Nicotiana species as surrogate host for studying the pathogenicity of Acidovorax citrulli, the causal agent of bacterial fruit blotch of cucurbits

Summary Bacterial fruit blotch (BFB) caused by Acidovorax citrulli is one of the most important bacterial diseases of cucurbits worldwide. However, the mechanisms associated with A. citrulli pathogenicity and genetics of host resistance have not been extensively investigated. We idenitfied Nicotiana benthamiana and Nicotiana tabacum as surrogate hosts for studying A. citrulli pathogenicity and non‐host resistance triggered by type III secreted (T3S) effectors. Two A. citrulli strains, M6 and AAC00‐1, that represent the two major groups amongst A. citrulli populations, induced disease symptoms on N. benthamiana, but triggered a hypersensitive response (HR) on N. tabacum plants. Transient expression of 19 T3S effectors from A. citrulli in N. benthamiana leaves revealed that three effectors, Aave_1548, Aave_2708, and Aave_2166, trigger water‐soaking‐like cell death in N. benthamiana. Aave_1548 knockout mutants of M6 and AAC00‐1 displayed reduced virulence on N. benthamiana and melon (Cucumis melo L.). Transient expression of Aave_1548 and Aave_2166 effectors triggered a non‐host HR in N. tabacum, which was dependent on the functionality of the immune signalling component, NtSGT1. Hence, employing Nicotiana species as surrogate hosts for studying A. citrulli pathogenicity may help characterize the function of A. citrulli T3S effectors and facilitate the development of new strategies for BFB management.


I N T RO D U C T I O N
Bacterial fruit blotch (BFB) of cucurbits is caused by the seedborne Gram-negative bacterium Acidovorax citrulli (formerly Acidovorax avenae subsp. citrulli ) (Schaad et al. , 2008). A. citrulli gained recognition after severe outbreaks occurred in watermelon fields in several USA states in the late 1980s. Subsequently, the pathogen has spread to many parts of the world, mainly by seed transmission, and has become a serious threat to the cucurbit industry worldwide (Burdman and Walcott, 2012). Several research groups have screened plant germplasm for resistance to BFB. While some cucurbit germplasm lines have been reported to be partially tolerant to A. citrulli (Bahar et al. , 2009a, Wechter et al. , 2011, to date, no cucurbit lines have been identified with complete resistance. Thus, all commercial cultivars of watermelon (Citrullus lanatus Thunb.) and melon (Cucumis melo L.) are susceptible to A. citrulli . Despite the economic importance of BFB, little is known about the molecular basis of A. citrulli -cucurbit interactions (Bahar and Burdman, 2010;Burdman and Walcott, 2012).
Two major evolutionary lineages of A. citrulli have been identified through DNA fingerprinting and multi locus sequence typing analysis (Feng et al. , 2009;Walcott et al. , 2000). Group I strains have a broad host range that includes several cucurbit species but primarily threaten commercially produced melon (Burdman et al. , 2005;Walcott et al. , 2004). M6, an A. citrulli strain isolated from melon plants in Israel (Burdman et al. , 2005), is representative of Group I and has become the model Group I strain for pathogenicity studies. In contrast, Group II strains are highly virulent on watermelon but less virulent on melon and other cucurbit species (Eckshtain-Levi et al. , 2014;Walcott et al. , 2004). The Group II representative strain, AAC00-1, was isolated from watermelon in the USA and its genome has been completely sequenced (GenBank accession NC_008752).
Annotation of the AAC00-1 genome revealed the presence of genes encoding components of a type III secretion system (T3SS). In many Gram-negative plant pathogenic bacteria, the T3SS is responsible for secretion of effector proteins into the cytosol of their hosts (Hueck, 1998). Most type III secreted (T3S) effectors characterized so far contribute to suppression of host immunity in susceptible plants ( Alfano and Collmer, 2004), while some are recognized by specific plant disease resistance (R ) genes to induce effector-triggered immunity (ETI) in resistant plants. Plant R genes mostly encode NB-LRR type proteins and their functionality frequently requires the conserved immune signalling component, SGT1 (Peart et al. , 2002). Mutation of T3SS genes from both Groups I and II A. citrulli strains abolished their ability to trigger hypersensitive response (HR) in non-host plants (i.e. tobacco and tomato) and pathogenicity in watermelon and melon plants (Bahar and Burdman, 2010;Johnson et al. , 2011;Ren et al. , 2009).
Analysis of the A. citrulli AAC00-1 genome also revealed at least 11 putative T3S effector genes, based on their homology to known effectors in other bacterial species (Eckshtain-Levi et al. , 2014). Comparative analyses of the 11 effector genes cloned from 22 A. citrulli strains indicated that Groups I and II strains of A. citrulli have evolved different T3S effector repertoires. Moreover, all assessed Group I strains were found to lack or have truncated Aave_2166 , Aave_2708 , and Aave_3062 genes (gene names are according to the AAC00-1 annotation) (Eckshtain-Levi et al. , 2014). It was proposed that differences in the repertoire of T3S effectors between the groups contribute to observed differences in host preferential association between Groups I and II A. citrulli lineages. However, thus far, none of these putative A. citrulli effectors have been functionally characterized in regards to their contribution to virulence on susceptible host plants or their ability to trigger ETI on resistant plants.
Model plant species have been used for studying important plant diseases (Glazebrook et al. , 1997;Goodin et al. , 2008). For instance, the tomato pathogen, Pseudomonas syringae pv. tomato (Pst ) strain DC3000, is pathogenic on certain genotypes of Arabidopsis thaliana (Dangl et al. , 1992;Dong et al. , 1991;Whalen et al. , 1991). Subsequently the molecular interactions between Arabidopsis and Pst DC3000 have been extensively studied, leading to a detailed understanding of the molecular mechanisms of bacterial virulence and plant immunity. Virus-host interactions have been studied for several decades in tobacco species, including Nicotiana benthamiana and Nicotiana tabacum , largely because of their susceptibility to various viruses (Goodin et al. , 2008). Other plant pathogens, including bacteria (Metz et al. , 2005, Wei et al. , 2007, fungi (Dean et al. , 2005;Rivas-San Vicente et al. , 2013), and oomycetes (Chaparro-Garcia et al. , 2011), have also been studied on N. benthamiana . In addition, both N. benthamiana and N. tabacum can be easily transformed by Agrobacterium tumefaciens , allowing efficient protein expression through Agrobacterium -mediated transient assays. Tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS) has been developed to study potential immunity genes in N. benthamiana (Zhu and Dinesh-Kumar, 2008). Recently, the draft genomes of N. benthamiana and N. tabacum have been determined (Bombarely et al. , 2012). The aforementioned factors make N. benthamiana and N. tabacum attractive model plants for studying pathogen-host interactions, and in particular A. citrulli -plant interactions.
The goals of this study were: (i) to compare the disease development of A. citrulli on N. benthamiana and N. tabacum along with its natural host watermelon plants; (ii) to determine if N. benthamiana could serve as a surrogate host for studying pathogenicity mechanisms of A. citrulli ; (iii) to characterize the virulence and avirulence functions of selected T3S effector genes of A. citrulli on N. benthamiana and N. tabacum . Here we report that N. benthamiana can function as a surrogate host for studying A. citrulli pathogenicity. One T3S effector, Aave_1548, was critical for A. citrulli viruence. Transient expression of three T3S effectors (Aave_1548, Aave_2708 and Aave_2166) triggered a non-host HR in N. tabacum that was dependent on the function of NtSGT1. This is the first report on functional characterization of A. citrulli T3S effectors. Employing Nicotiana species as surrogate hosts for studying A. citrulli pathogenicity may advance the understanding of the function of A. citrulli T3S effectors and facilitate the development of new strategies for BFB management.

