Emerging principles in plant chemical genetics

Emerging principles in plant chemical genetics

Review Emerging principles in plant chemical genetics Re´ka To´th1 and Renier A.L. van der Hoorn2 1 Department of Plant Developmental Biology, Max P...

350KB Sizes 1 Downloads 83 Views


Emerging principles in plant chemical genetics Re´ka To´th1 and Renier A.L. van der Hoorn2 1

Department of Plant Developmental Biology, Max Planck Institute for Plant Breeding Research, Carl-von-Linne Weg 10, 50829 Cologne, Germany 2 Plant Chemetics lab, Max Planck Institute for Plant Breeding Research, Carl-von-Linne Weg 10, 50829 Cologne, Germany; and Chemical Genomics Centre of the Max Planck Society, Otto-Hahn-Strasse 15, 44227 Dortmund, Germany

Chemical genetics is a powerful new discipline in plant science. Bioactive small molecules can be used to identify novel signalling nodes and unravel redundant networks. Observations made so far have revealed a series of principles in plant chemical genetics. These principles concern compound properties, such as bioactivation and bioavailability; and valuable approaches, like the use of derivatives and transcriptomics and successful ways of target identification. Together, these principles explain why the choice of the chemical library is important and instruct the design of future chemical genetic screens. ‘If you have a garden and a library, you have everything you need.’ Marcus Tullius Cicero The power of small molecules Small molecules are powerful tools to unravel biological processes. The use of small molecules to alter protein function and thereby explore biological roles of these target proteins is called ‘chemical genetics’ [1]. In principle, small molecules allow rapid, conditional and reversible alteration of biological functions; thereby the use of these molecules can generate different and often complementary information compared to classical genetic studies. Small molecules can also overcome limitations of genetic approaches such as lethality, genetic redundancy and pleiotropic effects observed in genetic mutants. Moreover, chemicals can distinguish between different isoforms of a gene product, e.g. caused by post-translational modification or the presence in a protein complex. A chemical genetics project involves three stages [2]. The first stage is the development of a robust assay in a small volume, preferentially in a microtiter plate format. The second, a screening stage, includes a primary screen to identify candidate compounds, and a secondary screen to verify the hits with an independent readout. The third stage encompasses phenotypic characterization and target identification. Compounds entering the third phase are already useful for biological studies, although target identification is essential to understand the precise mode-of-action of the compounds. Chemical genetics has become increasingly popular in plant science. The frequency of publications on screening chemical libraries, as well as the size of the screened Corresponding author: To´th, R. ([email protected]).

libraries is rapidly increasing (Table 1a), and there is more to come. Many of these studies have been summarized in excellent reviews [3–6]. Here we describe seven general principles that emerge from these initial studies. These principles provide a precedent and contain useful messages for scientists that consider embarking on a chemical genetic approach. Principle I: chemical genetics overcomes genetic redundancy The chemical genetic projects have provided excellent examples on how this approach can overcome problems associated with redundant signalling networks. Bioactive molecules (see glossary) can tackle the problem of redundancy in two ways (Figure 1): either the compound inhibits multiple components in the network and acts as a general antagonist, or the compound activates a specific component of the network as a specific agonist. This principle does not work the other way: specific antagonists would not stop the response from a redundant network, similar to single gene mutants; whereas a general agonist would trigger too many components to be identified by genetics. An example of a general antagonist is bikinin, which was identified in a screen for constitutive brassinosteroid (BR) responses [7]. Bikinin treatment of BR signalling mutants revealed that bikinin acts at the level of BIN2 (Brassinosteroid Insensitive 2), which is a GSK3-like Glossary Active moiety: the part of a small molecule that makes the molecule bioactive. Agonist: compound triggering a response. Antagonist: compound preventing or dampening a response. Bioactive: ability of a small molecule to modify a biological system. Bioactivation: activation of an inactive molecule by conversion inside the biological system. Bioavailability: the availability of a small molecule in a biological system at the site of action. Chemical genetics: selection and use of small bioactive molecules to investigate gene functions. Combinatorial chemistry: rapid synthesis of a large number of structurally related molecules from a set of building blocks. Mode-of-action (MOA): mechanism by which a small molecule acts or controls the organism. Rule-of-five: empirically determined rules to predict bioavailability of small molecules. Natural product: small molecules, mostly secondary metabolites, isolated from organisms, for example from bacteria, fungi, marine animals, or plants. Structure–activity relationship (SAR): analysis of compound derivatives to identify which moiety contributes to bioactivity.

