A unique virulence factor for proliferationand dwarfism in plants identified froma phytopathogenic bacteriumAyaka Hoshia,1, Kenro Oshimaa,1, Shigeyuki Kakizawaa,1, Yoshiko Ishiia, Johji Ozekia, Masayoshi Hashimotoa,Ken Komatsua, Satoshi Kagiwadab, Yasuyuki Yamajia, and Shigetou Nambaa,2

aDepartment of Agricultural and Environmental Biology, Graduate School of Agricultural and Life Sciences, The University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku,Tokyo 113-8657, Japan; and bDepartment of Clinical Plant Science, Faculty of Bioscience and Applied Chemistry, Hosei University, 3-7-2 Kajinocho, Koganei,Tokyo 184-8584, Japan

Edited by James L. Van Etten, University of Nebraska, Lincoln, NE, and approved February 24, 2009 (received for review December 20, 2008)

One of the most important themes in agricultural science is theidentification of virulence factors involved in plant disease. Here,we show that a single virulence factor, tengu-su inducer (TENGU),induces witches’ broom and dwarfism and is a small secretedprotein of the plant-pathogenic bacterium, phytoplasma. Whentengu was expressed in Nicotiana benthamiana plants, theseplants showed symptoms of witches’ broom and dwarfism, whichare typical of phytoplasma infection. Transgenic Arabidopsis thali-ana lines expressing tengu exhibited similar symptoms, confirmingthe effects of tengu expression on plants. Although the localiza-tion of phytoplasma was restricted to the phloem, TENGU proteinwas detected in apical buds by immunohistochemical analysis,suggesting that TENGU was transported from the phloem to othercells. Microarray analyses showed that auxin-responsive geneswere significantly down-regulated in the tengu-transgenic plantscompared with GUS-transgenic control plants. These results sug-gest that TENGU inhibits auxin-related pathways, thereby affect-ing plant development.

auxin � disease symptom � morphological change � phytoplasma

P lant pathogens affect both the magnitude and quality ofagricultural production; thus, one of the most important

goals in agricultural science is to understand the inductionmechanisms of plant disease. Plant pathogens can cause diversedisease symptoms in roots, leaves, f lowers, fruits, stems, andtubers. In particular, symptoms of disease involving cell death,such as necrotic spots, vascular necrosis, blight, and cankers,have been studied in detail (1). Several virulence factors knownto induce cell death in plants and the components of their relatedsignaling cascades have been identified (1). However, there aremany other types of disease symptoms in nature. For example,leaf malformation, leaf distortion, leaf rolling, small leaves,stunted growth, dwarfism, witches’ broom (the development ofnumerous shoot branches), shortened internodes, bolting (thegrowth of elongated stalks), f lower virescence (the greening offloral organs), or phyllody (leaf-like petals and sepals), all ofwhich induce abnormal organ formation in plants, can causedevastating agricultural losses.

One group of plant pathogens causes the development ofnumerous small and short branches that look like the nest ofTengu, a mythical, long-nosed Japanese goblin who lives in themountains and flies through the sky (see Fig. S1 A). Therefore,the disease caused by these pathogens is referred to as Tengu-su(Tengu’s nest) disease in Japan. The distinctive symptoms as-sociated with this disease have fascinated and troubled theJapanese people for some time, especially those involved inagricultural production. For example, Paulownia witches’ broom(‘‘Paulownia Tengu-su’’ in Japanese) disease, which is caused bya plant-pathogenic bacterium (phytoplasma), was first describedmore than 140 years ago in Japan (Fig. S1 B–D). The symptomsof Tengu-su include witches’ broom and dwarfism. Infected

plants produce large numbers of auxiliary or axillary shoots, andexhibit dwarfism (small f lowers and leaves with shortenedinternodes), resulting in a broom-like appearance (Figs. S1 B–Dand S2). Because this disease is accompanied by dramaticdevelopmental abnormalities, identification of its virulence fac-tor is important in terms of plant pathology, plant morphology,and plant physiology. However, the virulence factor involved inthe production of these symptoms remains unknown.

Phytoplasmas (class Mollicutes, genus Phytoplasma) are bac-terial plant pathogens that have had devastating effects on yieldsin diverse low- and high-value crops and plants worldwide (2).Coconut palm lethal yellowing disease, Paulownia witches’broom disease, and many other diseases caused by phytoplasmasinduce fatal damage in plants and crops all over the world.Phytoplasmas infect more than 700 plant species and bring aboutdramatic changes in plant development, including witches’broom, dwarfism, proliferation (the growth of shoots from floralorgans) (Fig. S2), phyllody, virescence, sterility of f lowers,bolting, purple tops (the reddening of leaves and stems), gen-eralized yellowing, and phloem necrosis (3, 4). Phytoplasmasreside within the sieve elements of plant phloem and within thecell in insects, and are transmitted from plant to plant by insectvectors. Phytoplasmas are pleiomorphic bacteria that lack cellwalls, and their cell size (0.1–0.8 �m in diameter) and genomesize (0.5–1.3 Mbp) are the smallest among bacteria (3, 4).Although the inability to culture phytoplasmas in vitro hashindered their characterization at the molecular level, the fullgenomic sequences of 4 phytoplasmas were recently determined(5–8). Analyses of these data indicate that phytoplasma has lostmany metabolic genes, suggesting that reductive evolution is aconsequence of being an intracellular parasite. An antigenicmembrane protein (Amp) of Candidatus Phytoplasma asterisOY strain interacts with the microfilaments of its insect hostMacrosteles striifrons, as well as with microfilament complexes incertain leafhopper species known to transmit OY, but not withthose in leafhoppers that are unable to transmit OY, suggestingthat the interaction between Amp and the insect microfilament

Author contributions: K.O., S. Kakizawa, and S.N. designed research; A.H. and Y.I. per-formed research; J.O., M.H., and S. Kagiwada contributed new reagents/analytic tools;A.H., K.O., S. Kakizawa, and Y.I. analyzed data; and A.H., K.O., S. Kakizawa, K.K., Y.Y., andS.N. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper have been deposited in the GenBankdatabase (accession no. AP006628). The microarray data reported in this paper have beendeposited in the Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo(accession no. GSE14564).

