Abstract
Plants prioritize growth over defense to gain a competitive advantage for limited resources, but change priorities to successfully fight infection and herbivory. Despite the importance of growth-defense tradeoffs in optimizing plant productivity in natural and agricultural populations, the molecular mechanisms that link growth and immunity remain unclear. Here, we demonstrate that growth-defense tradeoffs between pathways activated by BRI1, a steroid receptor, and FLS2, an innate immune receptor, are uncoupled in an Arabidopsis mutant (agg1 agg2) lacking two redundant heterotrimeric G-protein gamma subunits that form stable heterodimers with the Gβ subunit AGB1 to control one arm of the unfolded protein response (UPR) independently of ER stress. Growth inhibition from induced immunity in wild-type plants is likely caused by AGB1-AGG1/2 dimers interacting with nascent BRI1 and FLS2 proteins on the endoplasmic reticulum (ER) membrane and repressing an UPR response that is hardwired to promote BRI1 protein biogenesis and FLS2 protein degradation via autophagy. The ability to unlock and fine-tune growth-defense tradeoffs through UPR signaling provides a novel strategy to increase the natural defenses of crops while maintaining optimal plant productivity.
Author Contributions
C.B. contributed to characterization of agg1 agg2/XVE:FLS2 lines. T.C. contributed to Endo H digestion. S.A.L. generated agb1/35S:YFP-AGB1 lines, and contributed to qPCR of ire1a/b, Endo H digestion and MAPK activation of agg1 agg2 atg7/3 N.K.C. generated XVE:BRI1-RFP construct and agg1 agg2/XVE:FLS2 lines, and contributed to Co-IPs and pathogen assays. J.C.M. generated all other constructs, and contributed to characterization of agg1 agg2/XVE:FLS2 lines, Co-IPs and all other experiments. J.C.M. and N.K.C. interpreted the results and N.K.C. wrote the paper.
Significance Statement
Plants continually face a resource allocation dilemma – should they invest directly in growth to out-compete their neighbors for limited resources or invest directly in defense against herbivores and pathogens? Plants sense the competing demands of their surroundings through receptor proteins that are synthesized at high levels in ER. Here we show that plants use UPR to balance the synthesis of receptor proteins to favor defense over growth. This UPR response is mediated by heterotrimeric G-protein complexes whose signaling function in the ER is independent of their canonical functions at the plasma membrane. The uncoupling of growth and defense through UPR signaling provides a novel solution to the strong allocation tradeoffs against plant defensive traits as a consequence of plant domestication.
Introduction
Plants must maintain a precise balance between growth and defense in order to survive and reproduce, using a pool of limited resources (Huot et al. 2014). The tradeoff of shifting from growth to immunity upon detection of pathogens or herbivores has important ecological and agricultural consequences (Spoel and Dong, 2008; Dodds and Rathjen, 2010). The competing demands imposed on plants by their environments require mechanisms for sensing their surroundings and for effectively regulating the tradeoffs between growth and immunity. Over the past decade, a number of inhibitory crosstalks between individual pathways in growth and immunity have been characterized, including those involving the growth hormone brassinosteroids (BR)-perceiving transmembrane leucine-rich repeat-receptor kinase (LRR-RK) Brassinosteroid-Insensitive-1 (BRI1), bacterial flagellin-recognizing transmembrane LRR-RK Flagellin-Sensing-2 (FLS2), defense hormone salicylic acid (SA), and unfolded protein response (UPR) (Li and Chory, 1997; Gómez-Gómez and Boller, 2000; Rivas-San Vicente and Plasencia, 2011; Albrecht et al., 2012; Belkhadir et al., 2012; Nagashima et al., 2014; Jiménez-Góngora et al., 2015). The shared signaling components between BRI1- and FLS2-mediated signaling and between SA and UPR signaling have been characterized to mediate not only signal crosstalk between these pathways but also tradeoffs between growth and defense (Lozano-Durán et al., 2013; Fan et al., 2014; Meng et al., 2017).
The ER is the production and folding compartment for membrane proteins of the cell. Quality control mechanisms in the ER ensure that only properly folded proteins exit the ER via the secretory pathway, while improperly folded proteins exit the ER through ER-associated degradation (ERAD) or autophagy (Smith et al., 2011; Pu and Bassham, 2013). UPR is an evolutionarily conserved adaptive response triggered by the accumulation of unfolded proteins in the ER and aimed at restoring protein-folding homeostasis. However, the main function of UPR in vertebrates and plants is in growth and defense, where it acts as an anticipatory response that is activated well before the disruption of protein homeostasis and aimed at handling high folding loads that are part of normal physiology (Janssens et al., 2014; Bao and Howell, 2017). The UPR signaling pathway in Arabidopsis has three overlapping but independent arms (Ruberti and Brandizzi, 2014). One arm is mediated by two homologs of the evolutionarily conserved transmembrane ER kinase/RNA splicing factor IRE1 (IRE1A and IRE1B), whose primary target is the transcription factor bZIP60 mRNA (Koizumi et al., 2001; Deng et al., 2011). The second arm is mediated by ER membrane-associated transcription factors bZIP17 and bZIP28, which are functional homologs of ATF6 in metazoans (Liu et al., 2007). The last arm is mediated by the ER membrane-localized heterotrimeric G-protein β subunit AGB1 (Wang et al., 2007; Chen and Brandizzi, 2012) and possibly by other G-protein subunits, such as Gα subunits XLG1/2/3 and the Gγ subunits AGG1/2/3 (Chakravorty et al., 2015). AGB1 is required for UPR signaling under ER stress conditions (Wang et al., 2007; Chen and Brandizzi, 2012), but its UPR function in growth and defense has not yet been reported.