1360-1385/$ – see front matter ß 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.tplants.2009.11.005 Available online 24 December 2009



Trends in Plant Science Vol.15 No.2

Table 1. Properties of bioactive small molecules and chemical libraries (a) Chemical genetic screens in plant science a Year Screen 10.000 combinatorial 2004 10.000 combinatorial 2005 20.000 combinatorial 2007 120 bioactive 4000 LATCA 10.000 combinatorial 2008


10.000 combinatorial 3000 natural product 10.000 combinatorial 3000 LATCA 3300 bioactive 10.000 combinatorial 10.000 combinatorial

Hit Compounds A–D Gravacin Morlin Triclosan Cobtorin Hypostatin Pyrabactin Brassinopride Endosidin1 Proauxin Smex Methotrexate; 2,4-DNP WH7 Bikinin

(b) Molecular properties to predict bioavailability of small molecules Molecular property Molecular weight (Da) logP (octanol/water partition coefficient) Number of H-bond donor Number of H-bond acceptor Rotatable bonds

Mode-of-action Triggers auxin signalling Inhibits gravitropism Inhibits cellulose synthesis Blocks early immune response Disturbs microfibril–cellulose alignment Inhibits hypocotyl growth Inhibits seed germination Inhibits BR, triggers ethylene responses Blocks vesicle trafficking Triggers auxin signalling Enhances bacterial resistance Inhibits DNA synthesis; Blocks ATP synthesis Triggers auxin signalling Triggers BR signalling

Lipinski’s rule-of-fivec (Human drugs) [22] 500 5 5 10 ND b

Refs [46] [17] [36] [35] [42] [10,12] [19] [39] [27] [41] [34] [28] [7]

Tice’s rule-of-five (Herbicides) [23] 150–500 5 3 12 12

(c) Properties of commercially available small molecule libraries Bioactive library high Hit rate various Structural complexity various Chemical diversity very low Chance of chemical reactivity +++ SAR studies

Combinatorial (RO5) library low low low high +++

Natural product library medium high high very low +

Diverse oriented synthesis library low–medium high high low ++

Target identification. . . . . .by genetic insensitivity . . .by biochemistry . . .by candidate approach

+++ +++ +

+++ + +

+++ ++ +

+++ +++ +++

Abbreviations: BR, brassinosteroid; LATCA, Library of AcTive Compounds in Arabidopsis. a This table does not contain screens on other organisms and mode-of-action studies with bioactive compounds. b Lipinski rules do not include the analysis of the number of rotatable bonds. c Rule-of-five are not five rules, but every rule contains the number five (i.e. 5).

kinase that mediates phosphorylation of transcription factors BES1 (bri1-EMS-Suppressor 1) and BZR1 (Brassinazole Resistant 1). Although BIN2 was already identified by classical forward genetics [8], residual BES1 phosphorylation remained in the bin2 mutant, and even in a triplemutant impairing three BIN2 homologs [9]. It has been demonstrated that bikinin inactivates BIN2 and a subset of other GSK3-kinases by binding to their ATP pocket. The inhibition of multiple GSK3s by bikinin prevents BES1 phosphorylation and thereby triggers the complete BR response [7]. That bikinin is a general agonist that targets multiple GSK3 kinases is essential since a selective GSK3 inhibitor would not have caused a phenotype. Pyrabactin is an elegant example of a specific agonist that revealed the identity of the long-sought abscisic acid (ABA) receptor [10]. The ABA receptor has been elusive since the 1960s and many previously reported ABA receptors remained controversial [11]. Pyrabactin was identified from a chemical genetics screen for germination inhibitors [12]. Pyrabactin-insensitive mutants revealed insensitive alleles of PYR1 (Pyrabactin Resistance 1), which belongs to 82

a novel 14-member gene family. In the presence of pyrabactin, PYR1 interacts with protein phosphatase HAB1 (homolog of ABI1/ABI2), a key negative regulator of ABA responses. In contrast to pyrabactin, which is selective for PYR1 in seeds, ABA interacts with all PYR1-like homologs, which causes redundancy in the ABA signalling pathway [10]. Thus, by using a chemical that selectively activates part of the ABA signalling cascade, the signalling components could be identified by classical forward genetics. A non-selective agonist that would activate all receptors would not have lead to the identification of the ABA receptor. Selectivity to particular members of redundant protein families has also been useful for the target identification of two herbicides. Isoxaben was shown to be a cellulose biosynthesis inhibitor that targets two of the six cellulose synthases, which act in a hexameric complex. The selectivity of isoxaben was essential to identify cellulose synthase mutants from isoxaben insensitivity screens [13,14]. Similarly, DAS534 is an experimental herbicide from Dow Agro Sciences (www.dowagro.com) that reduces