1A.H., K.O., and S.K. contributed equally to this work.

2To whom correspondence should be addressed. E-mail: anamba@mail.ecc.u-tokyo.ac.jp.

This article contains supporting information online at www.pnas.org/cgi/content/full/0813038106/DCSupplemental.

6416–6421 � PNAS � April 14, 2009 � vol. 106 � no. 15 www.pnas.org�cgi�doi�10.1073�pnas.0813038106

complexes is involved in insect transmissibility (9). In addition tophytoplasma-insect interactions, some findings regarding phy-toplasma-plant interactions have been reported. In particular, ithas recently been reported that Ca. P. asteris AY-WB strainsecretes a protein that targets plant cell nuclei, which is thus oneof the candidate virulence factors of phytoplasma (10). However,the mechanism underlying its pathogenicity remains unknown.

Many Gram-negative bacterial pathogens that affect plantsand animals use a type-III secretion system (TTSS) to delivervirulence factors into the host cell. These prokaryotic virulencefactors often mimic eukaryotic proteins, allowing the pathogento modulate the biological systems of the host to promotebacterial invasion, multiplication, and dispersal (11). Diverseenzymatic activities are associated with these TTSS virulencefactors, including cysteine protease (12), SUMO protease (13),E3 ubiquitin ligase (14), protein phosphatase (15), and ADP-ribosyltransferase (16) activity. These virulence factors are gen-erally involved in the suppression of plant immunity (1). Thephytoplasma genome lacks genes that encode a TTSS (5).However, because phytoplasmas reside within the plant cell, theycan secrete proteins into plant hosts via the bacterial Sectranslocation system. The proteins secreted from phytoplasmamay function in the host cytoplasm like TTSS virulence factorsbecause phytoplasma is an intracellular parasite (4). In fact, ithas been recently reported that phytoplasma produces a proteinthat targets the nuclei of plant host cells (10).

In this study, we identified a protein secreted from phyto-plasma that induces symptoms of witches’ broom and dwarfismin plants, causing abnormalities in plant morphology. Thesesymptoms strongly resemble those of phytoplasma infection.

ResultsScreening for a Virulence Factor of Phytoplasma Disease. Becausephytoplasmas reside within the cells of their hosts, we assumedthat secreted proteins would be strong candidates for diseasevirulence factors. To investigate this hypothesis, we identifiedmore than 30 putative secreted proteins from the OY phyto-plasma genome (5) and expressed each of them in N. benthami-ana using a plant virus vector-mediated expression system (17).OY phytoplasma-infected plants show distinctive symptoms suchas witches’ broom and dwarfism; therefore, we tested whetherthe expression of any of our predicted phytoplasma secretedproteins would induce these characteristic symptoms.

We found that N. benthamiana plants expressing PAM765, 1of the 30 secreted proteins, developed clear symptoms of phy-toplasma infection, including witches’ broom and dwarfism. Inparticular, the number of shoots and leaves that emanated fromthe apical meristem was dramatically increased while plantheight was reduced (Fig. 1A and B). These phenotypes resembledtypical phytoplasma disease symptoms. In contrast, the pheno-type of N. benthamiana plants expressing the other secretedprotein, PAM486, was similar to the phenotype of the controlplant inoculated with empty viral vector (pCAMV) (Fig. 1 A).These results strongly suggest that PAM765 is the virulencefactor responsible for witches’ broom and dwarfism in plants.Surprisingly, PAM765 encodes a very small protein of 4.5 kDa.The mature protein, after the cleavage of its N-terminal signalpeptide, is only 38 amino acids in length (Fig. S3).

Transgenic A. thaliana-expressing PAM765 Exhibited Witches’ Broomand Dwarfism. To further analyze the effect of PAM765 on plantorgan formation and development, we produced transgenic A.thaliana plants (ecotype Col-0) that constitutively expressedPAM765. Wild-type plants inoculated with OY phytoplasma viasap-feeding of OY-infected insect vectors exhibited severe de-velopmental abnormalities, including symptoms of witches’broom and dwarfism (Fig. 2A, center and right). Interestingly,similar symptoms were observed in the PAM765-transgenic

plants (Fig. 2B, center and right, and Table S1). The severity ofthe symptoms varied across the transgenic lines. Among 87transgenic lines, 6 lines (6.9%) exhibited severe dwarfism (i.e.,short internodes) and produced sterile flowers (Fig. 2B, center);18 lines (20.7%) did not develop dwarfism, but instead exhibitedwitches’ broom (i.e., increased shoot branching) (Fig. 2B, right,Table S1). In contrast, 25 transgenic lines that expressed GUSalone showed no symptoms (0%) and were similar to plantsinoculated with healthy insects (Fig. 2 A, left, and B, left). Thereare significant differences between the number of abnormalplants of PAM765-transgenic lines and that of GUS-transgeniclines (Fisher’s exact probability test, P � 0.01, Table S1).

