• Open Access

Resistance and biomass in Arabidopsis: a new model for Salicylic Acid perception


*(fax (+34)963877859; e-mail ptornero@ibmcp.upv.es)


Salicylic acid (SA) is an essential hormone for plant defence and development. SA perception is usually measured by counting the number of pathogens that grow in planta upon an exogenous application of the hormone. A biological SA perception model based on plant fresh weight reduction caused by disease resistance in Arabidopsis thaliana is proposed. This effect is more noticeable when a chemical analogue of SA is used, like Benzothiadiazole (BTH). By spraying BTH several times, a substantial difference in plant biomass is observed when compared with the mock treatment. Such difference is dose-dependent and does not require pathogen inoculation. The model is robust and allows for the comparison of different Arabidopsis ecotypes, recombinant inbreed lines, and mutants. Our results show that two mutants, non-expresser of pathogenesis-related genes 1 (npr1) and auxin resistant 3 (axr3), fail to lose biomass when BTH is applied to them. Further experiments show that axr3 responds to SA and BTH in terms of defence induction. NPR1-related genotypes also confirm the pivotal role of NPR1 in SA perception, and suggest an active program of depletion of resources in the infected tissues.


Plants have a sophisticated defence system that is triggered or not depending on the nature of the pathogen. Some plant defences are specialized in necrotrophic pathogens (van Kan, 2006) while others are effective against biotrophic pathogens (Bent and Mackey, 2007). Salicylic acid (SA) is a key molecule in the triggering of plant defences against biotrophs. SA is also relevant for some developmental events (e.g. Vanacker et al., 2001 and Martinez et al., 2004). Despite the importance of this hormone in defence, little is still known about it. In Arabidopsis, (Arabidopsis thaliana) there have been developed transgenic lines that degrade SA (NahG, Friedrich et al., 1995) and also others that produce more SA (c-SAS and p-SAS, Mauch et al., 2001). Moreover, there are two mutants impaired in SA biosynthesis: eds5/sid1 (Nawrath et al., 2002) and sid2/ics1 (Wildermuth et al., 2001). Additionally, there are other mutants with less SA, e.g. eds1 and pad4 (Wiermer et al., 2005). SA biosynthesis is under a positive feedback loop; SA triggers the expression of EDS1 and PAD4, and these genes are required for the expression of the SA biosynthetic genes. The metabolism of SA is also under control (Shah, 2003). Most of the SA present in the plant is conjugated with glucose, forming a pool of temporary inactive SA that can be slowly released in an active form (Nawrath et al., 2005).

The scientific community has made an important effort to find the SA receptor. NPR1 is the only known gene that, when mutated, renders plants insensitive to SA, and yet it is not clear if it is the receptor itself. It has been found in at least four different screenings (Cao et al., 1994; Delaney et al., 1995; Glazebrook et al., 1996; and Shah et al., 1997), which indicates the essential role it plays in SA perception. NPR1 has been shown to accumulate in the cytosol and migrate to the nucleus upon SA perception. In the proposed model, SA triggers the expression of a thioredoxin that acts over NPR1 oligomers, rendering monomers that migrate to the nucleus (Tada et al., 2008). NPR1 is degraded by the proteosome in the nucleus, a process that is required for the activation of defence when SA is present (Spoel et al., 2009).

In parallel with the search for mutants, other biochemical approaches aimed at searching for proteins with SA binding have been adopted. Although some of the candidates have a strong affinity (Kumar and Klessig, 2003), it is likely that none of them is a conventional SA receptor, if such a thing exists.

An intriguing feature of plant defence is the resulting loss of fitness (Heil, 2002). It may seem intuitive that, upon a pathogen insult, the plant produces toxic compounds that negatively affect the plant, but other mechanisms can be proposed. For example, the plant may prioritize defence vs. development, redirecting all available resources to stop invasion. A third option is a ‘scorched earth defence’, i.e. the tissue where the pathogen is perceived is deprived of the elements that the microorganism requires (including oxygen, water, solutes, etc.). SA negatively regulates the effect of auxins, and this hormone is a good candidate to be the vehicle used to reduce plant development when a strong defence is triggered (Wang et al., 2007a).

To find the genes required for SA perception and its consequences, we have had to screen and accurately measure different Arabidopsis genotypes. The exponential growth of the pathogens used (Katagiri et al., 2002) has proved to be a handicap. The relationship between plant defence and development is also affected by the presence of the pathogen in the system. Therefore, we have developed a biological model for the perception of SA in Arabidopsis not based on pathogen inoculation but on the application of benzothiadiazole (BTH). BTH is a biotechnological development of the research in plant defence (Lawton et al., 1996), a chemical analogous to SA that triggers plant defence and biomass loss in a consistent and dose-dependent manner. With this system, we can analyse the natural and artificial variation of Arabidopsis in response to SA. Small differences were found in both cases. Arabidopsis ecotypes have shown no extreme behaviour, and only two mutants have been selected, axr3-1 (A semidominant allele, Ouellet et al., 2001) and npr1. Complementary experiments have proved that axr3 can perceive SA, confirming the unique role of NPR1 and related genes. We also have found that the presence of sni1 in the plant (Li et al., 1999) implies that NPR1 is relevant for a programmed down-regulation of plant metabolism which could affect the pathogen.


