BOTRYTIS -INDUCED KINASE1 modulates Arabidopsis resistance to green peach aphids via PHYTOALEXIN DEFICIENT4

One-sentence summary: Arabidopsis BOTRYTIS -INDUCED KINASE1 negatively regulates plant resistance against green peach aphid through PHYTOALEXIN DEFICIENT4-dependent hypersensitive response. ABSTRACT BOTRYTIS -INDUCED KINASE1 (BIK1) plays important roles in induced defense against fungal and bacterial pathogens in Arabidopsis thaliana . Its tomato homolog is required for host plant resistance to a chewing insect herbivore. However, it remains unknown whether BIK1 functions in plant defense against aphids, a group of insects with a specialized phloem sap-feeding style. In this study, the potential role of BIK1 was investigated in Arabidopsis infested with the green peach aphid, Myzus persicae . In contrast to the previously reported positive role of intact BIK1 in defense response, loss of BIK1 function adversely impacted aphid settling, feeding and reproduction. Relative to wild-type plants, bik1 displayed higher aphid-induced H 2 O 2 accumulation and more severe lesions, resembling a hypersensitive response (HR) against pathogens. These symptoms were limited to the infested leaves. The bik1 mutant showed elevated basal as well as induced salicylic acid and ethylene accumulation. Intriguingly, elevated salicylic acid levels did not contribute to the HR-like symptoms or to the heightened aphid resistance associated with the bik1 mutant. Elevated ethylene levels in bik1 accounted for an initial, short-term repellence. Introducing a loss-of-function mutation in the aphid resistance and senescence-promoting gene PHYTOALEXIN DEFICIENT4 ( PAD4 ) into the bik1 background blocked both aphid resistance and HR-like symptoms, indicating bik1 -mediated resistance to aphids is PAD4-dependent. Taken together, Arabidopsis BIK1 confers susceptibility to aphid infestation through its suppression of PAD4 expression. Furthermore, the results underscore the role of reactive oxygen species and cell death in plant defense against phloem sap-feeding insects.


INTRODUCTION
Aphids are specialized to feed and survive on phloem sap of their host plants. In contrast to chewing insects that cause extensive plant tissue damage, aphids have evolved to manipulate resource allocation within the host plant by converting the feeding site into a sink to deplete photoassimilates (Girousse et al., 2005). Their highly modified stylets navigate through plant tissues predominantly intercellularly before reaching phloem, causing very limited host cell damage. During probing and feeding, aphids secrete gelling and watery saliva (Tjallingii, 2006).
Gelling saliva forms the sheath enveloping the stylet along the pathway leading to the vascular bundle. The sheath limits damage to plant cells and avoids triggering extracellular defenses.
Watery saliva is thought not only to prevent clogging of phloem sieve elements and the food canal in aphid stylets due to protein coagulation, but also to modulate host cellular processes and mitigate host defense (Tjallingii, 2006;Will and van Bel, 2006;Will et al., 2007). Aphids make use of their stealthy feeding strategies and intimate associations with their hosts to disguise themselves and overcome plant defense, reminiscent of the deceptive strategies frequently employed by pathogens (Kaloshian, 2004;Walling, 2008).
During the long history of co-evolution, plants have developed sophisticated means to protect themselves against assaults from various herbivorous insects. Most plants are equipped with constitutive and induced defense mechanisms including physical barriers, such as trichomes and cell walls, and chemical defense, such as secondary metabolites. Despite the deceptive feeding style of aphids, the brief intracellular punctures along the stylet passage and secretions from salivation nevertheless can trigger responses in host plants (Tjallingii, 2006;Will and van Bel, 2006;De Vos and Jander, 2009;Bos et al., 2010). Plant defense responses can be classified as antibiosis, which curtails insect survival and reproduction, and/or antixenosis, which deters insect settling and herbivory. Transcriptomic studies suggest that phloem sap feeders modulate known defense signaling pathways, oxidative stress response, senescence, and plant metabolism and structure (Moran and Thompson, 2001;Zhu-Salzman et al., 2004;De Vos et al., 2005;Thompson and Goggin, 2006;Kusnierczyk et al., 2008). Plant response to aphids involves genes regulated by the major plant hormones salicylic acid (SA), jasmonic acid (JA), ethylene (ET) and abscisic acid (ABA), and genes encoding transcriptional regulators. Exogenous JA application al., 2009). Oxidative stress induced by insect herbivory is considered a component of soybean (Glycine max) resistance to invading corn earworm (Helicoverpa zea) (Bi and Felton, 1995).
Arabidopsis PHYTOALEXIN DEFICIENT4 (PAD4), a lipase-like protein essential for defense against microbial pathogens (Jirage et al., 1999), has been demonstrated to enhance plant resistance to green peach aphid (Myzus persicae) by promoting premature leaf senescence and cell death (Pegadaraju et al., 2005;Pegadaraju et al., 2007). Functional dissection further revealed that the molecular mechanism of PAD4 resistance against aphids is distinct from that against pathogens (Louis et al., 2012).
In this study, we examined the roles of several RL(C)Ks, including FLS2, EFR, BAK1, and BIK1, in Arabidopsis response to aphid infestation. We challenged these loss-of-function mutants with M. persicae, a phloem sap-feeding generalist, to evaluate aphid performance and plant response. bik1 plants displayed heightened antibiosis and antixenosis toward aphids, which was correlated with pronounced aphid-induced HR-like cell death. Further exploration of potential interactions between BIK1 and known defense pathways revealed that BIK1 modulated plant response to aphid infestation through its control of PAD4 expression.