In terms of their reproductive organs, the transgenic lines withwitches’ broom also had defects in phyllotaxis (leaf arrange-ment) such that 2 or more flowers grew from a single point onthe stem (Fig. 3B–D), and some of these lines produced sterileflowers (Fig. 3E). All of the abnormal phenotypes observed inthe PAM765-transgenic plants were the same as those observedin OY phytoplasma-infected A. thaliana (Fig. 3G–I). In contrast,transgenic plants that expressed the GUS gene alone weresimilar to the plants inoculated with healthy insects in that theyhad normal reproductive organs (Fig. 3A and F). These resultssuggest that PAM765 is the virulence factor that induces phy-toplasma-related witches’ broom, dwarfism and abnormal re-

Fig. 1. Identification of a virulence factor inducing phytoplasma diseasesymptoms. (A) N. benthamiana plants inoculated with A. tumefaciens har-boring empty vector (pCAMV) (Left), pCAMV-PAM765 (tengu) (Center), orpCAMV-PAM486 (Right). (Lower) Stems are highlighted with red lines; leafpetioles are highlighted with black lines. The center plant showed witches’broom disease symptoms (a dramatically increased shoot system). (B) Thenumber of leaves per plant following inoculation with the viral vector. Theerror bars indicate the SD. An asterisk indicates a significant difference (P �0.05). 1, pCAMV; 2, pCAMV-PAM765 (tengu).

Fig. 2. Comparison of 35S::PAM765 (tengu) transgenic plants and phyto-plasma (OY strain)-infected plants. (A) Phenotypes of the OY-infected plants.(Left) uninfected plant (control). (Center and Right) OY-infected plants. Thecenter and right plants show severe dwarfism and witches’ broom symptoms,respectively. (B) Phenotypes of the 35S::PAM765 transgenic A. thaliana lines.(Left) 35S::GUS transgenic line (control). (Center and Right) 35S::PAM765transgenic lines. The center and right transgenic lines show severe dwarfismwith short internodes and witches’ broom symptoms, respectively.

Hoshi et al. PNAS � April 14, 2009 � vol. 106 � no. 15 � 6417

PLA

NT

BIO

LOG

Y

productive organogenesis (i.e., Tengu-su disease). Thus, wedesignated PAM765 as phytoplasma tengu-su inducer (tengu).

TENGU Is Not a Silencing Suppressor. It has been reported that theconstitutive expression of a post-transcriptional gene-silencing(PTGS) suppressor of some plant viruses can induce develop-mental defects in plants (18). To investigate the possibility thatTENGU functions as a PTGS suppressor, we tested whetherTENGU could suppress PTGS of the GFP gene using a transientexpression assay (SI Text). Leaves of non-transgenic N.benthamiana infiltrated with A. tumefaciens carrying GFP(35S::GFP) exhibited strong fluorescence at 2–4 days post-inoculation (dpi); however, the level of f luorescence graduallydecreased after 4 dpi and finally disappeared by 5 dpi because ofthe induction of PTGS (Fig. S4). When both GFP and the PTGSsuppressor P19 of tomato bushy stunt virus (TBSV) wereco-expressed by agroinfiltration, strong fluorescence was ob-served at 7 dpi (Fig. S4), suggesting that the silencing of GFP wassuppressed by P19. On the other hand, because GUS does nothave PTGS suppressor activity, no GFP fluorescence was ob-served at 5 dpi when GFP and GUS were co-expressed. Likewise,when GFP and TENGU were co-expressed, no fluorescence wasobserved at 5 dpi (Fig. S4), suggesting that TENGU does notfunction as a PTGS suppressor.

TENGU Is More Highly Expressed in Plant Hosts than in Insect Hosts.Because phytoplasmas can infect both plant and insect hosts, wecompared the expression of tengu in a phytoplasma-infectedplant to that in a phytoplasma-infected insect by real-time PCRto gauge the importance of the tengu gene in each host. TotalRNA was extracted from OY-infected insects (M. striifrons) andplants (Chrysanthemum coronarium). Subsequently, real-timePCR was performed with the elongation factor Tu gene of OY(tufB) as an internal standard. The level of expression of tengu

in the plant host was approximately 5 times that in the insect host(Fig. 3J).

TENGU Is Transported from Phloem Tissue into Adjacent Tissues.Because TENGU is quite a small protein (�4.5 kDa), wehypothesized that it could be transported from cell to cellthrough symplasm. To examine this hypothesis, we used immu-nohistochemical analysis to investigate the localization ofTENGU protein in tissue. It has been reported that the intra-cellular localization of phytoplasma can be specifically detectedby immunohistochemical analysis using an antibody against Ampprotein, which is a phytoplasmal membrane protein (19). Sim-ilarly, the blue signal of Amp protein was specifically detected inthe phloem of the OY-infected plant by immunohistochemicalanalysis using anti-Amp antibody (Fig. 4 B and E), suggestingthat the localization of phytoplasma was restricted within thephloem. However, in an immunohistochemical analysis with ananti-TENGU antibody, the blue signal of TENGU protein wasobserved not only in phloem tissues but also in parenchyma andmeristem tissues (Fig. 4A). Surprisingly, quite a strong signal wasextensively detected in the tip region of the stem and thebranching region of axillary buds (Fig. 4 D and F). In addition,the TENGU signal was also detected in the apical meristem (Fig.4H). We confirmed that TENGU and Amp proteins were notdetected in healthy plants (Fig. 4C). These results indicate thatTENGU is transported from phloem tissues into parenchyma

