Consistency of Nicotiana attenuata's Herbivore- and Jasmonate-Induced Transcriptional Responses in the Allotetraploid Species Nicotiana quadrivalvis and Nicotiana clevelandii^,

Abstract

We examined the consistency of the native diploid Nicotiana attenuata (Na)'s herbivore-induced transcriptional changes in the two allotetraploid natives, Nicotiana clevelandii (Nc) and Nicotiana quadrivalvis (Nq), which are thought to be derived from hybridizations with an ancestral Na. An analysis of nuclear-encoded chloroplast-expressed Gln synthetase gene (ncpGS) sequences found strong similarity between Nc and Na and between N. trigonophylla and the two allopolyploids. All species were elicited with methyl jasmonate (MeJA), or were wounded and treated with either water, Manduca sexta oral secretions and regurgitant (R), or the two most abundant fatty acid amino acid conjugates (F) in R to simulate herbivory. The induced transcriptional responses in all three species were compared with a cDNA microarray enriched in Na genes. Na had the fastest transcriptional responses followed by Nc and then Nq. Na's R- and F-elicited responses were more similar to those from Nq, while the MeJA- or wound-elicited responses were more consistent in Nc. Treatment of wounds with the full cocktail of elicitors found in R elicits more complex responses than does treatment with F. The species differ in their elicited JA responses, and these differences are mirrored in the expression of oxylipin genes (LOX, HPL, AOS, and α-DOX) and downstream JA-elicited genes (TD). Elicitation decreases the expression of growth-related genes in all three species. We propose that this is a valuable system to examine the modification of complex, polygenic, adaptive responses during allopolyploid speciation.

Polyploidy has played a central role in the evolution of plants. More than 70% of all flowering plants have undergone one or more episodes of chromosomal doubling in their evolutionary history (Masterson, 1994; Song et al., 1995). Polyploid speciations are frequently associated with adaptive radiations and polyploid species appear to be better able to survive unfavorable conditions (Song et al., 1995). Many agricultural varieties are polyploid (wheat, oat, cotton, and tobacco) and express novel phenotypic qualities, including increased resistance to herbivores (Hilu, 1993; Shoemaker et al., 1996; Kellogg, 1998; Hieter and Griffiths, 1999). Given that adaptations to particular environments frequently involve complex polygenic traits that fit a plant to a particular environment (Halitschke et al., 2001; Hui et al., 2003; Kroymann et al., 2003), it is surprising that polyploid taxa are better adapted than their ancestor taxa. The genomic reorganization that occurs during polyploid speciation would be expected to disrupt complex adaptations, an expectation we will refer to as the merger paradox.

The relationship between adaptation and polyploidy is complex. Some polyploids are more resistant to herbivore and pathogen attack than are their diploid relatives (Schoen et al., 1992), but the effects of polyploidy on herbivore resistance are not always observed in a single lineage (Thompson et al., 1997; Nuismer and Thompson, 2001; Janz and Thompson, 2002). For example, a 4-year study of diploid and tetraploid individuals of the host plant Heuchera grossulariifolia, with its three most abundant species of phytophagous insects (Greya politella, Greya piperella, and Eupithecia misturata), did not reveal consistent associations between ploidy levels and insect resistance (Nuismer and Thompson, 2001). Allopolyploidy, the combination of two or more distinct nuclear genomes, is a common speciation mechanism in many taxa. In these taxa, the merger paradox is particularly acute, as the species-specific adaptations of the ancestor species are likely disrupted by the genomic reorganization that accompanies this form of instantaneous speciation. The genus Nicotiana contains many examples of allopolyploid speciation as well as one of the best-documented examples of a complex polygenic adaptation: the induced responses to herbivore attack.

Cytologically, Nicotiana attenuata (Na; synonymous with Nicotiana torreyana Nelson and Macbr.) is a 12-paired diploid species and is thought to be the common ancestor of two other North American tetraploid species, Nicotiana quadrivalvis (Nq; synonymous with Nicotiana Bigelovii; Chase et al., 2003) and Nicotiana clevelandii (Nc), according to their similarity in caryotypes and leaf, inflorescence, and trichome morphology (Goodspeed, 1954). Goodspeed also hypothesized that the other ancestor of the two tetraploids was an alatoid species. In agreement with this hypothesis, recently published genomic in situ hybridization (GISH) experiments used DNA probes from Na as well as Nicotiana sylvestris hybridized to the genomes of Nc and Nq, but the specificity of the probes is unclear (Chase et al., 2003). Phylogenies of the genus Nicotiana based on the internal transcribed spacer (ITS) regions and maturase K (matK) sequences (an analysis which did not include Nc) suggest that an ancestor of the species in N. section Trigonphyllae might have contributed to the genomes of the two tetraploids (Komarnyts'kyi et al., 1998; Aoki and Ito, 2000; Chase et al., 2003). Chloroplast-expressed Gln synthetase is a nuclear-encoded gene (ncpGS) that occurs as a single copy in most diploid taxa examined and is useful for phylogenetic studies (Emshwiller and Doyle, 1999). To examine Goodspeed's 1954 hypothesis, we cloned and compared the ncpGS chromosomal sequences from Na, Nq, and Nc, as well as from eight other wild species from the subgenus Petunioides.