bik1 exhibited increased resistance to green peach aphids
Plant defense response upon aphid infestation is often reflected by reduced offspring production (antibiosis) in a no-choice test with reduced feeding and body weight, or by nonpreference (antixenosis) in a choice test. To determine whether the several known RL(C)Ks, which play important roles in PAMP-triggered immunity, extend their function to aphidassociated defense response, we evaluated aphid performance on the loss-of-function mutants ( Fig. 1). Aphids infesting fls2, efr and bak1 mutants had fecundities comparable to that on the wild-type (WT) Col-0 plants (Fig. 1A). Likewise, no particular preference was detected among them ( Fig. 1C), suggesting that these RLKs may not play a major role in plant defense against aphids. Interestingly, on bik1, the amount of aphid progeny was on average about half that on WT plants (Fig. 1A). In agreement with this no-choice test result, aphids on bik1 excreted less honeydew (Fig. 1D), indicative of less food intake, and had less body weight (Fig. 1B) than those reared on WT. In the choice tests, approximately twice as many aphids preferred WT versus bik1 plants (Fig. 1C). Thus, BIK1 was a negative regulator of plant resistance to aphids. In addition, we confirmed that the heightened resistance in bik1 is indeed due to loss of BIK1 function via complementation experiments. Transgenic plants expressing BIK1 cDNA in bik1 mutant recovered the susceptibility to aphids in both choice and no-choice tests (Fig. 1E), verifying that the observed aphid resistance in bik1 was due to loss of BIK1 function.
Notably, bik1 mutant showed comparable size and biomass during the first 3 weeks of growth ( Fig. 1C; Table S1), when choice tests were performed. Later, bik1 mutant exhibited growth defect and were smaller than WT ( Fig. S1; Table S1). However, the antibiotic activity was unlikely due to their small stature, as inoculating six 2 nd instar nymphs and rearing them for 7 days on 4 to 5-week-old plants would by no means result in a population limited by space or nutrients.

Aphids induced hypersensitive response (HR)-like lesions in bik1
Despite an enhanced resistance to aphid infection, bik1 began to show apparent lesion spots approximately 5 days after aphid infestation, while no visible lesions were observed in fls2, efr and bak1 mutants or in WT ( Fig. 2A). With continued aphid infestation, all infested plants, regardless of the genotype, eventually displayed stunted growth, yellowing and necrosis with lesions spreading to the entire leaf and the whole plants. Notably, bik1 is not a lesion mimic mutant as no spontaneous lesions were observed without aphid infestation. Since bik1 plants are dwarfs, the number of aphids applied was adjusted by a ratio proportional to the rosette area. For plant symptom assessment, this ratio was applied for all genotypes exhibiting size differences relative to WT, to exclude potential misjudgment due to size discrepancies.
We further examined whether the aphid-induced lesion formation in the bik1 mutant resembles the features with an HR process that is often correlated with plant resistance against microbial pathogens (Lamb and Dixon, 1997;Heath, 2000). Using 3,3'-diaminobenzidine (DAB) staining, we observed that leaves of aphid-infested bik1 plants had much higher H 2 O 2 accumulation than any other genotypes examined ( Fig 2B). Likewise, more severe cell death was shown in aphid-infested bik1 leaves compared with WT and the other mutants by the trypan blue staining assay (Fig. 2C). In contrast, fls2, efr and bak1 mutants showed phenotypes similar to WT plants in either H 2 O 2 or cell death assays. Furthermore, we detected accumulation of autofluorescent phenolic compounds and deposition of callose at necrotic spots in aphid-infested bik1 plants (Figs. 2D,2E), which are also HR lesion-associated histological markers (Hunt et al., 1997;Luna et al., 2011;Williams et al., 2011). WT levels of H 2 O 2 and lesions upon aphid infection were restored in the bik1 BIK1 complementation line (Fig. 2). Taken together, the data indicate that aphid-induced lesions in bik1 were an HR-like response.
Since cellular H 2 O 2 accumulation precedes cell death (Hoeberichts and Woltering, 2003), earlier time points were chosen for DAB staining. Staining became apparent within 3 hours upon aphid infestation in bik1 leaves, but was absent from the infested WT leaves over the 24 hour course of the experiment (Fig. 3A). When aphids were caged on specific leaves, H 2 O 2 could only be detected in infested local leaves, not in uninfested systemic leaves (Fig. 3B), supporting our conclusion that the lesion formation in bik1 is an HR rather than a constitutive plant damage phenotype. Correlation between plant symptoms and aphid performance suggests that elevated H 2 O 2 accumulation and cell death in bik1 could be the defense mechanism compromising aphid fitness. BIK1 thus functions to counteract aphid-induced ROS production and cell death, distinct from its role in PAMP pathways.