A B C

D

E

F G H I

J

Fig. 3. Analysis of branching in 35S::PAM765 (tengu) transgenic plants andphytoplasma-infected plants. (A) A 35S::GUS transgenic plant (control). (B–E)Phenotypes of the 35S::PAM765 transgenic A. thaliana lines. (B–D) The35S::PAM765 transgenic plants exhibited defects in phyllotaxis (2 or moreflowers growing from a single point on the stem). (E) A 35S::PAM765 trans-genic plant with sterile flowers. (F) An uninfected plant (control). (G–I) Phe-notypes of the OY-infected plants. (G and H) The OY-infected plants exhibiteddefects in phyllotaxis, similar to (B–D). (I) An OY-infected plant with sterileflowers, as in (E). (Scale bars, 50 mm.) (J) The transcription of tengu wasexamined by quantitative real-time RT-PCR and the results were normalizedagainst the expression of tufB. The error bars indicate the SD. An asteriskindicates a significant difference (P � 0.05). 1, Phytoplasma-infected plants (C.coronarium). 2, Phytoplasma-infected insects (M. striifrons).

A B C

D F

E G

H

I

J

Fig. 4. Immunohistochemical detection of TENGU and Amp proteins inapical meristem tissue. (A and B) Apical meristem tissue sections from OY-infected plants were reacted with the anti-TENGU antibody (A) or the anti-Amp antibody (B). Bars, 1 mm. (C) An apical meristem tissue section fromhealthy plant was reacted with the anti-TENGU antibody. (Scale bar, 1 mm.) (D,F, and H) Enlarged section of (A). (E, G, and I) Enlarged section of (B). (D andE) Branching region of axillary buds. (Scale bar, 200 �m.) (F and G) Tip regionof stem. Bar, 200 �m. (H and I) Apical meristem. (Scale bar, 100 �m.) ph;phloem, pa; parenchyma. (J) Analysis of the subcellular localization of TENGU.Chimeric constructs (35S::GFP [left], 35S::GFP-tengu [center], and35S::tengu-GFP [right]) were transiently expressed in onion epidermal cells.

6418 � www.pnas.org�cgi�doi�10.1073�pnas.0813038106 Hoshi et al.

and meristem tissues, especially into the tip region of the stemand axillary buds.

To investigate the intracellular localization of TENGU, wetransiently expressed a TENGU-GFP fusion protein in onionepidermal cells (Fig. 4J). The fusion protein was localized in thecytoplasm regardless of whether GFP was fused to the N- orC-terminus of TENGU (Fig. 4J, center and right), suggestingthat TENGU would function in the cytoplasm of a plant cell. Thecytoplasmic localization of TENGU was also supported byprotein localization prediction using PSORT (20). In addition,subcellular localization signal sequences, such as a nuclearlocalization signal, were not identified in the secreted region ofTENGU by InterProScan (21).

TENGU Affects Plant Auxin Responses. To investigate the influenceof TENGU expression on the transcription profile of a hostplant, we identified differences in the gene expression profilesbetween tengu-transgenic and GUS-transgenic Arabidopsisplants using microarray analysis. A total of 373 genes wassignificantly up-regulated, with ratios of more than 2.0 (P �0.05), and 575 genes were significantly down-regulated, withratios of less than 0.5 (P � 0.05) in the tengu-transgenic plantscompared with the GUS-transgenic plants. Among these genes,the number of auxin-related genes was significantly high accord-ing to classification analysis based on gene ontology (24 genes,P � 0.05), most of which were down-regulated by the expressionof TENGU (Table 1). Auxin-related genes that were down-regulated in microarray analysis contained early auxin-responsive genes, that is, AUX/IAA family genes (IAA29, IAA7/AUX2), small auxin-induced RNA (SAUR) family genes(SAUR�AC1 and other 14 genes), and GH3 family genes (GH3.5/WES1), all of which are known to be induced after exposure toauxins (22). Additionally, the expression level of dormancy-associated protein 1 (DRM1) was approximately 5.7-fold lower inthe tengu-transgenic plants than in the GUS-transgenic plants. Ithas been reported that the DRM1 gene is up-regulated innon-growing axillary buds (dormant state) but down-regulatedin growing axillary buds (23). Moreover, the expression levels of2 pin-formed (PIN) genes (PIN7 and At5g01990) were approxi-

mately 2–3-fold lower in tengu-transgenic plants than in GUS-transgenic plants. Auxin is mainly synthesized in young mer-istems and leaves, and is transported to the root. PIN genesencoding components of the auxin eff lux machinery areinvolved in this polar auxin transport (24). Taken together,these results suggest that TENGU disrupts auxin signaling orbiosynthesis pathways.

