cDNA-AFLP Reveals a Striking Overlap in Race-Specific Resistance and Wound Response Gene Expression Profiles

Rapid Alterations in Cf9 Tobacco Transcript Levels in Response to Avr9

Cf9 tobacco cell cultures were treated with IF containing Avr9 (IF[Avr9⁺]) or lacking Avr9 (IF[Avr9⁻]) or pretreated with DPI for 5 min before adding IF(Avr9⁺). Cells were harvested after 30 min, and RNA was extracted and used for cDNA-AFLP. In experiments conducted simultaneously, Avr9-dependent AOS production was observed. This production could be inhibited by adding DPI, as shown by Piedras et al. (1998) (data not shown). Gene expression was surveyed in two independent sets of cells using 512 primer combinations for selective amplification. A section of a typical AFLP gel used in this analysis is shown in Figure 1A. The majority of bands showed no change in intensity between the three treatments. Changes in the intensity of individual bands did not affect others in the same lane, indicating that product accumulation was not affected by the concentration of individual substrates in the reaction. The AFLP fragments ranged in length from 50 to 400 bp; for each primer combination, 50 to 70 bands were observed, with ∼30,000 fragments being analyzed. Altered expression patterns were observed for 290 fragments (Figure 1B); of these, 273 were induced by Avr9. Only 4% of the changes in expression were inhibited by DPI, indicating that most rapid changes are independent of AOS.

Control experiments on 30 fragments confirmed that their differential expression required both Avr9 and Cf-9 (Figure 1C). No induction was observed in nontransformed tobacco cell cultures. Treatment with the chemically synthesized Avr9 peptide was sufficient to produce the differential expression observed after IF(Avr9⁺) treatment, confirming that Avr9 was responsible for the changes (Figure 1C). These controls demonstrate that the altered gene expression is Cf-9 and Avr9 dependent.

Compilation of Sequences from Induced and Repressed cDNA Fragments

The differentially expressed fragments were excised from the gels, reamplified by polymerase chain reaction (PCR), and sequenced. DNA sequences were obtained for 260 fragments. The other sequences were a mixture of PCR products and could not be directly sequenced. The sequences were compared with those in the GenBank database using the BLAST program (Altschul et al., 1997). Sequence similarity was found for 37 induced and five repressed sequences (Table 1). These sequences were similar to those of protein kinases, transcription factors, calcium binding proteins, a protein phosphatase, a 13-lipoxygenase, and other proteins. Their roles in signal transduction suggest that many Avr9-induced proteins have signaling functions. The sequences of the majority of the fragments showed no homology to known sequences. They may represent previously uncharacterized genes, or the AFLP fragments may have been too short to reveal significant homology. The sequence data for the fragments listed in Table 1 have been submitted to GenBank, and their accession numbers are listed in Table 1.

Time-Course and Inhibitor Studies of Selected cDNA Fragments

We focused our attention on 18 cDNA-AFLP fragments induced independently of AOS. These were selected based on their homology to known genes or because they were strongly induced. Their expression patterns were examined in Cf9 cells during a 4-hr time period after IF was added. All of the cDNA fragments were induced by IF(Avr9⁺) (Figure 2) but not by IF(Avr9⁻) (data not shown), but their patterns of expression varied in strength and timing. Fragments 1 and 111 were rapidly but transiently induced within 15 min, with maximum expression after 30 to 60 min; expression decreased after 2 to 4 hr. Fragment 4 showed very transient expression; it was induced 30 min after elicitation, but expression had decreased by 1 hr. The other 15 fragments were also induced rapidly, but their induction was more sustained and transcript levels remained high for at least 4 hr. These patterns were reproducible in at least two independent experiments.

Most of the genes were induced in the presence of DPI, suggesting that AOS are not intermediates in the pathway to gene induction and that an alternative pathway must be involved. The protein synthesis inhibitor cycloheximide (CHX) was used to test whether induction of these 18 genes required de novo protein synthesis. Cf9 cells were pretreated for 1 hr with 5 μg/mL CHX. IF was then added, and cells were harvested after another 30 min. Treatment with CHX did not inhibit Avr9-dependent gene induction; rather, it induced the expression of these genes in the absence of Avr9 (Figure 3). This suggests that new protein synthesis is not required to induce the 18 genes tested, reinforcing the idea that their induction is an early target of Cf-9. Induction by CHX alone is typical of many early-response genes and suggests the involvement of a short-lived repressor (Koshiba et al., 1995).