Trends in Plant Science

Vol.15 No.2

Figure 1. Chemical genetics overcomes genetic redundancy. Signalling networks containing structurally related redundant components (a) are difficult to address with classical forward genetic screens (b). General agonists that inhibit multiple components can block redundant networks (c), whereas a specific agonist that activates only one of the components in the absence of the signal can trigger a response from an otherwise redundant network (d).

growth by triggering auxin responses. A screen for DAS534-insensitive mutants revealed AFB5 [15], which belongs to a five-member subclade of F-box proteins that includes auxin receptor TIR1. In both cases, the selectivity of these bioactive small molecules to members of an otherwise redundant gene family facilitates studies of individual family members. Principle II: structure–activity relationship studies generates useful derivatives Bioactive compounds identified from a primary screen are not necessarily the best choice for the subsequent steps of a chemical genetics project. It is often desirable to increase the potency and reduce structural complexity of the molecule; to select structurally-related inactive molecules and antagonists; and to determine sites of the molecule that can be modified. Structure–activity relationship (SAR) studies describe the correlation between biological effects of compound derivatives and their structure. The derivatives can either directly be found in the screened library or obtained from other sources. As an example, an extensive SAR study lead to the identification of active moieties and the design of biotin-tagged derivatives of terfestatin A, an auxin antagonist, that does not bind to the auxin receptor TIR1 [16]. In some cases, SAR has been used to uncouple different phenotypes. For example, derivatives of gravacin, a gravitropism inhibitor that also causes altered localization of a vacuole membrane marker [17], resulted in molecules that had no gravitropic effects, but retained the effect on the tonoplast [18]. Similarly, brassinopride affects both BR and ethylene signalling, but some derivatives only altered ethylene signalling [19]. These observations are frequently interpreted as being caused by selectivity of derivatives for different pathways. However, these cases do not exclude that the derivatives act on the same pathway because different assays may have a different threshold for compound sensitivity, even though they belong to the same pathway. Furthermore, SAR studies resulted in the discovery of antagonists and ‘dead’ analogs. For example, larger extension of an alkyl chain on indole-3-acetic acid (IAA, the natural auxin), resulted in a series of derivatives that

prevent interactions of TIR1 with its targets, the AUX/ IAA proteins, thereby causing opposite effects of IAA (antagonists) [20]. SAR also uncovers inactive derivatives that do not act as antagonists but simply do not bind to the target protein. These ‘dead’ analogs can be essential controls in phenotyping experiments. Apyrabactin, for example, was used as a control for pyrabactin experiments [10]. Finally, SAR is important to remove false positive compounds. Small molecules containing for example aldehydes, ketones, epoxides, and alkyl halides, can be chemically reactive towards proteins [21]. Compounds containing these groups can be non-selective, and cause pleiotropic phenotypes. SAR detects the role of these reactive groups when the phenotype is only related to the presence of the reactive group. Principle III: bioactive compounds in plants comply with the ‘rule-of-five’ Bioactive compounds have to cross several barriers to reach their targets. Their availability in a biological system (bioavailability) depends on solubility, membrane permeability, and active uptake and transport within the organism. By comparing a large set of bioactive compounds, Lipinksi and colleagues defined these properties for orally taken human drugs and summarized them in the ‘rule-of-five’ (RO5) (Table 1b) [22]. Drug uptake is hampered if more than one of these parameters is out of range. However, these rules do not apply for natural products and drugs that enter through active uptake mechanisms. Bioavailability of compounds in plants has been barely studied in detail, despite being an important parameter in chemical genetics. The molecular properties have been analyzed for a set of herbicides [23,24], resulting in the ‘Tice’s rule-of-five’, which slightly deviates from Lipinski’s rules (Table 1b). Tice and colleagues analyzed the molecular properties of both post- and pre-emergence herbicides. Post-emergence herbicides are applied on the leaf surface and therefore have to cross the leaf cuticle before they are transported through the phloem. Pre-emergence herbicides are taken up by the root system and are transported through the xylem. The molecular properties of both herbicide groups are very similar, although a weak acidic 83


Trends in Plant Science Vol.15 No.2

Figure 2. Small molecules and their fate. (a)–(c) Conversion into bioactive molecules. (a) Conversion of sirtinol into the auxin-mimicking 2-hydroxy-1-naphthoic acid (HNC) involves several unidentified hydrolytic events and the oxidation of an aldehyde, which is catalyzed by an aldehyde oxidase and its cofactor MoCo. (b) Hydrolases are thought to release auxin-mimicking molecules from various chemicals that contain ester and amide bonds. (c) Hypostatin is glycoactivated by UDP glucosyltransferase HYR1, resulting in a toxic glucose–hypostatin conjugate. (d), (e) Complexity of molecules identified from natural product and combinatorial libraries. (d) Bikinin, pyrabactin, gravacin, cobtorin, morlin and brassinopride are simple molecules selected from in combinatorial libraries. (e) Yokolonide B and endosidin1/prieurianin are complex molecules identified from natural product libraries.