DiscussionIdentification of an Inducer of Witches’ Broom and Dwarfism. Severalphytopathogenic fungi and bacteria are known to cause witches’broom and/or dwarfism (25, 26), for example, the bacteriumRhodococcus fascians, which affects carnation plants (25), andthe fungus Taphrina wiesneri, which affects cherry trees (26). Ithas been demonstrated that these phytopathogens have biosyn-thetic genes for auxin or cytokinin, and produce these phyto-hormones to cause disease symptoms (25–27). Here, we showedthat a single protein from a plant-pathogenic bacterium induceswitches’ broom and dwarfism in plants (Figs. 1 A and 2B).Although some bacteria are known to induce histological symp-tom such as canker or gall (1), this is the first report of a bacterialvirulence factor that causes morphological abnormalities at thewhole plant level.

TENGU Induces Symptoms by Changing Auxin Responses. Microarrayanalysis indicated that the expression levels of many auxin-related genes were down-regulated in the tengu-transgenic plantcompared with the GUS- transgenic plant (Table 1), suggestingthat TENGU may suppress the auxin signaling and biosynthesispathways. Auxin is known to be involved in apical dominance,which is where an apical bud inhibits development of an axillarybud growth. Auxin biosynthesized in the apical bud is trans-ported to the root, and inhibits the growth of axillary buds (28).This growth inhibition of axillary buds is released by loss of ordamage to the apical bud (29). In this study, the expression levelsof early auxin-responsive genes, auxin efflux-related genes, anddormancy-associated genes were reduced in the tengu-transgenicplant (Table 1). It has been reported that the A. thaliana mutantof IAA/AXR2, which is an early auxin-responsive gene, exhibitedsevere dwarfism (30). Interestingly, this symptom greatly resem-bles that in the tengu-transgenic plant (Fig. 2). Taken together,these results suggest that TENGU suppresses plant auxin re-sponses, resulting in the growth inhibition of apical buds intengu-transgenic or OY-infected plants. Similarly, apical dor-mancy would be released by TENGU in tengu-transgenic orOY-infected plants, thus promoting the growth of axillary buds.It has been recently reported that auxin treatment inducedrecovery of phytoplasma-infected periwinkles (31). These find-ings imply that the symptoms caused by phytoplasma may beinvolved in an auxin-related pathway.

Some phytohormones, for example, auxin and cytokinin, arethought to be involved in morphological changes caused byphytoplasmas, such as witches’ broom or dwarf symptoms (3).However, this has not been experimentally demonstrated so far,and these disease symptoms were thought to be caused byindirect effects, for example, consumption of metabolites ofinfected plants by phytoplasmas. Our findings suggest thatTENGU could be a missing link between phytoplasmal symp-toms and phytohormones. However, many additional genesbesides the auxin-related genes were deregulated in the microar-ray analysis. It is unclear whether their deregulation was due toauxin imbalance or rather the deregulation of some auxinmetabolism related genes as a downstream effect of othermolecular event(s). Therefore, further analyses are needed toelucidate a direct connection between TENGU and auxin.

As stated above, some phytopathogenic fungi and bacteria areknown to produce phytohormones that cause disease symptoms(25–27). However, phytoplasmas do not have genes for the

Table 1. Auxin-related genes down-regulated intengu-transgenic plants as identified by microarray analysis

AGI number Fold change AGI number Fold change

AUX/IAA genes

At4 g32280 28.6 At3g23050 2.5Auxin responsive SAUR genes

At3 g53250 10.8 At4g38850 9.1At1 g75590 8.3 At5g18010 7.8At5 g18080 6.8 At5g18060 6.5At5 g18030 6.5 At1g29460 6.2At5 g18050 6.0 At5g18020 5.5At5 g03310 5.4 At1g29500 3.8At1 g29450 3.6 At1g29440 2.9At1 g29510 2.8 At3g03850 2.4

GH3 genes

At4 g27260 3.3Other auxin responsive protein

At5 g19140 3.9Dormancy-associated proteins

At1 g28330 5.7 At2g33830 5.6Auxin efflux carrier family proteins

At1 g23080 3.1 At5g01990 2.0

Genes with a ratio (GUS-transgenic plants/tengu-transgenic plants) of � 2and P � 0.05 are listed.

Hoshi et al. PNAS � April 14, 2009 � vol. 106 � no. 15 � 6419

PLA

NT

BIO

LOG

Y

biosynthesis of phytohormones (5). Instead, through secretion ofTENGU, phytoplasmas are thought to perturb the auxin path-way of the host plant. This represents a mechanism by whichpathogenic bacteria affect plant development.

TENGU Is Transported from Phloem Tissue and Localizes in Non-phloem Cells. It was previously shown that the size exclusion limit(SEL) of plasmodesmata between a sieve element and a com-panion cell ranges from 10 to 40 kDa (32), which is much higherthan the SEL of plasmodesmata between non-phloem cells suchas mesophyll cells (�1 kDa) (33). However, Imlau et al. showedthat GFP proteins (27 kDa), which are specifically produced inphloem cells under control of the phloem-specific AtSUC2promoter in A. thaliana and tobacco, could be transportedthrough plasmodesmata from the phloem into developing (sink)tissues, that is, young rosette leaves, petals, root tips, etc. (34).Therefore, because the molecular weight of TENGU is approx-imately 4.5 kDa, which is smaller than GFP, we assume thatTENGU can be transported into non-phloem cells throughsymplasm. In fact, TENGU proteins were detected not only inphloem tissues but also in parenchyma and meristem tissues viaimmunohistochemical analysis using an anti-TENGU antibody(Fig. 4A). This suggests that TENGU has the ability to betransported from phloem to non-phloem cells. In addition,TENGU proteins were strongly detected in the tip region ofstems and the branching region of axillary buds (Fig. 4D and F),and even in the apical meristem region (Fig. 4H), suggesting thatTENGU can be transported into apical buds. Because auxin isbiosynthesized in apical buds, TENGU may directly suppress theauxin biosynthetic pathway in the cells of apical buds. It has beendemonstrated that phytoplasmas selectively localize to thephloem, and are not observed in younger tissues of the apicalmeristem (19). Thus, although phytoplasmas cannot invadeapical buds, phytoplasmas may secrete TENGU to perturb plantmetabolism in apical buds by remote manipulation. Alterna-tively, TENGU may down-regulate the expression of auxinefflux-related genes (Table 1) or indirectly suppress the auxinsignaling pathway by inhibition of auxin transport, resulting in animpact on plant development.

