for Huperzia environmental adaptation since surrounding plants emit volatiles such as MeJA in response to stress (herbivory, pathogens, etc.) that could be perceived by Huperzia when converted to JA-Ile and alert the plant against an imminent attack. However, this may have also opened the door for COR-producing pathogens, since COR facilitates infection of plants with angiosperm-like COI1 but not with bryophyte-like COI1 (34). Therefore, further analyses of plant–pathogen interactions may help to understand why plants like Huperzia have retained evolutionarily angiosperm-like COI1 without producing JA-Ile.

The lack of JA or JA-Ile in H. selago defies the previous assumption that JA-Ile biosynthesis emerged in lycophytes, based on hormone measurements in S. moellendorffii (31). Further metabolomic analysis in additional lycophyte species should clarify whether JA-Ile is indeed synthesized in different lycophyte lineages or whether JA-Ile biosynthesis emerged in other tracheophytes. Once the origin of JA-Ile biosynthesis is established, it will be interesting to identify which type of JAZ is present in the corresponding plant lineage.

Our results suggest that during the tracheophyte evolution, there was a positive selection of JAZ mutations that restricted COI1/JAZ ligand binding to JA-Ile/COR. As a result, dn-OPDA, which is synthesized in all land plants, only acts as a hormone perceived by COI1/JAZ in bryophytes (containing COI1^V) and lycophytes (containing COI1^A and bryophyte-like JAZ). In euphyllophytes, the diversification of JAZ sequences resulted in the loss of dn-OPDA binding by the COI1^A/JAZ co-receptor, as shown in A. thaliana. The advantage of using JA-Ile as hormone instead of dn-OPDA might be related to the COI1-independent role of OPDA/dn-OPDA based on their properties as reactive electrophilic species (35) and to the energy gained after β-oxidations of these oxygenated polyunsaturated fatty acids (PUFAs). JA-Ile effect is exclusively mediated by COI1/JAZ co-receptor, whereas OPDA and dn-OPDA can induce transcriptional reprogramming independently on COI1 in both charophytes and land plants ((8, 35–39); Fig. 6). This COI1-independent effect of OPDA might explain why high concentrations of OPDA (50 µM) inhibit growth of AtJAZ3/Mpjaz-1 plants (Figs. 1–3), even if AtJAZ3 does not interact with MpCOI1 in the presence of dn-OPDA (Figs. 1 and 3). Alternatively, another possible explanation for the dn-OPDA to JA-Ile switch that has spread and dominated in all euphyllophytes is the higher polarity of JA and JA-Ile than OPDA/dn-OPDA, which might have facilitated the jasmonate mobility through the vasculature and their membrane permeability (8, 40).

Our previous work on jasmonate biosynthesis and signaling in the green lineage revealed that OPDA and dn-OPDA biosynthesis evolved before the emergence of the canonical jasmonate signaling pathway in land plants (8, 20, 21, 35, 41). We had previously proposed the dichotomy between bryophytes containing COI1^V and perceiving dn-OPDA as a ligand vs. tracheophytes containing COI1^A and JA-Ile as ligand, suggesting that the mutation in COI1 coevolved with JA-Ile biosynthesis (8). The presence of dn-OPDA in A. thaliana prompted us to hypothesize that dn-OPDA might also be a COI1/JAZ ligand in this plant. However, results in this work suggest that in COI1-dependent signaling in Arabidopsis, dn-OPDA is a mere precursor of JA-Ile and does not have functions as receptor ligand, since dn-OPDA does not induce the co-receptor complex formation between AtCOI1 and any of the AtJAZ. Our current work provides an updated scenario (Fig. 6) in which lycophytes (or Lycopodiales) could represent an intermediate evolutionary step where JAZ retained features from the common ancestor to bryophytes and lycophytes and enabled dn-OPDA perception by COI1^A/JAZ co-receptor prior to JA-Ile biosynthesis emergence. In other tracheophyte lineages (euphyllophytes) which produce JA-Ile, JAZ evolved to limit ligand specificity of the COI1^A/JAZ co-receptor to JA-Ile/COR. Our results highlight the role of JAZ in determining, together with COI1, ligand specificity and its evolutionary importance in the context of jasmonate biosynthesis and signaling.