Trends in Plant Science

Vol.15 No.2

property is common to herbicides transported through the phloem [23]. Importantly, nearly all bioactive compounds identified from chemical genetic screens (Table 1a) comply with the Tice’s rule-of-five (Supplementary Table S1), indicating that this rule can be used to predict bioavailability of small molecules in plants.

cluding Landsberg [12]. The gene responsible for hypostatin-sensitivity encodes an UDP-glycosyltransferase. This enzyme glucosylates hypostatin, generating a glc–hypostatin conjugate (Figure 2c). Glc–hypostatin is the bioactive form of the compound since its application inhibits growth also in accessions that are insensitive for hypostatin.

Principle IV: conversion into bioactive molecules is common in plants The fate of a small molecule in plants is not always predictable. Plants have the ability to selectively import and transport molecules to different organs or compartments, and to alter their structure and bioactivity through various modifications. Many small molecules are activated in biological systems (e.g. by hydrolysis or conjugation), a process known as bioactivation. In the case of sirtinol, a rather unexpected fate was revealed by SAR and a screen for insensitive mutants. Sirtinol activates auxin signalling but sirtinol-insensitive mutants concerned mostly genes required for the synthesis of the molybdenum cofactor MoCo [25,26]. Careful analysis of the activity of sirtinol derivatives revealed that sirtinol is catabolised into an auxin-mimicking molecule, and this conversion requires an alcohol dehydrogenase and its cofactor, MoCo [25] (Figure 2a). Conversion of small molecules to release an auxin is frequently observed. For example, amidases are suggested to cleave the amide bond in compounds 602 and WH7 [27,28], and esterases are thought to cleave the ester bond in compound A [29] (Figure 2b). These hydrolytic cleavage events generate small molecules that act as an auxin. Interestingly, the ‘pro-auxins’ can be used as a vehicle to transport the hormone into the cell since most auxins are not membrane permeable [27]. Bioactivation by hydrolysis is also common for herbicides, e.g. by carboxylesterase CXE12 [30]. Hypostatin undergoes a different kind of modification in plant cells. Hypostatin inhibits cell expansion in seedlings, but only in a few Arabidopsis thaliana accessions, in-

Principle V: transcriptome analysis defines targeted pathways Microarray analysis is a powerful tool to define which cellular pathways are affected by a bioactive compound. For example, transcriptional changes induced upon treatment with bikinin overlapped for 88% with those induced by BR treatment, placing the bikinin targets in the BR signalling cascade [7]. Furthermore, BR-biosynthesis inhibitor brassinazole causes transcriptome alterations that are almost perfectly the opposite of those induced by BR treatment, indicating that brassinazole specifically blocks the BR pathway [31]. Finally, a significant correlation in transcriptome profiles was found between seeds treated with pyrabactin and abscisic acid (ABA), indicating that pyrabactin targets components in the ABA signalling cascade [10]. Microarray analysis has also been useful to investigate possible off-target effects. Imazapyr, for example, is a herbicide targeting acetohydroxyacid synthase (AHAS), which catalyzes the first step in the biosynthesis of branched-chain amino acids. Point mutations in the AHAS-encoding gene, CSR1 (Chlorsulfuron Resistance 1), lead to herbicide resistance. Microarray analysis revealed that imazapyr treatment causes significant transcriptome changes which are absent in the csr1 mutant, demonstrating that CSR1 is the sole target of imazapyr [32]. Principle VI: library choice determines the future avenues The choice of the compound library is a crucial decision. This issue is generally underestimated and needs much

Box 1. Strategies of target identification There are many strategies for identification of small molecule targets. Below are the major strategies that are frequently used in chemical genetics [43,47]. Compound resistance screen Mutagenized organisms are exposed to the compound and resistant mutants are selected. The compound does not need to be modified. This method is successfully used but will not work if the target acts redundantly or causes lethality when deleted. Resistance screens will also identify non-targets such as downstream signalling steps and components that are required to transport or activate the compound.

transcription factor. The activation domain of the transcription factor is translationally fused to a plant cDNA library. An interaction of the compound with a plant protein (target) creates a three-hybrid complex that activates transcription of a reporter gene. This method can detect low-abundant targets but cannot identify protein complexes or post-translationally modified isoforms [48]. Phage display Proteins are presented on a phage surface and interacting phages are purified on the immobilized compound. This method can detect lowabundant targets but cannot detect protein complexes.