Plant-herbivore interactions have been intensively investigated using ecological, chemical, and molecular approaches in the Na-Manduca sexta system (Baldwin and Preston, 1999). Extensive analysis reveals that Na can recognize attack from the larvae of its specialist Sphingid herbivore, M. sexta, as evidenced by an endogenous jasmonic acid (JA) burst (Baldwin et al., 1994). Following the JA burst, proteinase inhibitor (PI) accumulation and the release of volatile organic compounds (VOCs) increased dramatically (Halitschke et al., 2001). PIs act as direct defenses, slowing the growth of the larvae and keeping them at a size which is vulnerable to generalist predators, while VOCs, as an indirect defense, attract generalist predators to the feeding larvae (Pohnert et al., 1999; Kessler and Baldwin, 2001, 2002; Glawe et al., 2003). In addition, a large-scale transcriptional reorganization (Schittko et al., 2000, 2001; Hermsmeier et al., 2001; Voelckel et al., 2001; Halitschke et al., 2003; Hui et al., 2003; Voelckel and Baldwin, 2003) accompanies the up-regulation of direct and indirect defenses elicited by M. sexta attack. The available evidence suggests that many of the responses are adaptive, increasing the fitness of plants attacked by M. sexta larvae. Moreover, because the responses are clearly polygenic, they provide an ideal system in which to study how complex polygenic adaptations are conserved or modified during allopolyploid speciation. To conduct such a comparative analysis, the responses must be elicited in a consistent and reproducible manner.

When herbivores attack plants, they cause wounding and introduce herbivore-specific elicitors into the wounds, which results in complex transcriptional and metabolic changes that are different from those caused by mechanical wounding or simple application of JA. Several types of elicitors have been found in the oral secretions and regurgitant (R) of the herbivores, such as β-glycosidase (Mattiacci et al., 1995) and Glc oxidase (Musser et al., 2002) as well as fatty acid-amino acid conjugates (FACs), which can alter plants' responses to wounds (Halitschke et al., 2001) and mimic herbivore-induced changes. Eight FACs in the R from M. sexta account for most of the herbivore-specific changes in Na defense-responses, including JA and PI bursts and the release of VOCs (Halitschke et al., 2001). Moreover, the treatment of wounds with only two (of the eight) FACs found in Manduca R at the R-equivalent concentrations elicits more than 55% of the R-specific transcriptional changes. Hence, Manduca-specific changes can be elicited by applying R to the wounds, and parts of the responses can be elicited by applying water or FACs to wounds, or by treating plants with the methyl ester of JA (MeJA).

M. sexta larvae, one of the three most abundant and damaging herbivores for Na, has also been found on the two allotetraploid descendants, Nq and Nc, in their native habitats (I.T. Baldwin, unpublished data). Comparisons of M. sexta recognition (JA burst) and resistance responses (PI, nicotine, VOCs, and other secondary metabolites) revealed that most of the M. sexta-induced responses are retained with modifications in Nq but lost in Nc (Lou and Baldwin, 2003). Here we provide a transcriptional analysis of the Manduca-specific responses in Na, Nc, and Nq using a cDNA-microarray containing 246 Manduca-R-induced Na genes.

To examine the Manduca-specific responses in the three species, the source-sink transition leaves from a plant at the rosette stage of growth were either (1) wounded and immediately treated with 20 μL water, 1:10 diluted R, or the two most abundant FACs (F) in the R; or (2) directly treated with MeJA in lanolin paste. The transcriptional responses were examined 24 h after elicitation in all three species. Since transcriptional elicitation by M. sexta R is a rapidly induced and waning response in Na (Halitschke et al., 2003), while R-elicited metabolic responses are slower in Nq and Nc than in Na (Lou and Baldwin, 2003), the M. sexta R- and F-specific transcriptional responses from all three species were also compared 2 h after elicitation. Since the induced responses occur in both local and systemic tissues, albeit with a delayed kinetic, we examined the transcriptional responses in both directly treated and systemic leaves 24 h after elicitation.

RESULTS

Molecular Support for Goodspeed's Phylogeny

With the primer pairs GS57, GS2, NQ40, and NQ41, we cloned and sequenced ncpGS genes from Na, Nq, Nc, and 8 other species closely related to Na. The sequence analysis of 8 clones from Nc contained two ncpGS isoforms, but the 12 clones from Nq all proved to be of the same isoform. Consensus lengths of approximately 650 nucleotides of ncpGS genes covering one exon and intron completely and another exon and intron partially are used for multiple alignments with ClustalW. To root the phylogenetic tree, the cDNA sequence from Solanum tuberosum was also included in which the introns were counted as indels. The distances were calculated according to the Jukes-Cantor one parameter evolution model. The phylogram obtained from the neighbor-joining algorithm was drawn by TreeCon. Two clades are recognizable in the neighbor-joining tree of Nicotiana: (1) one consisting of N. trigonophylla and two species from the N. section of Bigelovianae, Nq and Nc. In this clade, Nq and Nc are placed in the same subclade as sister taxa; and (2) one consisting of N. sylvestris, Nc, and all the examined species of N. section Acuminatae, which include Na. This clade is split again into two subclades, one consisting of only two species from the Acuminatae section, N. linearis and N. spegazzinii, and the rest of the Acuminatae section, with the addition of Nc. One ncpGS gene from Nc grouped closely with Na and had a bootstrap value of 100% (Fig. 1). This phylogeny is consistent with Goodspeed's hypothesis that all three species (Na, Nq, and Nc) share a recent common ancestor and suggests that a progenitor of N. trigonophylla is involved in the formation of Nq and Nc.