Aphids altered phytohormone contents and gene expression in bik1
Aphid-induced plant defense and cell death pathways are often regulated by certain plant hormones (De Vos et al., 2005). To determine whether the resistance to aphids conferred by loss of BIK1 function involved defense-related plant hormones, we measured SA, JA, ET and ABA levels in the presence and absence of aphid feeding in both WT and bik1 plants (Fig. 4A).
Elevated basal SA (consistent with Veronese et al. (2006)) and ET levels were detected in bik1, while JA and ABA contents were comparable in both genotypes. SA and ET levels increased in both WT and bik1 upon aphid infestation, and the levels of both hormones were higher in bik1 than in WT (Fig. 4A). No significant changes in JA and ABA were observed after aphid feeding.
Basal expression levels of the SA-signaling marker gene PR-1, and the ET/JA marker genes ERF1 and PDF1.2 were greater in bik1 compared to WT (Fig. 4B). Aphid infestation upregulated expression of these genes in both WT and mutant plants. In comparison, basal expression of the JA-regulated transcription factor MYC2 was similar in both genotypes and was not altered by aphid infestation in either genotype (Fig. 4B). These data imply that BIK1 may function as a negative regulator of SA and ET accumulation both in the presence and absence of aphid infestation, thereby suppressing expression of their responsive genes.

Resistance to aphids conferred by loss of BIK1 function was SA-independent
To assess the role that SA may play in bik1 resistance to aphids, bik1 sid2 and bik1 nahG plants were used for choice and no-choice tests (Fig. 5) (Delaney et al., 1994). In no-choice tests, the aphid numbers on bik1 sid2 or bik1 nahG plants paralleled those on bik1, and numbers on SA-deficient sid2 or nahG did not significantly differ from the WT (Fig.   5A). Similar results were obtained in choice tests (Fig. 5B), as well as from honeydew excretion assays (Fig. 5F). Apparently, reducing the SA level did not weaken aphid resistance in bik1, nor did it influence aphid response in WT. Therefore, elevated SA accumulation was not required for bik1 resistance to the aphid, in contrast to its requirement for bik1's resistance to a virulent strain of Pseudomonas syringae (Veronese et al., 2006).
To examine how SA impacted the aphid-triggered HR-like lesion formation, H 2 O 2 production and cell death in bik1, DAB and trypan blue staining were conducted on the SAdeficient plants. No correlations were observed between the SA status and lesion formation, H 2 O 2 production or cell death phenotypes (Figs. 5C, 5D, 5E), a result supporting previous studies showing that SA is not essential for aphid defense in Arabidopsis (Pegadaraju et al., 2005). In contrast, a correlation was observed between resistance to aphids and H 2 O 2 production as well as cell death occurrence. Notably, in terms of the plant size and morphology, bik1 sid2 and bik1 nahG were closer to WT than to bik1, yet they exhibited levels of H 2 O 2 production, cell death and aphid resistance comparable to bik1. Therefore, dwarfism was unlikely the cause of enhanced resistance to aphids in bik1. Heightened endogenous SA has been reported previously to confer bik1 with resistance to the bacterial pathogen PstDC3000 (Veronese et al., 2006).
Results from our study revealed differential function of SA in BIK1-mediated plant responses to bacterial pathogens versus phloem sap-feeding aphids.
Elevated ET signaling in bik1 increased aphid repellence during early stages of infestation Like SA, ET is known to play a key role in cell death and plant response to pathogens and insects (Dong et al., 2004;Cohn and Martin, 2005;Bouchez et al., 2007). To examine whether elevated ET has a role in aphid resistance in bik1, we pretreated plants with 1methylcyclopropene (1-MCP), an inhibitor of ET action that binds to the ET receptor. In choice tests, there was no significant difference in the number of aphids on 1-MCP-treated bik1 and WT plants 6 hr after aphid inoculation (Fig. 6), suggesting that 1-MCP may have compromised resistance in bik1. As time went on, however, 1-MCP-treated bik1 gradually regained their aphid repellence, presumably due to loss of 1-MCP function.
Since the 1-MCP effect was temporary, this pharmacological approach was limited to choice tests. To further investigate whether increased ET contributes to bik1 resistance to aphids, a genetic approach was used to impair ET signaling in bik1 and longer-term no-choice tests were performed. The bik1 mutant was crossed with two ET-insensitive mutants, ethylene insensitive 2-1 (ein2-1) and ein3-1 (Guo and Ecker, 2004;Broekaert et al., 2006). EIN2 (a transducer of ethylene signaling) and EIN3 (a primary ET-responsive transcription factor) are essential components of the ET signaling pathway. In no-choice tests, the bik1 ein2-1 double mutant showed resistance comparable to bik1 (Fig. 7A), suggesting that ET was not important in suppressing aphid reproduction in bik1, in agreement with honeydew secretion data (Fig. 7F).
However in choice tests, blocking ET signaling in bik1 (i.e. bik1 ein2-1) increased plant attractiveness to aphids (Fig. 7B), implying that elevated ET in bik1 contributed to its aphid repellence. Interestingly, when compared with bik1, bik1 ein2-1 was preferred more by aphids early on. As experiments continued, the difference in the number of aphids on each genotype became non-significant. Thus, the overall effect of ET on bik1-mediated aphid resistance appeared to be only temporary and rather subtle.
The bik1 ein2-1 double mutant maintained the small stature of the bik1 single mutant (Fig.   S1C). Feeding response in the bik1 ein2-1 double mutant, i.e. lesion formation, H 2 O 2 production and cell death upon aphid infestation, resembled those of bik1 (Figs. 7C, 7D, 7E). Similar results were obtained with bik1 ein3-1 plants (Fig. S2). Taken together, ET signaling in bik1 was mainly involved in aphid deterrence initially in choice tests, but appeared to play little role in cell death-mediated defense in bik1.