Affinity chromatography The compound is tagged and immobilized and used to purify interacting proteins. Inactive compounds are often used to remove false positives. This method can identify abundant target proteins and complexes with high binding affinity. Affinity chromatography is the most frequently used method in medical science for target identification.

Protein microarray A protein chip is screened with a fluorescent- or isotope-labelled small molecule. This method is sensitive and can identify low abundant proteins, but does not display protein complexes and post-translational modifications. A yeast protein microarray was successfully used [49] and generation of Arabidopsis protein microarrays has been initiated (http://plants.gersteinlab.org/).

Yeast-three-hybrid (Y3H) The compound is tagged with dexamethasone (DEX) or methotrexate (MTX) and applied to yeast cells, where it binds the DEX/MTX binding protein which is fused to the DNA-binding domain of a

Candidate approach Careful characterization of the phenotype and the compound structure points to a small, testable number of target candidates. Using only this approach can not exclude the existence of further targets.


Review consideration, because it has a large impact on the development of the project and the way of target identification [33]. The different types of small molecule libraries have different advantages and limitations (Table 1c). Bioactive libraries contain compounds with characterized biological functions. These libraries are specific (e.g. inhibitors of ion channels, lipid biosynthesis or kinases) or cover a broad-range of bioactivities. Since these libraries are relatively small, they are easy to screen, and perfect for setting up chemical genetic projects. Although these libraries usually do not reveal molecules with a novel mode-of-action, they have a high hit-rate and targets can be predicted. For example, a screen of a bioactive library for seed germination inhibitors identified inhibitors of translation, ATP biosynthesis and DNA synthesis [34]. In another screen, the fatty-acid synthetase type II (FASII) inhibitor triclosan was identified as a suppressor of early immune responses, suggesting a role for lipid biosynthesis in immunity [35]. Combinatorial libraries (also called RO5 libraries) consist of low-molecular weight compounds synthesized by combinatorial chemistry, tailored to increase the drug potential according to the Lipinski rules (Table 1b). These libraries have a relatively low hit-rate but are large in size (at least 10 000 compounds) and require high-throughput screening facilities. The advantage of these libraries is that they usually contain informative derivatives for SAR analysis and that fluorescent or biotinylated derivatives can be easily synthesized for target identification (Box 1). However, RO5 libraries tend to be biased towards watersoluble, heterocyclic compounds, which limit the possible modes-of-action. Furthermore, these compounds tend to become modified when presented to biological systems. Examples of bioactive compounds identified from combinatorial libraries are gravacin [18], morlin [36], hypostatin [12], bikinin [7] and pyrabactin [10,12] (Figure 2d). Natural products (NPs) isolated from organisms represent about 80% of the currently used drugs in pharmacy and biological studies [37]. This number illustrates the potential of NPs in chemical genetics. NP libraries represent large chemical diversity and complexity. They have a higher hit-rate than the RO5 libraries and higher probability of targeting novel protein classes. An important disadvantage, however, is that organic chemical synthesis of NP compounds is often unprecedented. This limits studies on SAR and almost excludes strategies of target identification that require chemical modification of the compound (Box 1). Yokolonide B [38] and endosidin1/ prieurianin [39], for example, are a challenge for synthetic chemists as they contain multiple-ring systems and functional groups and several chiral centres (Figure 2e). However, exactly these properties of the NPs increase their potential of being biologically active. Diversity Oriented Synthesis (DOS) libraries have not been used in plant science yet, but have a high potential because they combine advantages of the NP and RO5 libraries. DOS libraries consist of synthetic compounds inspired by scaffolds present in natural products [40]. These small molecules possess chemical properties distinct from other libraries and have potential to bind to yet untargeted protein classes [33]. Furthermore, DOS 86