Isolation of cDNA Clones Corresponding to 13 cDNA-AFLP Fragments

Sixteen out of the 18 AFLP fragments described above were used to screen a cDNA library for full-length clones (Table 2). The cDNA library was made from Cf9 tobacco cells harvested 30 min after the addition of IF(Avr9⁺). Before using the fragments for library screening, their identity was verified by repeating the AFLP technique using three selective nucleotides (rather than two) on the MseI primer and sequencing the band excised from the gel. Repeating the AFLP is important because no part of the gel is free from other PCR products, and it is likely that more than one fragment will be excised from the gel. Increasing the number of selective nucleotides decreased the number of fragments being amplified, thereby reducing the likelihood of amplifying contaminating fragments. The sequences of these fragments matched those from the original AFLP screen for all fragments used for library screening (data not shown). The reamplified fragments were used as probes to screen the library.

The results of the library screening are summarized in Table 2. We isolated and sequenced single clones for 13 AFLP fragments. Nicotiana tabacum is an allotetraploid originating from two distinct diploid species (N. sylvestris and N. tomentosiformis). The clones hybridizing with AFLP fragments 31, 75, 111, 137, and 180 and the AFLP fragments themselves had mismatched nucleotides, suggesting that these clones derived from a homolog of the cDNA from which the AFLP fragment originated. Clones with a 100% match at the nucleotide level with the AFLP fragment were obtained for the remaining eight fragments. When AFLP fragment 111 was used as a probe, two hybridizing clones were isolated, ACRE-111A and ACRE-111B, with a 99 and 94% match at the nucleotide level with the AFLP fragment, respectively. In some cases, cDNAs of different lengths with different polyadenylation sites were obtained. The genes were named ACRE genes, followed by the number corresponding to the AFLP fragment used for library screening.

Seven ACRE transcripts encoded proteins with similarity to sequences in the GenBank databases (percentage identity; amino acids in Table 2). The remainder showed no homology and represent previously unidentified gene products. ACRE-1, ACRE-111A, and ACRE-111B have amino acid sequences homologous to those of the ethylene response element binding protein (EREBP) transcription factors. ACRE-1 was similar to Arabidopsis ERF6 and tobacco EREBP4, which is induced by ethylene and wounding (Ohme-Takagi and Shinshi, 1995; Suzuki et al., 1998). The amino acid identity over the entire protein was low (49% for AtERF6) but much higher (78% for AtERF6) in the DNA binding domain, which is conserved throughout the EREBP family. ACRE-111A and ACRE-111B were most similar to Arabidopsis DREB1A (for dehydration-responsive element binding protein), which is induced by cold stress (Liu et al., 1998). Both ACRE-1 and ACRE-111A had small open reading frames (uORFs) upstream of the major ORF. ACRE-1 had three uORFs of 78, 9, and 18 bp, and ACRE-111A had two uORFs of 15 and 18 bp. uORFs have been described in other transcripts, and although present in <10% of eukaryotic mRNAs, they are more common in regulatory genes and have been shown to repress translation of the main ORF (Wang and Wessler, 1998).

ACRE-4 is 82% identical to the N resistance protein from tobacco. This homology extends throughout the N terminus of the protein but stops before the leucine-rich repeats. The predicted structure of ACRE-4 is similar to the truncated N polypeptide, Ntr, a product of alternative splicing (Whitham et al., 1994). No clone was identified encoding a full-length N-like protein.

ACRE-31 is 83% identical to CAST, a calcium binding protein of unknown function from potato (Gellatly and Lefebvre, 1993). It has three putative EF hand calcium binding domains.

ACRE-132 is homologous to a number of zinc finger proteins of the RING-H2 (for really interesting new gene) class E and is most homologous to the auxin-induced ATL2 gene product of Arabidopsis (Martínez-García et al., 1996). It has the conserved RING-H2 motif and a hydrophobic domain at the N terminus (predicted to be a membrane anchor), which are characteristic of this class of RING-H2 proteins (Saurin et al., 1996; Jensen et al., 1998).