R E S U LT S N. benthamiana was susceptible to A. citrulli infection while N. tabacum was resistant
To determine if A. citrulli strains are pathogenic on N. benthamiana , we inoculated 4-week-old plants by tissue infiltration and spray-inoculation methods. Infiltration of high concentrations of bacterial cell suspensions (~0.3 × 10 8 CFU/mL) into N. benthamiana leaves triggered a strong water-soaking-like cell death at the inoculation site by 2 days post-inoculation (dpi) (Fig. 1A). When infiltrated with lower concentrations (~0.3 × 10 5 CFU/mL), the inoculated leaves of N. benthamiana turned chlorotic and a black necrosis developed along the veins at 6 dpi (data not shown). At 9 dpi, brown necrosis developed on the main stem near the petiole of inoculated leaves. Plating of petiole and stem extracts confirmed the presence of A. citrulli (data not shown). As a control, N. benthamiana plants were inoculated with a suspension of Pseudomonas syringae pv. tabaci (Pta ) (Oh and Collmer, 2005) containing ~0.3 × 10 8 CFU/mL. Inoculation with Pta triggered a similar water-soaking-like cell death ( the M6 and AAC00-1 mutants impaired in the T3SS genes hrcV (Bahar and Burdman, 2010) and hrcC (Johnson et al. , 2011), respectively (Fig. 1A).
To determine if A. citrulli could infect other tobacco species, we infiltrated high concentrations (~0.3 × 10 8 CFU/mL) of M6 and AAC00-1 cell suspensions into N. tabacum leaves. Both M6 and AAC00-1 triggered a rapid and strong HR-like cell death 24 h after infiltration (Fig. 1B). In contrast, T3SS mutants M6-ΔhrcV and AAC00-1ΔhrcC failed to trigger HR (Fig. 1B). These results support that secretion of T3S effectors from A. citrulli trigger non-host HR in N. tabacum .
We further characterized the susceptibility of N. benthamiana and the resistance of N. tabacum to A. citrulli by the sprayinoculation method ( Fig. 1C and D). Four-week-old plants of  (1) AAC00-1, (2) AAC00-1-ΔhrcC , (3) M6, (4) M6-ΔhrcV, or (5) Pseudomonas syringae pv. tabaci (Pta ) 11528, (6) Pta -ΔhrcV at ~0.3 × 10 8 CFU/ mL. The plants were incubated at 25 °C for 3 days before photographs were taken. (C, D, E) Symptoms on N. benthamiana (C), N. tabacum (D), and watermelon (E) plants after spray-inoculation with different A. citrulli strains. Four-week-old plants were spray-inoculated with AAC00-1-ΔhrcC , AAC00-1 or M6 at ~0.2 × 10 8 CFU/mL. The inoculated plants were incubated at 30 °C for 2 days before photographs were taken. (F, G) Bacterial growth of A. citrulli strains on N. benthamiana (F) and N. tabacum (G) plants. Leaves of 4-week-old plants were infiltrated with AAC00-1-ΔhrcC , AAC00-1, M6 and Pta 11528 at ~0.3 × 10 5 CFU/ mL. Bacterial populations were monitored at 0 days and 6 days post-inoculation (dpi). Data represent means ± standard errors (SE) from one representative experiment of three with similar results, with three replicates per treatment. Different letters indicate significant differences (P ≤ 0.05) amongst treatments within each time point by Tukey-Kramer HSD test. both species were spray-inoculated with bacterial cell suspensions containing ~0.2 × 10 8 CFU/mL. Spray-inoculation of A. citrulli strains M6 and AAC00-1 onto N. benthamiana led to the development of necrotic lesions similar to those observed in the natural host, watermelon ( Fig. 1C and E). Interestingly, inoculation with the Group I strain M6 consistently resulted in greater disease severity than the Group II strain, AAC00-1. This observation is important considering that Group II strains are highly virulent on watermelon but weakly to moderately virulent on other cucurbits, while Group I strains are moderately to highly virulent in diverse cucurbits. Differences in disease severity between these two strains also correlated with differences in bacterial populations at 6 dpi ( Fig. 1F): at 6 dpi, M6 populations were significantly (P ≤ 0.05) higher than those of AAC00-1 in N. benthimiana . AAC00-1ΔhrcC did not induce disease symptoms on N. benthamiana plants (Fig. 1C) and its growth was significantly (P ≤ 0.05) impaired in N. benthamiana leaves relative to the wild-type strains (Fig. 1F).
In contrast to N. benthamiana , symptoms were not observed on N. tabacum plants spray-inoculated with different A. citrulli strains (Fig. 1D). These results were in agreement with the observed HR induced by A. citrulli wild-type strains in N. benthamiana. Also in agreement with these findings, in infiltration assays, populations of AAC00-1 and M6 were significantly (P ≤ 0.05) lower than those of the pathogenic control Pta at 6 dpi. Moreover, at this time, populations of these strains were about 2 to 3 orders of magnitude lower than those observed for the same strains in N. benthamiana ( Fig. 1F and G). Overall, virulence assays indicate that N. benthamiana is susceptible to infection by A. citrulli , while N. tabacum is resistant to this pathogen. Additionally, these responses are T3SS-dependent.
Transient expression of four T3S effectors induced water-soaking in N. benthamiana and cell death in N. tabacum In total, 19 putative T3S effector genes were cloned, 11 from AAC00-1 and eight from M6. Only eight effector genes were cloned from M6 because three, Aave_2166 , Aave_2708 and Aave_3062 are either absent or truncated in this strain (Eckshtain-Levi et al. , 2014). To determine if any of the 19 T3S effector genes contribute to the ability of A. citrulli to trigger water-soaking-like cell death in N. benthamiana and HR in N. tabacum , each effector gene was sub-cloned into the binary vector pEG101-SacB/R for Agrobacterium -mediated transient assays (Traore and Zhao, 2011). The pEG101-SacB/R vector has a yellow florescent protein (YFP) tag fused to the C-terminus of the expression cassette, which allows monitoring of effector expression in planta by fluorescence microscopy. Transient expression of three T3S effector genes from AAC00-1 (Aave_1548, Aave_2166 and Aave_2708 ), and one T3S effector gene from M6 (Aave_1548 ), triggered water-soakinglike cell death in N. benthamiana ( Fig. 2A) and HR-like cell death in N. tabacum (Fig. 2B). The water-soaking-like, cell death phenotype on N. benthamiana was observed 48 h after inoculation, while the HR-like cell death on N. tabacum plants was observed within 24 h after inoculation. Of the three genes cloned from AAC00-1, Aave_2166 triggered the strongest cell death on N. tabacum . Interestingly, Aave_1548 from M6 trigged stronger cell death compared to Aave_1548 from AAC00-1 (Fig. 2B). We have previously shown that the nucleotide sequence of the central region of Aave_1548 is highly variable between Groups I and II A. citrulli strains (Eckshtain-Levi et al. , 2014) (shown for AAC00-1 and M6 in Supplementary  Fig. S2). The expression of all effector-YFP fusion proteins was confirmed by flourescence microscopy (Fig. 2C) and Western blot analysis ( Fig. 2D and E).