TENGU Is Not a Silencing Suppressor. Previous studies have dem-onstrated that some PTGS suppressors might cause the distur-bance of miRNA function (18). Such PTGS suppressors mayaffect the expression of plant genes that are normally regulatedby miRNA and may also result in the induction of developmentaldefects (18). However, we demonstrated that TENGU does notpossess PTGS suppressor activity. Moreover, a viral vectorcontaining TENGU did not induce necrosis (Fig. 1 A), whereasviral vectors that contain PTGS suppressors induce severenecrosis in plants (35). These results strongly suggest thatTENGU is not a PTGS suppressor. Therefore, the developmen-tal changes induced by TENGU are probably not caused byinterference with the host’s miRNA pathways.

TENGU May Increase the Evolutionary Fitness of Phytoplasma. Thereason that phytoplasmas possess TENGU and induce morpho-logical changes in plants may be to increase their own evolu-tionary fitness by modifying plants. Disease symptoms of phy-toplasma-infected plants, including witches’ broom, phyllody,and virescence, have common characteristics: i.e., the aggressiveproduction of young and green organs. These characteristicsymptoms may be related to the life cycle of phytoplasmas.Because phytoplasmas are transmitted by insect vectors, sap-feeding by insects is one of the most important steps in thephytoplasmal life cycle (3, 4). Leafhoppers, which are the maininsect vector of phytoplasmas, prefer young and green/yellowtissues for feeding, as well as for laying eggs (4). Therefore,phytoplasmas that are able to increase the production of young

and green/yellow leaves in plants would increase their owntransmission efficiency by making the infected plants appearmore attractive to insects, in turn increasing their own survival(4). Thus, although speculative, witches’ broom, a characteristicsymptom of Ca. P. asteris-infected plants, may be the result of thephytoplasma manipulating the host to increase its own fitnessand extend its ecological niche. We demonstrated that tengu ishighly expressed in plant hosts (Fig. 3J), implying that it plays animportant role in supporting the existence of the phytoplasma inthe plant host. It is a very intriguing phenomenon that these hostcontrols are governed by a single protein (TENGU). However,our hypothesis may not apply to other phytoplasmas since manyof them do not induce witches’ broom. Further analysis of thisprotein will provide additional insight into the interactionsamong insects, plants, and phytoplasmas.

The discovery of virulence factors that induce cell death hasgreatly contributed to study of the mechanisms and signalingcascades involved in plant immunity. Likewise, the discoveryof TENGU will certainly contribute expanding progress inthe fields of plant pathology, plant physiology, and plantdevelopment.

Materials and MethodsIdentification of Secreted Proteins from OY Phytoplasma. Secreted proteinsgenerally have both a transmembrane region and a signal sequence at the Nterminus (4). We identified putative secreted proteins encoded in the OYgenome using the SOSUI (36) and Signal P (37) programs. Using this approach,more than 200 open-reading frames (ORFs) were predicted to have at least 1transmembrane region, of which more than 30 ORFs had a predicted signalsequence.

Transient Expression Assays. For Agrobacterium-mediated viral vector assays,the PAM765 (tengu) gene was amplified by PCR with KOD DNA polymerase(Toyobo) (Table S2), and the product was cloned into the binary potato virusX (PVX) vector pCAMV to produce pCAMV-tengu. The pCAMV is a derivativeof the binary vector pCAMBIA1301, which contains the PVX cDNA frompP2C2S (17) that was kindly provided by Dr. David Baulcombe (SainsburyLaboratory, Norwich, U.K.). A. tumefaciens strain EHA105 was transformedwith the vector to generate an agroinfiltration-ready clone. PVX agroinfec-tion assays were performed as described previously (38).

Inoculation of A. thaliana with Phytoplasma Strain OY. A. thaliana plants(ecotype Col-0) were grown and maintained in a growth chamber at 25 °C.Plants at the stage of 4 to 5 rosette leaves were covered with clear tubes, and5 OY-infected leafhoppers (M. striifrons) were released into the tube and thenremoved after 5 days.

Transgenic Plants. To create transgenic A. thaliana plants, the tengu gene wascloned into the binary plasmid vector pBI121 under the control of the cauli-flower mosaic virus (CaMV) 35S promoter. A. tumefaciens strain EHA105 wasthen transformed with the pBI121-tengu construct (35S::tengu).

A. thaliana plants (ecotype Col-0) were transformed using the floral dipmethod. To select transformed plants, sterilized T1 seeds were plated onkanamycin selection plates [Murashige and Skoog salt (Wako), MS vitamin(Sigma), 1% sucrose, 0.7% agar, and 50 �g/ml kanamycin]. The expression oftengu was confirmed by RT-PCR.