ACRE-75, ACRE-137, ACRE-146, ACRE-169, and ACRE-231 are similar to hypothetical proteins from Arabidopsis identified during genome sequencing. The ACRE-231 sequence also has 22% amino acid identity to glycosyl transferases of bacteria involved in lipopolysaccharide biosynthesis. ACRE-65, ACRE-75, ACRE-169, ACRE-180, and ACRE-194 encode small peptides of unknown function of 97, 139, 124, 102, and 156 amino acids, respectively. The sequences of the ACRE cDNA clones have been submitted to GenBank as accession numbers AF211527 to AF211540.

RNA Gel Blot Analysis Verifies Gene Induction in Cell Cultures

To verify the expression patterns observed after cDNA-AFLP analysis, RNA gel blot analysis was performed using probes derived from four ACRE cDNAs. For ACRE-31, ACRE-146, and ACRE-169, the full-length cDNAs were used as probes. To avoid cross-hybridization with related EREBPs, we used a 600-bp fragment from the 5′ end of the ACRE-1 cDNA. This fragment was used because it did not include the region encoding the conserved DNA binding domain. The induction pattern we observed in cell cultures by using this method was the same as that seen when cDNA-AFLP analysis was conducted (Figures 4A and 4B).

ACRE Gene Expression in Leaves: Transient Nonspecific Induction and Sustained Avr9-Dependent Induction

We investigated the expression patterns of ACRE-1, ACRE-31, ACRE-146, and ACRE-169 in leaves of nontransformed (data not shown) or Cf9 tobacco after injection with IF. Surprisingly, the ACRE genes were rapidly but transiently induced in leaves after 30 min, regardless of the genotype or elicitor used. The induction by Avr9 was stronger and/or more sustained (Figure 4C). Additional experiments were conducted in which the leaves were cut with a razor blade and either injected with water, MgCl_2, or MgSO₄ or not injected and harvested after 30 min. Induction of 18 AFLP fragments was investigated by using cDNA-AFLP analysis (Figure 5), and induction of ACRE-1, ACRE-31, ACRE-146, and ACRE-169 was also examined by RNA gel blot analysis (Figure 4D). Infiltration of leaves with any solution produced strong but transient (peaking after 30 min) induction of all AFLP fragments inspected. Cutting the leaves without infiltration induced all fragments, except for fragments 3, 65, 157, 169, and 192. Induction by cutting was in most cases weaker than that elicited by infiltration but provoked a similar response for fragments 1, 31, 44, 75, 111, and 194. These results are consistent with other data suggesting an overlap between the responses induced by different stresses, such as fungal elicitor, wounding, and flooding of the apoplast in plants (Romeis et al., 1999). Recent data from mammals also suggest that overlapping sets of genes are induced by the different signal pathways activated by growth factor receptors (Fambrough et al., 1999).

Hin1, Hsr203J, and EREBP1 Transcripts Also Accumulate in Response to Avr9

Using RNA gel blot analysis, we were able to determine whether other defense-induced genes were rapidly elicited in the Cf-9/Avr9 system. Hin1 (for harpin-induced) and Hsr203J (for hypersensitivity-related) are reported to be rapidly induced by bacterial harpins and Avr gene products during the hypersensitive response of tobacco (Pontier et al., 1994; Gopalan et al., 1996). The tomato homolog of Hsr203J is induced by Avr9 in Cf9 plants within 3 hr (Pontier et al., 1998).

In Cf9 tobacco cells, Hin1 and Hsr203J were expressed at a low constitutive level and were strongly induced by Avr9 (Figure 6A). Maximal induction of Hsr203J was after 2 hr. Its induction was later than that of the ACRE genes. The Hin1 transcript accumulated to the same level 30 min after the addition of either IF(Avr9⁻) or IF(Avr9⁺); however, induction by IF(Avr9⁻) was transient, whereas induction by IF(Avr9⁺) continued for at least 4 hr. The late induction of Hsr203J and the accumulation of Hin1 at 30 min after either treatment explain why they were not detected in our AFLP screen for rapidly induced genes.

