Biol. Bull. Sign up for etocs!
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hoover, C. A.
Right arrow Articles by Marsh, A. G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hoover, C. A.
Right arrow Articles by Marsh, A. G.
Related Collections
Right arrow Ecology
Right arrow Genomics
Right arrow Cnidarians
Biol. Bull. 214: 319-328. (June 2008)
© 2008 Marine Biological Laboratory

Transcriptome and Metabolite Responses to Predation in a South Pacific Soft Coral

Cindi A. Hoover1,3,*, Marc Slattery2, Nancy M. Targett3 and Adam G. Marsh3

1 The Broad Institute of MIT and Harvard, Cambridge, Massachusetts 02141
2 Department of Pharmacognosy, University of Mississippi, University, Mississippi 38677
3 College of Marine and Earth Studies, University of Delaware, Lewes, Delaware 19958

* To whom correspondence should be addressed. E-mail: choover{at}broad.mit.edu


   Abstract
TOP
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Literature Cited
 
Sinularia polydactyla, a dioecious, abundant soft coral in the South Pacific, exhibits biochemical phenotypic plasticity in secondary metabolite production in relation to predation intensity. However, it is unclear to what extent changes in secondary metabolites, such as 11β-acetoxypukalide, may result from specific, induced pathway activities at the level of gene expression. To investigate both chemical changes and differences in mRNA diversity in response to predation stress, artificial predation experiments were conducted in situ on colonies of S. polydactyla. Multivariate statistical analyses of coral biochemical metabolites and our kinetic transcriptome profiling technique indicate that that the induction of 11β-acetoxypukalide by predation stress likely results from the upregulation of either one dominant transcript or a very small set of transcripts, indicative of a targeted upregulation rather than a generalized, genetic stress response. Overall, this work establishes a routine method for integrating high-throughput transcriptome and metabolome data sets to allow for the identification of metabolites whose intracellular concentrations can be readily linked to gene expression events in response to specific treatments in non-model organisms.

Abbreviations: Hflr, a Shannon-Weaver–type entropy statistic that reflects complexity estimates among different samples on the basis of kinetic rates and maximun fluorescence • PCA, principal component analysis


   Introduction
TOP
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Literature Cited
 
In both marine and terrestrial environments, many sessile organisms that are vulnerable to sublethal grazing attacks can synthesize secondary metabolites to reduce the feeding activities of their predators (Cronin and Hay, 1996; Hammerschmidt, 1999; Hay, 2001; Schnitzler et al., 2001; Taylor et al., 2002; Thornton and Kerr, 2002; Deal et al., 2003; Ceh et al., 2005; Schoeppner and Relyea, 2005; Adie et al., 2007). In general, compounds synthesized in response to a wound stress can be described as a chemical defense phenotype, which can be either plastic or fixed—that is, inducible/adjustable in response to predation stress or continually existing at a constant level (Zangerl and Rutledge, 1996; Pavia and Toth, 2000; Taylor et al., 2002). Optimal defense theory predicts that fixed defenses should be allocated to tissues or areas of an organism that are the most valuable and that have the greatest probability of being attacked or eaten (Zangerl and Rutledge, 1996; Toth et al., 2005). However, when the attack probability is low or uncertain, it may be more cost effective to have an inducible or plastic form of defense (Herms and Mattson, 1992; Karban and Baldwin, 1997). Traditionally, these strategies for the regulation of chemical defense pathways have been well documented in terrestrial plant-herbivore systems (Stout et al., 1998; Thaler et al., 1999; Callaway et al., 2003). In marine environments, inducible and plastic chemical defenses have been documented in many soft-bodied, sessile invertebrate species where a motile predator can exert a frequent wound stress on a colony (La Barre et al., 1986; Pawlik, 1993; Assmann et al., 2000; Kubanek et al., 2000; Puyana et al., 2003; McClintock et al., 2004). Additionally, some soft coral species have been shown to exhibit plasticity in chemical defenses in response to changing environmental stimuli (Leone et al., 1995; Fleury et al., 2000); this includes Sinularia polydactyla (Enrenberg, 1834), the target species for this study (Slattery et al., 2001). The inducible, plastic defense strategy of S. polydactyla is also common to marine organisms in variable or unpredictable environments (Pavia and Toth, 2000; Taylor et al., 2002; Thornton and Kerr, 2002; Ceh et al., 2005).

S. polydactyla is abundant in shallow coral reef habitats in Guam, and synthesizes two terpenoid secondary metabolites, pukalide and 11β-acetoxypukalide (Slattery et al., 2001). The S. polydactyla colonies, predation intensity, and secondary metabolite production at our study sites are well characterized (Slattery et al., 1998, 1999, 2001; Kamel et al., 2004; Kamel and Slattery, 2005). The synthesis pathways for these two compounds are unknown but may involve the mevalonic acid pathway, a common terpene synthesis pathway. Temporal and spatial variation is observed in both metabolites, which are known for having feeding-deterrent and antimicrobial properties (Slattery et al., 2001). Variation in secondary metabolite content or phenotypic plasticity may be an adaptation to several environmental factors. Preliminary evidence suggests that intensity of predation by butterflyfishes, Chaetodon unimaculatus and C. melannotus, may be responsible for differential chemical synthesis (Farmer and Ryan, 1992; Slattery et al., 2001; Van Alstyne et al., 2001; Puyana et al., 2003).