An experimental model for SA perception

We are interested in unveiling the signal transduction that starts with SA application to Arabidopsis and results in the triggering of the plant defences. The amounts of SA that trigger plant resistance are close to phytotoxicity, and this is why BTH (Lawton et al., 1996) is commonly used in research. BTH is a chemical analogous to SA with no phytotoxicity. It is commercialized under different names (Actigard® and BION® among others, http://www.syngenta.com). The standard way of measuring SA perception is by means of a western blot of a defence marker (e.g. PR2, Cao et al., 1997), or by monitoring the growth of a inoculated pathogen (e.g. Pseudomonas syringae pv tomato DC3000 (Pto), Katagiri et al., 2002). Figure 1a shows the result of Pto inoculation. Pto is able to grow several orders of magnitude more in mock-treated plants than in BTH 350 μm treated plants. Although the procedure of inoculating and measuring Pto is straightforward (Tornero and Dangl, 2001), it is subject to important variations; small changes in the initial conditions can lead to a lack of reproducibility. Besides, factors that affect pathogen growth also affect Pto measurement. During the experiments, we noticed that BTH-treated plants have less biomass than the mock-treated ones (Figure 1b). This fact has been described in (Heil, 2002) and it is indicated in the label of the commercial product (http://www.syngenta.com). We repeated the experiments without pathogen inoculation and obtained the same results (data not shown); a single 350 μm application of BTH can reduce Arabidopsis biomass. Note that this effect is not exclusive of BTH. It has been described the relationship between the cost of fitness and resistance in absence of pathogen; for instance in the R-gene-mediated resistance (Tian et al., 2003) or constitutive defence mutant (Heidel et al., 2004). Conversely, plants deficient in SA level increase yield production (Abreu and Munne-Bosch, 2009), providing that no pathogen is present.

Figure 1.

 BTH increases disease resistance and decreases biomass accumulation. (a) Arabidopsis plants were pretreated with either mock or benzothiadiazole (BTH) 350 μm and then inoculated with Pseudomonas syringae pv tomato isolate DC3000 (Pto) 1 day later. Three days later, the bacteria (measured as Logarithm of colony forming units per plant) were measured. (b), Plant weight of (a) before bacterial extraction, in mg of fresh weight. (c), The same effect after considerable optimization that includes four treatments (see Experimental procedures). All panels show the average and standard deviations, and at least three independent experiments were performed with similar results.

As a single treatment lacked reproducibility and there was statistical overlapping between mock vs. BTH-treated plants, we tried different applications and treatments. Briefly, we applied BTH by imbibing, drenching, spraying, and in vitro culture (data not shown). The optimal method consists in spraying BTH four times for 2 weeks (see Experimental procedures) and recording plant weight when the plants are 3-weeks old. No special limit was observed after four treatments; up to eight treatments were applied during four weeks with no evident toxicity (data not shown). Increasing the number of treatments improves the difference between mock vs. BTH-treated plants when the plants have enough room to grow. When the treatment finished, Col-0 plants outgrew the treatment and were able to set seeds.

The results are more clearly shown when the ratio between BTH and mock-treated plants is used (Figure 1c). Different BTH concentrations were used to find the optimal option, starting with 350 μm (used in Figure 1a) and diluting by a factor of ten. BTH concentrations higher than 3.5 μm still produced a measurable effect on Arabidopsis biomass, whereas lower BTH concentrations failed to differentiate mock and BTH-treated plants. Therefore, 350 μm is the standard BTH concentration or ‘High BTH’, and 350 nm the subclinical BTH concentration or ‘Low BTH’.

Phenotypes of the model

The differences in plant biomass caused by High BTH treatments are numerically significant, corresponding to the strong phenotype of Figure 2a. Figure 2a shows Col-0 treated with mock (left) and High BTH (right) the difference in plant size is worth noting, whereas the number of leaves is similar. Therefore, a visual inspection can discern in most cases whether a genotype responds or not to BTH. This is a simple way to evaluate SA perception and characterize the response to BTH in the Col-0 ecotype. There were no observable macroscopic plant lesions, so we looked for microscopic lesions. Trypan blue staining pinpoints cell death and membrane damage (and fungi if present, Keogh et al., 1980). Figure 2b,c,d show the Trypan Blue staining of cotyledons from plants treated with mock, Low BTH, and High BTH treatments respectively. While subclinical BTH concentrations produced no measurable effects, standard amounts of BTH triggered Program Cell Death in a small number of cells. Callose depositions are a hallmark of defence induction, and are easily seen with aniline blue under ultraviolet light (Conrath et al., 1989). Therefore, cotyledons of mock, Low BTH and High BTH-treated plants were stained with aniline blue (Figure 2e,f,g, respectively. Figure 2h,i,j show the same cotyledons exposed to visible light). The result is that mock and subclinical BTH concentrations do not produce callose deposition. Standard BTH concentrations, on the other hand, produce abundant callose depositions, of several sizes and distributions. A 3.3′-diaminobenzidine stain showed no difference in Reactive Oxygen Species (ROS) at the time of the sampling (data not shown).

Figure 2.