In Cf9 tobacco leaves, there was also some constitutive expression of both genes, and the pattern of induction by IF(Avr9⁺) was similar to that observed in cells. The Hin1 transcript accumulated after 15 min and Hsr203J after 2 hr (Figure 6B). Some induction by infiltration of IF(Avr9⁻) and water or buffer was also observed after 30 min (Figures 6B and 6C).

To test further the role of EREBPs in Cf-9–mediated defense, we examined mRNA accumulation of EREBP1, which is induced in response to many stresses, including ethylene, salicylic acid, and wounding (Ohme-Takagi and Shinshi, 1995; Horvath et al., 1998; Suzuki et al., 1998). It is also induced by Pseudomonas syringae pv tabaci expressing the avrPto avirulence gene in tobacco plants expressing the Pto resistance gene (Zhou et al., 1997). EREBP1 was expressed at a low level in untreated Cf9 cell cultures (Figure 6A). The transcript levels 30 min after treatment with IF(Avr9⁻) or IF(Avr9⁺) were the same, explaining why EREBP1 was not identified in the original AFLP screen, despite having suitable restriction sites. In leaves, EREBP1 expression was induced within 30 min after leaf infiltration, which is consistent with the results of Suzuki et al. (1998) (Figures 6B and 6C). However, induction by Avr9 was stronger and more prolonged than induction by wounding (Figure 6B).

Advantages of cDNA-AFLP Analysis

By using cDNA-AFLP analysis, we were able to survey transcriptional changes with no prior assumptions about which genes might be induced or repressed. cDNA-AFLP analysis gave reproducible results that were confirmed using RNA gel blot analysis (Bachem et al., 1996, 1998). Amplification of fragments from constitutively expressed genes provided internal control bands with every primer combination. cDNA-AFLP analysis has advantages over other RNA fingerprinting techniques and DNA chip-based approaches (McClelland et al., 1995; Baldwin et al., 1999). DNA chips can only be used to analyze the expression of cloned genes (Kehoe et al., 1999), whereas cDNA-AFLP analysis can reveal altered expression of any gene that carries suitable restriction sites. With the enzyme combination chosen, ∼30,000 cDNA fragments were inspected, of which 290 were differentially expressed; therefore, ∼1% of the transcripts rapidly alter in abundance after the addition of the Avr9 elicitor. To visualize a cDNA by using an AFLP, it must contain ApoI-MseI fragments of 50 to 400 bp. We inspected the sequence of 50 tobacco mRNAs and found that 76% contained suitable ApoI-MseI fragments. This means that 76% of genes expressed could potentially be visualized by using this enzyme combination. Also, tobacco is an allotetraploid between two polymorphic species; therefore, if one copy of a gene cannot be visualized, then there is still a chance of seeing another copy. Thus, it seems likely that a large proportion (>75%) of the transcripts were inspected for their response to the addition of Avr9.

A Limited Role for AOS in Rapid Defense Gene Induction

Most of the early Cf-9– and Avr9-dependent changes in gene expression were not inhibited by DPI and therefore are not dependent on AOS. Studies with other systems have concentrated mainly on genes with roles in phytoalexin production or cellular protection. In soybean, H₂O₂ induces expression of genes required for cellular protection but not those that encode phenylalanine ammonia lyase or chalcone synthase, which are involved in phytoalexin synthesis (Levine et al., 1994). In parsley cell cultures challenged with the peptide elicitor Pep-13 from P. sojae, treatment with DPI, but not catalase, inhibited both phytoalexin production and the induction of many defense genes, indicating a requirement for superoxide rather than H₂O₂ (Jabs et al., 1997). It is feasible that although the oxidative burst is not required for rapid induction of ACRE genes, it may play a role in the expression of genes that respond at later times when AOS levels are higher.

ACRE Gene Products and Their Possible Roles in Signal Transduction

ACRE-65, ACRE-75, ACRE-169, ACRE-180, and ACRE-194 encode small peptides of <160 amino acids. A number of small peptides are induced during the defense response. Some of these have antimicrobial properties, for example, defensins and thionins (Broekaert et al., 1997). However, these small ACRE proteins are unlikely to be defensins or thionins because they are not cysteine rich and have no apparent signal peptide. ACRE-137, ACRE-146, and ACRE-231 are similar to hypothetical proteins predicted from the genome sequence of Arabidopsis. These ACRE proteins may play roles in signaling or other aspects of the defense response, but their function is unknown.