Aphid resistance and HR-like cell death in bik1 is PAD4-dependent
PAD4 is a lipase-like protein that, upon aphid feeding, promotes premature leaf senescence to suppress insect reproduction and colonization (Pegadaraju et al., 2005;Pegadaraju et al., 2007). Aphids induced PAD4 expression in both bik1 and WT (Fig. 8A). Compared to the WT plants, bik1 had much higher PAD4 basal expression. Consistently, a senescence marker gene, SENESCENCE ASSOCIATED GENE 13 (SAG13) regulated by PAD4 during aphid infestation (Weaver et al., 1998;Pegadaraju et al., 2005) shared a similar expression pattern with PAD4 (Fig. 8A). These results indicated that BIK1 suppresses PAD4 and senescence gene expression.
To learn whether potential interactions exist between BIK1 and PAD4 in cell deathmediated aphid resistance, we examined aphid performance on the bik1 pad4 double mutant. In no-choice tests, aphid numbers and body weight were both significantly higher on bik1 pad4 than on bik1 plants, and were comparable to WT (Figs. 8B, 8C). Honeydew excretion showed the same trend (Fig. 8H). Likewise, in choice tests, aphids showed a strong preference for bik1 pad4 when paired with bik1 ( Fig. 8D). Apparently, the antibiosis and antixenosis observed in bik1 diminished when the pad4 mutation was introduced. The pad4 mutant did not support more aphid growth than the WT plant, although it attracted more aphids in the choice test. Therefore, the suppression of aphid performance in bik1 was dependent on elevated basal PAD4 expression. Inactivation of PAD4 in bik1 blocked the cell death, indicating that PAD4 was required for hypersensitivity and aphid resistance resulting from loss of BIK1 function.
Interestingly, ET emission decreased in bik1 pad4 compared to bik1, both in the presence and absence of aphids ( Fig. 9). This observation suggested that PAD4 may positively regulate ET accumulation.

Loss of BIK1 function did not confer resistance to chewing insects
Unlike aphids, chewing insects massively damage the host cells during infestation. To assess the role of BIK1 in Arabidopsis defense against chewing insects, we performed bioassays using fall armyworm (Spodoptera frugiperda) neonate larvae placed on 4-week-old WT and bik1 plants ( Fig. S3). No significant weight and size differences were detected between larvae reared on the two genotypes (Figs. S3A, B). In addition, fall armyworm elicited comparable H 2 O 2 production on WT and bik1 plants (Fig. S3C). The data suggested that BIK1 has distinct roles in Arabidopsis response to two groups of insects that differ in their feeding behaviors. This observation is also different from a previous study showing that TPK1b, the tomato homolog of BIK1, enhances host plant resistance against tobacco hornworm (Manduca sexta) (Abuqamar et al., 2008).

DISCUSSION
Plants in the natural environment are constantly challenged by insect herbivory and pathogen infection. As a result, they have developed a plethora of sophisticated means to cope with diverse biotic stresses. Given the common features between plant responses to phloem sapfeeders and pathogens, we studied several PAMP/MAMP signal receptors for involvement in plant response to aphids using their loss-of-function lines. While FLS2, BAK1 and EFR did not seem to be associated with response to aphid infestation, BIK1 acted as a negative regulator of the defense response against aphids. This is in contrast to its positive role in resistance to fungal necrotrophs (Veronese et al., 2006) and flagellin-mediated immune responses (Lu et al., 2010).
Thus, the PAMP-recognition components did not seem to have a parallel role in perceiving or transmitting signals from invading aphids.

HR-like cell death could be pivotal for aphid resistance in bik1 plants
The bik1 mutant exhibited heightened resistance to aphids as well as enhanced local H 2 O 2 production and necrotic cell death upon aphid infestation (Figs. 1 and 2). As in plant-microbe interactions, cell death could be either considered a plant defense factor, or viewed as an effect of aphid manipulation of host nutritional quality (Goggin, 2007). Although bik1 plants displayed severe lesion formation, this aphid-induced symptom correlated with impeded aphid colonization, growth and reproduction. Thus, rather than a damage symptom, H 2 O 2 accumulation and cell death represent a major defense mechanism in bik1 to enhance resistance to aphids. These features were limited to aphid-infested bik1 leaves (  (Van Breusegem and Dat, 2006). Accordingly, the elevated ROS generated in bik1 may result in decreased quantity and quality of nutrients and antioxidants, causing damage to aphid tissues and ultimately reducing their fitness. Furthermore, it is plausible that H 2 O 2 -potentiated HR in infected and adjacent cells could limit photoassimilate flow to the feeding sites, although it is questionable how effective such an approach can be, given that aphids can move away from their feeding sites before a sufficient defense response is mounted. Nevertheless, poor aphid performance on bik1 plants relative to WT supported the hypothesis that rapid and potent HR-like cell death placed limitations on aphid infestation.