Trends in Plant Science Vol.15 No.2

libraries could have higher hit-rates than combinatorial libraries and contain derivatives for SAR studies. The LATCA library represents a special case. This ‘Library of AcTive Compounds in Arabidopsis’ (LATCA) was composed by Sean Cutler and colleagues (cutlerlab.blogspot.com/2008/05/latca.html) with the aim to provide a plant-specialized broad-range bioactive library to the plant research community. The LACTA library consists of known herbicides, plant hormones and common inhibitors, supplemented with 1600 compounds affecting yeast activities. Its true value lies in 2000 additional compounds that were selected from various libraries because they affect Arabidopsis hypocotyl growth. The LATCA library is ideal to validate a screening assay and determine if known pathways are involved. The LATCA library has an increased hit-rate, but it is limited by the way most compounds were pre-selected. Part of the LATCA library has been screened recently and the authors identified Smex and other compounds that reduce Pseudomonas syringae disease on Arabidopsis seedlings grown in liquid medium [41]. The LATCA library was also included in a screen for compounds that cause spherical swelling in tobacco (Nicotiana tabacum) cells [42], resulting in cobtorin, a novel inhibitor of the parallel alignment of microtubules with cellulose microfibrils. Principle VII: two proven routes of target identification Target identification is an integral part of a chemical genetics project as it is essential to understand the mode-of-action of the bioactive compound. Yet, target identification is also the bottleneck for most chemical genetic projects. There are several strategies for target identification [43] (Box 1). Affinity chromatography, yeastthree-hybrid (Y3H) and protein arrays require a fluorescent or biotinylated derivative of the compound synthesized by organic chemistry. Although these methods are frequently and successfully used in medical science, plant scientists have identified compound targets so far by only two other methods: (i) testing candidates and (ii) screening for compound-insensitive mutants. The candidate approach has been successful for target identification in several cases. The target of triclosan in Arabidopsis, for example, was predicted based on its known target in animals, the fatty-acid synthetase type II (FASII) [35]. The target of brassinazole was predicted because the triazole scaffold of brassinazole is known to inhibit cytochrome P450s. Indeed, brassinazole inhibits DWARF4 (DWF4), a cytochrome P450 required for BR synthesis [44]. The target of bikinin was indicated by the observation that bikinin affected all signalling events downstream of the BIN2 kinase [7]. Thus, the candidate approach is an effective way of target identification and requires a detailed characterization of the phenotype, and deep knowledge of the signalling pathway and possible mode-of-action. The second way of successful target identification comes from compound insensitivity screens. For example, pyrabactin targets PYR1, for which 14 mutant alleles were identified from a pyrabactin-insensitive screen [10]. In case of gravacin, the genetic screen for mutants with a normal gravitropic response in the presence of gravicin

Review identified P-glycoprotein 19 (PGP19) [18]. PGP19 is a transmembrane protein that transports auxin and thereby contributes to gravitropism. Binding studies for gravacin with membranes containing PGP19 indicate that gravacin binds to PGP19 and inhibits auxin transport, implying that PGP19 is a target of gravacin. However, gravacin probably has additional target(s) since pgp19 mutants exhibit only part of the gravacin-induced phenotype [18]. However, screens for compound insensitivity do not always reveal the target. Sirtinol-insensitive mutants, for example, included ten genes involved in downstream auxin signalling and four genes required for converting sirtinol into the bioactive auxin and only one mutant in the sirtinol target, the auxin receptor TIR1 [25]. Furthermore, screens for hypostatin insensitivity led to the identification of the glycoactivating enzyme, but not to the target of glc– hypostatin [12]. Nevertheless, these studies have been informative on the fate of the small molecules and the signalling cascades. Future of the plant chemical genetics Plant chemical genetics is still in its early days, but the first results illustrate that this field can make strong contributions to basic knowledge of plant science. However, the plant chemical genetic projects revealed a series of challenges that will have to be addressed in the future. (i) Expansion of chemical genetics beyond hormone signalling. Most of the chemical genetic screens are related to growth hormones, like auxin, BR and ABA. The bias for these pathways is explained by the simplicity of the screen: scoring for growth phenotypes. The next challenge is to screen for compounds that target other pathways, using sensitive reporter assays to study processes such as vesicle trafficking [18,39]. (ii) Identification of the targets. So far, targets were identified from screens for compound insensitive mutants and through a candidate approach, but each method has its limitations. Target identification by biochemical or proteomics approaches requires tagged compounds, SAR knowledge and synthetic chemistry (Box 1), therefore these approaches need new investments and collaborations with chemists. (iii) Collaborative networks and shared databases. Chemical genetics is interdisciplinary research that requires expensive resources such as facilities for highthroughput screening, chemistry, phenotypic characterization and target identification. The open access chemical databases that includes phenotypic screening results like ChemMine [45] (http://bioweb.ucr.edu/ChemMineV2) and use of the LATCA library could build collaborative networks of experts, essential to increase the success of chemical genetic projects in plant science. Acknowledgements We would like to thank Franziska Turck, Erich Kombrink, Brande Wulff and the anonymous reviewers for useful suggestions.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.tplants. 2009.11.005.

Trends in Plant Science

Vol.15 No.2

References 1 Zheng, X.S. et al. (2004) Genetic and genomic approaches to identify and study the targets of bioactive small molecules. Chem. Biol. 11, 609– 618 2 Blackwell, H.E. and Zhao, Y. (2003) Chemical genetic approaches to plant biology. Plant Physiol. 133, 448–455 3 Hicks, G.R. and Raikhel, N.V. (2009) Opportunities and challenges in plant chemical biology. Nat. Chem. Biol. 5, 268–272 4 Kaschani, F. and van der Hoorn, R. (2007) Small molecule approaches in plants. Curr. Opin. Chem. Biol. 11, 88–98 5 Raikhel, N. and Pirrung, M. (2005) Adding precision tools to the plant biologists’ toolbox with chemical genomics. Plant Physiol. 138, 563–564 6 Robert, S.p. and Raikhel Natasha, V. et al. (2009) Powerful Partners: Arabidopsis and Chemical Genomics. The Arabidopsis Book, The American Society of Plant Biologists, pp. 1–16 7 De Rybel, B. et al. (2009) Chemical inhibition of a subset of Arabidopsis thaliana GSK3-like kinases activates brassinosteroid signaling. Chem. Biol. 16, 594–604 8 Li, J. et al. (2001) BIN2, a new brassinosteroid-insensitive locus in Arabidopsis. Plant Physiol. 127, 14–22 9 Vert, G. and Chory, J. (2006) Downstream nuclear events in brassinosteroid signalling. Nature 441, 96–100 10 Park, S.Y. et al. (2009) Abscisic acid inhibits type 2C protein phosphatases via the PYR/PYL family of START proteins. Science 324, 1068–1071 11 Pennisi, E. (2009) Plant biology. Stressed out over a stress hormone. Science 324, 1012–1013 12 Zhao, Y. et al. (2007) Chemical genetic interrogation of natural variation uncovers a molecule that is glycoactivated. Nat. Chem. Biol. 3, 716–721 13 Desprez, T. et al. (2002) Resistance against herbicide isoxaben and cellulose deficiency caused by distinct mutations in same cellulose synthase isoform CESA6. Plant Physiol. 128, 482–490 14 Scheible, W.R. et al. (2001) Modifications of cellulose synthase confer resistance to isoxaben and thiazolidinone herbicides in Arabidopsis Ixr1 mutants. Proc. Natl. Acad. Sci. U. S. A. 98, 10079–10084 15 Walsh, T.A. et al. (2006) Mutations in an auxin receptor homolog AFB5 and in SGT1b confer resistance to synthetic picolinate auxins and not to 2,4-dichlorophenoxyacetic acid or indole-3-acetic acid in Arabidopsis. Plant Physiol. 142, 542–552 16 Hayashi, K. et al. (2008) Active core structure of terfestatin A, a new specific inhibitor of auxin signaling. Bioorg. Med. Chem. 16, 5331–5344 17 Surpin, M. et al. (2005) The power of chemical genomics to study the link between endomembrane system components and the gravitropic response. Proc. Natl. Acad. Sci. U. S. A. 102, 4902–4907 18 Rojas-Pierce, M. et al. (2007) Arabidopsis P-glycoprotein19 participates in the inhibition of gravitropism by gravacin. Chem. Biol. 14, 1366– 1376 19 Gendron, J.M. et al. (2008) Chemical Genetic Dissection of Brassinosteroid-Ethylene Interaction. Mol. Plant 1, 368–379 20 Hayashi, K. et al. (2008) Small-molecule agonists and antagonists of Fbox protein-substrate interactions in auxin perception and signaling. Proc. Natl. Acad. Sci. U. S. A. 105, 5632–6563 21 Rishton, G.M. (1997) Reactive compounds and in vitro false positives in HTS. Drug Discov. Today 2, 382–384 22 Lipinski, C.A. et al. (2001) Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv. Drug Deliver. Rev. 46, 3–26 23 Tice, C.M. (2001) Selecting the right compounds for screening: does Lipinski’s Rule of 5 for pharmaceuticals apply to agrochemicals? Pest Manag. Sci. 57, 3–16 24 Tice, C.M. (2002) Selecting the right compounds for screening: use of surface-area parameters. Pest Manag. Sci. 58, 219–233 25 Dai, X. et al. (2005) Genetic and chemical analyses of the action mechanisms of sirtinol in Arabidopsis. Proc. Natl. Acad. Sci. U. S. A. 102, 3129–3134 26 Zhao, Y. et al. (2003) SIR1, an upstream component in auxin signaling identified by chemical genetics. Science 301, 1107–1110 27 Savaldi-Goldstein, S. et al. (2008) New auxin analogs with growthpromoting effects in intact plants reveal a chemical strategy to improve hormone delivery. Proc. Natl. Acad. Sci. U. S. A. 105, 15190–15195 28 Christian, M. et al. (2008) Identification of auxins by a chemical genomics approach. J. Exp. Bot. 59, 2757–2767