Materials and Methods

Plant Material and Growth Conditions.

M. polymorpha subsp. ruderalis Takaragaike-1 (Tak-1) was used as wild type (20). Plants were kept under sterile conditions on 0.5 Gamborg’s B5 medium 1% agar under continuous light (50–60 μmol m⁻² s⁻¹) at 20°C. EF:AtCOI1/Mpcoi1-1 and EF:AtJAZ3-Citrine/Mpjaz-1 plants were grown on soil under continuous light supplemented with far-red light to induce gametangiophore formation. Plants were crossed to obtain F1 spores, which were sterilized with 0.25% sodium hypochlorite (Sigma) and 0.05% Triton X-100. F1 plants grown from F1 spores were genotyped to identify EF:AtCOI1/Mpcoi1-1/EF:AtJAZ3-Citrine/Mpjaz-1 plants (note that M. polymorpha thalli are haploid). A. thaliana 35S:MpCOI1-FLAG and 35S:AtCOI1-FLAG seeds were surface-sterilized with the chlorine gas method and grown on Murashige and Skoog (MS) medium supplemented with 0.6% agar for 2 wk under long-day conditions (16-h light/8-h dark) at 22°C. H. selago adult plants were grown on their original soil (Russian accession) or on Arabidopsis potting soil (Spanish and French accessions) under long-day conditions (16-h light/8-h dark) at 22°C and covered with a transparent plastic lid. Bulbils were sterilized as previously described (42) and grown on modified Moore’s medium (43). Despite our attempts to propagate H. selago through propagules (bulbils), we observed growth arrest in 2-cm plants grown on regular soil or young plants after radicle emergence when grown in vitro. In addition to its slow growth rate, H. selago generates a limited number of propagules for asexual reproduction, and we used most of them in our unsuccessful trials to grow young plants. The remaining bulbils from the Russian accession were directly grown on the original soil until the stems were ca. 5 cm for gene expression experiments. For growth inhibition experiments, M. polymorpha plants were grown from gemmae on 0.5 Gamborg’s B5 medium 1% agar supplemented or not with OPDA, JA-Ile, or COR for 2 wk. Plant pictures were taken with a NIKON D1-x digital camera. Plant area was measured with ImageJ software.

Plant Transformation.

M. polymorpha was transformed using the cut-thalli transformation method (44) and the Agrobacterium tumefaciens GV3101 strain.

Gene Identification and Alignments.

JAZ and COI1 sequences from lycophytes were retrieved using MpJAZ and MpCOI1 as query in the OneKP database (28, 29). Alignments were performed with MUSCLE, and only proteins with conserved TIFY and Jas domains were considered bona fide JAZ orthologs. Edward’s buffer (45) was used for gDNA extraction from H. selago adult plants to verify the HsCOI1 sequence by PCR and Sanger sequencing.

Cloning and Transformation.

Due to the incomplete JAZ sequences in the transcriptomic datasets, especially in the N and C termini of most lycophyte JAZ, we chose the longest JAZ sequence from H. myrsinites (CBAE2061511) and synthesized the corresponding coding sequence (cds SDS) (IDT technologies) from the codon encoding the first methionine upstream of the TIFY domain and including the attB tails. HmJAZ and AtJAZ3 chimera sequences were cloned into pKM596, pMpGW308, and pGADT7 (only for HmJAZ) via LR reaction (Invitrogen).

Protein Extraction and Pull-Down Assays.

Pull-down experiments were performed as previously described (8, 21, 25, 46). Proteins were extracted in SDS buffer β-mercaptoethanol using the same number of gemmalings.

Yeast Two-Hybrid Assays.

This experiment was performed as previously described (8, 47). Yeast colonies were grown for 3–10 d at 28°C.

Gene Expression Analysis.

Total RNA was extracted from H. selago bulbils grown on soil until the stems reached ca. 5 cm. Plants were incubated in liquid 0.5 Gamborg’s B5 medium containing the indicated hormones for 1 h. Alternatively, plants were wounded with forceps and samples collected after 1 h. Three biological replicates containing two plants each were used per treatment. RNA was extracted using the Favorgen Plant Total RNA Mini kit (21).

Hormone Measurements.

Branches of H. selago adult plants were wounded with forceps and incubated for the indicated time before flash freezing in liquid nitrogen. Hormones were measured as previously described (8).

Molecular Modeling.

The crystallographic structure of AtCOI1/AtJAZ1 forming a complex with JA-Ile (Protein Data Bank ID: 3OGL; (7)) was used to obtain the models of AtCOI1, MpCOI1, AtJAZ3, and MpJAZ (7). Proteins and ligands were prepared using modules incorporated within Maestro software (Maestro, Version 11.1, 2017; Schrödinger, LLC, New York, NY). MpCOI1 was constructed by homology modeling with Prime, while AtJAZ3 (ALPLARKASLARFLEKRK) and MpJAZ (ELPQARKASLARFLEKRK) were constructed using the Mutate Residue function. (Note that the MpJAZ model is equivalent to AtJAZ3^EQ.) Then, the protein structures were processed with the Protein Preparation Wizard module; hydrogen atoms were added, the bond orders were adjusted, and the protonation states for the protonable amino acid residues were set at physiological pH. The structures of JA-Ile and dn-cis-OPDA were sketched using the chemical editor, and then they were converted into three-dimensional ones and atomic partial charges were assigned using LigPrep.

Atomistic models of AtCOI1/AtJAZ3–JA-Ile and MpCOI1/MpJAZ–dn-cis-OPDA were obtained by applying two different molecular docking protocols. AtCOI1/AtJAZ3–JA-Ile was obtained using the first docking protocol (named dG here). The AtCOI1/AtJAZ3 complex was defined as receptor and JA-Ile was docked using Glide software (48). MpCOI1/MpJAZ–dn-cis-OPDA was obtained using the second docking protocol (named dGH here). The MpCOI1 structure was defined as receptor, dn-cis-OPDA was docked using Glide software, and then MpJAZ was docked in the cavity of the MpCOI1/dn-cis-OPDA complex using the HADDOCK program (49).

The extra-precision Glide mode (50) was used in dG and dGH protocols. For each Glide docking calculation, a grid box defined inside COI1 pocket was centered using as reference JA-Ile in the crystallographic structure. The best poses were selected by considering conformations with more favorable scoring values. Glide parameters were defined as in previous works (51, 52).

The dGH protocol was performed to evaluate whether the formation of MpCOI1/MpJAZ, in the presence of dn-cis-OPDA involves large conformational changes of MpJAZ with respect to the conformation of AtJAZ1 in the crystallographic structure. The docking with HADDOCK allows to model protein–protein complexes by considering flexibility and binding-induced conformational changes. HADDOCK performs a data-driven docking and enters the interface information by defining active and passive residues, where the active residues are those that have the highest probability of being part of the interaction interface, while the passive residues are in the vicinity of the active residues. We set as “active” all the MpCOI1 and MpJAZ residues that are large in size and are more exposed, while “passive” residues were automatically assigned by the system. Ten thousand rigid-body docking structures were created (iteration it0), followed by a semiflexible docking with only the interface being fully flexible (iteration it1). After this, the best 400 docked models according to HADDOCK score were refined in an explicit solvent (iteration itw). The final rank of clusters of poses was obtained by considering a 0.6-nm rmsd cutoff (fraction of common contacts were used as the clustering metric (53). The best cluster of poses was selected by considering the most favorable HADDOCK and Z-score values.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (TXT)

Dataset S02 (TXT)

Dataset S03 (TXT)

Dataset S04 (TXT)

Dataset S05 (TXT)

Dataset S06 (TXT)

Dataset S07 (TXT)

Dataset S08 (TXT)

Dataset S09 (TXT)

Dataset S10 (TXT)

Dataset S11 (TXT)

Dataset S12 (TXT)

Dataset S13 (TXT)

Dataset S14 (TXT)

Dataset S15 (TXT)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.