Chemical defenses can be induced through several mechanisms. Activation, or pre-formed defense, is often found when a quick response is necessary. This type of defense involves the storage of nontoxic or mildly toxic compounds that are mixed to form compounds with a greater deterrent effect (either by enzymatic reaction or by synergistic effects of the compounds) immediately after tissue damage and cellular disruption (Karban and Baldwin, 1997; Van Alstyne et al., 2001; Van Alstyne and Houser, 2003). Other pre-formed defenses occur when defensive compounds are quickly transported to the site of feeding damage (Dussourd and Eisner, 1987). Induced responses occurring on longer time scales (days to weeks) are often controlled by synthesis mechanisms (Karban and Baldwin, 1997). For example, methyl jasmonate, a molecule involved in the jasmonate synthesis pathway and a common plant secondary metabolite, induces the production of proteinase inhibitors in plants (Farmer and Ryan, 1990, 1992). Proteinase inhibitors disable endo- and exopeptidases, thus interfering with the digestion processes of many herbivores. Because secondary metabolites produced by marine and terrestrial organisms span many chemical classes, control of chemical induction may employ other biosynthetic pathways. Since synthesis of feeding-deterrent secondary metabolites is likely to be triggered by a change in a gene or group of genes, changes in gene expression after an experimental manipulation can often be profiled (Farmer and Ryan, 1992; Van Alstyne et al., 2001; Puyana et al., 2003).

Here we investigate transcriptome-level changes (i.e., changes in gene expression) in the soft coral Sinularia polydactyla, to assess the affect of experimental predation on both biochemical content and gene expression patterns. We employ a novel mRNA-pool profiling technique using an informatics-based analysis of kinetic profiles (Fielman and Marsh, 2005; Marsh and Fielman, 2005). Although other methods exist for profiling transcriptome changes (e.g., ESTs, SAGE, microarray), none are inexpensive or rapid and all depend upon having substantial genomic sequence available. This fact alone makes transcriptome profiling in non-model organisms rare in the literature. The method used in this study allows for quick, high-throughput analysis of sequence complexity in mRNA pools and has been used successfully to compare transcriptome-level differences in soft corals (Hoover et al., 2007b, c). In conjunction with the transcriptome analyses, chemical extracts from each sample were analyzed by HPLC to investigate changes in secondary metabolite levels. Examining phenotypic plasticity of secondary metabolites in S. polydactyla at a molecular level will allow us to describe the relationship between gene expression and defensive metabolite production. Here we attempt to provide direct evidence for changes in gene expression in response to predation as the primary mechanism influencing biochemical phenotypic plasticity.


   Materials and Methods
TOP
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Literature Cited
 
Predation experiment design
Five treatment-control pairs of Sinularia polydactyla colonies were selected and marked at the reef site, Piti Bomb Holes, Guam (13°25'NW; 144°55'E). Soft corals at this site are exposed to shallow waters (≤ 2 m) with a mean annual tidal range of 0.5 m (Slattery et al., 2001). S. polydactyla is a highly branched species. Each colony consists of multiple branches that separate into smaller branches, terminating with multiple fingerlike extensions. In 2004, the experiment ran for 25 days, divided into two periods around a mass spawning event. On days 1–3 of the experiment, five clippings were taken daily from the tips of each treatment coral. Tips are defined as the ends of the multiple extensions of a branch. These tissue pieces were placed in the RNA preservation solution RNAlater (Ambion) immediately after collection in an effort to preserve transcriptome changes that closely follow predation stress. A separate sample of tissue was harvested from a different branch of each treatment coral and frozen at –80 °C for chemical analysis, so as not to confound the biochemical analysis with the RNA analysis. After the first 3 days of the experiment, sample clippings were taken every other day for 7 days (days 5, 7, 9, etc.). Mass spawning occurred on day 13 of the experiment. The treatment corals were subjected to a second round of artificial predation after spawning, starting on day 16. Five clippings were again taken every day for 3 days, and then every other day for the next week. At the end of the experiment, a piece of tissue was harvested from each of the control (non-manipulated) colonies. These samples were frozen at –80 °C for chemical analyses and comparison to treatment samples. In addition, tissues from five randomly selected coral colonies were collected and preserved for gene expression and chemical analyses, serving as baseline reference samples. Using the same experimental design and sampling regime, the experiment was repeated in 2005, this time beginning after the mass spawning event and running for 12 days.

Bite scars appear on S. polydactyla as obvious light or white spots between 1 and 3 mm in diameter, and are visible for about 5 days after a predation event. The clippings taken in both sets of experiments simulate natural predation, with the quantity of tissue taken in each clipping mimicking a bite from a butterflyfish. To distinguish our experimental predation from naturally occurring fish predation, we often refer to this as "artificial predation." Care was taken to select healthy treatment and control coral colonies, excluding those with apparent damage or signs of severe predation.

Secondary metabolite analysis
Tissues were freeze-dried and ground into a powder, and secondary metabolites were extracted using an accelerated solvent extractor (ASE; Dionex 200). Each sample was extracted using 50:50 methanol/methylene chloride under high temperature and pressure. The resulting extracts were concentrated using a Buchi R-200 rotovapor, dried, and weighed. A 10-mg sample of crude extract was dissolved in 250 µl of 40:60 hexane/ethyl acetate, and 10 µl of diluted extract was injected into a Waters 2695 separations module with a Waters 2996 photodiode array detector. Each sample was run on a Phenomenex normal phase, silica column with 60:40 hexane/ethyl acetate at 0.5 ml min–1 for 25 min. Pure 11β-acetoxypukalide was used as a standard and was run under the same HPLC conditions as the experimental samples. Because crude extracts from the experimental samples were used, several peaks representing unknown compounds were present. Since 11β-acetoxypukalide is well characterized and known to discourage fish predation, our chemical analysis focused on changes in this compound as well as total differences across the entire chemical "fingerprint."

Statistical comparisons of 11β-acetoxypukalide concentrations were analyzed using Sidák and ANOVA tests in SPSS 13.0.

