The interplay of lipid acyl hydrolases in inducible plant defense

Abstract

Lipid acyl hydrolases (LAH) have received recently increased attention in the context of plant defense. Multiple structurally unrelated gene families have been annotated in Arabidopsis as encoding potential lipid deacylating enzymes with numerous members being transcriptionally activated upon biotic stress. Confirming in silico predictions, experimental data have illustrated the wide subcellular distribution of LAHs indicating they likely interact with distinct membrane systems to initiate specific cellular responses. While recombinant LAHs are active in vitro on a small set of polar lipids, precise knowledge of in vivo substrates and hydrolysis products is generally lacking. Functional analysis of a few LAHs has revealed their roles in initiating oxylipin biosynthesis, cell death execution, signaling or direct antimicrobial activity. The picture emerging is that pathogenic challenge triggers a complex network of lipid hydrolysis events across the cellular compartments resulting in changes in membrane structures and release of signal precursors involved in the building-up of an adequate immune response.

Beyond their structural role in delineating subcellular compartments, membranes perform prominent functions in the response of plant tissues to microbial attacks. Membrane-anchored receptor protein complexes form surveillance systems that sense the presence of non-self epitopes, also called pathogen-associated molecular patterns (PAMP). PAMP recognition events are transduced into sophisticated intracellular responses that trigger ultimately a multifaceted defense program. Among infection-induced defense reactions, host cellular membranes are targeted by a number of lipolytic enzymes including phospholipases (of the A, C and D type) and galactolipases that generate free fatty acids and various hydrolysis products.^1,2 In addition, lytic enzymes secreted by microbes at the leaf surface depolymerize the fatty-acid based cutin matrix, releasing elicitor-active and antimicrobial breakdown products.³ The action of these enzymes accounts for part of the many lipid-based signaling processes known to operate in inducible plant defense.^2,4–7 The degradative action of LAH may alter the capacity of the lipid bilayer to regulate exchanges between compartments, but many subsequent physiological responses are due to the novel biological properties of lysolipids or oxygenated fatty acid derivatives that are generated. Plant LAHs have been particularly examined in the context of the production of free oxylipins with prominent roles in defense responses⁶ and whose biosynthesis depends on the release of fatty acids or derivatives from membrane lipids. Targeted hydrolysis may activate early and transient signaling while long lasting, unspecific deacylation may compromise vital functions and commit tissue to cell death.

Stimulation of lipid hydrolysis in plant-microbe interactions was initially reported by monitoring enzyme activities or by following lipid breakdown in plant extracts.⁸ However, few enzymes could be identified by direct purification and formally associated with degradation of a given membrane system. Different activities, including phospholipase A (PLA), PLC and PLD, were found increased in various pathosystems but enzymes often proved low abundant and difficult to isolate.² Therefore, biochemical methods as a single approach only offered a limited access to lipolytic enzymes in vegetative plant tissues.

Genomic data mining in Arabidopsis has provided in the past decade new ways to tackle these questions more systematically, mostly with the tools of reverse genetics. The major known LAH gene families in Arabidopsis include the patatins, DAD1-like lipases, secreted PLA₂ (sPLA₂) and GDSL lipases^1,8 (Table 1). All families have members with described roles in the responses of plants to microbial or herbivore attacks. An additional family called PRLIP is strongly pathogen- or hormone-responsive,⁹ but no natural lipidic substrate is known nor are genetic data available for this family. We will review here briefly the current knowledge on the structure, regulation, enzymatic activity of several glycerolipid deacylating enzymes and their physiopathological impacts on inducible plant defense.

Table 1.

Examples of Arabidopsis LAH genes with reported roles in responses to wounding or microbial infections