ROS production, cell death and defense against aphids in bik1 required functional PAD4
While loss of BIK1 function promoted aphid-induced lesions, no lesions were formed without aphid infestation (Figs. 2, 3). Furthermore, the spread of the aphid-induced lesions in bik1 required continued aphid feeding (Data not shown). These data suggest that BIK1 does not directly repress but rather indirectly modulates a cell death pathway through an aphid-responsive component. We postulated that BIK1 may exert its negative regulation via PAD4, a lipase-like protein, for the following reasons: First, PAD4 regulates the activation of premature leaf senescence, i.e. a cell death-mediated resistance mechanism against aphids (Pegadaraju et al., 2005), consistent with the tight correlation between HR lesions and resistance we observed in bik1. Second, although PAD4 is involved in SA signaling, SA is not important for the defense against aphids conferred by PAD4, agreeing with our conclusion that bik1 resistance is SAindependent. Third, expression of PAD4 is induced in response to aphid feeding (Pegadaraju et al., 2005), potentially furnishing an aphid-triggered control point downstream of BIK1.
Experimental results demonstrated that PAD4 was required for bik1 resistance to aphids (Fig. 8).
It should be noted that although more aphids preferred pad4 plants over WT in the choice tests We propose that BIK1 modulates cell death and resistance to aphids through its control of PAD4 (Fig. 10). Removal of PAD4 function was sufficient to eliminate the strong HR-like cell death of bik1 and restore its susceptibility to aphids. Ectopic expression of PAD4 triggered more rapid cell death in aphid-infested leaves and stronger resistance to aphids than in WT (Pegadaraju et al., 2007). Inactivation of BIK1 repression in a sense resembles overexpression of PAD4. On the other hand, although aphid feeding induced PAD4 expression and localized cell death in WT plants, DAB staining revealed only marginal differences in H 2 O 2 production between the WT and the pad4 mutant (Fig. 8). These data suggest that in WT plants, BIK1 suppression most likely is the dominant control factor for cell death, prevailing over the stimulus from aphid feeding. It should be pointed out that high basal PAD4 expression alone, i.e. in the bik1 mutant without aphid feeding, was insufficient to result in cell death. Contrasting results of DAB staining of the bik1 mutant with and without aphid treatment appeared to support this assumption. It is possible that PAD4-mediated cell death is initiated and propagated by aphid oral secretion-triggered signaling cascades, which are predominantly repressed by BIK1.
It should be noted that bik1 is not the only mutant conferring PAD4-dependent aphid resistance. Loss of function of SUPPRESSOR OF SALICYLIC ACID INSENSITIVITY (SSI2), a desaturase, resulted in hyper-resistance to aphids, and the resistance required PAD4 as well (Louis et al., 2012). As with bik1, ssi2 resistance diminished in the ssi2 pad4 double mutant.
But unlike the bik1 mutant that expressed high basal PAD4 transcript, the ssi2 mutant did not show elevated PAD4 expression in the absence of aphid feeding. Thus, the role of PAD4 in aphid resistance could be regulated by distinct pathways; while bik1 may exert its resistance through releasing the suppression of PAD4 by BIK1, the interaction with SSI2 could be indirect.

Pleiotropic effects of BIK1
It is rather counterintuitive, at first glance, that a gene like BIK1 that confers plant susceptibility to invaders exists. A logical explanation could be that it plays an indispensable role in other processes, and/or is involved in multiple pathways in the plant where a balance has to be achieved through cross-talk. Constitutive defense is often associated with fitness costs, e.g. altered leaf morphology, stunted growth and decreased fertility (Heil and Baldwin, 2002).
Evidently, BIK1 is necessary for normal plant growth (Veronese et al., 2006) and seed production (Table S1). High levels of SA may be a major causal factor for the aberrant development and reduced growth of bik1 since SA depletion by sid2 and nahG largely restored the WT stature of bik1 plants ( Fig. 5; Fig. S1B). Furthermore, the defect in SA accumulation in pad4 could be responsible for the near WT plant form and leaf shape of the bik1 pad4 double mutant ( Fig. 8; Fig. S1D). Indeed, many lesion mimic mutants display altered plant morphology due to production of elevated levels of SA and its constitutive interaction with other pathways (Lorrain et al., 2003). Therefore, it is very likely that BIK1 regulates normal plant growth in part by controlling SA levels. Conversely, bik1 ein2-1 and bik1 ein3-1 double mutants suffered the same growth suppression and aberrant development as the bik1 single mutant, and did not show any phenotypic recovery ( Fig. 7; Fig. S1C). Therefore, despite the essential role of ET in plant development, it is unlikely that the elevated ET level contributed to the bik1 growth abnormality.
Notably, although BIK1 enhanced susceptibility to aphids, its presence did not block induction of effective aphid resistance genes but reduced their basal expression (Fig. 8). Perhaps, without BIK1 the penalty in general plant fitness imposed by maintaining a defense system in a no-pest environment outweighs an immediately available defense when plants are facing aphid attack. Besides plant development, BIK1 confers resistance to necrotrophic pathogens (Veronese et al., 2006) and is involved in activation of PAMP-triggered signaling pathways (Lu et al., 2010). Our current study showcased the crosstalk among signaling pathways involved in plant development and defense against insects versus pathogens.
In contrast to our results showing that BIK1 negatively regulated resistance to a phloem sap feeder and had no effect on a chewing insect, studies on the BIK1 homolog in tomato, TPK1b, indicate that TPK1b positively regulates plant resistance against herbivory of tobacco hornworm, also a chewing insect (Abuqamar et al., 2008). Since TPK1b rescues the phenotype of the Arabidopsis bik1 mutant, i.e. restoring its resistance to Botrytis, TPK1b and BIK1 are thought to perform similar functions in their respective species. The differential, even opposing functions exhibited by BIK1 and TPK1 suggests that the involvement of BIK1 in plant defense against insects could be shaped by specific insects through their distinct feeding styles and unique interactions with their host plants formed over the long history of coevolution.
Our study has drawn an important link between ROS production/cell death and plant resistance to aphids. However, uncoupling cell death from insect resistance has also been reported in studies with Medicago truncatula (Klingler et al., 2009). In these studies, it is clearly demonstrated that HR lesions are not required for resistance to the pea aphid (Acyrthosiphon pisum). In plant-pathogen interactions where the HR is often considered a major form of resistance, it has been shown that the Arabidopsis defense no death (dnd) mutant exhibits enhanced resistance against pathogen infection in the virtual absence of HR cell death (Yu et al., 1998). Further investigation is needed to establish whether the hypersensitivity is the basis for aphid resistance in bik1 plants. It also remains to be elucidated whether HR lesions directly cause plant defense or if they are the consequence of defensive biochemical reactions activated by aphids.