RNA extraction and precipitation
Soft coral tissues were preserved in RNAlater. Portions of tissue weighing 100 mg were weighed and placed in 1 ml of Trizol reagent (Invitrogen) overnight. Trizol reagent is a monophasic solution of phenol and guanidine isothiocyanate. RNA extraction is rapid and based on the single-step method developed by Chomczynski and Sacchi (1987). To prevent cross-contamination of samples, each individual was homogenized in Trizol, using a sterile microcentrifuge tube pestle. Extractions were carried out as described in the Trizol manual. Each RNA sample was then resuspended in DEPC-treated water and its integrity checked on an agarose gel. Intact samples were then ethanol-precipitated for future cDNA library synthesis.

cDNA library synthesis
Each RNA sample was quantified using a RiboGreen assay (Molecular Probes) run on a Fluoro-Star Optima fluorimeter. One microgram of RNA was ethanol-precipitated and used in library synthesis following the protocol presented in Marsh and Fielman (2005). First-strand synthesis was accomplished using a "switch-prime" method that yields single-stranded cDNA copies with unique oligo adapters on the 3' and 5' ends (Stirewalt et al., 2004). The RNA pellet was resuspended in a 10-µl mixture of 1x reverse transcriptase buffer, 1 µmol l–1 XDNA oligo (5'–GAG AGA CTC GAG AGA GAC CGT CTA CGA ATT CTG CAT C (T)24-3’), and RNasin (20 units). The XDNA oligo was annealed to the RNA transcripts by incubating each sample at 45 °C for 5 min, followed by a 5-min incubation at 37 °C. Next, 10 µl of a first-strand reaction mixture, containing 1x reverse transcriptase buffer, 50 mmol l–1 DTT, 2 mmol l–1 dNTPs, 1.0 µmol l–1 Xho-switch oligo (5'-AGA TGT CTC GAG GCG TAC CAG ATG ACG GG–3'), and 350 units of Superscript III reverse transcriptase (Invitrogen), was added to each sample. The entire mixture was then incubated at 37 °C for 12 h.

First-strand cDNAs were ethanol-precipitated after the 12-h incubation. Second-strand cDNA synthesis was accomplished by means of a linear PCR reaction using the Xho-switch nucleotide sequence site on the 3' end of first-strand cDNA molecules. Pellets from the above precipitation were resuspended in a 20-µl PCR reaction containing 0.25 mmol l–1 dNTPs, 1x Ex Taq reaction buffer, 0.5 µmol l–1 ZDNA-F oligo, and 0.5 units of Ex Taq DNA polymerase (Takara). These PCR reactions were amplified on a thermal cycler using 25 cycles of the following: 30 s at 94 °C, 15 s at 58 °C, 6 min at 72 °C. After linear amplification, cDNA libraries were amplified in a 50-µl PCR reaction containing 1 µmol l–1 ZDNA-F, 1 µmol l–1 XDNA-R, 0.5 mmol l–1 dNTPs, 1x Ex Taq reaction buffer, and 0.5 units of Ex Taq. Three microliters of the linear amplification reaction were used as the cDNA template for each reaction. A thermal cycler was used with the following profile for 35 cycles: 30 s at 94 °C, 15 s at 55 °C, 6 min at 72 °C.

Amplified cDNA libraries were ethanol-precipitated and resuspended in 50 µl of distilled water. A PicoGreen (Molecular Probes) assay was used to quantify the nucleic acid concentration, and libraries were checked visually on agarose gels.

Reannealing rate assays
For 2004 samples, reannealing rate assays, described below, were run in a 96-well-plate format on a quantitative PCR machine (ABI 7700); 2005 samples were run in a 384-well-plate format on the similar ABI 7900. All assays were run with 2x SSC, 18% formamide, 5% DMSO, 0.05% Tween-20, and 0.005x PicoGreen with 200 ng of cDNA in a final volume of 20 µl in each well. Four replicate wells were run for each sample. After being placed in the ABI machine, the samples were heat-denatured at 95 °C for 10 min. After denaturation, the temperature profile stepped down by 1 °C increments. During each 30-min temperature plateau, the fluorescence produced by the PicoGreen fluorophore in response to changes in the concentration of double-stranded DNA was measured. For the data presented here, the first temperature plateau occurred at 89 °C and continued down to 55 °C, for a total of 35 temperature plateaus. This type of assay requires 18 h to run and produces a data file containing raw fluorescence traces for each well at each temperature point (Hoover et al., 2007c). Comparisons of data run on both the ABI 7700 and the ABI 7900 showed no significant differences in overall gene expression patterns. Profiles from the ABI 7900 did have higher values of Hflr, a Shannon-Weaver–type entropy statistic that reflects complexity estimates among different samples on the basis of kinetic rates and maximun fluorescence, and appeared to have smoother transitions between each temperature step, but the relative transcriptome profile remained the same.

Data were collected from reannealing rate assays for both experimental and baseline (non-manipulated) samples from the start, midpoint, and end of the experiment for five individuals in each treatment, with four well duplicates for each individual (total n = 80).

Data analysis
Raw fluorescence data files were processed using a C++ program called RotKin, ver. 2.05a (ABI 7700) or 2.08 (ABI 7900), written for MS-Windows (Marsh and Fielman, 2005; Hoover et al., 2007c). This program sequentially separates the continuous data for each well into the experimental temperature plateaus, filters the fluorescence data to increase the signal-to-noise ratio, and normalizes each fluorescence data set. It then iteratively fits the fluorescence versus time data for each well at each temperature to a second-order kinetic regression, and finally transforms each data set into a frequency distribution of observations to calculate an information entropy statistic (Shannon, 1948; Weaver, 1948; Shpak and Churchill, 2000; Marsh et al., 2006).

The reannealing profiles are fit to a second-order kinetic regression, using an equation that describes fluorescence as a function of time. The kinetic-regression statistics are primarily used for quality control to automatically screen more than 3300 reannealing profiles that are produced on a single run with a 96-well plate. The fluorescence profiles for each well at each temperature are subsequently converted into a frequency distribution based on maximum fluorescence for that particular profile. These distributions are on a 0–100% scale, with 5% bin intervals. Information complexity of the reannealing curves can then be calculated from the frequency distribution, using a Shannon-Weaver–type entropy statistic: H'= –1 x {sum} pi x log2 (pi), where pi = the proportion of the ith observation in the total distribution. Thus, the Shannon-Weaver statistic can quantitatively represent the sequence complexity of gene transcripts in a cDNA library, which integrates information about both the distribution and abundance of mRNA sequences in a transcriptome (Marsh and Fielman, 2005; Hoover et al., 2007c). The resulting statistic, Hflr, estimates complexity, distribution, and abundance of mRNA transcripts in the transcriptome among different samples on the basis of kinetic rates and maximum fluorescence. Hflr is calculated using the following equation: Hflr = –Fmax x {sum} pi x log2 (pi), where Fmax is the maximum fluorescence in a fluorescence profile. Using these calculations based on kinetic rates and fluorescence (Hflr or complexity), the different treatment groups were examined for transcriptome-level differences at both the inter-individual and between-treatment groups levels.