Some of the induced genes are candidates for components of signaling pathways, which may be responsible for induction of other defense responses. Alternatively, they may encode proteins involved in pathways leading to gene expression that are induced by a positive feedback pathway to replace the protein as it is degraded. Protein phosphorylation cascades play an important role in many signaling pathways. The sequence homology of seven AFLP fragments to sequences of protein kinases and one to a phosphatase (Table 1) is therefore not unexpected.

AFLP fragment 44 is similar to cDNAs encoding 13-lipoxygenases, which are enzymes that catalyze an early step in the synthesis of jasmonic acid, which accumulates after wounding and herbivore attack (Creelman and Mullet, 1997). Jasmonic acid has been shown also to be important for resistance to fungal pathogens. For example, the Arabidopsis mutant coi1 (for coronatine insensitive), which is blocked in its response to jasmonic acid, shows enhanced susceptibility to Alternaria brassicicola and Botrytis cinerea (Thomma et al., 1998). The incompatible interaction between tobacco plants and P. parasitica var nicotianae is abolished in plants expressing antisense lipoxygenase (Rancé et al., 1998).

Calcium channel blockers, such as lanthanum, interfere with Cf-9 signaling, suggesting that calcium influx is required for the oxidative burst (Piedras et al., 1998). ACRE-31, a putative calcium binding protein with three EF hands, may be involved in signaling cascades, and this suggests that calcium is also involved in other responses.

ACRE-132 is a putative zinc finger protein of the RING-H2 subfamily. The cellular functions of RING proteins are diverse, but they are often involved in formation of multiprotein complexes (Saurin et al., 1996; Jensen et al., 1998). ACRE-132 has a hydrophobic domain at the N terminus that may function as a membrane anchor, so it might be part of a multiprotein complex anchored in the plasma membrane.

The sequence of ACRE-4 is 82% identical to Ntr, the product of the differentially spliced transcript N_L from the N resistance gene. The induction of an N gene homolog during Cf-9–mediated resistance is particularly intriguing. Both ACRE-4 and N contain a putative nucleotide binding site and a domain with homology to Toll and the interleukin-1 receptor (Whitham et al., 1994). The ratio of the alternatively spliced N transcripts changes after inoculation with tobacco mosaic virus (TMV) and plays a critical role in the TMV resistance response. The N_S transcript, which encodes the full-length N protein, is more prevalent before and during the first 3 hr of infection with TMV. The N_L transcript, which encodes the truncated N protein (Ntr), is more prevalent 4 to 8 hr after infection with TMV (Dinesh-Kumar and Baker, 2000). The induction of an N homolog by way of a Cf-9–dependent pathway suggests that the two classes of resistance gene may have some downstream signaling components in common and that there could be interaction between components of pathways activated by different classes of resistance protein.

