Secondary metabolism and defense

We selected three distinct PAL unigenes for analysis of secondary metabolite-related transcript, as this enzyme represents the first enzymatic step in the flavonoid synthesis pathway which contributes isoflavones, anthocyanins, condensed tannins and other secondary metabolic compounds in plants [32–34]. Flavonoids are often stored in plant tissues as 'pre-formed' defense compounds and may act as pathogen and herbivore deterrents [33]. The expression of PAL gene transcripts in addition to the secondary metabolites resulting from the flavonoid pathway are known to be important in plant defense against pathogens, herbivores and environmental stresses [32–34].

A chitinase and a beta-1,3-glucanase were selected to analyze defense-related transcription, as these transcripts represent members of the PR family of proteins, which have been widely implicated in plant resistance to pathogens [35–37]. Different forms of chitinase are involved in both active and passive defense responses in plants [37]. Glucanases have also been implicated in plant resistance to pathogens, and beta-1, 3-glucanases comprise part of the PR-2 group of pathogenesis-related genes [35].

The fact that PAL, chitinase and glucanase transcripts were all reduced in introduced tetraploids compared to native tetraploids (Figure 3A,B) might suggest that populations of plants from the introduced range will be less defended against herbivores than natives, as is generally predicted by the EICA hypothesis. Some studies suggest that constitutive or basal levels of defense-related transcripts in plants, similar to those analyzed in this study, can be used to predict pathogen susceptibility and induced defense responses [38, 39]. Very subtle genetic mutations, such as those in the Arabidopsis cpr (constitutive expressers of pathogenesis related genes) mutant, have been shown to increase basal levels of systemic acquired resistance, which in turn increase levels of pathogen resistance [38]. In addition, the over-expression of PR proteins in planta typically results in a phenotype of enhanced disease resistance [38, 40, 41]. Plants with high constitutive defenses may, however, also have a lower degree of defense induction than those with low constitutive defenses [10, 12].

Recent reports indicate that introduced C. stoebe plants are better defended against both generalist and specialist enemies than natives [31]. This observation, in combination with the current study, may suggest that introduced populations have a higher potential degree of defense induction. However, the current study only measured levels of genes that may be involved in constitutive defense. Thus, our results must be interpreted with caution with regard to ecological hypotheses of plant defense in biological invasions.

It is important to note here that the release of C. stoebe from specialist enemies has been considered an important factor in the invasive success of the weed, and this has spurred the introduction of a number of biological control species to North America over the past thirty years [9, 16, 42, 43]. Although many of these specialist herbivores have become established and widespread, C. stoebe densities have only been reduced in a few specific areas (e.g[44]), and the weed continues to expand its range at other sites [9, 23]. Interestingly, field observations in North America suggest that introduced C. stoebe experiences little pressure from generalist herbivores and pathogens (RM Callaway and WM Ridenour, personal communication), indicating that C. stoebe currently experiences a partial release from both specialist and generalist enemies in the introduced range.

In order to better understand defense responses in C. stoebe, future studies should monitor gene expression and physiological responses in tetraploid geo-cytotpyes when exposed to pathogens and herbivores. This would help determine if expression of genes involved in constitutive defenses are good predictors of pathogen and herbivore susceptibility in C. stoebe. In addition, it would be interesting to test the response of C. stoebe geo-cytotypes to a variety of generalist and specialist enemies at the level of gene expression.

Evolutionary capacity

The activity of transposable elements could facilitate evolution by reorganizing the genome, and may be one important aspect in this process [27, 28]. Therefore, we hypothesized that introduced populations of C. stoebe would have the highest expression of the transposable elements analyzed, potentially due to novel stresses encountered in the introduced range. However, this was not the case. In fact, native tetraploid populations had the highest expression rate of one CATCA En/Spm subclass transposable element (Figure 3C). The expression of RAD, which is involved in DNA recombination/repair [45], was also highest in native tetraploid populations, but was not significantly different from that of introduced populations (Figure 3D).

Although the expression of transposable elements could facilitate rapid evolution, transposition may not be adaptive and could cause deleterious genomic rearrangements as opposed to beneficial ones. In other studies, certain transposable elements have been detected in plants at specific growth stages or under conditions of biotic and abiotic stress [46, 47]; however, the biological role of active transposition currently remains unclear. Additionally, recent evidence suggests that epigenetic mechanisms such as DNA methylation and chromatin remodeling can play an important role in the regulation of gene expression in polyploids which may facilitate adaptive plasticity [48–50]. Similarly, paramutation (interactions between homeologous genetic loci) can also result in differential regulation of genes between polyploids and their diploid progenitors [48, 50]. Thus, although we did not detect the changes we predicted in expression of transposable elements, it is entirely possible that factors other than chromosomal rearrangement through transposition are responsible for the observed changes in gene expression.