For each of the samples, run in replicates of four, a mean Hflr value and standard error were calculated. Mean Hflr values were log-transformed and graphed against temperature (°C) to yield a transcriptome profile for each sample. Thus, differences in Hflr between two samples or groups indicate changes in the distribution and abundance in the pool of mRNA transcripts.

Changes in gene expression among the different treatment groups were further analyzed using a median normalization strategy. Here, a median Hflr value for all individuals being compared was calculated for each temperature. Next, the individual Hflr values were compared to the median at each temperature. These data were analyzed using a hierarchical clustering strategy with a Euclidean distance similarity metric in the program Cluster 3.0 (Eisen et al., 1998). The resulting dendrograms were visualized with Tree View 1.60 (Page, 1996). These dendrograms were based on median normalized data in a temperature-by-individual matrix that uncovered patterns that are not evident when looking at the raw profiles of gene expression.

A better understanding of the relationship between transcriptome profile and secondary metabolite data was achieved by integration of these two types of data using principal component analysis (PCA). PCA aims to find underlying dimensions or factors that can be used to explain complex data sets (Norusis, 1990). PCA searches for a linear combination of variables that extracts the maximum variance from the variables. Next, it removes this variance and seeks a second linear combination that explains the maximum proportion of the remaining variance. This process continues for each component. Data for PCA are in the form of ratios of the maximum value for each variable.

An integrated data matrix was compiled with all transcriptome complexity and metabolite concentrations (known and unknown) for the PCA analysis. The unique aspect of our analytical approach is that we can relate high-throughput data sets for both transcriptome and metabolome constituents.


   Results
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Literature Cited
 
Secondary metabolite data
Based on HPLC analyses, differences in the chemical "fingerprint" (HPLC profile) and in 11β-acetoxypukalide concentrations were found among treatment groups in both the 2004 and 2005 experiments. For example, an HPLC trace of a crude extract from an individual exposed to 24 days of artificial predation in the 2004 experiment shows obvious increases in secondary metabolite content compared to initial and control samples (Fig. 1). Although an obvious increase in 11β-acetoxypukalide concentration is observed, many peaks in the resulting HPLC traces represent unidentified compounds. Such unidentified compounds may also play a role in feeding deterrence or be biologically active, so changes in their concentration should be considered.


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Figure 1. Representative HPLC traces of crude extracts from control versus endpoint treatment groups (data plotted are from the 2004 experiment, in which endpoint = T24). Top = T24 control; bottom = T24 experimental. There is a marked increase in the feeding deterrent compound, 11β-acetoxypukalide, represented by the peak at about 5.5 min. Additional increases of unidentified compounds represented by peaks between 2.5 and 5 min suggest that artificial predation is causing a change in the secondary metabolite "fingerprint" between control and experimental samples.

 
In addition to changes across the entire secondary metabolite profile, differences in the known feeding-deterrent compound 11β-acetoxypukalide were also observed. This compound is represented by a peak with a retention time of about 5.5 min. Concentrations of 11β-acetoxypukalide in each sample were calculated on the basis of standard curves of pure compound analyzed by means of HPLC.

In both the 2004 and 2005 experiments, statistically significant differences in 11β-acetoxypukalide concentration were seen between groups exposed to artificial predation and those that were not manipulated (Fig. 2). For both years, 11β-acetoxypukalide concentrations were significantly higher in endpoint samples from treatment groups than in either initial (T0) or control samples. An ANOVA with Sidák tests for multiple comparisons was conducted for each experiment. For 2004 data, the ANOVA showed highly significant differences in 11β-acetoxypukalide concentration, with P ≤ 0.008. Multiple comparisons indicated that endpoint 11β-acetoxypukalide concentrations were significantly different from T0 and control samples (P ≤ 0.019). Control and T0 samples were not significantly different from each other, suggesting that the increase in secondary metabolite concentration is driven by increased predation pressure. Similarly, for 2005 data, the ANOVA resulted in a value of P ≤ 0.001, again indicating significant differences in 11β-acetoxypukalide concentration among different treatment groups. As in the 2004 experiment, multiple comparisons showed statistically significant differences in 11β-acetoxypukalide concentration in mid- and endpoint samples versus controls (P ≤ 0.015). Again, control and T0 samples were not significantly different from each other.


Figure 2
View larger version (12K):
[in this window]
[in a new window]

  		Figure 2. Histograms showing differences in 11β-acetoxypukalide concentration among different treatment groups for artificial predation experiments conducted in 2004 and 2005 (n = 5 for each group). Identical letters above the bars indicate no significant difference (P > 0.05) in ANOVA test. Bars show average concentration; error bars indicate standard error. Labels underneath bars indicate the timepoint at which the samples were taken.

 
Transcriptome profile data
Definitive molecular changes were observed between the start and end of each artificial predation experiment. S. polydactyla transcriptome profiles were characterized by a dominant peak at or around 64 °C. A second peak was observed around 77 °C (Fig. 3). Due to a temperature deviation of the ABI 7700 at 73 °C, data at that temperature point were removed from all analyses. That temperature point seemed to be a transition between high and mid-range temperatures for the ABI 7700 heating system; we did not see evidence of temperature fluctuation at any other temperatures. No temperature deviation was observed on the ABI 7900, so data from the full temperature range are used for 2005 samples.


Figure 3
View larger version (22K):
[in this window]
[in a new window]

  		Figure 3. A log scale plot comparing mean transcriptome profiles of T0 ({blacktriangleup}), T10 ({circ}), and T24 ({nabla}) samples from 2004 (n = 5 individuals per group, with 4 well duplicates for each indivdual). Error bars represent the standard error. As artificial predation continues, transcriptome complexity shifts from a dominant peak at 77 °C to a peak at about 64 °C. Additionally, the class of transcripts reannealing at 64 °C shows large changes in complexity from T0 to T24. Due to a temperature fluctuation at 73 °C, data from this point are omitted.