Importance of EREBP Transcription Factors in Defense

The amino acid sequences of ACRE-1 and ACRE-111A are 41 and 66% identical to those of EREBP4 and DREB1A, respectively. These are members of a large family of transcription factors thought to control expression of many genes during the defense responses induced by a number of biotic and abiotic stresses (Riechmann and Meyerowitz, 1998). Tobacco EREBP1 to EREBP4 bind GCC boxes in PR gene promoters, and their transcript levels are induced by C₂H₄ and wounding (Ohme-Takagi and Shinshi, 1995; Suzuki et al., 1998). DREB1A, DREB1B, and DREB1C bind the CRT/DRE sequence found in promoters of genes regulated by cold and drought (Liu et al., 1998). The rapid but transient induction of ACRE-1 and ACRE-111A by Avr9 suggests that they are required for induction of other defense response genes. ACRE-1 and ACRE-111A are likely to bind different promoter elements, the GCC box and CRT/DRE box respectively, based on their homology to known proteins. This suggests that they may regulate different sets of response genes required for plant defense. It has been suggested that EREBPs regulate the expression of the genes encoding them via interaction with GCC boxes in their promoters (Solano et al., 1998). Three tomato EREBPs, Pti4, Pti5, and Pti6, were identified as proteins that interact with the Pto resistance protein (Zhou et al., 1997). Pto is a protein kinase, which suggests that EREBP activity could be controlled by phosphorylation as well as by changes in the level of transcription. C₂H₄ plays an important role in pathogen resistance. For example, tobacco transformed with etr1-1 (for ethylene resistant) from Arabidopsis, conferring dominant C₂H₄ insensitivity, shows reduced basic PR gene expression and loses non-host resistance to soil-borne fungi (Knoester et al., 1998). C₂H₄ alone may not be sufficient for resistance; for A. brassicicola resistance in Arabidopsis, the C₂H₄ and jasmonic acid pathways must both be triggered (Penninckx et al., 1998). The translated sequence of another induced AFLP fragment is homologous to 1-aminocyclopropane-1-carboxylase (ACC) synthase, a key enzyme in C₂H₄ synthesis. C₂H₄ accumulates in Cf9 tomato 9 to 12 hr after injection with IF(Avr9⁺) (Hammond-Kosack et al., 1996). Increases in C₂H₄ also induce the expression of EREBP genes; therefore, this may amplify or prolong the Cf-9–mediated defense response by inducing the expression of the same or additional EREBP genes.

Other examples of rapid induction of transcription factors during the defense response have been reported. Treatment of parsley cells with the fungal elicitor Pep25 induces a rapid but transient increase in WRKY1 (named after the conserved amino acid sequence WRKY) and WRKY3 expression (Rushton et al., 1996). These bind elicitor-responsive W boxes in the parsley PR-1 promoter, suggesting that they play a key role in signal transduction. One AFLP fragment with homology to WRKY genes was identified during our survey of Avr9-induced genes, suggesting that WRKYs also play a role in Cf-9–mediated defense responses. Basic leucine zipper (bZIP) transcription factors are also involved in the defense response. Recently, several bZIP proteins were identified that interact with NPR1 (for nonexpressor of PR genes) and bind the PR-1 promoter, providing a direct link from a key signaling intermediate downstream of many R genes to a transcription factor (Zhang et al., 1999). We did not identify any bZIP transcription factors. However, the soybean bZIP, G/HBF-1, which binds the chalcone synthase promoter, is activated by phosphorylation and not by changes in expression (Dröge-Laser et al., 1997).

Functional Significance of Expression Profiling Data

Knowledge of the expression patterns of genes can provide important information about their function. The large sets of genome-wide expression data from DNA microarrays are often studied by using cluster analysis (Bassett et al., 1999). Cluster analysis uses statistical algorithms to group sets of genes with similar patterns of expression across a number of treatments or tissue types. In a clustering analysis of 2467 yeast genes of known function, many genes that share similar functions clustered together over several treatments (Eisen et al., 1998). When the analysis was performed using all of the ∼6200 yeast genes, the genes of known function clustered in the same pattern, but these clusters were expanded by the addition of uncharacterized genes. It has been inferred from such studies that the genes of unknown function that react in the same way as genes of known function have similar roles in the response to a particular stimulus. Furthermore, cluster analysis allows the grouping of different stimuli that elicit similar patterns of gene expression. This provides insight into the biological mechanisms of the responses to different stimuli but also raises new questions about how an organism decodes this to give a unique response.

In this study, we have performed a “poor man's cluster analysis.” A large set of genes was identified whose members respond to a gene-for-gene stimulus in cell cultures. These could be grouped according to whether they were induced or repressed by Avr9 and whether this change in expression required AOS. The induction of ACRE genes by Avr9 was verified in leaves; during these experiments, it was observed that ACRE gene expression was also transiently induced by mechanical stress. The induction of 18 AFLP fragments by mechanical stress was therefore examined by cDNA-AFLP analysis. Of these 18 genes that were induced with gene-for-gene specificity in cell cultures, all were induced by mechanical stress in leaves. This indicates that even these very early R gene–dependent transcriptional responses are not specific to R gene function but are shared by responses to other stresses. This observation is important for the interpretation of gene induction data from other plant pathogen systems and in response to abiotic stress.