 The continuous triggering of plant resistance produces a distinctive macroscopic and microscopic phenotype. (a) Macroscopic phenotype of plants either treated with mock (left) or with BTH 350 μm (right) at the same time as Figure 1c. BTH-treated plants have the same number of leaves as mock-treated plants, and are able to survive and set seeds. (b) (c) and (d) correspond to Trypan blue stains, unveiling cell death and membrane damage. (e), (f) and (g) show Aniline blue stains under ultraviolet light, which detects callose depositions. (h), (i) and (j) are the same micrographs under visible light. (b), (e) and (h) are from representative plants treated with mock, (c), (f) and (i) are from BTH 350 nm treated plants, and (d), (g) and (j) are from BTH 350 μm treated plants. Only BTH 350 μm produces microscopic cell death in few and isolated cells [dark blue staining outside the veins in (d)], and triggers plant defence, as observed in the callose depositions [fluorescent in (g)]. Three independent experiments were performed with similar results.

Salicylic acid is a hormone with a fine-regulated homeostasis. Thus, there is evidence of a positive feedback loop in SA synthesis and of negative regulation upon SA perception (Shah, 2003). There are two possible explanations for BTH effects on biomass: BTH could trigger the SA positive feedback loop increasing thus SA concentration, or it could be the responsible of the effect by itself. To check these hypotheses, the amounts of SA (free and total) in mock, Low BTH, and High BTH-treated plants were measured 3 days after the last treatment (Figure 3a, the order of magnitude of all these values are in agreement with reported concentrations (Defraia et al., 2008). Low BTH treatment does not change dramatically the amount of total and free SA, while High BTH treatment produces a decrease in free SA and a strong decrease in the total amount of SA. The subclinical amounts of BTH do not induce the expression of the marker PR1 (Figure 3b), a standard stress marker (Uknes et al., 1992), nor enough resistance to be detected in Pto growth curves (data not shown). Standard BTH concentrations induced a strong PR1 expression, even if the western blot was repeated with only mock and subclinical BTH treatments to avoid a possible signal masking because of the strong High BTH signal (data not shown).

Figure 3.

 SA accumulation and defence induction upon BTH application. (a) Quantification of SA upon mock, BTH 350 nm, and BTH 350 μm treatments as described in Figure 1c. Both free and total SA (i.e. glucosylated derivates released after hydrolysis plus the free SA) were measured, showing the average and standard deviations of three samples. (b) Western blot for PR1. This defence marker was immunodetected in samples from the same experiments as in (a). The arrow points to the expected size of PR1 (14 kDa). Three independent experiments were performed with similar results.

Natural variation and SA perception

Once the right conditions were set, we intended to assess whether Col-0 was the best ecotype to work with. Figure 4 shows the analysis of two sets of ecotypes and Col-0. Figure 4a corresponds to a nuclear core collection of 48 ecotypes (McKhann et al., 2004), while Figure 4b,c show a set of 96 ecotypes (Nordborg et al., 2005). Col-0 is a valid representative of the ecotypes tested; in the three panels it ranked in the middle of the ecotypes (between 40th and 56th percentiles) when ordered by percentage of plant fresh weight (PFW). Some ecotypes like Col-0, Ws-0, Laer-0 and No-0 were repeated with different stocks (e.g. Col-3, Col-4, Col-5, etc.), because they are the background of mutations or are used for mapping. None of them behaved in a different way (data not shown).

Figure 4.

 Arabidopsis ecotypes tested show a similar phenotype. Two collections of ecotypes were tested as described in Figure 1c. (a) Col-0 and the McKhann collection (McKhann et al., 2004) ranked for its percentage of fresh weight. (b) and (c) Col-0 and the Nordborg collection (Nordborg et al., 2005) were measured in two separate lots. The full names of the ecotypes shown, as well as other ecotypes tested, are listed in Table S1. None of the ecotypes tested shows an extreme behaviour under these conditions, and Col-0 ranked between 40th and 56th percentile in the three panels. Two independent experiments were performed with similar results.

Another option in Natural Variation is to search for quantitative trait loci (QTLs) in mapping populations. This can be performed even if the parentals have a similar behaviour, a phenomenon called transgression (Koornneef et al., 2006). We analysed seven recombinant inbreed lines (RILs) available at the beginning of this research searching for transgression. The RILs were: Col-0 × Nd-1 (Deslandes et al., 1998), Col-gl1 × Kas-1 (Wilson et al., 2001), Cvi-1 × Laer-2 (Alonso-Blanco et al., 1998), Laer-0 × Sha-0 (Clerkx et al., 2004), Bay-0 × Sha-0 (Loudet et al., 2002), Col-4 × Laer-0 (Lister and Dean, 1993) and Laer-0 × No-0 (Magliano et al., 2005) (Figure 5 and data not shown). There are three QTLs detected only in the mock-treated plant’s fresh weight (Colgl1 × Kas-1, Laer-0 × Sha-0 and Laer-0 × No-0, Figure 5b,d,e respectively). There is, however, no significant QTL specific of the response to BTH in terms of fresh weight.

Figure 5.