Plant ploidy

Although plant ploidy is often unaccounted for in comparisons of native and introduced populations, we found it to be a necessary and essential component for gene expression analyses. In native populations, we found lower expression of PAL2a, PAL2b and the transposable element in diploids compared to tetraploids, and all other genes examined showed similar relative expression (Figure 3, Table 1). The literature suggests that gene expression rates in polyploids tend to vary depending on plant species, ploidy, genetic background, and the genes examined; however, the phenomenon of gene dosage compensation appears to be common [49, 51–53]. This dosage effect results in gene or protein expression patterns in polyploids which are similar to their diploid progenitors. We did not necessarily expect to see this phenomenon in our plant populations because other studies involving ploidy and gene or protein expression have traditionally utilized plants with the same genetic background [49, 51, 52], whereas evidence suggests that C. stoebe plants within the native range harbor different genetic backgrounds [19, 20]. However, it appears that gene dosage compensation may be occurring to some extent in the native cytotypes of C. stoebe. Additionally, we observed increased expression of two PAL transcripts in native tetraploids compared to diploids, which may reflect increases in secondary compounds due to polyploidy as is seen in other plants [30].

Interestingly, native diploids exhibited similar expression profiles for nine of the ten total genes analyzed when compared to introduced tetraploids (Figure 3, Table 1), also suggesting gene dosage compensation. This result was rather surprising in that the diploid appears to be extremely rare (i.e., unsuccessful) in the introduced range, whereas the introduced tetraploid is a very problematic weed. Therefore, it is likely that other factors, such as plant performance characteristics, life cycle traits and the expression of other genes, are of greater importance in determining the success of tetraploids over diploids in the introduced range. Overall, the observed differences in gene expression between and within ploidies highlights the importance of using appropriate plant types when examining a particular species in both the native and introduced range.

Alternative gene roles and regulation

Genes similar to those selected in the current study have been detected in response to a variety of cues and conditions that do not necessarily reflect their primary annotation. For instance, many genes involved in defense response [54], flavonoid biosynthesis [34] and active transposition [46, 47] have been detected during particular points of plant growth and development. In this study we attempted to minimize any possible developmental differences in gene expression by sampling expanded, fully developed rosette leaves of similar age from all plants. All of the plants were grown in the same greenhouse environment and at the time of sampling remained in rosette form, none showing signs of bolting. If the genes tested here were expressed predominantly in response to developmental cues, it could be expected that expression of transcripts would be extremely similar across all geo-cytotypes, which is not what was observed.

Additionally, it is possible that the defense genes analyzed in this study are important for aspects other than plant defense against enemies. For instance, the production of certain flavonoids are thought to play important roles in photo-protection, frost hardiness and drought resistance [33], which could influence expression of PAL genes. C. stoebe occupies areas in both the native and introduced range that are often subject to these types of abiotic stress [21, 22, 24]. Thus, expression of PAL transcripts and resulting flavonoid accumulation may be important in both the biotic and abiotic stress response of the plant.

Field Sampling

Greenhouse experiment

In May 2006, five seeds from each maternal plant were placed in multi-pot trays in a mixture of sand (20%) and compost (80%, made from yard waste at the Botanical Garden in Fribourg, Switzerland). The greenhouse was not heated but temperatures stayed above 0°C in winter. One plantlet per mother plant was re-potted at eight weeks in 1 L pots of sandy soils (20% sand, 80% compost) in a naturally lit greenhouse supplemented with artificial light. The greenhouse was located near the University of Fribourg, Switzerland. Plants were watered regularly, but were not given nutrient solution. Number of leaves and longest leaf length were measured three times (10^th–14^th July 2006, 7^th–11^th August 2006, 27^th April–3^rd May 2007) before plants started bolting. Number of leaves multiplied by the longest leaf size was used as a non-destructive proxy for plant biomass, and is referred to subsequently as "biomass index". When the first flower opened (6^th July–23^rd August 2007), the date, number of stems, height of stems and number of buds larger than 5 mm were recorded for each plant. Survival, number of capitula per flowering plant and number of newly formed rosettes were estimated once the stem had senesced at the beginning of October 2007. The percent of flowering plants and percent plant mortality was calculated for each population. Previous studies on C. stoebe have indicated that although environmental maternal effects on offspring are detectable, they are relatively weak compared to other factors such as plant genotype and environmental conditions [55], therefore we do not expect maternal effects to confound the experimental results.

Tissue sampling

In November 2006 all plants remained in rosette form and had not bolted. One fully developed undamaged leaf was removed from each chosen plant using a razor blade. A few plants had minimal herbivore damage on the leaves, and these plants were avoided during tissue sampling. Four plants were sampled from each chosen population. Eight populations of North American tetraploids were sampled in addition to seven populations of Eurasian tetraploids and seven populations of Eurasian diploids (Table 2). Each leaf was immediately cut in half and the leaf tip was placed in a 5 mL vial containing RNAlater solution (Ambion, Austin TX). These samples were stored at -20°C for approximately four days, after which they were shipped on dry ice to Colorado State University. Upon arrival samples were placed at -20°C for storage.