Aave_1548 and Aave_2166 are important for A. citrulli virulence on N. benthamiana plants
To determine if any T3S effector genes contribute to A. citrulli virulence on N. benthamiana , Aave_1548 and Aave_2166 were mutated in AAC00-1 and M6 by marker-exchange mutagenesis. N. benthamiana plants were inoculated with Aave_1548 mutants and complemented strains using tissue infiltration for phenotype observation and spray-inoculation for population growth assessment. When inoculated at a high concentration (~0.3 × 10 8 CFU/ mL), the Aave_1548 mutant strains triggered weaker cell death in N. benthamiana than the wild type and complemented strains (Fig. 3A). Similarly, at 6 dpi, AAC00-1-ΔAave_1548 grew significantly (P ≤ 0.05) less than both the wild type and the Aave_1548 complemented strain on N. tabacum (Fig. 3B). M6Δ-Aave_1548 mutant also grew to significantly (P ≤ 0.05) lower populations than the wild-type strain in N. benthamiana (Fig. 3C). These results suggest that effector Aave_1548 significantly contributes to the virulence of both M6 and AAC00-1 on N. benthamiana ( Fig. 3A-C).
The AAC00-1Δ-Aave_2166 mutant triggered a weaker cell death response than the wild-type strain when infiltrated at high concentrations into N. benthamiana leaves (Fig. 3A). At 3 dpi, the Aave_2166 mutant grew to significantly (P ≤ 0.05) lower populations than wild-type AAC00-1 (Fig. 3B). However, at 6 dpi, AAC00-1-ΔAave_2166 grew to a significantly (P ≤ 0.05) higher population than the wild-type strain on N. benthamiana (Fig. 3B). We were unable to generate a complemented strain for this mutant. Attempts to amplify and clone a DNA fragment containing the full length of this gene were unsuccessful, probably because of its high G + C content.
that there are multiple T3S effectors or other components of A. citrulli that trigger HR on N. tabacum in a T3SS-dependent manner. These results suggest that two T3S effector genes, Aave_1548 and Aave_2166 , contribute substantially to A. citrulli virulence and its ability to infect N. benthamiana plants.
M o l ec u l a r P l a n t Pat h o lo gy (2019) 2 0 (6), Silencing the NtSGT1 gene in N. tabacum compromised its resistance to A. citrulli At least three A. citrulli T3S effectors elicited HR in N. tabacum , but T3SS mutants of A. citrulli failed to induce HR. Therefore, we hypothesized that this HR is the result of specific recognition of T3S effector(s) by unknown cognate plant R gene(s). The function of many R genes depends on the presence of the conserved immune signalling component, SGT1 (Peart et al. , 2002). It was also reported that pepper SGT1 interacts with the Xanthomonas effector AvrBsT, a homologue of Aave_2166, that is required for AvrBsT-triggered cell death in pepper (Kim et al. , 2014). Therefore, we attempted to silence NtSGT1 using RNAi to determine if it is required to elicit HR in N. tabacum in response to A. citrulli . Two independent NtSGT1 -RNAi transgenic lines were used for this study, and NtSGT1 silencing was confirmed by Reverse Transcription-Polymerase Chain Reaction (RT-PCR) (Fig. 4A). Transgenic and non-transgenic N. tabacum plants were used for inoculation and in planta bacterial growth assays with A. citrulli strains. Inoculation with AAC00-1 and M6 failed to trigger HR cell death in the RNAi-NtSGT1 transgenic N. tabacum plants (Figs 1B and 4B). Compared to the phenotypes observed with non-transgenic N. tabacum plants, transient expression of Aave_2166 and Aave_1548 from AAC00-1, as well as Aave_1548 from M6 failed to trigger the cell death phenotype on the RNAi-NtSGT1 plants (Figs 2B and 4C), even though the fluorescent signal of the effector-YFP fusion proteins was detected (data not shown).
To further test whether silencing NtSGT1 in N. tabacum could suppress its resistance to A. citrulli , the population dynamics of AAC00-1 and M6 on RNAi-NtSGT1 plants were monitored. AAC00-1 induced weak lesions on inoculated leaves at 9 dpi, while M6 failed to induce lesions by 9 dpi (Fig. 4D and E). Despite this difference, RNAi-NtSGT1 plants supported population growth of both AAC00-1 and M6 in comparison to wild-type N. tabacum plants ( Fig. 4F and G). These results support that the HR response observed in N. tabacum may be caused by the recognition of one or more effectors produced by A. citrulli expressing unknown R gene(s) that require the function of NtSGT1 .