Figure 1.

Neighbor-joining tree of the ncpGS sequences from selected North American Nicotiana species to test Goodspeed's hypothesis for the relationships among the diploid N. attenuata and the two allopolyploids, N. clevelandii and N. quadrivalvis. The chloroplast Gln synthetase gene from S. tuberosum was included to root the phylogenic tree. The branch lengths (bar) and bootstrap (bold, as percentages of 100 replicates) and the chromosome number (n value) of each species (from Goodspeed, 1954) are given. GenBank accession numbers: AF302113 (S. tuberosum); AY566989 (N. sylvestris); AY568077 (N. corymbosa); AY568079 (N. pauciflora); AY568078 (N. acuminate); AY568082 (N. linearis); AY568081 (N. spegazzinii); AY568084 (N. quadrivalvis); AY568083 (N. trigonophylla); AY568080 (N. miersii); AY566990 (N. clevelandii 1); AY566991 (N. clevelandii 2); and AY183657 (N. attenuata).

Twenty-Four Hour Transcriptional Responses

The 24-h transcriptional responses elicited by either MeJA treatment or wounds (W) treated with water (W + W), R (W + R) or F (W + F) are examined separately in the treated leaves and the untreated systemic leaves from all three species (Table I; Supplemental Table I, available at www.plantphysiol.org). The analysis of the 24 hybridized arrays from the 24 h treatments revealed that 217 (88% of 246) genes were differentially expressed with 91.2% (198), 65% (141), and 55.3% (120) being differentially regulated in Na, Nq, and Nc, respectively. Of these 217 genes, 150 genes were regulated by the different treatments in both diploid and allotetraploid species, and 67 genes were elicited only in the allotetraploid (19) or diploid species (48). The MeJA treatment allowed the 150-μg dose of MeJA to diffuse from the lanolin paste into the leaf over a protracted period of time (Zhang and Baldwin, 1997); hence we expected this treatment to produce the largest transcriptional response at the 24 h harvest. MeJA elicitation differentially regulated 133, 81, and 70 genes in the treated leaves and 33, 63, and 24 genes in the untreated systemic leaves of Na, Nq, and Nc, respectively. Of the MeJA-elicited responses, 51.1% and 45.1% (local) or 36% and 30% (systemic) from Na were also observed in Nq and Nc. Most of those elicited in the treated leaves of allotetraploid (84%, Nq and 86%, Nc) were also elicited in the treated leaves of Na. However, fewer of the MeJA-elicited genes from allotetraploid species (19%, Nq and 41%, Nc) in systemic leaves were elicited in Na (Table I).

Table I.

Number of genes that are significantly regulated (up or down) by treatments with MeJA, R, F, or both R and F (R/F) 2 h (locally treated leaves) and 24 h (locally treated and systemic leaves) after elicitation in Na, Nq, and Nc

Time	Leaf Position	Elicitor	Species	Number of Genes
								Na	Nq	Nc
			Na	64	16	26
		R	Nq	16	21	14
			Nc	26	14	55
			Na	93	15	12
2h	local	F	Nq	15	24	9
			Nc	12	9	29
			Na	41	8	7
		R/F	Nq	8	11	7
			Nc	7	7	12
			Na	56	9	3
		R	Nq	9	45	5
			Nc	3	5	42
			Na	81	26	14
		F	Nq	26	57	8
			Nc	14	8	39
	local		Na	43	5	1
		R/F	Nq	5	15	1
			Nc	1	1	22
			Na	133	68	60
		MeJA	Nq	68	81	42
			Nc	60	42	70
24h			Na	89	6	8
		R	Nq	6	20	3
			Nc	8	3	35
			Na	55	14	10
		F	Nq	14	32	7
			Nc	10	7	37
	systemic		Na	23	1	0
		R/F	Nq	1	14	1
			Nc	0	1	14
			Na	33	12	10
		MeJA	Nq	12	63	18
			Nc	10	18	24