LAH family (number of genes in Arabidopsis)	Member name (AGI)	Confirmed subcellular localization	Transcriptional upregulation	in vitro substrate	Known functions	References
patatin (10)	pPLA-IIα (At2g26560)	cytosolic	fungal and bacterial infections	MGDG>DGDG>PC	host cell death execution	La Camera et al. 2005;13 La Camera et al. 200916
pPLA-IIIβ (At3g54950)	membrane-bound	fungal and bacterial infections	ND	ND	La Camera et al. 200513
	pPLA-I (At1g61850)	cytosolic/chloroplastic	ND	MGDG>DGDG>PG	basal JA production, antifungal resistance	Yang et al. 200715
DAD1 (12)	DAD1 (At2g44810)	chloroplast	expressed in anthers; wound-induced in leaves	PC>MGDG>TG	JA production in anthers; late JA production in wounded leaves	Ishiguro et al. 200119
	DGL (PLA1-Iα-1; At1g05800)	lipid bodies	wound-induced in leaves	DGDG>PC> MGDG>TG	minor role in wound-induced JA biosynthesis?	Hyun et al. 2008;22 Ellinger et al. 201023
	PLA1-Iγ1 (At1g06800)	chloroplast	fungal and bacterial infections	galactolipids≈ phospholipids≈ triglycerids	ND	Seo et al. 2009;20 T. Heitz, unpublished
PLA1-Iγ2 (At2g30550)	chloroplast	fungal and bacterial infections	galactolipids≈ phospholipids≈ triglycerids	ND	Seo et al. 2009;20 T. Heitz, unpublished
GDSL (108)	GLIP1 (At5g40990)	extracellular space	A. brassicicola, ethylene	synthetic esters	signaling; antifungal activity	Oh et al. 2005;30 Kwon et al. 200931
GLIP2 (At1g53940)	?	salicylic acid	synthetic esters	antifungal, antibacterial resistance	Lee et al. 200932

ND, not determined.

Occurrence and Diversity

Most LAH known to date (except sPLA₂) are serine hydrolases with a conserved G-x-S-x-G motif where the central Ser forms a catalytic triad with conserved Asp and His residues. This minimal esterase motif is also shared by other hydrolytic enzymes cleaving ester and amide bonds. Distinct additional structural features are specific to each LAH gene family, resulting in a set of LAH varying in their substrate preference and catalytic properties. A distinction can be made between broad spectrum deacylating enzymes such as patatins that hydrolyze both sn-1 and sn-2 positions of galactolipids and phospholipids but are not active on storage triglycerides, and more specific enzymes such as sPLA₂ that have an sn-2 positional preference on phospholipids.⁸

A number of recent studies of individual genes has confirmed that members of a given LAH gene family are targeted to different subcellular compartments and therefore the complement of LAHs is specific to each compartment. This is particularly true upon wounding or pathogenic assault where increased lipolysis occurs by activation of pre-existing enzymes and rapid appearance of induced isoforms. A few examples are illustrated in Figure 1 showing the distribution of four pathogen-induced LAHs of the patatin and DAD1-like family by LAH-GFP fusion visualization. pPLAIIα is primarily cytosol-localized (Fig. 1A), while pPLA-IIIβ is prone to associate with cellular membranes (Fig. 1B). In contrast, PLA₁-Iγ1 and PLA₁-Iγ2 are targeted to chloroplasts, but fusion proteins label distinct subchloroplastic domains, likely reflecting interactions with different membrane systems (Fig. 1C and D).

Figure 1.

Examples of subcellular localization of Arabidopsis pathogen-induced LAHs as visualized by LAH-GFP fusions observed under confocal microscope. Red signal is due to chlorophyll autofluorescence. (A, C and D) mesophyll cells. (B) epidermal cell. (A) pPLA-IIα; (B) pPLA-IIIβ, inset: close-up showing labeling of cortical cytoplasm. (C) PLA₁-Iγ2. (D) PLA₁-Iγ1. Scale bar: 20 µm. (C and D) insets (10 µm) show close-up views of a 4 µm optical section across a chloroplast.

Functional Analysis of LAH Gene Families in Inducible Plant Defense

Patatins.

Patatins are non-specific acylhydrolases that were known first as abundant storage proteins in potato tubers.¹⁰ This organ expresses a large number of closely related patatin genes, complicating genetic analysis. The patatin-related family, whose nomenclature was recently updated using the acronym pPLA¹¹ comprises ten genes in Arabidopsis, with encoded proteins falling into three subclasses.^12,13 Interestingly, there is evidence for involvement in defense responses for a member of each subclass.

pPLA-I (the single member of subclass I, Table 1) is a constitutive enzyme in Arabidopsis that shares structural resemblance with calcium-independent iPLA₂α, a member of the so-called animal PNPLA protein family displaying a patatin domain.¹⁴ pPLA-I has a modular structure with leucine-rich and Armadillo repeats in addition to the catalytical region. Recombinant pPLA-I bears preferential galactolipase over phospholipase activity that may be modulated in planta by interaction with regulatory protein partners.¹⁵ pPLA-I was shown to hydrolyze in vitro both oxylipin- and non-oxylipin-containing galactolipids. However, the in vivo relevance of this observation is unknown, as lipid profiling of Botrytis-infected leaves did not evidence differences in MGDG or DGDG hydrolysis between wild-type and pPLA-I knock-out plants. Interestingly, pPLA-I-deficient plant lines displayed increased fungal colonization upon Botrytis inoculation, but this enhanced susceptibility was not associated with altered increase of jasmonic acid (JA). Instead, ablation of pPLA-I reduced basal JA levels to about half the WT levels. By which mechanism this change affects antifungal resistance is unknown.

Subclass II of patatins (Table 1) comprises five genes that are most similar in sequence and size to potato tuber patatins.^12,13 Transcript profiling established that pPLA-IIα is the only subclass member that is strongly responsive to all necrotizing pathogens tested including fungi and bacteria. At least in the case of Botrytis cinerea infection, induction was COI1-dependent.¹³ pPLA-IIα encodes a cytoplasmic galacto/phospholipase (Fig. 1A) that is responsible for most extractible galactolipase activity in infected leaves. Using silenced and overexpressing plant lines, it was found that its expression increases the severity of symptoms and favors colonization of leaves by fungal and bacterial pathogens. These results and altered sensitivity to paraquat pointed to the conclusion that the primary role of pPLA-IIα is to promote host cell death execution. Accordingly, pPLA-IIα expression positively affected resistance to an obligate pathogen like cucumber mosaic virus whose spread requires living host cells.¹⁶ Furthermore, silencing or overexpressing pPLA-IIα in the lesion mimic mutant vad1 background reduced or enhanced, respectively, developmentally triggered cell death. We performed broad spectrum oxylipin profiling in B. cinerea-infected plants modified for pPLA-IIα expression and found an impact of this LAH on α-dioxygenase-generated oxylipins but not on jasmonates that are lipoxygenase-derived.¹⁶ Altogether it appears from these results that pPLA-IIα is not a signaling enzyme but catalyzes a late, non-specific hydrolysis of membrane lipids that commits tissues to cell death and affects differentially resistance to pathogens. These two examples suggest that members of the patatin LAH family can have opposite effects on pathogen resistance, possibly via distinct molecular mechanisms. pPLA-IIα is also induced by water deficit and it was hypothesized to contribute to the galactolipid breakdown in leaves upon drought stress.¹⁷

The subclass III of patatins (Table 1) whose four members are less similar to potato tuber patatins has not been studied in detail. A common feature of the encoded proteins is a modified serine hydrolase motif and accordingly no enzymatic activity has been described for these isoforms, although this aspect would deserve deeper investigation. An early indication of biological activity was the report that the overexpression of the subclass III protein pPLA-IIIδ resulted in a pleiotropic developmental phenotype with stiff inflorescence stems, thicker leaves and larger seeds.¹⁸ Interestingly, pPLA-IIIβ appeared as the sole subclass member to be strongly induced by B. cinerea and Pseudomonas syringae infection in parallel to pPLA-IIa.¹³ T-DNA inactivation or ectopic overexpression of pPLA-IIIβ did not alter resistance to these pathogens (Heitz T., unpublished) and thus no hint for possible function has been obtained from these experiments.

DAD1-like LAHs.

A second important LAH family was uncovered by forward screening of male sterile mutants. This search identified the deficient in anther dehiscence 1 (dad1) mutant that is impaired in pollen development and anther dehiscence.¹⁹ DAD1 encodes a plastidial PLA₁ that initiates JA synthesis in stamens, this latter hormone coordinating gene expression and water loss in maturating anthers. Most importantly, DAD1 defined a novel family of intronless genes encoding putative lipolytic enzymes (Table 1) that can be sorted into three subclasses according to the structure of their predicted N-termini. Subclass I isoforms including DAD1 were predicted to be chloroplast-targeted, while four isoforms with no transit peptide are presumably cytosolic and one may be addressed to mitochondria.

The in vitro enzymatic properties of the seven subclass I members were recently described²⁰ and based on their activity towards various lipids could be further divided into three subgroups. A first group displayed PLA-specific activity, a second PLA and galactolipase activities and the third had broad activity towards phosphatidylcholine, galactolipids and triglycerides. These distinct substrate preferences are consistent with LAH-GFP fusions labeling different structures in chloroplasts (Fig. 1C and D) that likely vary in lipid composition. They also point towards specific roles of these acylhydrolases in plastidial lipid homeostasis.