D I SC U SS I O N
Despite the significant economic threat that BFB poses to cucurbit crop production worldwide, limited information is available about the mechanisms of pathogenicity of the causal agent, A. citrulli (Bahar and Burdman, 2010;Burdman and Walcott, 2012). An accurate understanding of the molecular host-pathogen interactions involved in pathogenicity would greatly enhance efforts to manage this disease, particulary as it pertains to durable-host-resistance development. This report demonstrates that N. benthamiana can function as a surrogate host for studying A. citrulli pathogenicity and virulence, while N. tabacum may contain non-host R genes that can recognize A. citrulli T3S effectors to trigger disease resistance (Fig. 1). By employing Agrobacterium -mediated transient assays, bacterial mutagenesis and growth curve assays on N. benthamiana plants, we identified Aave_1548 as the first A. citrulli T3S effector that contributes significantly to the virulence of A. citrulli (Fig. 3).
N. benthamiana is an excellent model plant species for studying molecular plant-microbe interactions. In this study, we demonstrated that the symptoms caused by A. citrulli on N. benthamiana leaves are similar to those observed on its natural hosts. Spray-inoculation with cell suspensions containing ~10 8 CFU/mL of A. citrulli onto N. benthamiana leaves resulted in necrotic lesions similar to those reported for cucurbit plants ( Fig.  1C and E) (Hopkins and Thompson, 2002). By 9 dpi, brown lesions were observed on petioles and stems of inoculated tobacco leaves, as a possible result of systemic infection of A. citrulli . The ability of A. citrulli to infect and spread via vascular tissue was previously reported in melon seedlings (Bahar et al. , 2009b) and squash leaves (Makizumi et al. , 2011). 1993.88 ± 93 ‡ 0 *Hybrid cantaloupe/muskmelon seeds (cv. Athena) were used for inoculations with AAC00-1 and AAC00-1 mutants, while melon cv. Ofir seeds were used for inoculations of M6 and M6 mutants. All inoculated and control seeds were incubated in bacterial cell suspensions (~1 × 10 6 and 1 × 10 7 CFU/mL for Athena and Ofir seeds, respectively). Control seeds were subjected to the same treatment but with water) for 2 h. Infected seeds were blot dried and planted, four seeds/per pot with two pots for each treatment. The weights of the seedlings were measured at 12 days post-inoculation and compared to the control (H 2 O). † In plant weight values, different letters indicate significant differences between treatments (P ≤ 0.05) by Tukey-Kramer HSD test. ‡ Disease severity was scored using a scale of 0 to 7, based on plant weight values of inoculated plants relative to the average shoot weight of non-inoculated controls (Bahar et al. , 2009b): 0, weight higher than 90% of average control weight; 1 to 5, weight equal to 76%-90%, 61%-75%, 46%-60%, 31%-45% and 16%-30% of average control weight, respectively; 6, weight equal to or lower than 15% of average control weight; 7, dead seedling. Data was presented as mean ± standard error, n = 8. In this study, the A. citrulli T3S effector, Aave_1548, was identified as a significant virulence effector when inoculated on N. benthamiana leaves. While differences in virulence between strain M6 and M6-ΔAave_1548 could also be detected on melon, inoculation of watermelon leaves with AAC00-1 wild-type and mutant strains did not reveal significant contributions by Aave_1548 nor Aave_2166 (data not shown) to virulence. Therefore, our results suggest that N. benthamiana can be a valuable tool for studying the role of T3S effectors in A. citrulli virulence.
We recently showed that all tested Group I strains of A. citrulli , including M6, lack functional effector genes Aave_2166 and Aave_2708 (Eckshtain-Levi et al. , 2014). Group I strains were also shown to possess a copy of Aave_3062 that is truncated in the centre of the open reading frame (ORF) (Eckshtain-Levi et al. , 2014). In our study, AAC00-1-ΔAave_2166 reduced bacterial growth in N. benthamiana leaves at 3 dpi, but enhanced population growth at 6 dpi. Some T3S effectors can suppress the activity of other T3S effector proteins when co-expressed in plant cells (Guo et al. , 2009). It is possible that functional Aave_2166 and/ or Aave_3062 have an inhibitory effect on other T3S effectors, such as Aave_1548, which could mask their phenotype in Group II strains in cucurbits other than watermelon. Nevertheless, transient co-expression of Aave_2166-AAC00-1, Aave_1548-AAC00-1 and Aave_1548-M6 in pairs failed to trigger different phenotypes compared to individual expression of each effector (data not shown). Further investigation is needed to determine the role Aave_2166 plays in AAC00-1 virulence relative to M6 on N. benthamiana and whether this effector is involved in the observed restriction of cucurbit host range in Group II strains.
R -gene-mediated resistance can be an indication of the loss of avirulent T3S effector genes (Kearney et al. , 1988;Vera Cruz et al. , 2000;Yang et al. , 2005). It is also possible that N. benthamiana and some cucurbit species have evolved weak R genes that can partially recognize Aave_2166, Aave_2708 and Aave_3062 effectors produced by Group II strains. The fact that Group I strains do not express these effectors in a functional state could partially explain the broad cucurbit host range of these strains relative to Group II. In future studies, it would be interesting to express these T3S effector genes in M6 to determine if they reduce virulence on N. benthamiana and cucurbits other than watermelon.
Aave_2166 is a homologue of avrBsT , a gene that encodes proteins with acetyltransferase enzyme activity, that was originally cloned from Xanthomonas campestris pv. vesicatoria (Xcv ) (Ciesiolka et al. , 1999). In Xcv , AvrBsT is a critical virulence effector that may target important Arabidopsis or pepper immune signalling components (Cheong et al. , 2014;Kim et al. , 2010Kim et al. , , 2014Kim et al. , , 2010Kim et al. , , 2014Szczesny et al. , 2010;Thieme et al. , 2005). In this study, we demonstrated that A. citrulli Aave_2166 triggers non-host HR in N. tabacum ; however, deletion of Aave_2166 did not significantly compromise the virulence of A. citrulli in melon seedling transmission assays (Table 1). Interestingly, an additional T3S effector gene from A. citrulli , Aave_2708 , also encodes a protein belonging to the acetyltransferase family of effectors that is only present in Group II strains of A. citrulli (Eckshtain-Levi et al. , 2014). This raises the possibility that Aave_2166 and Aave_2708 may have redundant functions when translocated into plant cells. It will be interesting to test if double knockout mutants of Aave_2166 and Aave_2708 significantly increase AAC00-1 virulence on cucurbit hosts other than watermelon. A recent report suggests that the Xcv effector, AvrBsT interacts with pepper SGT1, and silencing of pepper SGT1 suppressed AvrBsT-triggered cell death (Kim et al. , 2014). In the current study, N. tabaccum SGT1 was also required for Aave_2166-triggered cell death (Fig. 4B). It remains to be seen whether NtSGT1 also directly interacts with Aave_2166.