Comparisons of transcriptional responses to R, F, and water treatments of wounds allow one to estimate how the wound response is modified by either the complete cocktail of elicitors found in R or F in the cocktail. Genes were defined as being specifically regulated by R or F only if (1) there was at least a ±0.2 difference in the expression ratio of genes that were significantly regulated by W + W, W + R, and W + F treatments; or (2) the genes were significantly regulated by only one of the three treatments using the three criteria (see “Materials and Methods”). This analysis revealed that 56, 45, and 42 genes from the treated leaves compared to 89, 20, and 35 genes from systemic leaves are significantly regulated by R in Na, Nq, and Nc, respectively. The comparison of R- and F-elicited transcriptional changes in all three species showed that both F-induced local (32.1%, Nq and 17.3%, Nc) and systemic (25.5%, Nq and 18.2%, Nc) responses from Na are more similar to those from two allotetraploids than are R-induced local (16%, Nq and 5.4%, Nc) and systemic (6.7%, Nq and 9%, Nc) responses. In other words, the responses to the complex cocktail of elicitors in Na are substantially less consistent than are the responses to components of the cocktail; hence, F elicitors account for more R-elicited responses in Na than they do for those responses in the allopolyploids. Of the R-elicited responses, a minority of those from the treated (20%, Nq and 7.1%, Nc) or systemic leaves (30%, Nq and 22.9%, Nc) of the allotetraploids were also elicited in Na. F elicitation differentially regulated 81, 57, and 39 genes from the treated leaves, and 55, 32, and 37 genes from systemic leaves in Na, Nq, and Nc, respectively. Of the F-elicited responses, 45.6% (Nq) and 35.9% (Nc) or 43.8% (Nq) and 27% (Nc) of those elicited in the treated or systemic leaves from the allotetraploids were also regulated in Na. F elicitation can account for 76.7% (43) and 25.8% (23) of the R-specific local and systemic elicitation observed in Na. In contrast, in the two tetraploids F elicitation accounts for only 33% (15) and 70% (14) of the R-elicited transcripts from the local and systemic leaves in Nq and only 52.4% (local) and 40% (systemic) of R-regulated transcripts in Nc (Table I).

To provide a global analysis of the Na-specific induced transcriptional responses in Nq and Nc, we log-10 transformed the expression ratios and conducted a principal component analysis (PCA). A two-dimensional presentation of the PCA accounted for about one-half of the variation in the data; axes 1 and 2 explain 55.1% and 59.2% of the total variance observed in treated (Fig. 2A) and systemic (Fig. 2B) leaves, respectively. This analysis revealed that the responses in treated leaves are more distinct than those observed in systemic leaves. The local responses of the three species were largely resolved along axis 2 (accounting for 17.7% of the variance; Fig. 2A), while no clear species-specific pattern was observed in the systemic (and younger) leaves. MeJA elicitation, as expected, produced transcriptional imprints (transient changes in the transcriptome) that are distinct in both local and systemic leaves, but again the differences are clearer in the elicited leaves. The transcriptional imprints resulting from the treatment of wounds with R or F modifies the wound-induced response in all three species, but the differences are more apparent in the systemic tissues.

Figure 2.

A PCA of the logarithmic expression ratios from 24 cDNA microarrays hybridized with fluorescently labeled cDNA isolated from Na, Nq, and Nc. At the rosette stage of growth, node 0 leaves from each species were wounded and immediately treated with: (1) 20 μL water, 1:10 diluted R, or F in R; or (2) directly treated with MeJA in lanolin paste. The treated leaf (A- local) at node 0 as well as the next younger leaf at position −1 (B- systemic) were harvested for mRNA extraction 24 h after treatment. Identical leaves were also harvested from noninduced control plants. Microarrays were hybridized with cDNA reverse-transcribed from mRNA pooled from 20 independently treated plants and Cy3-labeled (elicited plants) or Cy5-labeled (control plants). The MeJA treatment does not include wound responses, while water, R, and F treatments include wound responses. Together, axes 1 and 2 of the PCA account for 55.1% (local) and 59.2% (systemic) the variance. White circles, squares, and black squares represent Na, Nq, and Nc, respectively.

Most differences in the PCA associations between elicited and systemic leaves can be attributed to differences in the waning and waxing of the responses between the species. Na's rapid responses to R, F, and water treatments of wounds had largely waned in the local tissues by 24 h and as expected, the PCA clearly points up the different effects of the longer lasting MeJA elicitation among the three species; however, the R, F, and water responses were clearly still differentiating in the systemic tissues at 24 h. The speed of responses observed in Nc in local tissues followed Na's, while that of Nq was the slowest.

In summary, MeJA elicitation leads to more distinctive transcriptional changes compared to R or F elicitation in all three species. MeJA-elicited local responses from two allotetraploids are more similar to those from Na than are MeJA-elicited systemic responses. In comparison to F-elicited responses, the R-elicited responses are less similar among three species. Local recognition of R and F elicitation was faster in comparison to the systemic responses in all three species and, furthermore, Na showed the fastest R- and F-elicited responses (followed by Nc and then Nq).

Two-Hour Transcriptional Responses

From the analysis conducted at 24 h, it is clear that by then elicited recognition responses had largely waned to control levels, particularly in the fast-responding diploid Na. To remove the effect of wounding from the transcriptional signatures and to capture the earlier responses, we hybridized arrays with samples taken from locally elicited leaves 2 h after the elicitation in which mechanical wounds had been treated with either R or F, with samples taken at the same time from plants whose wounds had been treated only with water (for R), or triton-containing water (for F).

This analysis from the three different species revealed that the expression of 138 transcripts (56.3% of 246) was differentially regulated: 84% (116), 24.6% (34), and 50.7% (70) in Na, Nq, and Nc, respectively. Of these 138 regulated transcripts, 58 genes were regulated by R or F treatments in all 3 species, while 80 genes were elicited only in the allotetrapolyploids (23) or in Na (57). The comparison of R- and F-elicited transcriptional changes in all 3 species showed that R-induced responses (25%, Nq and 40.6%, Nc) from Na are more similar to those from the 2 allotetraploids than are F-induced responses (16.1%, Nq and 12.9%, Nc); moreover, the R- and F-induced responses from Nq (76.2%, R and 62.5%, F) are more similar than those from Na compared to Nc (47.3%, R and 41.4%, F). Treatment of F accounted for 64% (41) of the R-specific elicitation observed in Na. Similar analyses of F- and R-induced genes in Nc and Nq revealed that F treatment accounted for 21.8% (12 of 55) and 52.4% (11 of 21) of the total R- induced genes, respectively (Table I).