Trends in Plant Science Vol.15 No.2

29 Sungur, C. et al. (2007) The small molecule 2-furylacrylic acid inhibits auxin-mediated responses in Arabidopsis thaliana. Plant Cell Physiol. 48, 1693–1701 30 Gershater, M.C. et al. (2007) Role of a carboxylesterase in herbicide bioactivation in Arabidopsis thaliana. J. Biol. Chem. 282, 21460–21466 31 Goda, H. et al. (2002) Microarray analysis of brassinosteroid-regulated genes in Arabidopsis. Plant Physiol. 130, 1319–1334 32 Manabe, Y. et al. (2007) CSR1, the sole target of imidazolinone herbicide in Arabidopsis thaliana. Plant Cell Physiol. 48, 1340–1358 33 Shelat, A.A. and Guy, R.K. (2007) The interdependence between screening methods and screening libraries. Curr. Opin. Chem. Biol. 11, 244–251 34 Bassel, G.W. et al. (2008) Elucidating the germination transcriptional program using small molecules. Plant Physiol. 147, 143–155 35 Serrano, M. et al. (2007) Chemical interference of pathogen-associated molecular pattern-triggered immune responses in Arabidopsis reveals a potential role for fatty-acid synthase type II complex-derived lipid signals. J. Biol. Chem. 282, 6803–6811 36 DeBolt, S. et al. (2007) Morlin, an inhibitor of cortical microtubule dynamics and cellulose synthase movement. Proc. Natl. Acad. Sci. U. S. A. 104, 5854–5859 37 Li, J.W. and Vederas, J.C. (2009) Drug discovery and natural products: end of an era or an endless frontier? Science 325, 161–165 38 Hayashi, K. et al. (2003) Yokonolide B, a novel inhibitor of auxin action, blocks degradation of AUX/IAA factors. J. Biol. Chem. 278, 23797– 23806 39 Robert, S. et al. (2008) Endosidin1 defines a compartment involved in endocytosis of the brassinosteroid receptor BRI1 and the auxin

40 41 42

43 44

45 46

47 48


transporters PIN2 and AUX1. Proc. Natl. Acad. Sci. U. S. A. 105, 8464–8469 Tan, D.S. (2005) Diversity-oriented synthesis: exploring the intersections between chemistry and biology. Nat. Chem. Biol. 1, 74–84 Schreiber, K. et al. (2008) A high-throughput chemical screen for resistance to Pseudomonas syringae in Arabidopsis. Plant J. 54, 522–531 Yoneda, A. et al. (2007) Chemical genetic screening identifies a novel inhibitor of parallel alignment of cortical microtubules and cellulose microfibrils. Plant Cell Physiol. 48, 1393–1403 Terstappen, G.C. et al. (2007) Target deconvolution strategies in drug discovery. Nat. Rev. Drug Discov. 6, 891–903 Asami, T. et al. (2001) Selective interaction of triazole derivatives with DWF4, a cytochrome P450 monooxygenase of the brassinosteroid biosynthetic pathway, correlates with brassinosteroid deficiency in planta. J. Biol. Chem. 276, 25687–25691 Girke, T. et al. (2005) ChemMine. A compound mining database for chemical genomics. Plant Physiol 138, 573–577 Armstrong, J.I. et al. (2004) Identification of inhibitors of auxin transcriptional activation by means of chemical genetics in Arabidopsis. Proc. Natl. Acad. Sci. U. S. A. 101, 14978–14983 Sleno, L. and Emili, A. (2008) Proteomic methods for drug target discovery. Curr. Opin. Chem. Biol. 12, 46–54 Kley, N. (2004) Chemical dimerizers and three-hybrid systems: scanning the proteome for targets of organic small molecules. Chem. Biol. 11, 599–608 Huang, J. et al. (2004) Finding new components of the target of rapamycin (TOR) signaling network through chemical genetics and proteome chips. Proc. Natl. Acad. Sci. U. S. A. 101, 16594–16599

Plant Science Conferences in 2010 The 5th EPSO Conference: Plants for Life 18–22 April, 2010 Olos Polar Center (Lapland), Finland http://www.epsoweb.org/catalog/conf2010.htm 21st International Conference on Arabidopsis Research 6–10 June 2010 Yokohama, Japan http://arabidopsis2010.psc.riken.jp/ BANFF Conference on Plant Metabolism 24–28 June, 2010 Banff, Canada http://www.ucalgary.ca/plantmetabolism/ XVII Congress of the Federation of European Societies of Plant Biology (FESPB) 4–9 July, 2010 Valencia, Spain http://www.geyseco.es/fespb/principal.php?seccion=welcome