Heterotrimeric G-proteins are the most commonly used signal transducers in eukaryotic cells. They transduce signals at the cytosolic surfaces of the plasma membrane (PM) and endoplasmic reticulum (ER) as Gαβγ heterotrimers, Gβγ heterodimers or individual Gα subunits (Weiss et al., 1997; Kaydamov et al., 2000; Wang et al., 2007; Hewavitharana and Wedegaertner, 2012; Giannotta et al., 2012; Klayman and Wedegaertner, 2017). Gβγ dimers and Gαβγ trimers are thought to assemble on the ER membrane (Dupre et al., 2007; Marrari et al., 2007), and are anchored to membranes by virtue of lipid modifications on the Gα and Gγ subunits (Wedegaertner 1998; Adjobo-Hermans et al., 2006; Zeng et al., 2007). Arabidopsis has at least four Gα (GPA1, XLG1/2/3), a Gβ (AGB1), and three Gγ (AGG1/2/3) subunits (Chakravorty et al., 2015; Maruta et al., 2015; Thung et al., 2012), and likely more non-canonical G-protein subunits yet to be discovered (Lee and Assmann, 1999; Chakravorty et al., 2011). All known G-protein subunits are involved in some aspect of growth and development (Lease et al., 2001; Ullah et al., 2001; Ullah et al., 2003; Trusov et al., 2007; Ding et al., 2008; Chakravorty et al., 2011). In addition, a subset – Gα GPA1, Gβ AGB1 and Gγ AGG3 – are involved in BR signaling in sugar-responsive growth (Peng et al., 2018), and all save two – XLG1 and AGG3 – are involved in flagellin signaling (Ishikawa 2009; Zhang et al., 2008; Zhu et al., 2009; Liu et al., 2013; Torres et al., 2013; Maruta et al., 2015; Miller et al., 2016). Recently, AGB1 has been shown to be a shared component in BR and flagellin signaling via interactions with corresponding receptors BRI1 and FLS2, presumably at the PM, and with the BR transcription factor BES1 in the nucleus, downstream of receptor signaling (Liang et al., 2016; Peng et al., 2018; Zhang et al., 2018). The essential role of G-proteins in both BR and flagellin signaling pathways suggests that they may function as rate-limiting factors between both pathways to mediate tradeoffs between growth and immunity.
Here, we show that AGB1 and Gγ subunits AGG1 and AGG2 work together in UPR signaling to mediate growth-defense tradeoffs that are upstream of BR and flagellin signaling and independent of ER stress. Specifically, they interact with nascent FLS2 and BRI1 proteins at the ER membrane and repress an UPR response that is hardwired to promote BRI1 protein biogenesis and FLS2 protein degradation via autophagy.
Results
Increased BRI1 signaling likely contributes to enhanced growth of agg1 agg2
While the loss-of-function agb1 mutant exhibited growth defects due to its involvement in multiple hormone signaling pathways, (Ullah et al., 2003; Chen et al., 2004; Pandey et al., 2006; Trusov et al., 2006), the loss-of-function agg1 agg2 double mutant exhibited increased vegetative growth and faster transition to inflorescence development compared to WT and agb1 (Fig. 1A; SI Appendix, Fig. S1A). To identify the Gγ subunit(s) interacting with AGB1 in the crosstalk between BR and flagellin signaling, we performed a time course of seedling growth and BRI1 protein expression under normal growth conditions. The loss-of-function agg1 agg2 mutant exhibited increased BRI1 protein expression relative to WT after day 6 post-germination, correlating with its increased growth after day 8 (Fig. 1B–D; SI Appendix, Fig. S2A). Although the loss-of-function agb1 mutant exhibited normal seedling growth (Fig. 1C), it displayed slightly increased BRI1 protein expression after day 6 (Fig. 1C–D). To confirm our genetic results, we used the estrogen receptor-based XVE system to drive AGG1-RFP expression in agg1 agg2 (agg1 agg2/XVE:AGG1-RFP). Induced expression of AGG1-RFP reduced BRI1 protein expression to WT level or below in 9-day-old seedlings (Fig. 1D).
To investigate whether increased BRI1 protein expression in agg1 agg2 and agb1 is due to increased transcription and/or translation, we measured BRI1 gene expression during a 24-hr time period in 4-hr increments and measured BRI1 protein expression in 5.5-day-old seedlings pretreated with the protein translation inhibitor cycloheximide (CHX). agg1 agg2 and agb1 plants exhibited increased BRI1 expression relative to WT throughout the time period (SI Appendix, Fig. S2B), consistent with the observed BRI1 protein expression pattern (Fig. 1D) and indicative of G-protein-dependent transcription. In addition, CHX treatment inhibited developmental upregulation of BRI1 protein expression in agg1 agg2 and agb1 (Fig. 1E). These results indicate that increased transcription and translation are likely responsible for the increased BRI1 protein expression in G-protein mutants. BRI1-overexpression has been demonstrated to confer enhanced BR signaling in plants (Belkhadir et al., 2012), whereas mutants impaired in BRI1 signal transduction, such as bak1, exhibited reduced growth (Fig. 1C; Clouse et al., 1996; Li et al., 2002). Our findings suggest that BR signal transduction is increased in the agg1 agg2 mutant, contributing to its increased growth under normal growth conditions.