The PCA analysis indicates that the R- or F-induced transcriptional changes from Na differ distinctly from those belonging to the two allotetraploids as is readily seen along ordination axis 1, which accounts for 52.9% of the variance in the data (Fig. 3). While the R-treated plants of all three species cluster similarly along axis 1, the F treatments in Na are displaced along the vector in a positive direction, while in both Nc and Nq, the F-treated samples are displaced in a negative direction. The second axis (accounting for 16.1% of the variance) clearly distinguishes the transcriptional profiles of the two allopolyploids.

Figure 3.

A PCA of the logarithmic expression ratios from six cDNA microarrays hybridized with fluorescently labeled cDNA isolated from Na, Nq, and Nc, respectively. At the rosette stage of growth, node 0 leaves from each species were wounded and immediately treated with 20 μL of water, 1:10 R, F in the R, or 0.0005% triton solution, respectively. The treated leaves were harvested for mRNA extraction 2 h after treatment. Microarrays were hybridized with cDNA reverse transcribed from mRNA pooled from 20 independently treated plants and Cy3-labeled (plants elicited with R or F) or Cy5-labeled (plants elicited with water or triton). In these hybridizations, the wound-induced responses were removed. Axes 1 and 2 of the PCA account for 79% of the variance. White circles, squares, and black squares represent Na, Nq, and Nc, respectively.

The PCA presented in Figure 3 contrasts the transcriptional changes elicited by the complex cocktail of elicitors found in R with those elicited by only two components of the cocktail (F) among the three species. Interestingly, the transcriptional responses are greater in response to F compared to R treatments across the three species, and the differences between F and R treatments within a species are greatest in Na (followed by Nc and then Nq). Two hours after the elicitation, all three species responded to R and F treatments; Na shows the largest induced changes followed by Nc and then Nq. In comparison to F-elicited responses, R-induced responses from Na are more similar to those from Nq and Nc. The transcriptional responses elicited by both R and F treatments from Na are more consistent in Nq than Nc.

Individual Genes

Genes involved in oxylipin signaling, antimicrobial defense, remodeling of metabolism, and transcriptional machinery are shown by this study to be regulated. Continuous release of MeJA in the plants leads to significant up-regulation of genes that encode for enzymes involved in the oxylipin signaling cascade (13-lipoxygenase, LOX; allene oxide synthase, AOS; hydroperoxide lyase, HPL; alpha-dioxygenase, α-DIOX) locally in all three species, while the significant up-regulation of all three genes in systemic leaves was observed only in Nq (Fig. 4). MeJA elicted a 2-fold increase of the expressions of LOX and AOS in Nq compared to those from Na, while the expressions of HPL and LOX from Nc were lower than those from Na. Transcripts involved in the oxylipin-signaling cascade were also regulated by R or F treatments in all three species, but the regulation differed in timing among the three species as the expression of HPL, LOX, and AOS in Na and Nc shows; these genes were already up-regulated by R or F at 2 h, while the up-regulation of these genes in Nq was observed only 24 h after elicitation. The expression of LOX in Nc was up-regulated only by R or F at 2 h, while its expression in Nq was up-regulated by F at 24 h. HPL showed strong up-regulation by R or F in Nc at 2 and 24 h, whereas this pattern was observed in Nq only 24 h after elicitation. In addition, the R- or F-elicited expression of dehydration-induced protein and unknown genes such as RE166 and RE172 also differed in timing among the three species, showing the fastest response in Na, followed by Nc and then Nq. In addition, a 2-fold F-induced increase of the expression of HPL, LOX, and AOS in Nq compared to that of Na and Nc was also observed.

Figure 4.

Mean (± sd) of the logarithmic expression ratios from microarrays with eight replicate cDNA spots of Na sequences of: genes mediating oxylipin cascade (LOX, HPL, AOS, and α-DOX); an example of JA-elicited genes (TD); and an example of a growth-related gene (RUBISCO SSU: RbcsS). Microarrays were hybridized with fluorescently labeled probes derived from Na, Nq, and Nc. At the rosette stage of growth, node 0 leaves (local) from each species were wounded and immediately treated with either 20 μL water (wound + H₂0), 1:10 diluted R (wound + R) or F in the R (wound + F), or directly treated with MeJA in lanolin paste (MeJA). The wound + H₂0, wound + R, wound + F, or MeJA samples were hybridized against the untreated control samples, while R or F samples were hybridized against the wound + water or wound + triton treatments.

Threonine deaminase (TD), which is involved in the first step of branched chain amino acid biosynthesis in plants and microorganisms, also shows very strong MeJA-induced responses. In Na, TD shows a strong local response to MeJA with a 29-fold increase. Compared to those from noninduced plants, the expressions of TD are stimulated by MeJA locally up to 8- or 9-fold and systemically up to 4- and 16-fold in Nc or Nq, respectively (Fig. 4).