The functional data reported for DAD1 motivated a number of promising studies to explore the possibility that some of the DAD1-related LAHs could function in the initiation of stress-induced JA biosynthesis in leaves. JA is derived from linolenic acid, an abundant fatty acid in plastidial glycerolipids.²¹ As the core enzymes catalyzing the synthesis up to 12-oxo-phytodienoic acid, the precursor of JA, are all chloroplast-localized, fatty acid precursors are likely mobilized from plastidial lipid membranes by a LAH residing in this organite. DAD1 deficiency was initially reported not to affect wound-induced JA accumulation in leaves.¹⁹ A recent study described an Arabidopsis activation line that overexpresses DGL, or PLA₁-Iα-1, in the nomenclature defined by Ryu (2004),¹ a close relative of DAD1, resulting in a stunted growth phenotype and increased JA content.²² According to these authors, overexpression of other Arabidopsis DLP genes does not lead to elevated JA levels. DGL encodes a chloroplast-targeted enzyme with galactolipase and PLA₁ activity and is induced earlier than DAD1 upon leaf wounding. Data collected from silenced and overexpressing lines pointed to cooperative and partially redundant functions of these two LAHs in wound-induced JA production, DGL being proposed as having a role in basal levels and early response and DAD1 in the later phases. The conclusions that DGL and DAD1 are necessary and sufficient for JA production in Arabidopsis leaves were recently questioned by Ellinger et al.²³ who failed to identify abolished basal levels or altered early wound-induced JA levels in DGL-deficient or dad1 plants. A closer examination also revealed that DGL is targeted to lipid bodies rather than to chloroplasts. A quadruple mutant defective in the expression of four DAD1-like genes accumulated about 30% of wild-type basal and early wound-induced JA levels, indicating significant functional redundancy in this gene family. In addition, single mutants in 18 novel lipases with chloroplast localization, predicted by a bioinformatic approach, did not show strong differences with wild-type wound-induced JA levels.²³ Functional homologs of DAD1 and DGL were searched for in Nicotiana attenuata and the silencing of one homolog, GLA-I, encoding a chloroplastic LAH with activity on MGDG, phosphatidylcholine and TAG, was found to reduce strongly the levels of JA elicited by fatty acid conjugates generated in insect oral secretions upon feeding.²⁴ Curiously, GLA-I mRNA levels were downregulated upon elicitor application, showing that transcriptional activation may not be necessary for early, minute-scale wound responses.

We have also examined the response of DAD1-related genes to microbial infections. Using the B. cinerea and P. syringae pathosystems, we determined that transcript levels of the PLA₁-Iγ1 and PLA₁-Iγ2 were the most pathogen-responsive (Fig. 2) while PLA₁-III was only Botrytis-responsive. However, knock-out lines in PLA₁-Iγ1, PLA₁-Iγ2 and PLA₁-III did not exhibit significant changes in resistance levels to these pathogens nor did the two former mutants display strongly altered induced JA profiles compared to wild-type (Heitz T., unpublished). Therefore, and despite intensive efforts, no strong evidence is available pointing to prominent roles of the DAD1-like family of LAHs in mediating antimicrobial resistance.

Figure 2.

Real-time RT-PCR analysis of the expression of PLA₁-Iγ2 and PLA₁-Iγ1 upon (A) inoculation with Botrytis spores or (B) infiltration of virulent (Pst) or avirulent (Pst avrRpm1) strains of Pseudomonas syringae pv. tomato. Uninoculated plants were taken as reference condition. Dpi, days after inoculation.

It should be noted that these studies focused on the impact of mutations in genes encoding chloroplastic LAHs on jasmonate levels, but broader oxylipin/lipid analysis is needed to reveal their possible roles in general lipid homeostasis in this organite. Given the high number of chloroplast-targeted LAHs, these organites are major sites for glycerolipid hydrolysis upon stress responses: while galactolipids and phospholipids are the predominant glycerolipids in plastidial membranes, Arabidopsis also accumulates high levels of esterified oxylipins (mostly to galactolipids) known as arabidopsides, in response to wounding or infection.^25,26 Plastoglobules are subplastidial lipoproteic particles that also contain neutral lipids such as TAG or steryl-esters.²⁷ This structural lipid diversity opens the possibility that some of the many uncharacterized plastidial LAHs may mediate important functions in stress responses by hydrolyzing other lipids than galacto- and phospholipids.

GDSL lipases/esterases.

This latter large family was named after their conserved motif present in the catalytic site which is distinct from the widespread GxSxG serine hydrolase motif discussed above.²⁸ They have a flexible active site that can accommodate many substrates and characterized members display lipase, protease, thioesterase or other esterase activities.²⁹ Therefore, only certain members are true lipolytic enzymes and enzymatic data of several reports should be interpreted with caution. A first Arabidopsis member, GLIP1, was isolated in a study investigating the secreted proteome of salicylic acid-treated cultured cells.³⁰ The GLIP1 gene is also induced in plants by fungal infection and glip1 knock-out lines are susceptible to Alternaria infection. GLIP1 displays direct fungitoxic activity and seems also to affect signaling as glip1-1 exhibits aberrant regulation of some defense genes. Further properties were inferred from analysis of GLIP-overexpressing plants. Such lines have elevated basal defense transcript levels and their phloem exudates contain a defense- and resistance-inducing activity that is absent in glip1 exudates.³¹ GLIP2, a second isoform out of a family of seven members, was recently analyzed. GLIP2 is mostly expressed in roots and its inactivation alters several auxin responses and also impairs resistance to the necrotrophic bacterial pathogen Erwinia carotovora.³² While these observations point to original roles in disease resistance, it should be noted that enzymatic evaluation of these two GLIP proteins was carried out using synthetic short-chain p-nitrophenyl-esters substrates. Therefore, these results should be taken provisionally as proof of esterase activity, until a real lipid substrate can be identified.