Quantitative Real-time RT-PCR. Total RNA from OY-infected insects (M. strii-frons) and plants (C. coronarium) was extracted with ISOGEN (Nippon Gene)and reverse-transcribed with the High-Capacity cDNA Reverse TranscriptionKit (Applied Biosystems) according to the manufacturer’s instructions. Quan-titative real-time RT-PCR assays were performed by the 7300 Real-Time PCRSystem (Applied Biosystems) with tufB as an internal standard (Table S2).

Determination of the Subcellular Localization of TENGU. The subcellular local-ization of TENGU was investigated using the following constructs: CaMV35S::tengu-GFP, CaMV 35S::GFP-tengu, and CaMV 35S::GFP. The tengu genewas cloned downstream of the CaMV 35S promoter to produce CaMV35S::tengu. The GFP gene was cloned downstream or upstream of tengu in theCaMV 35S::tengu construct to produce CaMV 35S::tengu-GFP (C-terminalfusion) and CaMV 35S:: GFP-tengu (N-terminal fusion), respectively. Eachvector was bombarded into onion epidermal cells using the Helios Gene Gun

6420 � www.pnas.org�cgi�doi�10.1073�pnas.0813038106 Hoshi et al.

System (Bio-Rad). Following the incubation of the cells overnight in the dark,the level of GFP fluorescence was examined by confocal laser scanning mi-croscopy (LSM5 PASCAL; Carl Zeiss).

Immunohistochemical Analysis. Immunohistochemical analysis was performedaccording to a previously described method with some modifications (19).Tissues including the apical meristem were excised from OY-infected andhealthy plants. These tissues were fixed, embedded in Paraplast Plus (Sher-wood Medical), and sectioned to 10-�m thickness with a microtome. Anti-TENGU and anti-Amp IgG were used with an alkaline phosphatase-mediatedreporter system to detect TENGU and Amp proteins in each tissue. Thesetissues were observed by Axio Imager microscopy (Carl Zeiss).

Microarray Analysis. For microarray analysis, 3 independent transgenic lineswere harvested from 2-week-old tengu- or GUS-transgenic plants. Total RNAwas isolated from each plant with ISOGEN (Nippon Gene), and used for thepreparation of Cy3-labeled cRNA probes. Samples were subjected to microar-ray experiments using an Arabidopsis 3 (4 � 44K) Oligo Microarray (Agilent

Technologies). All microarray experiments were performed according to thesupplier’s manual. The slides were scanned with a Microarray Scanner (AgilentTechnologies) at a 5-�m resolution, and extraction and image analysis soft-ware (Feature Extraction version 9.5.3.1; Agilent Technologies) was used tointegrate each spot’s intensity. We used the GeneSpring GX 9.0 (AgilentTechnologies) for filtering, normalization, and statistical and gene ontologyanalyses. Statistical significance of gene expression was tested using a t testcombined with a Benjamini and Hochberg false discovery rate multiple cor-rection algorithm. We selected statistically significant genes (P � 0.05) only iftheir fold change was �2.0.

ACKNOWLEDGMENTS. We thank Dr. D. Baulcombe (The Sainsbury Labora-tory, Norwich, U.K.) for providing the binary P19 construct and the PVX cDNA,and Dr. A. Shinkai (Koibuchi College of Agriculture, Ibaraki, Japan) for pro-viding photos of phytoplasma infected plants. This work was supported byGrants-in-Aid for Scientific Research from the Japan Society for the Promotionof Science (category ‘‘S’’ of Scientific Research Grant 16108001) and by Pro-gram for Promotion of Basic Research Activities for Innovative Bioscience(PROBRAIN).

1. Abramovitch RB, Anderson JC, Martin GB (2006) Bacterial elicitation and evasion ofplant innate immunity. Nat Rev Mol Cell Biol 7:601–611.

2. Bertaccini A (2007) Phytoplasmas: Diversity, taxonomy, and epidemiology. Front Biosci12:673–689.

3. Christensen NM, Axelsen KB, Nicolaisen M, Schulz A (2005) Phytoplasmas and theirinteractions with hosts. Trends Plants Sci 10:526–535.

4. Hogenhout SA, et al. (2008) Phytoplasmas: Bacteria that manipulate plants and insects.Mol Plant Pathol 9:403–423.

5. Oshima K, et al. (2004) Reductive evolution suggested from the complete genomesequence of a plant-pathogenic phytoplasma. Nat Genet 36:27–29.

6. Bai X, et al. (2006) Living with genome instability: the adaptation of phytoplasmas todiverse environments of their insect and plant hosts. J Bacteriol 188:3682–3696.

7. Tran-Nguyen LT, Kube M, Schneider B, Reinhardt R, Gibb KS (2008) Comparativegenome analysis of ‘‘Candidatus Phytoplasma australiense’’ (subgroup tuf-Australia I;rp-A) and ‘‘Ca. Phytoplasma asteris’’ strains OY-M and AY-WB. J Bacteriol 190:3979–3991.

8. Kube M, et al. (2008) The linear chromosome of the plant-pathogenic mycoplasma’Candidatus Phytoplasma mali’. BMC Genomics 9:306.

9. Suzuki S, et al. (2006) Interaction between the membrane protein of a pathogen andinsect microfilament complex determines insect-vector specificity. Proc Natl Acad SciUSA 103:4252–4257.