Genes involved in the remodeling of transcriptional machinery and metabolism are also regulated in response to different elicitors in different species. The transcripts encoding photosynthetic enzymes (N. sylvestris mRNA for the small subunit of ribulose bisphosphate carboxylase; tobacco ribulose-1,5-bisphosphate carboxylase small subunit pseudogene) tend to be down-regulated in all three species and the expression patterns are retained up to 24 h after elicitation. RB493, which shares the sequence similarity with Zymomonas mobilis rrnB operon and 23S ribosomal RNA gene, were locally up-regulated by R/F in both Na and Nq, and systemically regulated in Nc (Fig. 4).

DISCUSSION

The species Na, Nq, and Nc are morphologically similar annuals whose distributions in North America differ. Based on the caryotypes and morphology of leaf, inflorescence, and trichomes, Goodspeed deduced that Nq and Nc are members of Nicotiana section Bigelovianae which originated from amphidiploidy involving Na and an ‘alatoid’ progenitor such as N. sylvestris (Goodspeed, 1954). Phylogenetic relationships in the genus Nicotiana were investigated by parsimony analyses of plastid matK and internal transcribed spacer (ITS) regions of nuclear ribosomal DNA (nrDNA). The possible progenitor of N. section Bigelovianae is still not clear, although the patterns of relationship in Nicotiana are largely congruent (Goodspeed, 1954; Komarnyts'kyi et al., 1998; Aoki and Ito, 2000). In accord with the published data, the phylogenic analysis of ncpGS splits the N. section Acuminatae into two clades. N. linearis is placed together with N. spegazzinii and the ancestor of Na is supposed to be the common origin of N. corymbosa, N. pauciflora, N. acuminate, and N. miersii. In light of all the published data, as well as of the two ncpGS genes from Nc that belong to two clades, we can conclude that neither Nq nor Nc is autotetraploid and that an ancestor of Na is involved in the formation of both allopolyploids, Nq and Nc. This conclusion further supports Goodspeed's phylogenetic hypothesis.

N. sylvestris, according to Goodspeed's hypothesis, represents the other line of parents for Nq and Nc. Although N. sylvestris is placed in the N. section Alatae (Goodspeed, 1954) mostly based on flower morphology and distribution, its chromosomes physically resemble those of N. section Acuminatae. Both ITS and ncpGS analyses suggest that N. sylvestris is closely related to Na. Furthermore, the genomic DNA from N. sylvestris and Na hybridize to the same chromosome set in Nc and Nq. These results all suggest that species other than Na and N. sylvestris must be involved in the speciation of N. section Bigelovianae and that an ancestor of N. trigonophylla may be involved in the formation of N. section Bigelovianae. Polyploid speciation is frequently associated with adaptive radiations and polyploids can exhibit an enhanced ability to survive under unfavorable conditions (Song et al., 1995). How this occurs is not well understood, but polyploidy and its associated duplication of genes are thought to offer opportunities for the evolution of a new function in the redundant copy of the gene. To compensate for the dosage effects, recent studies have demonstrated that extra copies are frequently silenced or lost due to mutations, or otherwise epigenetically regulated (Pichersky et al., 1990; Comai et al., 2000; Soltis and Soltis, 2000; Wendel, 2000; Kashkusha et al., 2002). Unlike autotetraploidy (in which the same genome is duplicated), the allopolyploid caryotype is fixed when the ancestral genomes that are combined in the new species are sufficiently different to prevent tetrasomic recombination. Our examination of the induced transcriptional profiles from two allotetraploids, Nq and Nc, in comparison to their diploid ancestor Na provides an evolutionarily-motivated analysis which reveals three insights into ecological adaptation on the transcriptional level.

First, the examination of their transcriptional changes in response to herbivore-specific elicitor R shows that all three plants can recognize M. sexta attack, but the recognition of herbivore attack on the transcription level differs in timing among the three species. The quickest activation of gene expression is found in Na; 64 and 56 transcripts from local leaves are significantly regulated by R at 2 and 24 h. Nq, on the other hand, shows the slowest responses and these may last longer, as evidenced by the fact that 21 and 45 transcripts from local leaves are regulated by R at 2 and 24 h. In addition, herbivore-specific recognition is much faster in leaves that are locally treated as compared to the responses in systemic untreated leaves. The largest difference among transcripts occurs in Na followed by Nc, then Nq in systemic leaves, a pattern which is exactly the opposite of that which characterizes the locally-treated leaves 24 h after the elicitation.

Second, the transcriptional regulation of Na is more similar to that in Nq than in Nc, although Nq is slower to recognize herbivore attack. The comparison of R-induced transcriptional changes among the three species at 2 h revealed that 76.2% and 47.2% of R-induced transcripts from Nq and Nc are regulated similarly to those from Na. The difference in the consistency of Na-specific R recognition between two allotetraploids is not due to the plant's inability to react to R. About 47% of MeJA-elicited transcriptional changes from Na are observed in Nq and Nc, while 84% of MeJA-elicited transcripts from both tetraploids are regulated similarly to those from Na. Furthermore, the expression profiles of LOX, AOS, and HPL involved in the synthesis of jasmonic acid, as well as α-DIOX, are coordinated in response to different elicitors from all three species. This coordinated regulation is correlated with induced changes in the jasmonate pool in the different species (Lou and Baldwin, 2003). Jasmonic acid, a fatty-acid-derived signaling molecule, is known to be involved in several aspects of plant biology, including pollen and seed development as well as defense against attack from microbial pathogens and insect pests (Dong, 1998; Reymond and Farmer, 1998; Bostock, 1999; Kunkel and Books, 2002; Thaler et al., 2002).