Two other gene families have been clearly linked to defense responses and some members could mediate functions via non-classical modes of action. EDS1 and PAD4 were isolated by genetic screens from mutant plants with increased sensitivity to biotrophic pathogens. While no canonical lipolytic activity could be evidenced, these proteins form complexes that transduce redox signals in response to stress and activate SA signaling.³³ sPLA₂-α is a member of four secreted PLA₂ in Arabidopsis and was found to interact with AtMYB30, a transcription factor positively regulating hypersensitive cell death. Upon interaction, sPLA₂-α is relocalized to the nucleus where it represses AtMYB30-mediated defense. Accordingly, sPLA₂-α mutation or overexpression leads to increased or decreased anti-bacterial resistance respectively.³⁴ Other non-classical albeit important functions of LAH in defense may be revealed in the future.

Conclusions and Future Challenges

While we have now a better view of the genetic repertoire encoding lipid deacylating enzymes in the Arabidopsis genome, we are still far from understanding how these players are set in motion upon pathogenic attack. Genome mining has revealed a plethora of genes whose products exhibit recognizable features of lipases/esterases, but an initial task is to identify which of these proteins bear genuine LAH activity. To fill this functional gap between poor genome annotation and functional characterization, activity-based proteomics offers interesting potentials. This approach allows to display lipolytic activities in complex proteomes by labeling proteins exhibiting active site features with chemical probes.³⁵ Mass spectrometry-based identification of active proteins then opens the door to functional studies. Successful application of this strategy was reported in the search of serine hydrolases activated in response to Botrytis infection³⁶ and has a high potential for identifying additional LAHs.

Once an unambiguous lipolytic activity has been assigned to a novel gene product, its exact occurence and subcellular site of action need to be precisely defined. To this aim, transcriptional data and reporter fusion proteins are reasonable starting points but may not be sufficient to understand the regulation of lipid hydrolysis. LAHs are prone to post-translational activation by calcium, pH changes or interactions with protein partners. Fatty acid release by glycerolipases takes place at the interface between hydrophilic and hydrophobic environments in a way that challenges classical enzymology. An other difficulty that complicates the analysis of mutant plants arises from the need to monitor discrete hydrolysis events catalyzed by individual LAHs when several enzymes from different families can generate the same products in a given tissue. A new sophisticated way to look at lipid contents of tissues is lipid profiling by mass spectrometry using Electrospray Ionisation Collision-Induced Dissociation (ESI-CID).³⁷ This technique allows to quantify precisely distinct phospholipid and galactolipid molecular species varying in their fatty acid composition. Specific changes detected in a given species can inform on the nature of membrane system that underwent hydrolysis.

Given the occurrence of an extended repertoire of unrelated LAH genes, it can be predicted that a large spectrum of other lipid types or derivatives exert their biological functions after being released by LAHs. For example, phytoprostanes are biologically active fatty acid derivatives that are formed nonenzymatically in plant membranes under oxidative stress.^38,39 Phytoprostanes occur both as free or lipid-esterified forms and their mobilization is believed to depend on uncharacterized LAHs whose action may initiate membrane repair and stress signaling.

In the field of jasmonate biosynthesis, recent data showed that, at least in Arabidopsis, major pools of stress-induced OPDA and dn-OPDA, the precursors of JA, occur as galactolipid-esterified forms.^25,26 Interestingly, metabolic profiling of wounded Arabidopsis leaves identified partially hydrolyzed lysoarabidopsides, providing indirect evidence that an LAH was acting on Arabidopsides in vivo,⁴⁰ and could be a rapid means to mobilize JA precursors. Therefore, it is currently unclear to which extent JA may originate from arabidopside cleavage rather than from conversion of free linolenic acid.

Finally, there is a need to define better the cellular fate of hydrolysis products for each individual hydrolysis event. For instance, the possibility that some LAHs feed fatty acids into b-oxidation pathways to fullfill high energy demand in attacked plants has not been explored yet.