Aave_1548 is a homologue of HopW1-1, originally identified in Pseudomonas syringae pv. maculicola strain ES4326 (Guttman et al. , 2002). HopW1-1 was shown to induce defence responses and resistance in Arabidopsis ecotype Ws, with increased SA accumulation (Lee et al. , 2008). More recently, Kang and colleagues showed that HopW1-1 promotes virulence in Arabidopsis Col-0 ecoype by disrupting the actin cytoskeleton, which is associated with inhibition of endocytosis and trafficking of specific proteins to vacuoles (Kang et al. , 2014). In the current study, Aave_1548 significantly contributed to A. citrulli virulence in N. benthamiana Fig. 4 Silencing of NtSGT1 in Nicotiana tabacum suppressed the hypersensitive response (HR) phenotype triggered by Acidovorax citrulli strains and individual T3S effectors. (A) Validation of NtSGT1 gene silencing by Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) analysis. NtSGT1 gene specific primers flanking the region that is not part of DNA fragment used for the RNAi construct was employed for RT-PCR analysis. NtActin primers were used as internal control. cDNAs isolated from two independent NtSGT1 -RNAi lines (NtSGT1-RNAi -1 and NtSGT1-RNAi -3) and the wild-type N. tabacum (Nt ) plants were used for RT-PCR. (B) HR was not triggered in NtSGT1 -RNAi lines inoculated with A. citrulli strains. Two independent NtSGT1 -RNAi lines were inoculated with AAC00-1, M6 and AAC00-1-∆hrc C strains at ~0.3 × 10 8 CFU/mL (shown for one of the lines). Pictures were taken at 2 dpi. (C) Agrobacterium -mediated transient expression of T3S effectors did not trigger HR in NtSGT1 -RNAi plants. Agrobacterium tumefaciens strains carrying the effector genes Aave_2166 , Aave_1548 and Aave_2708 from strain AAC00-1 or a GFP gene were inoculated on NtSGT1 -RNAi plants at the concentration of ~0.3 × 10 8 CFU/mL. Pictures were taken at 2 dpi. (D, E) A. citrulli strain AAC00-1 triggered disease lesions on NtSGT1 -RNAi plants (D) while M6 did not (E). The NtSGT1 -RNAi transgenic plants were infiltrated with A. citrulli strains at a concentration of ~0.3 × 10 5 CFU/mL. Pictures were taken at 9 dpi. (F, G) In planta growth of A. citrulli populations on NtSGT1 -RNAi and wild-type N. tabacum plants. AAC00-1(F) and M6 (G) were infiltrated-inoculated onto NtSGT1 -RNAi plants at the concentration of ~0.3 × 10 5 CFU/mL. Data represent means ± standard errors (SE) of one representative experiment of three with similar results, with three replicates per treatment. Different letters indicate significant differences between treatments (P ≤ 0.05) within each time point by Tukey-Kramer HSD test.  Table 1). Future studies should determine whether A. citrulli Aave_1548 promotes virulence through interference with the actin cytoskeleton in N. benthamiana and cucurbit hosts. Interestingly, the M6 Aave_1548 triggerred a stronger HR in N. tabacum than AAC00-1 Aave_1548. Amino acid sequence alignment between Aave_1548 sequences of Groups I and II strains revealed a small central domain with high polymorphism (14 amino acid differences in a 45-amino acid region; Fig. S2) (Eckshtain-Levi et al. , 2014). We are currently performing site directed mutagenesis to idenitify the critical amino acid residues that are responsible for differences in HR observed between the two Aave_1548 homologue.
N. tabacum is naturally resistant to A. citrulli (Schaad et al. , 1978). There are several explanations for the non-host resistance of N. tabacum to A. citrulli : (i) the structure of N. tabacum leaves could render a physical barrier for the establishment of A. citrulli populations; (ii) recognition of conserved pathogen associated molecular patterns (PAMPs) by tobacco pattern recognition receptors (PRRs) and subsequent initiation of PAMP-triggered immunity (PTI); and/ or (iii) the presence of R genes that may recognize one or more effectors and initiate ETI. Our data showed that at least three A. citrulli T3S effectors triggered HR in N. tabacum . Moreover, silencing of NtSGT1 , a putative common R-protein chaperone, compromised the resistance of N. tabacum to A. citrulli strain AAC00-1 (Fig. 4C). Therefore, ETI contributes, at least partially, to the non-host resistance of N. tabacum to A. citrulli . Interestingly, the Group II strain AAC00-1, but not Group I strain M6, induced weak lesions on inoculated leaves of the NtSGT1 -RNAi plant at 9 dpi ( Fig. 4D and E). Comparing the whole genome sequences of AAC00-1 and M6 may help identify other virulence factors that constribute to the host ranges of Groups I and II strains.
In this study, deletion of Aave_2166 and Aave_1548 from A. citrulli did not completely abolish the non-host resistance phenotype of N. tabacum . Deletion analysis may have been ineffective because multiple A. citrulli T3S effectors could be recognized by as yet unidentified N. tabacum R gene(s). Importantly, isolation and transfer of non-host resistance genes from N. tabacum to cucurbits could potentially offer a novel strategy for controlling BFB. Future studies should focus on identifying other T3S effector genes in A. citrulli and to generating A. citrulli mutants with multiple deletions of T3S effector genes to determine whether some combinations of impaired effector genes abolish non-host resistance in N. tabacum . Such information could help focus screens for non-host resistance genes against A. citrulli .
In summary, our data support that Nicotiana species can be used as surrogate hosts for studying A. citrulli pathogenicity. Agrobacterium -mediated transient assays of individual A. citrulli T3S effectors on N. benthamiana and N. tabacum can allow us to characterize the biological functions of T3S effectors and may facilitate the development of new strategies for BFB management.