Third, the recognition of R- and F-elicitation is different among the three species. The comparisons of R- and F-elicited transcriptional changes in the three species indicate that F treatments account for a majority of R-induced transcriptional changes in Na (2 h: 64%; 24 h: 77%), but not in Nq (2 h, 52%; 24 h, 33%) or in Nc (2 h, 22%; 24 h, 52%). With its full complement of elicitors, R brings out more complex transcriptional responses, which are both positive and negative. R treatments differentially regulate F-elicited transcripts by 56% (2 h) and 47% (24 h), 54% (2 h) and 74% (24 h), and 59% (2 h) and 44% (24 h) in Na, Nq, and Nc, respectively. The proportion of R-induced transcripts accounted for by F treatments increases with the timing of elicitation in Na (2 h, 64%; 24 h, 76.8%) and Nc (2 h, 22%; 24 h, 52%) but not in Nq (2 h, 52%; 24 h, 33%), indicating that F elicits longer-lasting or slower responses in Na and Nc than does R, while in Nq, F treatment leads to fast responses.

In addition, the coordinated regulation between the up-regulated genes involved in the remodeling of transcriptional machinery (DH017, DH219, and RB493) and down-regulated photosynthesis genes is also observed in the three species. This observation is consistent with a more general response to stress in which the remobilization of limiting resources can minimize the fitness consequences of losing tissues to herbivores, or activate metabolically demanding defense responses in all three species (Karban and Baldwin, 1997; Voelckel et al., 2001; Heil and Baldwin, 2002; Strauss et al., 2002; Zangerl et al., 2002; Hui et al., 2003; Zavala et al., 2004).

The ecological significance of the retention of MeJA and larval R recognition abilities in the two allopolyploid species remains unknown, but it likely provides important insights into the maintenance or modification of polygenic adaptive traits during allopolyploid speciation. Nq and Nc descended from an ancestor of modern Na due to polyploidy that happened several million years ago. The differences among the levels of consistency of Na-specific MeJA-, R- or F-induced transcriptional responses in the two allopolyploids may be due to the chromosomal rearrangements that occurred during speciation (allopolyploidy) as has been shown in wheat (Ozkan et al., 2001; He et al., 2003). Alternatively or perhaps in addition, the differences may be due to the fact that Na, Nq, and Nc are distributed in different habitats and therefore are likely subjected to different selection pressures. Nq is found in coastal California; Nc has a more southern distribution in both coastal and interior Baja California and southern California; Na is found throughout the Great Basin Desert and adjacent areas as an ephemeral member of the post-fire community in sagebrush and pinyon-juniper habitats. Hence, differences between the levels of consistency of Na-specific induced transcription profiles in two allopolyploids may also be due to the effects of natural selection during the interval since the speciation occurred. It is important to note that the Na genome has likely undergone evolutionary change since the hybridization events. Since the genes on the array are derived from a modern Na genome, some of the differences in transcriptional signatures among the species may reflect differences between the modern Na cDNAs and the Na genome that were incorporated into the other species at the time of speciation, which has not likely been subjected to the same selection pressures. Moreover, the differences in the speed of responses among the species are inferred from analyzing responses at a few time points. It is quite possible that, in nature, all the genes that respond to these treatments in Na are also capable of responding in the other species in the context of their own natural habitats. The experiments were conducted under the same growth conditions and with plants in the same developmental stage, but these growth conditions have been optimized for Na and environmental signals important for the regulation of Nc and Nq genes may have been lacking.

These evolutionary hypotheses can be falsified by determining whether the M. sexta-induced responses in Na, Nq, and Nc are indeed defensive (e.g. increase plant fitness in environments with herbivores) and by comparing the responses in newly created allopolyploid species (synthetic hybrids from Na and N. trigonophylla). Furthermore, the fact that larger-scale transcriptional changes can be elicited by either R or F and the coordinated regulation of genes involved in the signaling and remodeling of the metabolism in all three species point to the existence of unknown trans-activating factors and corresponding cis-elements. These elements may be involved in the reorganization of the complex herbivore-specific transcription changes in plants and may have been under strong selection in the allotetraploids.

MATERIALS AND METHODS

Plant Growth

Nicotiana sylvestris, Nicotiana corymbosa, Nicotiana pauciflora, Nicotiana acuminata, Nicotiana linearis, Nicotiana spegazzinii, Nq, Nicotiana trigonophylla, Nicotiana miersii, and Nc seeds were kindly supplied by Dr. Verne A. Sisson (Oxford Tobacco Research Station, Oxford, NC) and originated from collections made by Dr. H. Goodspeed (Goodspeed, 1954). Na Torr. Ex Watts (an inbred line collected from the DI Ranch, Utah (Baldwin et al., 1998) seeds were germinated in smoke-treated soil and plants were grown in individual 1-L pots as previously described (Kruegel et al., 2002). All plants were grown in the glasshouse at 26°C to 28°C under 16 h light supplemented by Philips Sun-T Agro 400 or 600 W Na lights (Eindhoven, The Netherlands).