10. Bai X, et al. (2009) AY-WB phytoplasma secretes a protein that targets plant cell nuclei.Mol Plant Microbe Interact 22:18–30.

11. Stebbins CE, Galan JE (2001) Structural mimicry in bacterial virulence. Nature 412:701–705.

12. Shao F, Merritt PM, Bao Z, Innes RW, Dixon JE (2002) A Yersinia effector and aPseudomonas avirulence protein define a family of cysteine proteases functioning inbacterial pathogenesis. Cell 109:575–588.

13. Hotson A, Chosed R, Shu H, Orth K, Mudgett MB (2003) Xanthomonas type III effectorXopD targets SUMO-conjugated proteins in planta. Mol Microbiol 50:377–389.

14. Janjusevic R, Abramovitch RB, Martin GB, Stebbins CE (2006) A bacterial inhibitor ofhost programmed cell death defenses is an E3 ubiquitin ligase. Science 311:222–226.

15. Espinosa A, Guo M, Tam VC, Fu ZQ, Alfano JR (2003) The Pseudomonas syringae typeIII-secreted protein HopPtoD2 possesses protein tyrosine phosphatase activity andsuppresses programmed cell death in plants. Mol Microbiol 49:377–387.

16. Fu ZQ, et al. (2007) A type III effector ADP-ribosylates RNA-binding proteins and quellsplant immunity. Nature 447:284–288.

17. Baulcombe DC, Chapman S, Cruz SS (1995) Jellyfish green fluorescent protein as areporter for virus infections. Plant J 7:1045–1053.

18. Kasschau KD, et al. (2003) P1/HC-Pro, a viral suppressor of RNA silencing, interferes withArabidopsis development and miRNA function. Dev Cell 4:205–217.

19. Arashida R, et al. (2008) Cloning and characterization of the antigenic membraneprotein (Amp) gene and in situ detection of Amp from malformed flowers infectedwith Japanese Hydrangea phyllody phytoplasma. Phytopathology 98:769–775.

20. Horton P, et al. (2007) WoLF PSORT: Protein localization predictor. Nucleic Acids Res35:W585–W587.

21. Quevillon E, et al. (2005) InterProScan: Protein domains identifier. Nucleic Acids Res33:W116–W120.

22. Hagen G, Guilfoyle T (2002) Auxin-responsive gene expression: Genes, promoters andregulatory factors. Plant Mol Biol 49:373–385.

23. Stafstrom JP, Ripley BD, Devitt ML, Drake B (1998) Dormancy-associated gene expres-sion in pea axillary buds. Planta 205:547–552.

24. Petrasek J, et al. (2006) PIN proteins perform a rate-limiting function in cellular auxinefflux. Science 312:914–918.

25. Manes CLD, et al. (2004) Phenotypic alterations in Arabidopsis thaliana plants causedby Rhodococcus fascians infection. J Plant Res 117:139–145.

26. Nishida H, Sugiyama J (1993) Phylogenetic relationships among Taphrina, Saitoella,and other higher fungi. Mol Biol Evol 10:431–436.

27. Crespi M, Messens E, Caplan AB, van Montagu M, Desomer J (1992) Fasciation inductionby the phytopathogen Rhodococcus fascians depends upon a linear plasmid encodinga cytokinin synthase gene. EMBO J 11:795–804.

28. Mori Y, Nishimura T, Koshiba T (2005) Vigorous synthesis of indole-3-acetic acid in theapical very tip leads to a constant basipetal flow of the hormone in maize coleoptiles.Plant Sci 168:467–473.

29. Cline M (1997) Concepts and terminology of apical dominance. Am J Bot 84:1064–1069.

30. Timpte C, Wilson AK, Estelle M (1994) The axr2–1 mutation of Arabidopsis thaliana isa gain-of-function mutation that disrupts an early step in auxin response. Genetics138:1239–1249.

31. Perica M (2008) Auxin-treatment induces recovery of phytoplasma-infected periwin-kle. J Appl Microbiol 105:1826–1834.

32. Kempers R, vanBel AJE (1997) Symplasmic connections between sieve element andcompanion cell in the stem phloem of Vicia faba L have a molecular exclusion limit ofat least 10 kDa. Planta 201:195–201.

33. Goodwin P (1983) Molecular-size limit for movement in the symplast of the Elodea leaf.Planta 157:124–130.

34. Imlau A, Truernit E, Sauer N (1999) Cell-to-cell and long-distance trafficking of thegreen fluorescent protein in the phloem and symplastic unloading of the protein intosink tissues. Plant Cell 11:309–322.

35. Pruss G, Ge X, Shi XM, Carrington JC, Vance VB (1997) Plant viral synergism: thepotyviral genome encodes a broad-range pathogenicity enhancer that transactivatesreplication of heterologous viruses. Plant Cell 9:859–868.

36. Hirokawa T, Boon-Chieng S, Mitaku S (1998) SOSUI: classification and secondarystructure prediction system for membrane proteins. Bioinformatics 14:378–379.

37. Bendtsen JD, Nielsen H, von Heijne G, Brunak S (2004) Improved prediction of signalpeptides: SignalP 3.0. J Mol Biol 340:783–795.

38. Takahashi S, et al. (2006) The efficiency of interference of Potato virus X infectiondepends on the target gene. Virus Res 116:214–217.

Hoshi et al. PNAS � April 14, 2009 � vol. 106 � no. 15 � 6421

PLA

NT

BIO

LOG

Y