 There are no significant QTLs in the tested populations specific to SA perception. Plants were treated with either mock or BTH 350 μm as described in Figure 1c. The output showed is the likelihood of a QTL (in logarithm of odds; LOD, in the Y axis) in a particular region of the genome (X axis). The horizontal line shows the threshold of significance. The continuous lines show the QTLs for mock, and the dotted line the QTLs for BTH treatment. The populations analysed were (a) Col-0 × Nd-1, (b) Col-gl1 × Kas-1, (c) Cvi-1 × Laer-2, (d) Laer-0 × Sha-0, (e) Laer-0 × No-0, (f) Col-4 × Laer-0.

Most signal transductions do not affect SA perception

The next step was to analyse the wealth of information generated in the form of mutants. SA biosynthesis is regulated by a positive feedback loop, so the mutations related to SA were the first objective. Thus, we assayed the mutant that failed to perceive SA; npr1, mutants of SA biosynthesis; eds5 and sid2, transgenic lines with altered SA content (NahG less SA, and c-SAS more SA); and mutants with a down-regulation of SA biosynthesis, eds1 and pad4 (Figure 6a). Only npr1 failed to respond to SA. This clear result prompted us to keep npr1 as a negative control, and to extend the list of mutants in defence (Figure 6a,b). Then, we tested mutants in basal resistance (either more resistant or more susceptible), Systemic Acquired Resistance (SAR), specific resistance, and non-host resistance (see Table S1). None of the tested mutants in defence, except npr1, differed from the wildtype (wt) in their response to BTH.

Figure 6.

 Analysis of a collection of mutants points out only to two candidates for SA perception. All the mutants were tested as described in Figure 1c. The complete list is shown in Table S1. (a) and (b) show defence mutants, (c) corresponds to mutations in Auxin and Light signalling, (d) mutations in Abscisic Acid, Ethylene and Reactive Oxygen Species, and (e) mutations in Jasmonic Acid and/or response to necrotrophs. 35FM stands for 35S:FMO1, 35DIR is 35S:DIR1, phyA/B is the double phyA phyB, 35HAB is 35S:HAB1, abi/hab is abi1-2 hab1-1, 35ERF is 35S:ERF1, atrb-D is atrboh-D, and atrb-F is atrboh-F. Three independent experiments were performed with similar results.

Salicylic acid signal transduction has been reported to crosstalk with several signal transductions (Lopez et al., 2008), being Jasmonic Acid, Ethylene, Abscisic Acid, Auxins, Light and ROS the most commonly cited. Therefore, the response to BTH of a representative set of mutants in each of these pathways was measured. For the Auxin pathway, 19 mutants were tested (Figure 6c), and only axr3 did not respond to BTH in a consistent manner. Note that the allele used in this work is axr3-1, a semidominant mutation that enhances the stability of the protein (Ouellet et al., 2001). Mutants in other pathways, like Light (Figure 6c), Abscisic Acid (Figure 6d), Ethylene (Figure 6d), ROS (Figure 6d) or Jasmonic Acid and/or response to necrotrophs (Figure 6e), had a response to BTH similar to that of wt. A complete list of the mutants tested is provided in Table S1.

axr3 and npr1 show a distinct response to SA

The conclusion of Figure 6 and other data not presented is that from a total of 98 mutants tested, only two did not respond to BTH; npr1 and axr3. NPR1 is a gene clearly involved in SA perception, but the result of axr3 was unexpected. While it was tempting to discard axr3 because of the small size of this mutant, other small mutants like cpr1 (Bowling et al., 1994), showed percentages of fresh weight in the same order of magnitude as the wt (Figure 6a). Therefore, a detailed characterization of axr3 in terms of response to SA and BTH was performed. Figure 7a shows Pto growth in Col-0, npr1 and axr3 pretreated with mock or High BTH. BTH is clearly able to trigger defence in axr3, as opposed to the effect caused in npr1. The levels of the PR1 protein were determined by western blot (Figure 7b) in plants either treated with mock or BTH 350 μm and proved to be basically the same. While npr1 fails to induce this defence marker upon High BTH, axr3 is able to increase the expression of this defence protein. Note that in axr3 plants there is a small but detectable amount of PR1 even in the mock treated ones.

Figure 7.

axr3 can sense SA and BTH. (a) Pto growth in Col-0, npr1 and axr3 treated either with mock or BTH 350 μm, as described in Figure 1a. The resistance triggered by BTH is significant (Student’s t-test, P < 0.05) in both Col-0 and axr3, not in npr1. (b) Western blot for PR1 of the same experiment as (a), but prior to bacterial inoculation. The arrow points to the expected size of PR1 (14 kDa). (c) Phenotype of the same genotypes in MS plates supplemented with SA 500 μm. npr1 plants do not perceive SA and therefore can not avoid accumulation to toxic levels. Col-0 and axr3 can perceive SA and are able to grow in this medium. Three independent experiments were performed with similar results.