 
Comparisons of transcriptome profiles showed noticeable differences in the distribution and abundance of mRNA transcripts between samples collected at the beginning of the experiment and those that underwent from 10 to 24 days of artificial predation. Comparisons of 2004 samples from timepoints T0, T10, and T24 showed considerable changes in the class of transcripts reannealing at 64 °C (Fig. 3). The profile shown represents the mean of 20 transcriptome profiles; n = 5 individuals per group with four well duplicates for each individual. Before artificial predation, transcriptome profiles had two main peaks, at 64 and 77 °C. When samples were subjected to continued artificial predation, their transcriptome profiles shifted to having a larger peak at 64 °C. Furthermore, the magnitude of this peak increased throughout the experiment, suggesting a large upregulation in transcripts reannealing at 64 °C. In a comparison of complexity statistics at 64 °C, response ratios of Hflr between T0 and T24 were markedly different, with the T24 group having a 29-fold increase from T0. Likewise, obvious differences were observed in comparisons of individuals at T0 and T12 (n = 20 transcriptome profiles in each group) in the 2005 experiment at 64 °C with a 1.6-fold increase in complexity from T0 to T12.

In Figure 4, which illustrates gene expression patterns explored using median normalization plots, the lighter colors indicate values above the median, or upregulation, and darker colors indicate values below the median, or downregulation. Samples from T0 formed a distinct cluster from endpoint samples (T24 and T12 for the 2004 and 2005 experiments, respectively). This may indicate that S. polydactyla responds to artificial predation or continued intense predation by upregulating certain classes of genes.


Figure 4
View larger version (37K):
[in this window]
[in a new window]

  		Figure 4. (Upper panel) Comparison showing the gene expression changes over the 24-day artificial predation experiment in 2004. (Lower panel) Comparison showing the gene expression changes over the 12-day artificial predation experiment in 2005. Both diagrams were created using median normalization data clustered with Tree View 1.60. Temperatures (°C) run from high to low across the top of the matrix; sample labels run down the right side of the matrix. Light shades indicate values above the median, or upregulation. Dark colors indicate a value below the median, or downregulation. Correlation values shown at the nodes.

 
Integrating secondary metabolite and transcriptome data
Exposure to experimental predation resulted in significant increases in 11β-acetoxypukalide from baseline control samples. Similarly, profiles of mRNA pool complexity resulted in distinct clustering of samples exposed to artificial predation versus those that were not exposed. These observations raised the question of whether 11β-acetoxypukalide concentration was correlated with the Hflr value at each temperature. Twelve samples from the 2004 artificial predation experiment were used to calculate correlation coefficients between 11β-acetoxypukalide concentration and mRNA pool complexity data. Using a nonparametric test, Kendall’s coefficient of rank, secondary metabolite concentration was found to be correlated to Hflr at 65–68 °C with correlation coefficients of 0.38 to 0.66 (Fig. 5), which means that 60% to 80% of the variance in 11β-acetoxypukalide concentrations was accounted for by the mRNA pool complexity of the 65–68 °C kinetic transcript classes. These correlation coefficients were statistically significant, with P values ranging between 0.04 and 0.002. Principal component analysis of data collected from the artificial predation experiment resulted in distinct clusters of baseline, T0, and T24 (endpoint) samples (Fig. 6). Here, the first two components account for 61% of the sample variance. High correlations (r2 = 0.66) between 11β-acetoxypukalide concentration and gene expression at 64 °C suggest that the transcripts reannealing at this temperature are involved in defensive metabolite production.


Figure 5
View larger version (15K):
[in this window]
[in a new window]

  		Figure 5. Graph of nonparametric correlation coefficients of 11β-acetoxypukalide and Hflr value at each temperature. Individuals from start, endpoint, and control groups from the 2004 predation experiment were included (n =12) in the correlation calculation. The correlation coefficients seen at temperatures between 65 and 69 °C were highly significant (indicated by *) with P values = 0.002 to 0.04. The quantity of 11β-acetoxypukalide is correlated (r = 0.66) with the peak in gene expression profiles seen at about 65–66 °C.

 

Figure 6
View larger version (9K):
[in this window]
[in a new window]

  		Figure 6. Principal component analysis of data from the artificial predation experiment. Distinct clusters are formed for each treatment group. T0 is represented by the circles, T24 by squares, and baseline by triangles.

 

   Discussion
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Literature Cited
 
Artificial predation as a trigger for secondary metabolite production is not prevalent in marine systems, and chemical defenses have not been induced in some previous artificial predation experiments (Hartley and Lawton, 1987, 1991). Here we present data that supports induction of chemical defenses in the soft coral Sinularia polydactyla. Although small inter-individual variations do occur, the largest changes in transcriptome complexity occurred, for all the samples analyzed, in the class of transcripts reannealing at about 64–65 °C. The overall similarity of gene expression patterns of different treatment groups suggests that responses to predation stress may be controlled by a small number of gene activation events. This suggests (1) a greater potential for inter-individual variation between activity states in individuals independent of the level of predation stress and (2) a greater degree of plasticity because gene on/off states may not be a strict binary function. In addition, the correlation between secondary metabolite concentration and Hflr values at 64–65 °C may indicate that transcripts reannealing at these temperatures are responsible for regulation or production (or both) of defensive secondary metabolites. Furthermore, the distinct clustering of predation-exposed corals from non-manipulated corals in the PCA plot supports the hypothesis that predation intensity is responsible for the observed changes in metabolite and transcriptome complexity.