E X PE R I M E N TA L PROC E D U R E S Bacterial strains, plasmids, primers and plant material
Bacterial strains and plasmids used in this study are listed in Table 2. N. benthamiana PI 555478 and N. tabacum cv. Samsun-NN plants were propagated from seed in a growth chamber programmed for 16 h light (140 µmol/m 2 /s cool white fluorescent irradiance) at 28 °C and 8 h dark at 25 °C. Hybrid cantaloupe/muskmelon seeds (cv. Athena) were inoculated with A. citrulli AAC00-1 and derived mutants, while melon cv. Ofir were inoculated with A. citrulli M6 and derived mutants. Watermelon cv. Sugar Baby was inoculated with all A. citrulli strains. Primers used for amplification of ORFs of T3S effector genes from strains AAC00-1 and M6 are listed in Supplementary Table S1. Other primers used in this study are listed in Supplementary Table S2.

Agrobacterium -mediated transient assays in Nicotiana plants
The ORFs of T3S effector genes were PCR amplified from AAC00-1 and M6 genomic DNA. The amplified DNA fragments were cloned into the Topo ENTR/D vector (Invitrogen, Carlsbad, CA) according to manufacturer's instructions. The sequences of all cloned genes, including the construct described below, were verified by DNA sequencing at the core facility of the Virginia Bioinformatics Institute (Blacksburg, VA). The effector genes were sub-cloned into the plant expression vector pEG101-SacB/R (Traore and Zhao, 2011) by LR reaction (Invitrogen). The pEG101-SacB/R vector has a C-terminal YFP fusion. After subcloning, the effector genes were fused to the N-terminus of the YFP gene as described (Traore and Zhao, 2011). Recombinant plasmids were electroporated into Agrobacterium tumefaciens strain GV2260 as described previously (Traore and Zhao, 2011). Transient expression assays in N. benthamiana or N. tabacum plants were performed as described previously (Traore and Zhao, 2011). Briefly, the A. tumefaciens strains were streaked onto Luria-Bertani (LB) agar supplemented with appropriate antibiotics and incubated at 28 °C for 2 days. Bacterial cells were harvested and resuspended in induction buffer (10 mM MgCl 2 , 10 mM MES [pH 5.6] and 100 µM acetosyringone) and incubated for 3 h at room temperature. Bacterial cell suspensions were then adjusted to OD 600 = 0.6 (~6 × 10 8 CFU/mL) and infiltrated through stomata into intercellular spaces of fully expanded N. benthamiana leaves using a 1 mL, blunt-end syringe. The

Western blot analysis
The effector genes cloned in pEG101-SacB/R could express the effecor-YFP-HA fusion proteins, which can be detected with anti-HA antibodies by Western blot. At 20 h after Agrobacterium infiltration, the transiently expressed effecor fusion proteins were extracted by grinding three leaf discs (19 cm diameter) in 100 µL 3 × Laemmli buffer containing 16% β-mercaptoethanol. Also, 20 µL protein samples were separated on a 10% SDS-PAGE gel, and blotted to a PVDF membrane. The blot was probed with anti-HA-HRP (Sigma, St. Louis, MO; H6533; 1:500), and the signal was detected with using SuperSignal ® West Pico Chemiluminescent Substrate (Thermo Scientific, Waltham, MA).  The chemiluminescent signals were exposed to autoradiography film (Genesee Scientific, San Diego, CA) using a Kodak film processor (Kodak, A Walsh Imaging, Inc, Pompton Lakes, NJ). Another 20 µL protein samples were also separated on a 10% SDS-PAGE gel, and stained with comassie blue to confirm the equal loadings.

Knockout of T3S effectors by marker-exchange mutagenesis and mutant strain complementation
To knockout A. citrulli T3S effectors, 1.4 kb and 1.5 kb DNA fragments from regions flanking the ORFs of effector genes Aave_2166 and Aave_1548 , respectively, were amplified using the following primers: Aave2166_f1/Aave2166_R2 (upstream region), and Aave2166_f5/Aave2166_R6 (downstream region) for gene Aave2166 ; and Aave1548_f1/Aave1548_R2 (upstream region), and Aave1548_f5/Aave1548_R6 (downstream region) for gene Aave_1548 . A kanamycin (Km) resistance gene (nptII ) was amplified from pDSK519-GFP (Matthysse et al. , 1996) with primers kan_for and kan_rev. The two flanking fragments from both genes were then fused to the nptII gene by overlap PCR (Higuchi, 1990). The derived cassettes were cloned into the PCR8/GW-Topo vector (Invitrogen), and cloned into the suicide vector pLVC18L-Des (Staskawicz et al. , 1987;Zhao et al. , 2011) using LR clonase (Invitrogen). The derived construct was then mobilized into A. citrulli strains by tri-parental mating as previously described (Ditta et al. , 1980). Double crossover mutants ( Supplementary  Fig. S1A) were selected using marker-exchange mutagenesis as previously reported (Zhao et al. , 2011). The mutant genotypes were confirmed by PCR analysis (Supplementary Fig. S1B, C and D). A. citrulli strains that contained impaired genes were selected on nutrient agar (NA) medium (Thermo Fisher Scientific Inc, Waltham, MA) supplemented with rifampicin (100 µg/mL) and kanamycin (50 µg/mL). To complement the knockout mutants of Aave_1548, the Aave_1548 effector gene, including its native promoter and full ORF, was amplified by PCR using primers Aave_1548comp for and Aave_1548comp rev. The PCR products were cloned into the PCR8/GW-Topo vector (Invitrogen), and then sub-cloned into the broad host range vector pVSP61-Des-GM (Century et al. , 1995). Successful transformants were selected on NA supplemented with gentamycin at 50 µg/mL. The derived constructs were then mobilized into A. citrulli by conjugation. An M6 mutant impaired in Aave_1548 was generated as previously described (Bahar et al. , 2009b). An internal fragment of Aave_1548 (which does not span the 3′ and 5′ ends of this gene) was PCR amplified with primers Inter1548F and Inter1548R. The PCR product was cloned into pTZ57R/T (Thermo Fisher Scientific Inc), verified by sequencing, excised with appropriate restriction enzymes, and then cloned into pJP5603 (Penfold and Pemberton, 1992) conferring kanamycin resistance. Transformation into strain M6 was carried out as described previously (Bahar et al. , 2009b). Putative mutants were selected on NA with kanamycin and verified by Southern blot analysis.
Virulence assays on N. benthamiana , N. tabacum and watermelon plants A. citrulli strains AAC00-1, M6, the derived mutants and complemented strains, as well as Pseudomonas syringae pv. tabaci (Pta 11528) were grown on NA supplemented with rifampicin (100 µg/mL) at 28 °C for 48 h. For inoculation, bacterial cells were harvested from the plates and suspended in 10 mM MgCl 2 . The inoculum concentrations were adjusted to OD 600 = 0.3 (~0.3 × 10 8 CFU/mL) for the HR assay and (~0.2 × 10 5 CFU/mL) for the infiltration-based virulence assay. For spray-inoculation, bacterial concentrations were adjusted to OD 600 = 0.2 (~2 × 10 7 CFU/mL) and 0.01% of Silwet L-77 (LEHLE SEEDS, Round Rock, TX, USA) was added to the cell suspensions. Before inoculation, 4-week-old N. benthamiana, N. tabacum, and 3-week-old watermelon plants were covered with plastic bags for 12 h to maintain high relative humidity. At least three plants have been used for each inoculation in all experiments. The plants were then spray-inoculated using a atomizing minibottle and covered with a plastic bag for 24 h. Inoculated plants were maintained in a growth chamber as described above. Two hours after inoculation, spray-inoculated leaves were harvested, soaked in 5% hydrogen peroxide solution for 2 min, then washed in distilled water at least three times before leaf discs (0.33 cm 2 ) were collected. Bacteria were extracted from nine leaf discs collected from three plants ground in the tube and populations were determined by tenfold serial dilution of cell suspensions followed by spread plating 100 µL aliquots onto NA media amended with rifampicin. Bacterial populations were also determined 3 days and 6 days post-inoculation using the same procedure.