Significance of Common Responses to Different Stimuli

The ACRE genes were induced by cutting leaves with a razor blade and infiltrating liquid into the apoplast and by using Avr9 (Figure 5), suggesting a surprising degree of overlap between the responses elicited by Cf-9/Avr9 and other stresses. Other evidence for overlap in responses comes from studies on the activation of MAP kinases. Two MAP kinases, wound-induced protein kinase and salicylic acid–induced protein kinase, are rapidly induced in Cf9 tobacco by Avr9 (Romeis et al., 1999). These are also induced by a number of other stresses, including wounding, salicylate, and TMV infection (Scheel, 1998; Zhang and Klessig, 1998; Romeis et al., 1999). In mammalian cells, the numerous signal pathways activated by growth factor receptors induce many of the same genes, and the transcriptional responses to different growth factors overlap (Fambrough et al., 1999; Pawson and Saxton, 1999). The final responses to Avr9 and to mechanical stress are different. Both MAP kinase and ACRE gene induction are more transient after mechanical stress than after elicitation with Avr9, suggesting that the differences in final outcome may be achieved by differences in the duration and magnitude of the activation of the same pathway components and gene inductions.

A model for the role of ACRE genes during the defense response is proposed in Figure 7. Recognition of Avr9 in Cf9 plants results in the rapid production of AOS and induction of ACRE genes. These appear to be mediated by two different pathways because induction of most ACRE genes does not require AOS. The pathway leading to ACRE gene induction is also activated by wounding. ACRE proteins are candidates for components of signaling pathways involved in the activation of later defense responses.

This comprehensive survey of Cf-9–dependent gene induction has identified a number of genes that are induced during the plant defense response independently of AOS. There are two routes for further work. First, ACRE genes need to be studied to determine their function in the plant defense response by using reverse genetics. Second, the transcriptional regulation of the ACRE genes needs to be investigated. What transcription factors regulate ACRE gene expression, and how is Cf-9–dependent activation of these transcription factors achieved? The answers to these questions will provide further insight into inducible cellular defense mechanisms.

Plant and Cell Culture Conditions and Treatments

Plants (Nicotiana tabacum cv Petite Havana and transgenic Cf-9–expressing tobacco) were grown as described by Romeis et al. (1999). Intercellular washing fluid (IF) was obtained from cultivar Petite Havana plants expressing Avr9 (IF[Avr9⁺]) or nontransformed (IF[Avr9⁻]) (De Wit and Spikman, 1982; Hammond-Kosack et al., 1994). Leaves were infiltrated with IF from a syringe barrel through small razor blade cuts. At given times, two leaf panels were cut from the injected area and frozen in liquid nitrogen.

Tobacco cell cultures were maintained as described by Piedras et al. (1998). Before use, cells were washed, and aliquots of 2 g of cells in a 40-mL assay buffer (5 mM MES, pH 6.0, 175 mM mannitol, 0.5 mM CaCl₂, and 0.5 mM K₂SO₄) were dispensed into single flasks and left to equilibrate for 3 hr at 22°C. To each flask, 150 μL of IF(Avr9⁺) or IF(Avr9⁻) was added, and at times given in the figures, two 2-g aliquots of cells were harvested by filtration and frozen in liquid nitrogen. The synthetic Avr9 peptide was used at a final concentration of 10 nM. To study the effects of inhibitors, we pretreated cells before elicitation for 5 min with 1 μM diphenyleneiodonium (DPI) or for 1 hr with 5 μg/mL (17.7 μM) cycloheximide (CHX). Compounds were from Sigma.

Measurement of Reactive Oxygen Species

Aliquots (0.2 mL) of all cell samples were tested for active oxygen species production using the ferricyanide-catalyzed oxidation of luminol (Schwacke and Hager, 1992).

RNA Extraction and cDNA Synthesis

Total RNA was extracted according to Logemann et al. (1987). Poly(A) RNA was isolated from 150 μg of total RNA using oligo(dT) coupled to paramagnetic beads (Dynal A.S., Oslo, Norway). cDNA was synthesized using Expand reverse transcriptase (Boehringer Mannheim), RNase H, and DNA polymerase I.

Template Preparation, Amplified Fragment Length Polymorphism Reactions, and PAGE Analysis of Products

The template for cDNA-AFLP was prepared according to Vos et al. (1995) and Bachem et al. (1996), using ApoI and MseI. Sequences of primers and adapters used for amplified fragment length polymorphism (AFLP) reactions are as follows (N can be any nucleotide): ApoI adapter top strand, 5′-CTCGTAGACTGCGTACC-3′; ApoI adapter bottom strand, 5′-AATTGGTACGCAGTCTAC-3′; MseI adapter top strand, 5′-GACGATGAGTCCTGAG-3′; MseI adapter bottom strand, 5′-TACTCAGGACTCAT-3′; ApoI preamplification primer, 5′-CTCGTAGACTGCGTACCAATT-3′; ApoI selective amplification primers, 5′-GACTGCGTACCAATT(C/T)NN-3′; MseI preamplification primer, 5′-GACGATGAGTCCTGAGTAA-3′; and MseI selective amplification primers, 5′-GATGAGTCCTGAGTAANN-3′. AFLP reactions were performed according to Bachem et al. (1996), except that a mixture of Taq and Pfu DNA polymerases (160:1 units) was used for polymerase chain reaction (PCR) in a final volume of 10 μL. Selective amplification products were separated on a 5% polyacrylamide gel run at 100 W until the bromophenol blue reached the bottom. Gels were dried onto 3MM Whatman paper (Whatman, Maidstone, UK) and positionally marked before exposing to Kodak Biomax film (Sigma) for 2 days.

Isolation and Sequencing of Fragments

Films were aligned with markings on the gels. The bands of interest were marked, cut out with a razor blade, and incubated in 150 μL of TE (10 mM Tris, pH 7.5, and 1 mM EDTA, pH 8.0) overnight at 37°C. AFLP fragments were recovered by PCR under the same conditions as used for the preamplification. The dideoxy termination method (Sanger et al., 1977) was used for sequencing, and sequences were determined using an automated sequencer. Nucleotide sequences and translated sequences were compared with nucleotide and protein sequences of the GenBank nonredundant databases and sequences of the expressed sequence tag databases by using the BLAST sequence alignment program (Altschul et al., 1997).

Construction and Screening of cDNA Library

RNA was extracted from Cf9 tobacco cells 30 min after adding IF(Avr9⁺). Poly(A) RNA was isolated from 1.25 mg on oligo(dT) cellulose columns (Pharmacia). cDNA was synthesized and inserted into the Uni-ZAP XR λ insertion vector (Stratagene, La Jolla, CA). After packaging (Gigapack III Gold; Stratagene), the library was plated at a density of 50,000 plaques per 20 × 20-cm plate. Plaque lifts were screened with the reamplified AFLP fragments, which were radiolabeled with ³²P-dCTP using an oligolabeling kit (Pharmacia). Hybridization was performed following the protocol of Church and Gilbert (1984). Filters were washed four times for 15 min with 100 mL of 1% SDS and 1 × SSC (0.15 M NaCl and 0.015 M sodium citrate) at 65°C.

RNA Gel Blot Analysis

Aliquots (5 to 10 μg) of total RNA were separated by electrophoresis on a denaturing 1.4% agarose gel (100 mL of denaturing gel contained 1.4 g of agarose and 5.4 mL of 12.3 M formaldehyde [37% w/w]) followed by blotting onto nylon membrane (Hybond-N; Amersham, UK) (Sambrook et al., 1989). The RNA was cross-linked to the membrane using a UV Stratalinker (model 2400; Stratagene). Probe labeling and hybridization were performed as given above. The membranes were washed with 0.1% SDS and 2 × SSC twice for 10 min and subsequently with 0.1% SDS and 0.5 × SSC twice for 30 min at 65°C.

Isolation of Hin1, Hsr203J, and EREBP1 cDNAs from Tobacco

The Hin1, Hsr203J, and EREBP1 cDNAs were amplified from tobacco cDNA by PCR using specific primers. PCR products were ligated into the pGEM-T EASY vector (Promega, Southampton, UK), and the identity of the clones was verified by sequencing. Clone inserts were used as probes on RNA gel blots. For the EREBP1 probe, a 239-bp C-terminal fragment was used. This fragment did not contain the DNA binding domain.