The marked increase in transcriptome complexity at 64 °C and the correlation to secondary metabolite production suggest that an inducible gene expression event is occurring in response to increased predation. The sharp transcriptome complexity increase indicates that the gene or genes that produce the transcripts reannealing at 64 °C are being upregulated. Thus changes in the metabolic pathways involved in the stress response (increased metabolite production) seem to be targeted, not general, such that stress activates one specific pathway rather than multiple pathways. Although the exact mechanism responsible for the upregulation of these genes is unknown, one possibility is that tissue loss triggers the production of polypeptide growth factors. These growth factors may bind to receptors in the genes involved in the secondary metabolite pathway and ultimately increase transcription of genes that produce secondary metabolites. Certainly, there may be a relationship between genes controlling growth or wound healing and those controlling secondary metabolite production. In fact, some may argue that the gene upregulation we observed was actually that of repair genes. However, the correlation we found between 11β-acetoxypukalide and Hflr supports the idea of upregulation of transcripts involved in producing defensive metabolites. Furthermore, the transcriptome profile peak at 77 °C is evident in all coral samples, including baseline controls. Considering that these corals are constantly exposed to environmental stressors necessitating cellular repair, it is more plausible that the 77 °C peak is associated with expression of repair genes, whereas the 64 °C peak is associated with metabolite production. Future studies should elucidate which genes are being expressed in Sinularia to produce defensive secondary metabolites.

Additionally, differences in secondary metabolite production in S. polydactyla may be sex-related. A recent study indicated significant differences in defensive diterpene production between male and female clones of the soft coral Sarcophyton glaucum (Fleury et al., 2006). This finding presents interesting questions for future studies of S. polydactyla. Perhaps a clearer picture of secondary metabolite production could be gained by comparing male and female clones of S. polydactyla at both the biochemical and gene expression levels.

Previous attempts to identify the genes responsible for 11β-acetoxypukalide regulation focused on isolating and measuring changes in 3-hydroxy-3-methylglutaryl-CoA (HMG CoA reductase), a central enzyme in common pathways for terpene synthesis (Karban and Baldwin, 1997). These attempts were unsuccessful at isolating or identifying an orthologous gene of HMG CoA reductase. As an alternative method, the correlation of 11β-acetoxypukalide concentration to a class of mRNA transcripts presents another potential avenue for identifying functionally relevant genes. The isolation of cDNA fractions that are kinetically labile at 64–65 °C could produce a small subset of gene targets that may be related to the synthesis of secondary metabolites, all without any a priori knowledge of what specific metabolic functions may be represented in that pool (i.e., without having to generate a list of specific target genes for designing oligo primers or probes). Isolation of these transcripts may provide us with a few highly represented transcripts that could indicate what the most active metabolic pathways might be after the induction of chemical defenses in response to predation.

The reannealing rate assay method used in this study has advantages in that it is rapid, relatively inexpensive, and high-throughput. For work with natural populations of non-model organisms, it is very important to be able to sample at a statistically significant scale. Few studies have looked at gene expression patterns in non-model organisms; those that have, used sample sizes of less than 20 individuals, primarily because of limitations in handling microarray-based measurements of gene expression (Oleksiak et al., 2001, 2002; Hashimoto et al., 2004). For a marine invertebrate like S. polydactyla, few sequence data are available, making microarray work difficult. In this study, we have profiled the transcriptome complexity of more than 50 individual samples, allowing us to describe significant changes in gene expression patterns in response to an artificial predation stress, and providing a detailed assessment of the interindividual variance in transcriptome activities among individuals. This ability to screen mRNA samples for specific characteristics indicative of transcriptional activation is important for subsequent studies to identify specific genes involved in response pathways.

An additional advantage of the method we have developed here is the statistical integration of the resulting transcriptome profile data with other biochemical response data sets. Here, secondary metabolite levels were highly correlated with the transcriptome data at 64–65 °C. Relationships between transcriptome and secondary metabolite data were validated by the linkage that was shown to exist for 11β-acetoxypukalide, a known inducible feeding deterrent in this soft coral. However, this method also revealed other potentially important metabolites worth further characterization because of the apparent linkage to gene expression activities associated with predation stress. Our cross-correlations of molecular and biochemical data in response to an ecological-level process (predation-wound stress) make this transcriptome profiling approach highly applicable for research focusing on the molecular ecology of any non-model organism collected from field populations.

The relatively recent ability to apply genomics-based assays to non-model organisms, such as S. polydactyla, opens enormous potential to better understand the impact of environmental changes on coral reef habitats and marine ecosystems. Given recent interest in marine and plant-derived secondary metabolites for biomedical use and drug development (Fenical, 2006), a better understanding of the relationship between gene expression and the pathways controlling secondary metabolite production is necessary. Therefore, new techniques like the one we describe here offer an inexpensive, high-throughput mechanism to integrate transcriptome and metabolite data (for review, see Hoover et al., 2007a). The application of genomics-based assays to ecological studies, especially those investigating the impacts of environmental stressors, will allow us to identify the direct linkages that exist between molecular/genetic and biochemical/physiological levels of biological organization.


   Acknowledgments
 
We thank P. Schupp for providing laboratory space and logistical support at the University of Guam Marine Laboratory. We also recognize D. Gochfeld, A. Palmer, F. Camacho, and E. Pinder for their assistance with field collections. For their help with secondary metabolite extractions and analyses, we thank S. Ankisetty, S. Elshahawi, and D. Burt. This work was supported by a grant from the NOAA National Institute for Undersea Technology (A.G.M.), a University of Delaware Presidential Fellowship (C.A.H), and a University of Delaware Graduate Scholars Fellowship (C.A.H.).


   Footnotes
 
Received 19 October 2007; accepted 30 January 2008.