Melon seed transmission assays
Melon cv. Ofir and hybrid cantaloupe/muskmelon (cv. Athena) seeds were used for inoculation of M6-and AAC00-1-derived strains, respectively. Seeds were incubated for 2 h with gentle agitation in 50 mL tubes containing 10 mL suspensions of A. citrulli (~1 x 10 6 and ~1 × 10 7 CFU/mL for AAC00-1-derived and M6-derived strains, respectively). As a control, melon seeds were incubated in sterile distilled water for 2 h. Following incubation, seeds were air-dried in a laminar flow for 4 h. Seeds were sown in 600 mL pots filled with peat (one seed per pot). The pots were kept in a greenhouse at 26 °C-28 °C for 12 days and then the above ground parts of the seedlings were collected and their fresh weight was determined. We have previously shown that seedling fresh weight directly correlates with disease severity in A. citrulli seed transmission assays (Bahar et al. , 2009b). Disease severity was quantified using a 0 to 7 scale, based on the fresh weight of inoculated plants relative to the average weight of non-inoculated control plants (Bahar et al. , 2009b) where: 0 = weight higher than 90% of average control weight; 1 to 5 = weight equal to 76%-90%, 61%-75%, 46%-60%, 31%-45% and 16%-30% of average control weight, respectively; 6 = weight equal to or lower than 15% of average control weight; 7 = dead seedling.
Developing an RNAi construct for silencing the NtSGT1 gene in N. tabacum A partial NtSGT1 gene fragment was amplified by PCR from the cDNA of N. tabacum (cv. Samsun-NN) using primers NtSgt1-H3-RI and NtSgt1-Xba-Sal. The PCR product was digested with Xba I and Hind III to obtain the anti-sense strand DNA or Eco RI and Sal I to obtain the sense DNA fragment, and sub-cloned into the corresponding restriction sites of Topo-Cannibal vector as described previously (Xu et al. , 2011). The derived Topo-Entry construct consisted of a DNA fragment carrying an anti-sense-NtSGT1 DNA fragment, a castor bean intron spacer sequence, and a sense-NtSGT1 DNA fragment. The DNA fragment was released from the Topo-Cannibal vector using Not I, and subcloned into plasmid pORE-E2 (Coutu et al. , 2007) after digestion with Sal I and filled-in with the Klenow enzyme. The derived construct was named pORE-E2-SGT1-RNAi.

Generating N. tabacum-RNAi-NtSGT1 transgenic plants
The pORE-E2-SGT1-RNAi construct was transferred into A. tumefaciens strain LBA4404 by electroporation (Traore and Zhao, 2011). Agrobacterium-mediated leaf disc transformation was performed as described previously (Horsch et al., 1989). Briefly, fully expanded leaves of 4-week-old N. tabacum plants (cv. Samsum-NN) were collected and surface sterilized with 10% sodium hypochlorite. Disinfected leaves were cut with a sterilized razor blade into small leaf discs (about 1 cm 2 ) that were inoculated by incubation in an Agrobacterium suspension (OD 600 = 0.1, ~1 × 10 8 CFU/mL). Inoculated leaf discs were incubated on MS medium (PhytoTechnology Laboratories, Shawnee Mission, KS, USA) in the dark at 25 °C for 2 days-3 days. After co-cultivation, the transformed leaf discs were washed with liquid MS medium and cultured on MS medium supplemented with kanamycin (300 µg/mL), carbenicillin (500 µg/mL), 1-naphthaleneacetic acid (100 mM), and 6-benzylaminopurine (50 mM) for 4 weeks. Regenerated transgenic shoots were further selected on MS medium supplemented with kanamycin (100 µg/mL) and carbenicillin (300 µg/mL). Fully rooted transgenic plants were transplanted into soil and maintained in a growth chamber under conditions of 16 h of light daily at 28 °C, and 8 h dark at 25 °C as described above.

Monitoring the silencing of NtSGT1 by Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR)
Total mRNAs of the pORE-E2-SGT1-RNAi transgenic, and nontransgenic control plants were extracted using the TRIzol reagent (Invitrogen) according to manufacturer's instructions. The total RNA samples were treated with DNase I (Zymo Research, Irvine, CA, USA). First-strand cDNA synthesis was performed using the DyNAmo cDNA Synthesis Kit (Thermo Scientific Inc., Pittsburgh, PA, USA). NtSGT1 transcripts were detected by qPCR with primers NtSGT1-Cter For and NtSGT1-Cter Rev, which amplify an NtSGT1 fragment outside the region used for developing the RNAi construct. The N. tabacum actin gene served as a reference and was amplified using primers Nt_actin For and Nt_actin Rev. PCR was performed using one cycle at 98 °C (3 min), followed by 25 cycles at 98 °C (1 min), 55 °C (1 min), and 72 °C (1 min), and by one cycle at 72 °C (7 min). Similar PCR conditions were also used for checking the mutants of A. citrulli strains ( Supplementary Fig. S1). The PCR products were separated on 0.8% agrose gel, stained with ethodium brominde, and visualized using the Gel-Document Image System (Bio-Rad, Hercules, CA, USA).

Statistical analysis
To determine if there were significant differences in virulence between bacterial strains in in planta assays, treatment means were analysed by the Tukey-Kramer HSD multiple comparison procedure using the JMP software (Version 7, SAS Inst., Cary, NC, USA).

SU PP O R T I N G I N F O R M AT I O N
Additional supporting information may be found in the online version of this article at the publisher's web site:  Table S2. (B and C) The Aave_1548 and Aave_2166 mutants were genotyped using primers flanking the effector gene (B, Aave_1548 ; C, Aave_2166 ). The nptII gene fragment that was used to replace the open reading frames (ORFs) of the effector genes is larger than the replaced ORFs thus, amplification with flanking regions of the mutated effectors give bigger bands than those amplified from the wild type strain. (D) The presence of the nptII gene was confirmed in the mutants using nptII (kan) specific primers (Supplementary Table S2).

Fig. S2
Amino acid sequence alignment of Aave_1548 effectors from A. citrulli AAC00 1 and M6. A small domain that is highly polymorphic between the two Aave_1548 homologues is highlighted in yellow.

Table S1
List of the putative AAC00 1 T3S effector genes, their annotation, and primers used to amplify the open reading frames (ORFs) from AAC00 1 and M6. Table S2 List of additional primers used in this study.