Increased FLS2 protein turnover contributes to enhanced growth of agg1 agg2 under defense-inducing conditions
BRI1-overexpression also confers reduced FLS2-mediated immune responses (Belkhadir et al., 2012). To investigate whether transient growth-defense tradeoffs are affected in the agg1 agg2 mutant upon flagellin perception, we measured seedling growth inhibition (SGI) in response to the active epitope of bacterial flagellin, flg22. Consistent with BRI1-overexpression plants (Belkhadir et al., 2012), agg1 agg2 exhibited a reduced flg22-induced growth inhibition response relative to WT after day 5 (Fig. 2A; SI Appendix, Fig. S3A-B), whereas induced expression of AGG1-RFP restored flg22-induced growth inhibition in agg1 agg2 to WT level (Fig. 2A). Although agb1 exhibited a normal growth inhibition response (Fig. 2A), we used the 35S system to drive YFP-AGB1 expression in agb1 (agb1/35S:YFP-AGB1). Constitutive expression of YFP-AGB1 increased flg22-induced growth inhibition in agb1 relative to WT (Fig. 2A). As a control, we measured FLS2 protein expression in agg1 agg2 and agb1 and found that they exhibited reduced FLS2 protein expression (Fig. 2B), whereas induced expression of AGG1-RFP restored FLS2 protein expression to or above that of WT (Fig. 2B). To confirm these results, we measured FLS2 protein expression during a 24-hr time period in 4-hr increments. agg1 agg2 exhibited decreased FLS2 protein expression relative to WT throughout most of the day with peak reductions occurring at night (Fig. S2C). To investigate whether reduced FLS2 protein expression in agg1 agg2 is due to transcription, translation and/or protein degradation, we measured FLS2 gene expression during a 20-hr period in non-treated plants and measured FLS2 protein expression during a 16-hr period in plants pretreated with CHX. agg1 agg2 and agb1 exhibited increased FLS2 expression relative to WT 4 hr after dawn and reduced FLS2 expression 4 hr after dusk, which does not correlate with the observed FLS2 protein expression pattern (SI Appendix, Fig. S3C–D) and is indicative of G-protein-dependent transcription. In addition, CHX treatment revealed that FLS2 protein expression was more reduced in agg1 agg2 relative to WT at an earlier time point (Fig. 2D), indicative of increased protein degradation. These findings suggest that increased FLS2 protein turnover contributes to the enhanced growth observed in agg1 agg2 under defense conditions.
To determine whether the increased turnover in agg1 agg2 affects newly synthesized FLS2 proteins on the ER membrane or mature FLS2 proteins on the PM, we performed aqueous two-phase partitioning of total membrane proteins and measured FLS2 abundance among microsomal membrane (MM) and PM proteins. agg1 agg2 exhibited a decrease in FLS2 protein expression at both membrane populations (SI Appendix, Fig. S3E). Since membrane proteins on the PM must exit the ER, this result suggests that nascent FLS2 proteins at the ER membrane are targeted for turnover in the agg1 agg2 mutant.
To confirm that increased BRI1 signaling and/or increased FLS2 turnover in agg1 agg2 contributes to reduced FLS2-mediated immune responses, we measured activation of mitogen-associated kinases (MAPKs) and callose deposition at the cell wall in response to flg22. agg1 agg2 exhibited reduced flg22-induced MAPK activation and callose deposition responses relative to WT (SI Appendix, Figs. S4 and S5), whereas induced expression of AGG1-RFP restored MAPK activation in agg1 agg2 to WT level (SI Appendix, Fig. S4C) and increased callose deposition 2-fold (SI Appendix, Fig. S5). On the other hand, agb1 exhibited a normal flg22-induced MAPK activation response (SI Appendix, Fig. S4A). agb1 also exhibited a reduced callose deposition response that could not be rescued by constitutive expression of AGB1 (SI Appendix, Fig. S5), likely due to poor 35S promoter expression in seedling leaves (Kamo et al., 2000).
Growth and defense are uncoupled in agg1 agg2 mutant
BR signaling has been shown to antagonize FLS2-mediated immune responses downstream of receptor signaling (Albrecht et al., 2012; Belkhadir et al., 2012; Lozano-Durán et al., 2013; Fan et al., 2014). agg1 agg2 exhibits enhanced growth under growth-inducing conditions and defense-inducing conditions, consistent with the increased and decreased expression of respective BRI1 and FLS2 in this mutant (Figs. 1 and 2). To investigate whether the negative crosstalk between BR and flagellin signaling is preserved in agg1 agg2, we measured three FLS2-mediated immune responses (i.e., defense response gene transcription, callose deposition and seedling growth inhibition) in response to the BR hormone 24-epibrassinolide (BL). Consistent with a previous report (Albrecht et al., 2012), MAPK-activated transcription of defense response genes CYP81F2 and CYP82C2 (Boudsocq et al., 2010) was reduced in WT seedlings in response to co-treatment with 1 μM BL and 100 nM flg22 compared to flg22 treatment alone (Fig. 3A). Similarly, callose deposition and growth inhibition were reduced in WT plants and unchanged in BRI1 signaling mutant bak1 in response to co-treatment with BL and flg22 (Fig. 3B–C). By contrast, agg1 agg2 exhibited no changes in defense gene transcription, smaller reductions in callose deposition relative to WT, and no changes in growth inhibition in response to BL and flg22 (Fig. 3A–C), whereas agb1 resembled WT plants in all three FLS2-mediated immune response (Fig. 3A–C). These results indicate that growth and defense are uncoupled in the agg1 agg2 mutant.
Interestingly, agg1 agg2 and agb1 exhibited reduced growth in response to 1 μM BL alone (Fig. 3C). High BR concentrations and/or signaling can inhibit growth (Clouse et al., 1996; Müssig et al., 2003; González-García et al., 2011), indicating that an appropriate intensity of BR signaling is important for optimal plant growth. To investigate whether increased BRI1 signaling in agg1 agg2 is responsible for its increased insensitivity to exogenous BL application, we measured hypocotyl lengths in response to 1 and 10 μM BL. At 1 μM BL, the concentration reported to result in nearly full BL responsiveness in this tissue (Clouse et al., 1996), agg1 agg2 and agb1 exhibited a normal hypocotyl elongation response, whereas at 10 μM BL, they exhibited reduced BL responsiveness (SI Appendix, Fig. S6A–B). By contrast, induced expression of AGG1-RFP and constitutive expression of YFP-AGB1 restored WT level of BL sensitivity in respective agg1 agg2 and agb1 plants treated with 10 μM BL (SI Appendix, Fig. S6B). These results lend further support to increased BRI signaling occurring in G-protein mutants.
AGG1 and AGG2 are involved in UPR signaling in the absence of ER stress
AGB1 is enriched in the ER, where it functions as an UPR sensor during ER stress (Wang et al., 2007; Chen and Brandizzi, 2012). To investigate whether UPR signaling is affected in the agg1 agg2 mutant, we measured seedling growth inhibition in response to ER stress induced by tunicamycin (Tn). Consistent with previous reports on the agb1 mutant (Chen and Brandizzi, 2012; Chakravorty et al 2015), agg1 agg2 is hyper-responsive to long-term (14 days) Tn-induced ER stress (Fig. 4A). In fact, dose-response curves indicate that agg1 agg2 is more sensitive to Tn-induced ER stress than UPR signaling mutants agb1 and ire1a ire1b (Fig. 4A). We then looked for changes in expression of UPR-activated genes in response to short-term (5 hr) Tn-induced ER stress and observed increased expression of IRE1A and IRE1B, their spliced target bZIP60s, and G-protein genes AGB1, AGG1 and AGG2 (Fig. 4B). Furthermore, IRE1A, IRE1B and bZIP60s were expressed normally in agg1 agg2 and agb1 mutants in response to short-term Tn-induced ER stress, whereas AGG1 and AGB1 were downregulated in ire1a ire1b mutant (Fig. 4B), indicative of IRE1A/B-dependent transcription.
IRE1A/B and AGB1 have been shown to respectively upregulate and downregulate the gene expression of the chaperone and IRE1A/B ligand BIP3 in response to long-term (3 days) Tn-induced ER stress (Chen and Brandizzi, 2012). Consistent with this finding, combined BIP1– 3 (BIP) protein expression was reduced in ire1a ire1b relative to WT and agb1 independent of ER stress, whereas ER stress-induced expression of folding catalyst PDI was unchanged in agg1 agg2, agb1 and ire1a ire1b (SI Appendix, Fig. S7). Furthermore, agg1 agg2 exhibited increased BIP3 expression in response to short-term (5 hr) Tn-induced ER stress (Fig. 4B), whereas BIP3 expression was unchanged in WT, ire1a ire1b, and agb1 (Fig. 4B). More importantly, ire1a ire1b exhibited normal BRI1 and FLS2 protein expression (Figs. 1C and 2B), and agg1 agg2 and agb1 exhibited no upregulation of IRE1A/B, BIP3 and bZIP60s gene expression or PDI protein expression in the absence of ER stress (Fig. 4B;SI Appendix, Fig. S7). Thus, IRE1A/B and G-proteins have distinct UPR functions independent of ER stress. Altogether, these findings suggest that AGB1 and AGG1/2 work together to promote FLS2 biogenesis and repress BRI1 biogenesis through UPR signaling in the absence of ER stress.
AGB1-AGG1/2 interact with FLS2 and BRI1 at the ER membrane
AGB1 forms obligate heterodimers with AGG1 and AGG2 (Mason and Botella, 2000; Adjobo-Hermans et al., 2006; Chakravorty and Botella, 2007), and interacts with FLS2 and BRI1 in vivo (Liang et al., 2016; Peng et al., 2018). To investigate whether AGB1 and/or AGG1 interact with nascent FLS2 and BRI1 proteins at the ER membrane, we first detected AGG1, AGB1, BRI1 and FLS2 proteins that were C-terminally tagged with GFP or RFP in Nicotiana benthamiana leaves. Consistent with a previous report (Adjobo-Hermans et al., 2006), AGG1 co-localized with AGB1 at the PM (SI Appendix, Fig. S8A), and with FLS2 at plasmolysis-induced Hechtian strands, which are PM fragments still attached to the cell wall and separated from the cytosol (SI Appendix, Fig. S8B). AGG1 also co-localized with AGB1 and FLS2 at the highly reticulated ER membrane and with the ER protein marker HDEL (Fig. 4A; Gomord et al., 1997). The ER localization of AGG1 was further validated in agg1 agg2/XVE:AGG1-RFP plants (SI Appendix, Fig. S8C). AGG1 and AGB1 also co-localized with BRI1 at the ER membrane (SI Appendix, Fig. S8D). These results confirmed that the C-terminal tag did not disrupt AGG1’s lipid modification and subsequent localization of AGG1 and its partner AGB1 to membranes.
We then immunoprecipitated AGG1-GFP and AGB1-GFP proteins from PM and MM protein extracts of N. benthamiana leaves as well as YFP-AGB1 and AGG1-RFP proteins from PM and MM protein extracts of Arabidopsis seedlings. FLS2-RFP and native FLS2 protein co-immunoprecipitated with respective AGG1-GFP and AGG1-RFP at the PM and MM (Fig. 4B–C). Native FLS2 also co-immunoprecipitated with YFP-AGB1 at the PM and MM in Arabidopsis (Fig. 4C). Similarly, BRI1-RFP co-immunoprecipitated with AGG1-GFP and AGB1-GFP at the PM and MM in N. benthamiana (SI Appendix, Fig. S9). We were unable to confirm the BRI1-AGB1/AGG1 interactions in Arabidopsis due to the relatively weak antibody for the native BRI1 protein. Our findings suggest that AGG1 and AGB1 work together in UPR signaling to mediate growth-defense tradeoffs that involve direct interactions with nascent FLS2 and BRI1 proteins at the ER membrane.
Combination of agg1 agg2 and atg7/3 promotes robust growth and defense
To investigate whether FLS2 protein is being targeted by UPR-associated protein degradation processes in agg1 agg2, we measured FLS2 protein expression in seedlings pretreated with chemical inhibitors of ERAD and autophagy. Co-treatment of proteasome inhibitor MG132 and autophagy inhibitor E-64D (Oh-ye et al., 2011) or concanamycin A (Con A) (Yoshimoto et al., 2004), or that of E-64d and Con A restored FLS2 protein expression in agg1 agg2 to or above WT level, whereas single treatments did not (SI Appendix, Fig. S10A). We then measured the flg22-induced seedling growth inhibition response in the presence of these inhibitors. Co-treatment of MG132 and E-64d restored growth inhibition in agg1 agg2 to WT level, whereas single treatments did not (SI Appendix, Fig. S10B). Furthermore, MG132 and E-64d co-treatments did not significantly affect growth in WT or agg1 agg2 plants relative to mock treatment (SI Appendix, Fig. S10C), whereas treatments with Con A at nano-molar concentrations proved toxic and were thus removed from analysis. These data suggest that UPR-associated degradation of nascent FLS2 proteins involves autophagy and/or ERAD.
To confirm these findings, we knocked out the two autophagy-requiring ubiquitin-like conjugation systems in the agg1 agg2 mutant by introducing loss-of-function mutations in the E1-like ATG7 and E2-like ATG3 genes by intermutant crosses (Klionsky 2005; Ohsumi 2001; Kim et al., 2012). We then measured flg22-induced growth inhibition and FLS2 protein expression in the agg1 agg2 atg7 and agg1 agg2 atg3 triple mutants. Flg22-induced growth inhibition and FLS2 protein expression were both restored in agg1 agg2 atg7 and agg1 agg2 atg3 to or greater than WT levels (Fig. 6A–B). To confirm that the recovered FLS2 proteins were not retained in the ER, we digested the proteins with endoglycosidase H (Endo H) enzyme to cleave off their ER-specific glycans. FLS2 proteins that exit the ER will acquire Golgi-specific glycans that are resistant to Endo H digestion. Wild-type FLS2 proteins were partially deglycosylated upon Endo H digestion, whereas FLS2 proteins produced in the mns1 mns2 mns3 mutant lacked Golgi-specific glycans (Liebminger et al., 2009), and thus were fully deglycosylated (SI Appendix, Fig. S11A). FLS2 proteins in the agg1 agg2 atg7 and agg1 agg2 atg3 plants were partially deglycosylated (SI Appendix, Fig. S11A), indicating that they have exited the ER en route to the PM.
We then investigated whether inhibition of autophagy is sufficient to restore growth-defense tradeoffs in the agg1 agg2 mutant. agg1 agg2 exhibited reduced seedling flg22-induced MAPK activation response and adult leaf resistance to the virulent bacterial pathogen Pseudomonas syringae pv. tomato DC3000 (Pto DC3000), as well as enhanced seedling, vegetative and reproductive growth relative to WT (Fig. 6D; SI Appendix, Figs. S1B and S11B). By contrast, agg1 agg2 atg7 and agg1 agg2 atg3 exhibited normal flg22-induced MAPK activation response and anti-bacterial defense, as well as normal growth and development relative to WT and atg7/3 single mutants (Fig. 6C-D; SI Appendix, Figs. S1 and S11B). Furthermore, agb1 exhibited reduced anti-bacterial defense, whereas the double mutants agb1 atg7 and agb1 atg3 exhibited normal bacterial resistance and adult development relative to WT and atg7/3 single mutants (SI Appendix, Figs. S1 and S12). These findings indicate that AGB1 and AGG1/2 work together in UPR signaling to mediate growth-defense tradeoffs that involve repression of FLS2 protein degradation by autophagy.
UPR is hardwired to promote FLS2 protein degradation in the absence of ER stress
To investigate whether UPR in growth-inducing conditions requires protein ‘triggers’ for selective protein degradation, we used the XVE system to drive expression of a FLS2 transgene in agg1 agg2 (agg1 agg2/XVE:FLS2). To our surprise, we obtained three independent lines (#2–4), whose FLS2 protein expression was knocked-down to undetectable levels with or without FLS2 induction, whereas their seedling growths were unchanged upon FLS2 induction (Fig. 7A; SI Appendix, Fig. S13). We also obtained one line (#1) whose FLS2 protein expression and growth were increased to and greater than WT levels, respectively, in the absence of FLS2 induction (Fig. 7A; SI Appendix, Fig. S13). This finding lends further support to the uncoupling of growth and defense in the agg1 agg2 mutant. Leaky expression of the XVE system has been reported in rice (Okuzaki et al., 2011) and appears to be sufficient to activate further increases in FLS2 production and/or degradation. To confirm that FLS2 production and subsequent degradation were activated in agg1 agg2/XVE:FLS2 lines #2–4, we measured their FLS2 protein expression after gene induction and co-treatment with E-64d and Con A and found FLS2 protein expression to be restored to WT level for all three lines (Fig. 7B). Taken together, these data suggest that in the absence of AGG1/2 and ER stress, UPR is hardwired to promote BRI1 protein biogenesis and FLS2 protein degradation via autophagy while at the same time responsive to transient and minute increases in FLS2 protein expression (Fig. 7C).
Discussion
A key innovation of our study was the development of the agg1 agg2 atg7/3 mutants and the agg1 agg2/XVE:FLS2 transgenic lines. The removal of functionally redundant Gγ subunits caused hyperactivation of BRI1 protein biogenesis and FLS2 protein degradation and the uncoupling of growth and defense, while the removal of autophagy restored FLS2-mediated immune responses in agg1 agg2 mutant. As a consequence, even though agg1 agg2 plants exhibited enhanced growth and reduced defenses, agg1 agg2 atg7/3 plants were able to grow and defend well at the same time. Transient FLS2 expression in agg1 agg2 uncovered an anticipatory UPR-mediated response that appears hardwired to promote growth over defense and is actively repressed by G-proteins that directly interact with FLS2 at the ER. Although traditionally viewed as an adaptive response triggered by the accumulation of unfolded proteins in the ER, we show that UPR is also an anticipatory response that is activated well before the disruption of protein homeostasis. Our results provide the first evidence that Gβγ dimers mediate growth-defense tradeoffs through UPR signaling. Furthermore, their signaling function in the ER is independent of their canonical functions in the Gαβγ heterotrimers at the PM. The ability to unlock or fine-tune growth-defense tradeoffs through UPR signaling provides a novel strategy to combine plant traits in ways that can have practical applications in biotechnology and agriculture.
A central premise underlying current views of growth-defense balance between BR and flagellin signaling is that the defense-defense antagonism is largely unidirectional (favoring growth over defense) and indirect (Lozano-Durán and Zipfel, 2015). This signaling architecture presumably serves to prevent autoimmunity and severe growth retardation from prolonged and/or de-regulated activation of immune receptors so that plants can excel in obtaining limited resources from their competitors. The ability of agg1 agg2 plants to grow robustly at the expense of defense provides evidence that the growth-defense antagonism can be reversed to promote faster activation of immune receptors and overcome domestication-related tradeoffs against defensive traits. Furthermore, AGB1-AGG1/2 heterodimers bind directly to BRI1 and FLS2 proteins at the ER membrane. In yeast and mammals, the binding of unfolded proteins to IRE1 directly activates UPR under ER stress-inducing conditions (Gardner and Walter, 2011; Karagöz et al., 2017). Similarly, the binding to BRI1 and FLS2 to AGB1-AGG1/2 may tune the homeostatic functions of UPR under conditions that favor growth and are independent of ER stress. Whether G-proteins in the ER also mediate growth-defense tradeoffs under defense-inducing conditions remains to be tested.
METHODS
Plant materials and growth conditions
Surface-sterilized seeds of Arabidopsis thaliana accession Columbia-0 (Col-0) were stratified for at least 2 days and sown in 12-or 24-well microtiter plates sealed with parafilm. Each 12-or 24-well plate contained 12 and 5 seeds, respectively, with 1 and 0.5 mL of filter-sterilized 0.5X MS liquid (pH 5.7–5.8) [4.43 g/L Murashige and Skoog basal medium with vitamins (Murashige and Skoog, 1962) (Phytotechnology Laboratories, Shawnee Missions, KS), 0.05% (w/v) MES hydrate, 0.5% (w/v) sucrose], respectively. Alternatively, surface-sterilized and stratified seeds were sown on MS agar plates [0.5X MS, 0.75% (w/v) agar (PlantMedia, Chiang Mai, Thailand)] sealed with parafilm. Unless otherwise stated, sample-containing plates were placed on grid-like shelves over water trays on a Floralight cart (Toronto, Canada), and plants were grown at 21°C and 60% humidity under a 12-hr light cycle (70–80 μE m-2 s-1 light intensity). Unless otherwise stated, media in microtiter plates were exchanged for fresh media on day 7. For bacterial infection experiments, Arabidopsis plants were grown on soil [3:1 mix of Fafard Growing Mix 2 (Sun Gro Horticulture, Vancouver, Canada) to D3 fine vermiculite (Scotts, Marysville, OH)] at 22°C daytime/18°C nighttime with 60% humidity under a 12-hr light cycle (100 μE m-2 s-1 light intensity). Nicotiana benthamiana plants were grown on soil [3:1 mix] on a Floralight cart at 22°C under a 12-hr light cycle (100 μE m-2 s-1 light intensity) for 4 weeks.
The following Col-0 T-DNA insertion lines and mutants were obtained from the Arabidopsis Biological Resource Center (ABRC, Columbus, Ohio): agb1-1 (CS3976), agb1-2 (CS6535), agg1-1c (CS16550), agg2 (SALK_039423), agg1-1c/agg2-1 (CS16551), atg7 (SAIL_11_H07), fls2 (SAIL_691_C4).
Vector construction and transformation
To generate estradiol-inducible C-terminally tagged GFP and RFP (XVE:X-G/RFP) and 35S:YFP-AGB1 DNA constructs, attB sites were added via PCR-mediated ligation to the coding sequences of cDNAs, and the modified cDNAs were recombined into pDONR221 entry vector and then into pABindGFP, pABindRFP (Bleckmann et al., 2010) and pB7WGY2 (Karimi et al., 2002) destination vectors, according to manufacturer’s instructions (Gateway manual; Invitrogen, Carlsbad, CA). XVE:AGG1-RFP, XVE:FLS2-RFP, and 35S:YFP-AGB1 constructs were introduced into agg1-1c agg2-1 or agb1-2 plants via Agrobacterium-mediated floral dip method (Clough and Bent, 1998), and transformants were selected on agar media containing 15 μg/mL hygromycin B (Invitrogen) or 15 μg/mL glufosinate (Cayman Chemical, Ann Arbor, MI). Transgene expression was induced 48 hr (or 5-6 days for growth assays) after elicitation with 20 μM β-estradiol (2 mM stock solution in DMSO; Sigma-Aldrich, St. Louis, MO). Transient expression of XVE:X-G/RFP constructs in Nicotiana benthamiana leaves was performed as previously described (Bleckman et al., 2010) with the following modification: transformed Agrobacterium strains were grown in LB medium supplemented with 50 μg/mL rifampicin, 30 μg/mL gentamycin, kanamycin 50 μg/mL and 100 μg/mL spectinomycin, in the absence of a silencing suppressor, to an OD600 of 0.7. Transgene expression was induced 10 hr (for co-immunoprecipitation) and 4-8 hr (for microscopy) after spraying with 20 μM β-estradiol and 0.1% Tween-20.
BL-induced hypocotyl elongation
Seedlings were grown on MS agar supplemented with 1 or 10 μM 24-epibrassinolide (BL; Phytotechnology Laboratories). Sample-containing agar plates were placed vertically on a Floralight cart under a constant light cycle (140–180 μE m-2 s-1 light intensity). Hypocotyl lengths were measured from images of 5-day-old seedlings using NIH ImageJ.
Flg22-induced seedling growth inhibition
Three-day-old seedlings in 24-well microtiter plates were elicited with water or 100 nM flg22 (QRLSTGSRINSAKDDAAGLQIA; Genscript, Nanjing, China) for 6 days. Fresh weights were measured from 9-day-old seedlings that were dried between paper towels for a few seconds.
Flg22-induced Callose Deposition
9-day-old seedlings were elicited with 1 μM flg22 for 16-18 hr. Alternatively, 9-day-old seedlings were treated with DMSO, 100 nM BL, or 1 μM BL 6 hours prior to flg22 elicitation. Callose deposition staining was performed as previously described (Clay et al., 2009). Callose deposits were viewed on a Zeiss (Oberkochen, Germany) AxioObserver D1 fluorescence microscope under UV illumination with Filter Set 49 (excitation filter 365 nm; dichroic mirror 395 nm; emission filter 445/50 nm). Callose deposits were quantified using NIH ImageJ.
Flg22-induced MAPK activation
9-day-old seedlings were elicited with 100 nM flg22 for 5, 15, and/or 30 min. MAPK activation assay was performed as previously described (Lawerence et al., 2017). 20 μl of supernatant was loaded onto a 10% SDS-PAGE gel, and the separated proteins were transferred to PVDF membrane (Millipore) and probed with phosphor-p44/p42 MAPK (Cell Signaling Technology, Danvers, MA) and MPK3 antibodies (Sigma-Aldrich, St. Louis, MO) at 1:2000 dilution in 5% (w/v) nonfat milk in 1X PBS. The combined signal intensities of phosphorylated MPK3/4/6 were quantified using NIH ImageJ and normalized to that of total MPK3 (loading control).
Total protein extraction, SDS-PAGE, and western blotting
Total protein was extracted from snap-frozen seedlings into 80 μL of extraction buffer [50 mM Tris-Cl (pH 7.5), 50 mM DTT, 4% (w/v) SDS, 10% (v/v) glycerol] using a 5-mm stainless steel bead and ball mill (20 Hz for 3 min). Samples were centrifuged briefly, incubated at 95°C for 10 min, and centrifuged at 12,000 x g 19 for 8 min to precipitate insoluble material. Endo H treatment was performed as previously described (Lawrence et al., 2017). 5 or 10 μL of extract were loaded onto a 8.5% SDS-PAGE gel, and the separated proteins were transferred to PVDF membrane (Millipore, Billerica, MA), stained with Ponceau S for labeling of total protein (loading control), and probed with either BRI1 (Agrisera), FLS2 (antigen: CTKQRPTSLNDEDSQ; Genscript), RFP (MBL International, Woburn, MA), GFP (Roche, Basel, Switzerland), BIP (Enzo Life Sciences, Farmingdale, NY), PDI antibodies at 1:500 (BRI1), 1:1000 (FLS2, RFP, GFP) and 1:5000 (BIP, PDI) dilutions in 5% (w/v) nonfat milk in 1X PBS. Signal intensities of immuno-detected proteins were quantified using NIH ImageJ and normalized to that of loading control.
Aqueous 2-phase partitioning and immunoprecipitation
Microsomal membrane (MM) and plasma membrane (PM) proteins were isolated from 250 mg of snap-frozen plant tissue using Minute Plasma Membrane Protein Isolation Kit for Plants (Invent Biotechnologies, Plymouth, MN). Membrane protein pellets were extracted into 250 μL of extraction buffer [50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% NP-40, 10% glycerol, 2 mM EDTA, 5 mM DTT, complete-mini protease inhibitor cocktail (Roche), 5 μM AEBSF] for 1 hr at 4°C with rocking and clarified at 8,000 x g for 10 min. Twenty microliters of extract was set aside as input. Membrane proteins were immunoprecipitated with 2.5 μL of antibody for 4 hr at 4°C with rocking followed by 25 μL of 50% slurry of Protein A/G magnetic beads (EMD Millipore, Burlington MA) for 1 hr at 4°C with rocking, and washed 3x with 350 μL of wash buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl). Enrichment of ER membrane proteins in the MM protein extracts was confirmed by western blotting using BIP antibody.
RNA isolation and quantitative PCR (qPCR)
Total RNA was extracted into 1 mL of TRIzol reagent (Invitrogen) according to manufacturer’s instructions. 2 μg of total RNA was reverse-transcribed with 3.75 μM random hexamers (Qiagen, Hilden, Germany) and 20 U of ProtoScript II (New England Biolabs, Boston, MA). The resulting cDNA:RNA hybrids were treated with 10 U of DNase I (Roche) for 30 min at 37°C, and purified on PCR clean-up columns (Macherey-Nagel, Düren, Germany). qPCR was performed with Kapa SYBR Fast qPCR master mix (Kapa Biosystems, Wilmington, MA) and CFX96 or CFX384 real-time PCR machine (Bio-Rad, Hercules, CA). The thermal cycling program is as follows: 95°C for 3 min; 45 cycles of 95°C for 15 sec and 53°C or 55°C for 30 sec; a cycle of 95°C for 1 min, 53°C for 1 min, and 70°C for 10 sec; and 50 cycles of 0.5°C increments for 10 sec. Biological replicates of control and experimental samples, and three technical replicates per biological replicate were performed on the same 96-or 384-well PCR plate. Averages of the three Ct values per biological replicate were converted to differences in Ct values relative to that of control sample. Pfafflmethod (Pfaffi2001) and calculated primer efficiencies were used to determine the relative fold increase of the target gene transcript over the housekeeping eIF4AI gene transcript for each biological replicate. Primer sequences and efficiencies are listed in Supplementary Table 1.
Confocal microscopy
Live epidermal root cells of 5-day-old Arabidopsis seedlings and 4-week-old N. benthamiana leaves were imaged using a 40X 1.0 numerical aperture Zeiss water-immersion objective and a Zeiss LSM 510 Meta confocal microscopy system. GFP and RFP were excited with a 488-nm argon laser and 561-nm laser diode, respectively. GFP and RFP emissions were detected using a 500-550 nm and 575-630 nm filter sets, respectively. Plasmolysis was induced by 5-10 min treatment of N. benthamiana leaf strips with 0.8 mannitol, and co-localization of GFP/RFP-tagged proteins to Hechtian strands was made visible by over-exposing confocal images using ZEN software.
ER stress induction
For qPCR and western blots, 9-day-old seedlings were treated with 5 μg/mL tunicamycin (0.5 mg/mL stock solution in DMSO; Sigma-Aldrich) or solvent control (DMSO) for 5 hr. For growth inhibition, seedlings were grown on MS agar supplemented with 0, 25, 50, and 100 ng/mL tunicamycin for 14 days.
Inhibition of protein translation and degradation
For BRI1 westerns, 5-day-old seedlings were treated with DMSO (mock) or 50 μM cycloheximide for 6 hr. For FLS2 westerns, 9-day-old seedlings were treated with 50 μM cycloheximide for 0, 2 and 16 hr. 9-day-old seedlings were also treated with DMSO (mock), 50 μM MG132 (50 mM stock solution in DMSO; Selleck Chemicals, Houston, TX), 20 μM E-64d (20 mM stock solution in DMSO; Cayman Chemical), and/or 2 μM concanamycin A (200 μM stock solution in DMSO; Santa Cruz Biotechnology) for 24 hr. For growth inhibition, 3-day-old seedlings were elicited with 100 nM flg22 or water for 2 days, and then treated with DMSO, 50 nM MG132, and/or 20 nM E-64d for 4 days.
Bacterial infection
Pathogen assays on 4-to 5-week-old adult leaves were performed as previously described (Chezem et al., 2017).
ACKNOWLEDGEMENTS
We thank N. Koizumi for ire1a-2/ire1b-2 (SALK_018112, GABI_638B07) mutant, B. Bartel for atg3 mutant, F. Ausubel for bri1-116 mutant, R. Simon for pABindGFP and pABindRFP vectors, and C. Somerville for PDI antibody. The work was supported by T32 GM007223 (to J.C.M), T32 GM007499 (to S.A.L), Yale University Elizabeth Brown Fellowship (to T.C.), and Yale College Dean’s Research Fellowship (to C.B.).
Author Contributions
C.B. contributed to characterization of agg1 agg2/XVE:FLS2 lines. T.C. contributed to Endo H digestion. S.A.L. generated agb1/35S:YFP-AGB1 lines, and contributed to qPCR of ire1a/b, Endo H digestion and MAPK activation of agg1 agg2 atg7/3 N.K.C. generated XVE:BRI1-RFP construct and agg1 agg2/XVE:FLS2 lines, and contributed to Co-IPs and pathogen assays. J.C.M. generated all other constructs, and contributed to characterization of agg1 agg2/XVE:FLS2 lines, Co-IPs and all other experiments. J.C.M. and N.K.C. interpreted the results and N.K.C. wrote the paper.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