   Literature Cited
TOP
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Literature Cited
 

    Adie, B., J. M. Chico, I. Rubio-Somoza, and R. Solano. 2007. Modulation of plant defenses by ethylene. J. Plant Growth Regul. 26: 160–177.
    Assmann, M., E. Lichte, J. R. Pawlik, and M. Kock. 2000. Chemical defenses of the Caribbean sponges Agelas wiedenmayeri and Agelas conifera. Mar. Ecol. Prog. Ser. 207: 255–262.
    Callaway, R. M., S. C. Pennings, and C. L. Richards. 2003. Phenotypic plasticity and interactions among plants. Ecology 84: 1115–1128.[ISI]
    Ceh, J., M. Molis, T. M. Dzeha, and M. Wahl. 2005. Induction and reduction of anti-herbivore defenses in brown and red macroalgae off the Kenyan coast. J. Phycol. 41: 726–731.[ISI]
    Chomczynski, P., and N. Sacchi. 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate phenol chloroform extraction. Anal. Biochem. 162: 156–159.[ISI][Medline]
    Cronin, G., and M. E. Hay. 1996. Induction of seaweed chemical defenses by amphipod grazing. Ecology 77: 2287–2301.[ISI]
    Deal, M. S., M. E. Hay, D. Wilson, and W. Fenical. 2003. Galactolipids rather than phlorotannins as herbivore deterrents in the brown seaweed Fucus vesiculosus. Oecologia 136: 107–114.[ISI][Medline]
    Dussourd, D. E., and T. Eisner. 1987. Vein-cutting behavior—insect counterploy to the latex defense of plants. Science 237: 898–901.[Abstract/Free Full Text]
    Eisen, M. B., P. T. Spellman, P. O. Brown, and D. Botstein. 1998. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95: 14863–14868.[Abstract/Free Full Text]
    Farmer, E. E., and C. A. Ryan. 1990. Interplant communication —Airborne methyl jasmonate induces synthesis of proteinase-inhibitors in plant leaves. Proc. Natl. Acad. Sci. USA 87: 7713–7716.[Abstract/Free Full Text]
    Farmer, E. E., and C. A. Ryan. 1992. Octadecanoid precursors of jasmonic acid activate the synthesis of wound-inducible proteinase-inhibitors. Plant Cell 4: 129–134.[Abstract/Free Full Text]
    Fenical, W. 2006. Marine pharmaceuticals past, present, and future. Oceanography 19: 111–119.
    Fielman, K. T., and A. G. Marsh. 2005. Genome complexity and repetitive DNA in metazoans from extreme marine environments. Gene 362: 98–108.[ISI][Medline]
    Fleury, B. G., J. C. Coll, E. Tentori, S. Duquesne, and L. Figueiredo. 2000. Effect of nutrient enrichment on the complementary (secondary) metabolite composition of the soft coral Sarcophyton ehrenbergi (Cnidaria : Octocorallia : Alcyonaceae) of the Great Barrier Reef. Mar. Biol. 136: 63–68.
    Fleury, B. G., J. Coll, and P. Sammarco. 2006. Complementary (secondary) metabolites in a soft coral: sex-specific variability, inter-clonal variability, and competition. Mar. Ecol. 27: 204–218.
    Hammerschmidt, R. 1999. Phytoalexins: What have we learned after 60 years? Annu. Rev. Phytopathol. 37: 285–306.[ISI][Medline]
    Hartley, S. E., and J. H. Lawton. 1987. Effects of different types of damage on the chemistry of birch foliage, and the responses of birch feeding insects. Oecologia 74: 432–437.[ISI]
    Hartley, S. E., and J. H. Lawton. 1991. Biochemical aspects and significance of the rapidly induced accumulation of phenolics in birch foliage. Pp. 105–132 in Phytochemical Induction by Herbivores, D. W. Tallamy and M. J. Raupp, eds. John Wiley, New York.
    Hashimoto, K., T. Shibuno, E. Murayama-Kayano, H. Tanaka, and T. Kayano. 2004. Isolation and characterization of stress-responsive genes from the scleractinian coral Pocillopora damicornis. Coral Reefs 23: 485–491.[ISI]
    Hay, M. E. 2001. Seaweed chemical ecology: its importance to individuals, populations, and communities. J. Phycol. 37: 22–23.[Medline]
    Herms, D. A., and W. J. Mattson. 1992. The dilemma of plants—to grow or defend. Q. Rev. Biol. 67: 283–335.
    Hoover, C. A., M. Slattery, and A. G. Marsh. 2007a. A functional approach to transcriptome profiling: linking gene expression patterns to metabolites that matter. Mar. Biotechnol. 9: 411–419.[Medline]
    Hoover, C. A., M. Slattery, and A. G. Marsh. 2007b. Gene expression profiling of two related soft corals, Sinularia polydactyla and S. maxima, and their putative hybrid at different life-history stages. Comp. Biochem. Physiol. D 2: 135–143.
    Hoover, C. A., M. Slattery, and A. G. Marsh. 2007c. Profiling transcriptome complexity and secondary metabolite synthesis in a benthic soft coral, Sinularia polydactyla. Mar. Biotechnol. 9: 166–178.[Medline]
    Kamel, H. N., and M. Slattery. 2005. Terpenoids of Sinularia: chemistry and biomedical applications. Pharm. Biol. 43: 253–269.
    Kamel, H. N., F. R. Fronczek, N. H. Fischer, and M. Slattery. 2004. A novel metabolite from the hybrid soft coral Sinularia maxima x Sinularia polydactyla: a biosynthetically mixed skeleton linking cembrane and africanane terpenoids. Tetrahedron Lett. 45: 1995–1997.
    Karban, R., and I. T. Baldwin. 1997. Induced Responses to Herbivory. University of Chicago Press, Chicago.
    Kubanek, J., J. R. Pawlik, T. M. Eve, and W. Fenical. 2000. Triterpene glycosides defend the Caribbean reef sponge Erylus formosus from predatory fishes. Mar. Ecol. Prog. Ser. 207: 69–77.
    La Barre, S. C., J. C. Coll, and P. W. Sammarco. 1986. Defensive strategies of soft corals (Coelenterata: Octocorallia) of the Great Barrier Reef. II. The relationship between toxicity and feeding deterrence. Biol Bull. 171: 565–576.[Abstract/Free Full Text]