PCR Amplification, Cloning, and Sequencing of ncpGS

Using young leaves, total genomic DNA was extracted according to the cetyl-trimethyl-ammonium bromide method (Richards, 1997). Two pairs of primers GS57 (5′-AGA C/T CC TTT CCG TGG A/G/C/T GG-3′), GS2 (5′-CAG CTC CGG TTC CAC AGT AGT-3′) and NQ 40 (5′-TAA AGG CCA CTT CAC ATT TT-3′), NQ 41 (5′-TGA ATG TTT GAT CTC ATT GG-3′) were designed to amplify Gln synthetase isoforms. Amplification was performed in the total volume of 50 μL including 10 ng of template genomic DNA, 1× PCR buffer, 1.5 mm MgCl₂, 0.2 μm of each primer, 200 μm of each dNTP, and 1.25 units of Taq DNA polymerase. The PCR conditions were as follows: an initial denaturation step at 94°C for 5 min, 30 cycles at 94°C for 1 min, 57°C for 1 min (50°C during the first two cycles for the primer pair GS57/GS2) and 72°C for 2 min, and a final extension step at 72°C for 10 min. The PCR fragments were gel-purified and cloned into pCR2.1 TOPO (in Escherichia coli TOP10F') or pGEM-7Zf(+) vector. Sequencing was conducted using an ABI PRISM 377 automated sequencer (Sunnyvale, CA). Each clone was sequenced in both sense and antisense directions and at least four clones were sequenced for each fragment.

Treatments

For microarray analysis, the node 0 leaves (source-sink transition) of 20 individual, developmentally synchronized plants per treatment in the rosette stage of growth from Na, Nq, and Nc, were wounded by creating three rows of puncture wounds on each leaf half (Ohnmeiss and Baldwin, 1994). Wounds were immediately treated with either 20 μL of deionized water, M. sexta R (diluted in 1:10 deionized water) from fourth to fifth instar larvae, or a synthetic mixture of the two most abundant FACs in M. sexta R: N-linolenoyl-l-Gln and N-linolenoyl-l-Glu (0.012 mm N-linolenoyl-l-Gln and 0.034 mm N-linolenoyl-l-Glu). Since the FACs were dissolved in R-equivalent concentrations in an aqueous solution of 0.005% Triton X-100, we applied 0.0005% (1:10 dilution) triton solution to the wounds to control the possible effects of triton. Nonwounded plants were treated with 150 μg MeJA in 20 μL of lanolin paste, which was applied to adaxial lamina. The treated or next younger position leaves (position −1) were harvested at 2 or 24 h after the treatment and frozen in liquid nitrogen. The identical leaves from uninduced plants were harvested at 24 h as control.

cDNA-Microarray Analysis

Two PCR-amplified fragments from each of 246 herbivore-regulated genes isolated by either differential-display reverse transcription-PCR or cDNA-amplified fragment-length polymorphism from Na were amino-linked on either strand and spotted on epoxy-coated slides as previously described (Hermsmeier et al., 2001; Halitschke et al., 2003; Hui et al., 2003). Leaf samples were homogenized in liquid nitrogen and the total RNA was isolated according to the methods described in Hermsmeier et al. (2001). Twenty-four cDNA microarrays were competitively hybridized with the Cy3-labeled cDNA from a 24 h harvest of treated node 0 leaves or −1 leaves and Cy5-labeled cDNA generated from identical leaves from uninduced plants. To monitor the herbivore-specific recognition responses without the potentially confounding effects of the wound response, we hybridized an additional six microarrays with Cy3-labeled cDNA from leaves from the three species elicited by treating wounds immediately with either R or FACs for 2 h and Cy5-labeled cDNA from leaves elicited by treating wounds immediately with water or triton-containing solution. Hence, in these six arrays, the wound-response was removed from the analysis.

The cDNA microarrays were scanned and evaluated with methods described and verified in previous publications (Halitschke et al., 2003; Hui et al., 2003). Briefly, an array-specific normalization factor was calculated from the middle 75% of the distribution of expression ratios (ERs). The ratios of normalized ERs of each individual spot and the mean of the four replicate spots for each cDNA (two for each gene) were calculated. A transcript was defined as being differentially regulated only if the following three criteria were met: the sum of normalized background corrected Cy3 and Cy5 signals was higher than 1,000; both individual ERs were equal to or exceeded the arbitrary thresholds for differential expression (1.25 and 0.75), respectively; and both individual expression ratios were significantly different from one as determined by t test analysis of the eight replicate spots. The expressions of genes are summarized in Supplemental Table I in which the expression patterns are represented as up, down, or -, depending on whether or not the three criteria are met.

To compare the full transcriptional responses of three species to different treatments at different harvest times, the logarithmic expression ratios of all transcripts from all 30 cDNA-microarrays were analyzed by detrended correspondence analysis. Since the longest gradient from DCA analysis was shorter than 3, a principal components analysis, a linear type of ordination model, was performed to examine the interarrays distances (Canoco for Windows 4.5, Microcomputer Power, Ithaca, NY; Yeung and Ruzzo, 2001; ter Braak and Smilauer, 2002).

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers AF302113, AY566989, AY568077, AY568079, AY568078, AY568082, AY568081, AY568084, AY568083, AY568080, AY566990, AY566991, and AY183657.