Plant growth and aphid rearing
Arabidopsis thaliana was grown in LP5 potting medium (Sun Gro Horticulture, Bellevue, WA) in environmental chambers at 23ºC (day) /21ºC (night), 65% relative humidity (RH) and 12L/12D photoperiod with a photosynthetic photon flux density of 85 µMoles m -2 s -1 . For plant damage evaluation, histochemical assays and aphid no-choice tests, 4 to 5-week-old plants were used. For plant gene expression analyses and hormone measurements, as well as for aphid choice tests, 3 to 4-week-old plants were used.
Phloem sap-feeding green peach aphids M. persicae (a tobacco-adapted red lineage, kind gift from Dr. Georg Jander, Boyce Thompson Institute for Plant Research, Cornell University, NY) were cultured on cabbage (Brassica oleracea) and maintained in an environmental chamber at 21ºC, 65% RH, and 12L/12D photoperiod (63 µMoles m -2 s -1 ). All insect treatments and bioassays were performed in this chamber.

Insect bioassays
Aphid no-choice and choice tests were performed to assess the antibiotic and antixenotic resistance of different Arabidopsis genotypes. For the no-choice tests, 6 age-synchronized second instar nymphs (within 24 hr) were placed on 4-week-old plants. The total aphid population (adult and nymph) on each plant was counted 7 days after infestation. Each genotype had at least 10 replicates. For the choice tests, 35 adults were released at an equal distance between two plants of different genotypes. The number of adult aphids settled on each plant was recorded 6 and 24 hours after releasing. At least 10 pairs of plants were used in each comparison.
All experiments were repeated at least three times and a representative data set was presented.
To obtain the average adult aphid body weight, adult aphids were transferred to WT or bik1 plants and removed 24 hrs later to produce age-synchronized progenies. Ten days later, the new generations of adults reared on Arabidopsis genotypes were collected and were weighed as 6 groups of 10 aphids each.
Eggs of fall armyworm, purchased from Benzon Research Inc (Carlisle, PA), were incubated in a growth chamber (27°C and 65% RH). Newly hatched larvae were transferred to 4-week-old WT or bik1 plants. Plants were replaced once a week to ensure sufficient food supply.
Larvae reared on Arabidopsis genotypes were weighed after feeding for 16 or 22 days. At least 30 larvae were measured for each genotype.

Ninhydrin staining and quantification of aphid honeydew
Honeydew production served as an indicator of insect feeding activity. To determine honeydew secretion, Whatman filter papers, protected by a plastic membrane to avoid absorbance of water from soil, were placed under Arabidopsis plants of various genotypes infested by 30 adult aphids. These filter papers were collected 1, 2 and 3 days after aphid infestation, soaked in 0.1% ninhydrin in acetone, and dried in a 65ºC oven for 30 min.
Honeydew stained by ninhydrin was shown as purple spots (Kim and Jander, 2007).
To quantify the honeydew stains, the filter papers were cut into pieces and stains were extracted into 1 mL of 90% methanol for 1 h at 4ºC with continuous agitation. After centrifugation at 6,000 g for 1 min, the absorbance of the supernatant was measured at 500 nm (Nisbet et al., 1994). Methanol (90%) served as a blank. Four to five-week-old Arabidopsis plants were infested with adult aphids taking into consideration the variation of the rosette size of each genotype. Accordingly, 48 aphids were placed on WT, fls2, efr, bak1-3, bak1-4, bik1+BIK1, sid2, nahG, ein2-1, ein3-1 and pad4 (sizes comparable to WT), 12 on bik1, bik1 ein2-1 and bik1 ein3-1 (one quarter the size of WT), and 24 on bik1 sid2, bik1 nahG and bik1 pad4 (one half size of WT). Plants were examined daily to identify symptoms of yellowing and lesion formation. Digital images were taken of representative leaves at 6-days post aphid infestation. Leaves obtained in the same manner were subjected to histochemical assay (see below). For every experiment, eight plants or more of each genotype were used. All experiments were repeated at least 3 times.
Leaves at 6-days post infestation, as well as control leaves, were collected and vacuuminfiltrated with DAB solution (1 mg/ml DAB, in pH 3.5 water) in a 6-well titer plate. After an overnight incubation in the same solution in darkness, the leaves were destained in 95% ethanol until they turned clear. Images were then captured with a digital camera.
To determine local and systemic ROS accumulation, aphids were placed in clear plastic cups (4 cm diameter, 4 cm height) with mesh cloth replacing the bottoms for ventilation. Twenty insects were used for WT, and 10 for bik1. The cage was fitted around the leaf petiole between the cap and the cup, and sealed with cotton to avoid wounding as well as aphid escape, restricting the aphids onto one 4-week-old Arabidopsis leaf for the desired time (Kim and Jander, 2007). Caged leaves without aphids served as controls. After treatments, the cages were removed and leaves were excised for DAB staining. Arabidopsis leaves were fixed in buffer containing 10% formaldehyde, 5% acetic acid, and 50% ethanol at 37°C overnight. Slightly translucent leaves were then washed in 95% ethanol several times until clear, rinsed twice in water, and then stained for 4 hr or longer in the dark with 0.01% aniline blue in 150 mM K 2 HPO 4 (pH 9.5). Callose deposits were visualized with an Olympus IX-81 microscope at 10x magnification under UV illumination with a broadband DAPI filter set.

JA, SA and ABA measurements
For SA, JA and ABA measurements, 3-week-old plants were infested with aphids (30 per plant). Two days later, treated or control plants were ground to a fine powder in liquid nitrogen.
For each sample replicate, ground tissue (60 mg) and a mixture of stable isotope-labeled hormones including 10 ng 2 H 4 -SA, 3.8 ng 13 C 2 -JA, and 1 ng of 2 H 6 -ABA were added to a 5 mL glass tube with 500 µL of methanol at 55 o C, and extracted by vortexing three times during a 10 min incubation. The tissue was re-extracted with 500 µL methanol, and then once with 500 µL of 80% ethanol warmed to 55 o C, centrifuging and pooling the cleared supernatants after each extraction. The pooled extracts were dried and the residue was resuspended in 800 µL of chloroform and partitioned against 1 mL of H 2 O adjusted to pH 9.0 with NH 4 OH. The aqueous fraction was recovered, adjusted to pH 5.0 with acetic acid and partitioned against 1 mL of ethyl acetate. The organic fraction was transferred to a Reactivial, dried, and then methylated with was maintained at 230 o C and the quadrupole was heated to 150 o C. The ion source was operated in electron impact mode and both scan and selected ion data were acquired. Two ions were monitored for each hormone, and the larger fragment was used for peak area quantification 124,152,156;195,224,226;166,190,194 m/z).

ET measurement and 1-MCP treatment
Three-week old Arabidopsis were infested with aphids (30 per plant) for 2 days. Shoots were excised, weighed and kept in 10 mL-syringes with 3-way stopcocks to seal them. One hour later, 1 mL of headspace gas was injected into a Photovac 10SPlus gas chromatograph (Photovac, Markham, Ontario, Canada). At least 6 individual plants were averaged for each treatment.
Each experiment was repeated at least three times. ET was quantified by integration of peak area, relative to an authentic standard (Finlayson et al., 2007).
1-methylcyclopropene (1-MCP) gas was generated by dissolving a solid formulation of a proprietary 1-MCP α-cyclodextrin complex (AgroFresh) in 0.1N NaOH in a flask fitted with a septum. The mass of the 1-MCP α-cyclodextrin complex used was calculated to produce 1000 ppm 1-MCP gas in the headspace of the flask. An aliquot of the concentrated 1-MCP gas was then injected into a desiccator to give a final calculated concentration of 1 ppm. Plants in the desiccator thus were subjected to 1-MCP treatment. After 1 hr exposure to 1-MCP, plants were brought to a normal environmental atmosphere. This procedure was repeated every 12 hr for 5 days to maintain the effect of 1-MCP, followed by aphid choice tests. Control plants were handled in the same manner without 1-MCP gas.

Quantitative RT-PCR
Plant samples were harvested, frozen and ground in liquid nitrogen to a fine powder.
Total RNA was extracted with TRIzol Reagent (Invitrogen, Carlsbad, CA) and then treated with Mastermix (BioRad, Hercules, CA) according to the manufacturer's protocol. Primers were designed using PerlPrimer software (Marshall OJ, 2004), and their quality was examined using NCBI Primer Blast. Arabidopsis UBQ10 (AT4G05320) served as an internal control for data normalization. qRT-PCR was run on an ABI Prism 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). Controls using untranscribed RNA confirmed that there was no genomic DNA contamination. Dissociation curve analyses were applied to check amplification specificity. The mean fold change in gene expression was calculated as described previously (Zhu-Salzman et al., 2003).

Statistical analysis
SPSS 16.0 software (SPSS Inc, Chicago IL) was used for analyses of all data. The nochoice tests of aphid performance among genotypes were analyzed by one-way ANOVA.
Tukey's multiple range test analysis was used for pairwise comparisons of the difference between treatments for mean separation (P < 0.05). The Chi-square test was applied to the aphid choice tests (P < 0.05).

Accession numbers
Sequence data from this article can be found in the Arabidopsis Genome Initiative or One-way ANOVA was applied to no-choice tests and the Chi-square test was used to analyze data derived from choice tests. Body weight and honeydew secretion data were analyzed by independent samples t-test. Bars represent means ± standard error (SE). Statistical significance for treatment effects is marked *(P < 0.05), **(P < 0.01) or ***(P < 0.001). Means with different letters were significantly different (P < 0.05).    Ninhydrin staining of honeydew after 48 hr aphid feeding. All experiments were performed as described in in Materials and Methods. Bars represent means ± SE. Statistical significance for treatment effects is marked * (P < 0.05), ** (P < 0.01) or *** (P < 0.001). Means with different letters were significantly different (P < 0.05). Figure 6. 1-MCP temporarily attenuates bik1 deterrence of aphids.
Choice tests between 3-week-old WT and bik1 plants in the presence and absence of 1-MCP.
Settled aphids were recorded 6 and 12 hr after aphid infestation. Application of 1-MCP began 5 days prior to choice tests, and was reapplied every 12 hrs to prevent the loss of its effectiveness.

Figure 7.
Elevated ET increases bik1 repellence against aphids but shows no effect on aphid reproduction or on aphid-induced plant hypersensitive response. Means with different letters were significantly different (P < 0.05).  ET production by WT, bik1, bik1 pad4 and pad4 plants measured before or after 48 hr aphid infestation as described in Materials and Methods. Bars represent means ± SE from at least 6 individual plants. Different lowercase letters indicate significant differences between genotypes by one-way ANOVA and Tukey's multiple range test (P < 0.05). Different uppercase letters indicate significant differences between treatments by an independent samples t-test (P < 0.05). Based on the intensity of DAB staining, the BIK1 suppression is presumably much stronger than the aphid induction, illustrated by thicker lines in the graph. BIK1 also suppresses SA and ET accumulation. SA has no direct influence on resistance to aphids. ET increased host repellence early on, possibly prior to significant ROS production.  Means with different letters were significantly different (P < 0.05). Figure S3. Loss of BIK1 function did not confer Arabidopsis resistance to fall armyworm.

SUPPLEMENTAL FIGURE LEGENDS
(A) Comparison of larval body weight after 16 d or 22 d feeding on WT or bik1 plants (n = 30).
Different letters indicate significant differences between samples (P < 0.05). (B) Images of representative larvae feeding on each genotype. (C) Images of DAB-stained WT and bik1 plants after aphid and fall armyworm feeding.

SUPPLEMENTAL MATERIALS
The following materials are available in the online version of this article.   were then collected and were weighed as 6 groups of 10 aphids each. (C) Choice tests. Threeweek old plants were used. At this developmental stage, no apparent size differences were observed between genotypes including the WT vs. bik1 pair. Settled aphids were counted 6 hr after releasing 35 adults in between two plants of the tested genotypes. Each test was comprised of 10 replicates. Inset image of the shoot phenotypes of 3-week old, uninfested WT and bik1.

Supplemental
(D) Aphids on bik1 excreted less honeydew than those reared on WT. Quantity of honeydew secretion was correlated with the area and intensity of ninhydrin stains (left) and with OD500 values (right). (E) Expression of BIK1 cDNA confers WT levels of aphid susceptibility to bik1.
One-way ANOVA was applied to no-choice tests and the Chi-square test was used to analyze data derived from choice tests. Body weight and honeydew secretion data were analyzed by independent samples t-test. Bars represent means ± standard error (SE). Statistical significance for treatment effects is marked *(P < 0.05), **(P < 0.01) or ***(P < 0.001). Means with different letters were significantly different (P < 0.05).     Hormone measurements were performed as described in Materials and Methods. Data were analyzed by independent samples t-test (P < 0.05). Different lowercase letters indicate significant differences between genotypes within the same treatment. Different uppercase letters indicate significant differences between treatments within the same genotype. (B) Relative expression of SA, JA and ET marker genes, PR1, MYC2, ERF1 and PDF1.2 in response to aphid feeding at 0 and 48 hr time points. Data were analyzed by one-way ANOVA. Tukey's multiple range test analysis was used for pairwise comparisons of the difference between treatments for mean separation (P < 0.05).   Settled aphids were recorded 6 and 12 hr after aphid infestation. Application of 1-MCP began 5 days prior to choice tests, and was reapplied every 12 hrs to prevent the loss of its effectiveness.
Control plants were subjected to the same manipulation without 1-MCP. Statistical significance for treatment effects is marked * (P < 0.05), ** (P < 0.01) or *** (P < 0.001).   Representative leaf images of 4 to 5-week-old plants (C), DAB staining (D, H2O2 indicator) and trypan blue staining (E, cell death indicator) before (top panel) or after aphid infestation (bottom panel). (F) Ninhydrin staining of honeydew after 48 hr aphid feeding. All experiments were performed as described in in Materials and Methods. Bars represent means ± SE. Statistical significance for treatment effects is marked * (P < 0.05), ** (P < 0.01) or *** (P < 0.001). Means with different letters were significantly different (P < 0.05).   ET production by WT, bik1, bik1 pad4 and pad4 plants measured before or after 48 hr aphid infestation as described in Materials and Methods. Bars represent means ± SE from at least 6 individual plants. Different lowercase letters indicate significant differences between genotypes by one-way ANOVA and Tukey's multiple range test (P < 0.05). Different uppercase letters indicate significant differences between treatments by an independent samples t-test (P < 0.05).   Based on the intensity of DAB staining, the BIK1 suppression is presumably much stronger than the aphid induction, illustrated by thicker lines in the graph. BIK1 also suppresses SA and ET accumulation. SA has no direct influence on resistance to aphids. ET increased host repellence early on, possibly prior to significant ROS production.