An interesting feature of plants mutated in npr1 is that they fail to regulate the levels of SA (Cao et al., 1997). When growing npr1 in MS plates supplemented with SA 500 μm, the cotyledons are bleached and the plant is unable to grow (Figure 7c). The easiest interpretation is that npr1 fails to perceive SA, and therefore it is unable to trigger SA degradation and SA accumulation has deleterious effects. Col-0 and axr3 plants, on the other hand, grow in plates containing SA 500 μm (Figure 7c).

npr1-related genes and SA perception

Then, we focused on npr1 and related genes. The previous experiments were repeated with npr1-1, but there are 11 alleles of npr1 (Cao et al., 1994; Delaney et al., 1995; Glazebrook et al., 1996; and Shah et al., 1997). We assayed four of them (Figure 8), and––with some variation––all the alleles tested show no response to BTH in terms of PFW. There are no mutants with an increasing sensitivity to SA; therefore, the next best candidates are the transgenics that overexpress NPR1. 35S:NPR1 is an overexpression of NPR1, and the plants can more strongly perceive BTH, as reported in the literature (Cao et al., 1998). 35S:NPR1:HBD is a version of NPR1 fused to the glucocorticoid receptor in a npr1 background (Wang et al., 2005). The result is a protein not subjected to the nuclear vs. cytoplasm traffic, vital to its function in SA perception (Dong, 2004). Figure 8 shows that the mere presence of NPR1 in the cytosol is not enough to trigger response to BTH and the nuclear localization is required.

Figure 8.

 Behaviour of genotypes related to NPR1 in SA perception. Alleles of npr1, loss of function of related genes and transgenic plants that overexpress the protein were assayed as described in Figure 1c. 3xtga stands for the triple mutant tga6 tga2 tga5 and NPR1HBD for the transgenic 35S:NPR1:HBD. Three independent experiments were performed with similar results.

There are five genes in Arabidopsis with a high homology to NPR1 (Liu et al., 2005). NPR3 and NPR4 have been reported to play a key role in plant defence (Zhang et al., 2006) and mutations in BOP1 and BOP2 affect the identity of the floral organs and the shape of the leaves (Ha et al., 2007; McKim et al., 2008). Plants from these two double mutants respond to BTH in the same way as in wt (Figure 8).

Regarding its biochemistry, NPR1 has been shown to interact in yeast two-hybrid with two sets of proteins, TGAs (Zhang et al., 1999) and NIMINs (Weigel et al., 2001), and in vivo with some of them. T-DNA insertions in TGA1 and TGA7 show small but consistent differences between these mutants and wt in their response to BTH (Figure 8). This small effect is more noticeable when a triple mutant tga6 tga2 tga5 is used (Zhang et al., 2003), and the plants show an intermediate macroscopic phenotype (data not shown).

The npr1 phenotypes are quite straightforward, which has led to a number of suppressor screenings. One of these suppressors is sni1 (Li et al., 1999), and the double sni1 npr1 regains the ability to activate defences upon BTH application. Interestingly, the double sni1 npr1 does not behave as a suppressor in our system (Figure 8). We also tested T3 seeds from insertions in the homologue NPR2, the interactors NIMIN1, NIMIN2, NIMIN3, TGA3 and TGA4 and the suppressors SSI2 (Shah et al., 2001) SON1 (Kim and Delaney, 2002), and SNI1 but no npr1-like phenotype was observed in the segregating families (data not shown).


SA perception and plant fitness

Salicylic acid is a necessary hormone in plants for full resistance against biotrophic pathogens such as Pto. While the amount of SA can be measured in the laboratory (Defraia et al., 2008), for the quantification of SA perception we usually rely on the growth of the pathogen we are interested to start with. This is a potential circular problem, since we use a tool to answer a question that affects the tool.

Another potential problem is the nature of the pathogen. Pathogens like Pto grow exponentially, and small differences in the input lead to considerable differences in the output. For example, note the difference between the growths of Pto in Figure 1a vs. Figure 7a. There are alternatives, like immunodetection of defence markers (Uknes et al., 1992; Figure 2 and Figure 7), or measurement of phytoalexin accumulation (Glazebrook and Ausubel, 1994). These alternatives can produce quantitative data, but are not suitable for high throughput assays.

One side effect of several resistances is their negative effect on plant fitness (Heil, 2002). In general, the more resistant an individual is the less fit it is to compete when the pathogen is not present. There are several hypotheses to explain this fact. The first one is that, as the plant produces molecules that eventually stop the growth of the pathogen, it is plausible that the same molecules affect the plant. Other alternative is an economic consideration; the triggering of defence genes involves the use of resources that have to be obtained from normal plant growth. A somewhat related argument is the ‘scorched earth’ defence, where the plant tries to stop the infection by starving the pathogen.

Resistance and fresh weight are inversely correlated

In the case of BTH, a single application can produce measurable effects in terms of plant fresh weight (Figure 1b). This subtle effect (measured 4 days after a single BTH treatment) was optimized for measurement and screen. While different ways of applying BTH produce visible differences, the best condition for our goals is to spray the plants with BTH four times on separate dates (see Experimental procedures). This procedure provides us with an accurate quantification of genotypes such as the mutants and ecotypes described above (Figure 1c). But most importantly, it gives us a tool to do a screening (Figure S1). There is a correlation between size and fresh weight that can be used to search for new mutants in a high-throughput fashion, and the next generation can be retested with two simple measurements (fresh weight of mock vs. BTH-treated plants). In principle, this model is analogous to screen for mutants in auxin perception with plates of 2,4-D (Maher and Martindale, 1980). 2,4-D is more stable and has a stronger effect on the plant than the endogenous auxin, like BTH vs. SA. The main differences are that BTH does not work in plates, and it is not lethal. But in both cases we can recover mutations impaired in the perception of the hormone by using an analogue and a set of extreme conditions (Mockaitis and Estelle, 2008).

To use the biological model, several steps must be taken. First we need to characterize plant response in terms of macroscopic, microscopic and molecular phenotypes, to be sure that the observed effects on fresh weight correspond to the activation of plant defences. Second, the choice of the ecotype to be used, because Col-0 may not be the best background. And third, there is the question of genetic specificity; the biological model proposed should not mislabel mutants that affect the growth of the pathogen as a mutant in SA perception (e.g. cpr1,Figure 6a), and it should correctly label npr1 as defective in SA perception.

The response to BTH in terms of PFW is dose-dependent (Figure 1c). The highest BTH concentration tested is 350 μm, a concentration frequently used in Arabidopsis (Lawton et al., 1996). To put it in context, this corresponds to approximately nine times the recommended dose for Pto infection in tomato (http://www.epa.gov), but it is between two and six times lower than SA concentrations used in Arabidopsis (1 mm, e.g. Cao et al., 1994; 2 mm, e.g. Aviv et al., 2002). The loss of fresh weight can be detected as low as 3.5 μm, but not at 350 nm. Low BTH is unable to trigger cell death or callose deposition (Figure 2c,f). High BTH, on the other hand, is able to cause cell death in a small number of cells (Figure 2d), as it is also reported in the literature for SAR (Alvarez et al., 1998) and labelled as micro-HRs. Although a plausible hypothesis was that these micro-HR sites are similar, we did not observe any oxidative burst (data not shown). It is therefore possible that the micro-HRs are different, and while in SAR they are caused by oxidative burst, the cell death shown in Figure 2d is caused by other effector. Another alternative is that in our model a transient oxidative burst occurs immediately after the treatments, but it disappears when the tissue is stained (3 days after the last treatment). In any case, the small number of cell deaths observed does not account for the difference in PFW, and it seems an effect rather than a cause of resistance. While, it has been reported that BTH by itself does not strongly trigger callose depositions, a second mock treatment (water infiltration) after BTH had the ability to do so (Kohler et al., 2002). Consistently with this result (we sprayed the plant several times), there is a strong callose staining with High BTH.

The next step was to analyse the molecular events that occur in this system. The amount of SA is under the control of feedback loops, positive in SA biosynthesis and negative in SA accumulation (Shah, 2003). Therefore, it was relevant to measure the amount of SA in this system. High BTH produced a strong reduction in the amount of total SA (Figure 3a). There is a small reduction in the amount of free SA, but it is clear that the plant responded to High BTH with a reduction of the conjugated form of SA (mainly glucoside) (Nawrath et al., 2005). Therefore, this constitutes additional evidence in favour of a negative feedback loop that regulates the accumulation of SA. The other piece of evidence is the amount of SA in npr1 (Cao et al., 1997). This mutant has more SA than the wt, both in mock and pathogen-inoculated plants. Another form of this phenotype is the low tolerance of npr1 plants to SA in vitro (Figure 7c). It cannot detect SA, and therefore it cannot avoid SA accumulation and toxicity. The same mechanism impaired in npr1, is triggered continuously in the plants treated with High BTH in Figure 2, in an effort to maintain the homeostasis of the SA levels. The result is that the levels of total SA are depleted. It has been reported that a single application of BTH increase SA content (von Rad et al., 2005). We speculate that this discrepancy is due––as it happens with the callose staining and with the fresh weight––to the repetition of treatments (four in our case vs. one), rather than the difference of age (8 days in the first treatment vs. 4–6 weeks) or growth conditions.

The detection of the defence marker PR1 (Figure 3b) and Pto growth (Figure 1a and data not shown) confirms that fresh weight loss and disease defence are closely correlated, as low concentrations that do not produce fresh weight loss, do not trigger defence. Correspondingly, high concentrations are able to produce both phenotypes.

SA perception in natural variation

Before starting the search for new mutants, the best genotype has to be chosen. Col-0 is the ecotype most widely used for mutant screening (http://www.arabidopsis.org), but it could be an extreme ecotype in response to BTH. Figure 4 shows that Col-0 is a representative Arabidopsis ecotype, because it ranks between the 40th and 56th percentile among the collections tested. Another reason for these experiments was to search for natural variation, but there is no extreme ecotype in the response to BTH.

We also searched for transgression in seven RILs (Figure 5 and data not shown), but found none. The three QTLs found are only relevant to the differences in growth when a mock treatment is applied, but there is no difference in the response to BTH. This does not mean that there are not variations in the SA response (van Leeuwen et al., 2007), but that none was both significant and specific to SA perception with the populations and system under study.

SA perception in defence and signalling mutants

From the comprehensive list of mutants tested, there is no evidence of desensitization. That is, mutants that have more SA than their corresponding wt (e.g. c-SAS and cpr1) are still able to respond to exogenous BTH applications (Figure 6). A direct consequence is that we can assay genotypes that are more resistant to bacteria and unequivocally discriminate if it is due to an enhanced SA perception. So far we have found no evidence for such genotype, with the exception of 35S:NPR1 (see below). Regarding the different kind of defences, mutants in SAR, basal, specific (or gene-for-gene) and non-host resistance were tested and found not to be different from the wt, with the exception of npr1, as discussed below.

The more we study plant biology, the clearer it becomes that everything is interconnected. If two decades ago plant defence and development could be seen as two separate programs, evidence in the last years reveals a much more intricate signal network with complex interactions. Thus, there are reports on the interactions between SA and Auxins, Light perception, Ethylene, Jasmonic Acid, Abscisic Acid and ROS, among others (reviewed by Lopez et al., 2008). Mutations in pathways different from Auxin do not have a measurable impact on SA perception when measured as described. Regarding Auxins, only axr3 does not respond to BTH in weight, and there is no visible difference between mock and BTH-treated plants (data not shown). AXR3 belongs to the family of IAAs, genes that are rapidly induced with auxins, and behave as activators or repressors of the auxin response (Reed, 2001). The allele of axr3 used is a semidominant mutant that stabilizes the protein, causing an increase in auxin perception and phenocopying the overexpression of the wt protein (Ouellet et al., 2001). It is tempting to speculate that AXR3 is the link between defence and development.

In favour of this hypothesis, there are solid evidences of the interaction between SA and Auxins (Wang et al., 2007a), and the overexpression of AXR3 reproduces the axr3 phenotype (Reed, 2001). Thus, the phenotype that responds to BTH could be explained by an increase in the amount of the AXR3 protein. However, this hypothesis has serious drawbacks. AXR3 is slightly repressed under pathogenic conditions (http://www.genevestigator.com), which does not fit with a prominent role in the response to BTH. Mechanistically, exogenous Auxin applications reduces SA perception (Wang et al., 2007a). But axr3 has Auxin hypersensitivity, so instead of sensing more SA, it should perceive less SA, which contradicts the model. A closer examination proves that axr3 is indeed able to perceive SA and BTH, as measured by Pto growth, western blot of PR1, and tolerance to SA in plates (Figure 7). This perception is slightly attenuated (Figure 7a,b), as expected by the interaction between Auxins and SA.

The second hypothesis is that the small size of the plant does not allow it to lose weight, as it is already at minimal levels. The average weight of axr3 in mock is less than Col-0 with BTH in Figure 6c, while in other replicates both weights were similar (data not shown). The difference with respect to the first hypothesis is that the small fresh weight is not related to a defence mechanism. The results of our experiments support this second hypothesis.

NPR1-related genotypes mark the relationship between plant defence and development.

NPR1 is a gene necessary for SA perception (Figure 6), among other roles in plant defence (Pieterse and Van Loon, 2004) and development (Vanacker et al., 2001). The extreme npr1 phenotype in response to BTH (Figure 6) is not allele specific, because the available alleles behave in the same way. It is worth mentioning that the npr1-3 allele is still functional for the so-called Induced Systemic Resistance (Pieterse and Van Loon, 2004). In fact, the overexpression of NPR1 fused to the glucocorticoid receptor (35S:NPR1:HBD in Figure 8) reproduces the same phenomenon, i.e. a functional NPR1 protein that is unable to migrate to the nucleus. Therefore, the response to BTH is dependent on the NPR1 protein acting in the nucleus. The overexpression of NPR1 increases sensitivity to SA and its analogues in terms of pathogen growth and defence markers (Cao et al., 1998; Friedrich et al., 2001), and we can reproducibly detect this enhanced SA perception (Figure 8).

In the Arabidopsis genome there are five genes with high homology to NPR1 (Liu et al., 2005). Certain functional redundancy could exist in the genes of this family; therefore, we assayed loss of function mutations in these genes. Fortunately, there are two double mutants available, npr3 npr4 and bop1 bop2, and none of them is consistently different from wt. In the case of NPR2, T3 seeds from a T-DNA insertion (Table S1) were found to be like wt (data not shown). Therefore, there is no measurable functional redundancy, at least in a NPR1 wt background.

NPR1 interacts with proteins from two families of genes, TGAs (Zhang et al., 1999) and NIMINs (Weigel et al., 2001). TGAs are a subclass of the family of bZIP transcription factors (Jakoby et al., 2002) that physically interact with promoters of PR genes (Johnson et al., 2003). The transcriptional activation of these PR genes is dependent on the interaction of the TGAs with NPR1 (Després et al., 2000). There are reports about functional redundancy in this family (Jakoby et al., 2002).

In our model, T-DNA insertions in tga1 and tga7 have a small but measurable phenotype (Figure 8), but the best indication of the significant role of this gene family in SA perception is the phenotype of the triple mutant tga6 tga2 tga5 (Zhang et al., 2003; Figure 8). In this case the phenotype is visible to the naked eye (data not shown). T3 seeds from T-DNA insertions in TGA3 and TGA4 (Table S1) were phenotypically similar to wt (data not shown).

NIMINs are a family of three small genes, and their proteins interact in vitro with NPR1. Mechanistically, NIMIN genes would act as repressors of SA signalling (Weigel et al., 2001). T3 seeds from T-DNA insertions in NIMIN1, NIMIN2 and NIMIN3 (Table S1) were found to behave like wt (data not shown).

NPR1 is the only gene necessary for SA perception, and several suppressor screenings have been carried out to identify other players (Li et al., 1999). T3 seeds from T-DNA insertions in SSI2, and SNI1 behave like wt (data not shown). Interestingly, the double sni1 npr1 does not behave as a suppressor in our system (Figure 8). SNI1 encodes a nuclear protein rich in leucine and it is assumed to be a negative SAR regulator (Li et al., 1999). We were able to confirm the suppression of the npr1 phenotype by sni1 in Pto growth curves (data not shown), but not in weight.

An obvious hypothesis is that the signal that goes from SA to NPR1 is genetically divided into two; one is repressed by SNI1 and activates defence genes (e.g. PR1), causing the measurable reduction of the infection. The other one is SNI1 independent, and reduces the growth of the plant. This branching could be achieved through different signal thresholds, since sni1 induces defence at lower concentrations of SA analogues (Li et al., 1999) both in wt and in npr1. In any case, the evidence that a genotype produces defence (PR gene expression included) with no loss of fresh weight contradicts the first two hypotheses presented to explain the interaction between plant defence and development (‘defence is toxic’, and ‘defence is expensive’). Thus, the third hypothesis (‘scorched earth defence’) is favoured by the results presented here. In other words, the plant has two programs: active synthesis of defences and active depletion of nutrients.

Experimental procedures

Inoculation and plant treatment

For all the experiments, Arabidopsis thaliana was sown in small pots, kept at 4 °C for 3 days and then transferred to growing conditions under a short-day regime [8 h of light (150 μmol/m2/s) at 21 °C, 16 h of dark at 19 °C)]. The treatments, inoculations, and sampling started 30 min after the initiation of the artificial day to ensure reproducibility. Pseudomonas syringae pv. tomato DC3000 (Pto) containing pVSP61 (empty vector) were maintained as described (Ritter and Dangl, 1996). The bacteria were grown, inoculated and measured as described (Tornero and Dangl, 2001) with minor changes. Trypan Blue and Aniline Blue staining were performed as described (Conrath et al., 1989; Tornero et al., 2002, respectively). For all the experiments, three independent treatments were performed (three independent sets of plants sown and treated on different dates), only two in the case of the large collection of ecotypes.

BTH and fresh weight

Benzothiadiazole (BTH, CGA 245704), in the form of commercial product (Bion® 50 WG, a gift from Syngenta Agro SA, Spain) was prepared in water for each treatment and applied with a household sprayer. When indicated, a mock inoculation of distilled water was performed. The treatments were conducted on the 8th, 11th, 15th, and 18th day (day 0 is when plants are transferred to growing conditions), and the weight of the plants recorded on the 21st day. For each genotype and treatment, 15 plants were weighed in three groups of five. The mock treatment was considered to have a value of 100, and the average and standard deviation of the percentage of the fresh weight resulting from the BTH treatment are represented.

Western blot

Immunodetection of PR1 protein was carried out as described (Wang et al., 2005), using an Amersham ECL Plus Western Blotting Detection Reagents (GE HealthCare, Little Chalfont, UK). The second antibody was a 1:25 000 dilution of Anti-Rabbit IgG HRP Conjugate (Promega, Madison, USA). Chemiluminescent signals were detected using a LA-3000 Luminescent Image Analyser (Fujifilm Life Science, Stamford, CT, USA).

SA in plates and in planta

Arabidopsis seeds were surface-sterilized for 10 min in 70% ethanol and for 10 min in commercial bleach. Then, five washes were performed with distilled water and the seeds were distributed on agar plates. The medium contains 0.5× Murashige and Skoog salts (Duchefa BV, Haarlem, the Netherlands), 0.6% (w/v) Phyto Agar (Duchefa), 2% (w/v) sucrose, with or without SA 500 μm (final concentration). The result was evaluated 10 days after transferring to growing conditions. For the measurement of SA in planta, three samples of 250 mg were frozen in liquid nitrogen. SA extraction was performed as described by (Mayda et al., 2000).

QTL mapping

Plants of Arabidopsis were treated with either mock or benzothiadiazole (BTH) 350 μm as described above. The weight of five plants per line and the genotype of each line were used as input for the program WinQTLCart (Wang et al., 2007b), that calculates the probability that a QTL is link to a particular region of the genome. The populations analysed were: Col-0 × Nd-1, 98 lines (Deslandes et al., 1998); Col-gl1 × Kas-1, 115 lines (Wilson et al., 2001); Cvi-1 × Laer-2, 50 lines (Alonso-Blanco et al., 1998); Laer-0 × Sha-0, 114 lines (Clerkx et al., 2004); Bay-0 × Sha-0, 162 lines (data not shown, Loudet et al., 2002); Col-4 × Laer-0, 85 lines (Lister and Dean, 1993); and Laer-0 × No-0, 135 lines (Magliano et al., 2005).


This work was supported by a BIO2006-02168 grant of Ministerio de Ciencia e Innovación (MICINN) to PT, a JAE-CSIC Fellowship to JVC and a FPI- MICINN to AD. We thank the English translation service of the Universidad Politécnica de Valencia and the Statistics Service (CTI-CSIC). We appreciate the seeds provided by a great number of colleagues and the BTH provided by Syngenta. Thanks also to Carlos A. Blanco for advice in QTL mapping, and to José León and Pablo Vera for useful advice.