RNA extraction and cDNA synthesis

Approximately 100 mg of each leaf sample (leaf tip) was removed from the RNAlater solution and quickly blotted on filter paper to remove excess liquid. Tissue was immediately frozen in liquid nitrogen and pulverized using a disposable pestle. RNA was isolated using Trizol reagent with its associated protocol (Invitrogen, Carlsbad CA). RNA pellets were resuspended in 30 μL RNase free water, and total RNA was quantified using a NanoDrop spectrophotometer (Wilmington DE). RNA samples were all diluted to the same concentration using RNase free water. RNA was treated with DNase to remove any genomic DNA contamination, and concentrations were re-evaluated using a NanoDrop spectrophotometer (Wilmington DE). Equal amounts of RNA from each sample were then individually translated into cDNA using reverse transcriptase, following a protocol from Invitrogen (Carlsbad CA). Samples were randomized in their preparation, such that RNA from plants from the same population (four plants tested per population) would not all be extracted on the same day.

Quantitative PCR

Candidate unigenes were chosen from the C. stoebe EST library based on a keyword search using the PLAN database (Table 3, [15, 59]). Gene specific primers were designed to amplify a 200–600 basepair region of each candidate C. stoebe unigene sequence (Table 3). Initially, specific primer sets were designed for a wide array of genes potentially involved in constitutive defense or secondary metabolism. However, many resulted in either poor amplification or amplification of multiple C. stoebe cDNAs, so these were not used in the final Q-PCR analysis. Successful primer sets included those for three distinct transcripts of phenylalanine ammonia lyase (PAL1, PAL2a and PAL2b), a chitinase, a glucanase, a transposable element and a DNA repair enzyme (Table 3). Amplification of each of these transcripts resulted in a single band visualized using agarose gel electrophoresis and each reaction produced a single peak in the Q-PCR melting temperature (Tm) curve, suggestive of a single product. An additional transposable element was successfully amplified in preliminary experiments, but was expressed to a very low extent in the experimental plant samples.

When multiple unigenes had the same annotation, nucleotide sequences were aligned using the DNA alignment program in CLC Free Workbench (Cambridge MA) to determine similarities. Unigenes with over 90% similarity (after removing the terminal 100 bases in case of sequencing error) were grouped together under one annotation, and primers were designed to the alignments. When the ESTs were originally clustered to form unigenes, they had to have an overlap of at least 40 bp and at least 94% sequence identity to be clustered together. The reason some unigenes were grouped in this analysis, but not in the original clustering analysis, is likely due to sequencing errors at the terminal (3') ends of the ESTs, which exhibited the largest amount of variability. In this analysis the terminal 100 bp of sequence was removed, such that only the most reliable sequence information was included. In addition, a few single base changes within similar ESTs were identified and these may represent either sequencing errors or natural polymorphisms. In addition, three potential housekeeping genes were analyzed as controls: actin (C. stoebe unigene 01058, top BLAST hit AAP73454, Gossypium hirsutum) cytochrome c oxidase (originally designed for Solanum tuberosum cv Cara, [60]), and ubiquitin (originally designed for Nicotiana). All primer sets amplified a single product from C. stoebe cDNA.

All reactions were run and analyzed using the BioRad iCycler software (Hercules CA). A standard curve was created for each primer set using serial dilutions (concentrations of 5–625 ng/μL) of cDNA prepared from leaves of a greenhouse-grown C. stoebe plant (fresh tissue was frozen in liquid nitrogen, and RNA extraction and cDNA synthesis followed the protocol above), and negative controls using water instead of template were run for all reactions. The optimal annealing temperature for all primer sets was determined empirically, with all sets working well at an annealing temperature of 55°C. All PCR reactions had a final volume of 20 μL and contained 10 μL of 2× Jumpstart cyber green reaction mix, 0.2 μL 1 μM flourescein, 2.4 μL 25 mM MgCl₂, 0.2 μL 10 μM forward primer, 0.2 μL 10 μM reverse primer, 2 μL template (20 ng/μL) and 5 μL sterile H₂O. Reactions conditions for PCR were as follows: 95°C 30 seconds, 55°C 30 seconds, 72°C 30 seconds, for 40 cycles.

For each sample, total RNA (ng/μL) was estimated using the appropriate standard curves and normalized using the geometric mean of actin, cox, and ubiquitin, as suggested in Vandesompele et al. (2002) [61]. Any expression levels that fell below the standard curve for either the gene of interest or the three housekeeping gene standards were removed from the analysis.