    Leone, P. A., B. F. Bowden, A. R. Carroll, and J. C. Coll. 1995. Chemical consequences of relocation of the soft coral Lobophytum compactum and its placement in contact with the red alga Plocamium hamatum. Mar. Biol. 122: 675–679.
    Marsh, A. G., and K. T. Fielman. 2005. Transcriptome profiling of individual larvae of two different developmental modes in the poecilogonous polychaete Streblospio benedicti (Spionidae). J. Exp. Zool. B 304B: 238–249.
    Marsh, A. G., Y. Zeng, and J. Garcia-Frias. 2006. The expansion of information in ecological systems: emergence as a quantifiable state. Ecol. Inform. 19: 107–116.
    McClintock, J. B., M. O. Amsler, C. D. Amsler, K. J. Southworth, C. Petrie, and B. J. Baker. 2004. Biochemical composition, energy content and chemical antifeedant and antifoulant defenses of the colonial Antarctic ascidian Distaplia cylindrica. Mar. Biol. 145: 885–894.
    Norusis, M. J. 1990. Factor analysis: procedure FACTOR. Pp. 321–342 in SPSS Introductory Statistics Student Guide. SPSS, Inc., Chicago.
    Oleksiak, M. F., K. J. Kolell, and D. L. Crawford. 2001. Utility of natural populations for microarray analyses: isolation of genes necessary for functional genomic studies. Mar. Biotechnol. 3: S203-S211.[Medline]
    Oleksiak, M. F., G. A. Churchill, and D. L. Crawford. 2002. Variation in gene expression within and among natural populations. Nat. Genet. 32: 261–266.[ISI][Medline]
    Page, R. D. M. 1996. TreeView: an application to display phylogenetic trees on personal computers. Comput. Appl. Biosci. 12: 357–358.[Free Full Text]
    Pavia, H., and G. B. Toth. 2000. Inducible chemical resistance to herbivory in the brown seaweed Ascophyllum nodosum. Ecology 81: 3212–3225.[ISI]
    Pawlik, J. R. 1993. Marine invertebrate chemical defenses. Chem. Rev. 93: 1911–1922.
    Puyana, M., W. Fenical, and J. R. Pawlik. 2003. Are there activated chemical defenses in sponges of the genus Aplysina from the Caribbean? Mar. Ecol. Prog. Ser. 246: 127–135.
    Schnitzler, I., G. Pohnert, M. Hay, and W. Boland. 2001. Chemical defense of brown algae (Dictyopteris spp.) against the herbivorous amphipod Ampithoe longimana. Oecologia 126: 515–521.[ISI]
    Schoeppner, N. M., and R. A. Relyea. 2005. Damage, digestion, and defence: the roles of alarm cues and kairomones for inducing prey defences. Ecol. Lett. 8: 505–512.[ISI]
    Shannon, C. E. 1948. A mathematical theory of communication. AT & T Tech. J. 27: 379–423.
    Shpak, M., and G. A. Churchill. 2000. The information content of a character under a Markov model of evolution. Mol. Phylogenet. Evol. 17: 231–243.[ISI][Medline]
    Slattery, M., C. Avila, J. Starmer, and V. J. Paul. 1998. A sequestered soft coral diterpene in the aeolid nudibranch Phyllodesmium guamensis Avila, Ballesteros, Slattery, Starmer and Paul. J. Exp. Mar. Biol. Ecol. 226: 33–49.[ISI]
    Slattery, M., G. A. Hines, J. Starmer, and V. J. Paul. 1999. Chemical signals in gametogenesis, spawning, and larval settlement and defense of the soft coral Sinularia polydactyla. Coral Reefs 18: 75–84.[ISI]
    Slattery, M., J. Starmer, and V. J. Paul. 2001. Temporal and spatial variation in defensive metabolites of the tropical Pacific soft corals Sinularia maxima and S. polydactyla. Mar. Biol. 138: 1183–1193.
    Stirewalt, D. L., E. L. Pogosova-Agadjanyan, N. Khalid, D. R. Hare, P. A. Ladne, O. Sala-Torra, L. P. Zhao, and J. P. Radich. 2004. Single-stranded linear amplification protocol results in reproducible and reliable microarray data from nanogram amounts of starting RNA. Genomics 83: 321–331.[ISI][Medline]
    Stout, M. J., K. V. Workman, R. M. Bostock, and S. S. Duffey. 1998. Specificity of induced resistance in the tomato, Lycopersicon esculentum. Oecologia 113: 74–81.[ISI]
    Taylor, R. B., E. Sotka, and M. E. Hay. 2002. Tissue-specific induction of herbivore resistance: seaweed response to amphipod grazing. Oecologia 132: 68–76.[ISI]
    Thaler, J. S., A. L. Fidantsef, S. S. Duffey, and R. M. Bostock. 1999. Trade-offs in plant defense against pathogens and herbivores: a field demonstration of chemical elicitors of induced resistance. J. Chem. Ecol. 25: 1597–1609.[ISI]
    Thornton, R. S., and R. G. Kerr. 2002. Induction of pseudopterosin biosynthesis in the gorgonian Pseudopterogorgia elisabethae. J. Chem. Ecol. 28: 2083–2090.
    Toth, G. B., O. Langhamer, and H. Pavia. 2005. Inducible and constitutive defenses of valuable seaweed tissues: consequences for herbivore fitness. Ecology 86: 612–618.[ISI]
    Van Alstyne, K. L., and L. T. Houser. 2003. Dimethylsulfide release during macroinvertebrate grazing and its role as an activated chemical defense. Mar. Ecol. Prog. Ser. 250: 175–181.
    Van Alstyne, K. L., G. V. Wolfe, T. L. Freidenburg, A. Neill, and C. Hicken. 2001. Activated defense systems in marine macroalgae: evidence for an ecological role for DMSP cleavage. Mar. Ecol. Prog. Ser. 213: 53–65.
    Weaver, W. 1948. Science and complexity. Am. Sci. 36: 536.[Medline]
    Zangerl, A. R., and C. E. Rutledge. 1996. The probability of attack and patterns of constitutive and induced defense: a test of optimal defense theory. Am. Nat. 147: 599–608.[ISI]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hoover, C. A.
Right arrow Articles by Marsh, A. G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hoover, C. A.
Right arrow Articles by Marsh, A. G.
Related Collections
Right arrow Ecology
Right arrow Genomics
Right arrow Cnidarians

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS