Transcriptome profiles of hybrid poplar (Populus trichocarpa × deltoides) reveal rapid changes in undamaged, systemic sink leaves after simulated feeding by forest tent caterpillar (Malacosoma disstria)


  • Ryan N. Philippe,

    1. Michael Smith Laboratories, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    2. Department of Botany, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    Search for more papers by this author
  • Steven G. Ralph,

    1. Michael Smith Laboratories, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    2. Department of Biology, University of North Dakota, Grand Forks, ND 58202-9019, USA
    Search for more papers by this author
  • Shawn D. Mansfield,

    1. Department of Wood Sciences, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    Search for more papers by this author
  • Jörg Bohlmann

    1. Michael Smith Laboratories, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    2. Department of Botany, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    3. Department of Forest Sciences, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
    Search for more papers by this author

Author for correspondence:
Jörg Bohlmann
Tel: +1 604 8220282


  • Poplar has been established as a model tree system for genomic research of the response to biotic stresses. This study describes a series of induced transcriptome changes and the associated physiological characterization of local and systemic responses in hybrid poplar (Populus trichocarpa × deltoides) after simulated herbivory.
  • Responses were measured in local source (LSo), systemic source (SSo), and systemic sink (SSi) leaves following application of forest tent caterpillar (Malacosoma disstria) oral secretions to mechanically wounded leaves.
  • Transcriptome analyses identified spatially and temporally dynamic, distinct patterns of local and systemic gene expression in LSo, SSo and SSi leaves. Galactinol synthase was strongly and rapidly upregulated in SSi leaves. Genome analyses and full-length cDNA cloning established an inventory of poplar galactinol synthases. Induced changes of galactinol and raffinose oligosaccharides were detected by anion-exchange high-pressure liquid chromatography.
  • The LSo leaves showed a rapid and strong transcriptome response compared with a weaker and slower response in adjacent SSo leaves. Surprisingly, the transcriptome response in distant, juvenile SSi leaves was faster and stronger than that observed in SSo leaves. Systemic transcriptome changes of SSi leaves have signatures of rapid change of metabolism and signaling, followed by later induction of defense genes.


Poplar (Populus spp.) trees are ecological keystone species found throughout the northern hemisphere, often inhabiting riparian or plains environments (Whitham et al., 1996). Throughout their lifetime (in some cases up to several hundred years), poplars are exposed to a large variety of insect pests, that inherently have the capacity to evolve at a much faster pace than the long-lived tree species. To cope with the unpredictable array of possible herbivores, poplars deploy a suite of constitutive and inducible, as well as direct and indirect defenses (Philippe & Bohlmann, 2007; Ralph, 2009). Induced defenses allow plants to allocate limiting resources for growth, development and reproduction when not under herbivore stress, which is of benefit for plant fitness (Baldwin, 1998; Mauricio, 1998; Strauss et al., 2002). Poplars can induce defense responses systemically, that is throughout the plant (Parsons et al., 1989; Arimura et al., 2004; Babst et al., 2009), thus providing undamaged tissues and organs with induced resistance to herbivory (Havill & Raffa, 1999).

Several recent studies have investigated the transcriptome responses in the damaged leaves of poplars challenged by real or simulated insect herbivory (Christopher et al., 2004; Lawrence et al., 2006; Ralph et al., 2006; Miranda et al., 2007). Major & Constabel (2006) also compared damaged poplar leaves with undamaged systemic leaves of similar developmental stage (source leaves) and found extensive overlap in these gene expression profiles. Other work demonstrated the importance of source–sink relationships for induced defense in poplars and the heterogeneity of responses between the metabolically distinct leaf groups (Arnold & Schultz, 2002; Arnold et al., 2004; Babst et al., 2008). Recently, Babst et al. (2009) identified overlapping transcript profiles between systemic source and sink leaves of poplars in response to herbivory when a single time-point (22 h after treatment) of the defense response was analysed.

In order to identify spatial and temporal patterns of locally and systemically induced defense responses in sink and source leaves of hybrid poplar (P. trichocarpa × deltoides), we investigated transcriptome changes in leaves of different age and source/sink status over a time-course of 2–24 h after simulated insect attack. Oral secretions (OS) of forest tent caterpillars (FTC, Malacosoma disstria) induce gene expression in poplar and function as authentic mimics of insect herbivory when added to mechanical wounds (Major & Constabel, 2006). We report the analysis of transcriptome profiles of local (treated) source leaves (LSo), systemic (untreated) source (SSo) leaves, and systemic (untreated) sink (SSi) leaves in response to OS application. The results of this study highlight a rapid response in SSi leaves that is distinct compared to profiles from LSo and SSo leaves.

Materials and Methods

Plant and insect materials

All experiments were done with hybrid poplar (P. trichocarpa × deltoides, H11-11). Saplings were propagated, maintained in the glasshouse, treated, and harvested as described in Ralph et al. (2006). Source leaves (leaf plastochron index LPI 9+; Larson & Isebrands, 1971) and juvenile sink leaves (LPI 0–5) were collected for microarray analysis from trees of 150–200 cm in height (Fig. 1a,b). Rearing conditions for M. disstria Hübner (FTC) larvae were as described in Ralph et al. (2006). Collection of FTC OS is described in Philippe et al. (2009) (Fig. 1c). Unless otherwise stated, all reagents and solvents were from Fischer Scientific (Pittsburgh, PA, USA), Sigma-Aldrich (St. Louis, MO, USA), EM Science (Darmstadt, Germany) or Invitrogen (Carlsbad, CA, USA).

Figure 1.

 Plant treatment and sampling. (a) The five lowermost fully-expanded, nonsenescing mature source leaves (LSo) were treated with mechanical wounding followed by the application of forest tent caterpillar (FTC) oral secretions (OS). Leaves were sampled from separate trees 2 h, 6 h or 24 h post-treatment. In addition, the five acropetally adjacent systemic mature source leaves (SSo) and the uppermost juvenile sink leaves (SSi) were also collected. (b) Photograph of LSo, SSo and SSi leaves. The leaves are arranged in the same vertical order as found along the tree axis. Note: SSo and SSi leaves are separated by up to 100–150 cm. Bar , 10 cm. (c) Photograph of FTC OS collection, showing regurgitant collecting on the larvae’s mouth near the opening of a glass capillary connected to a vacuum system. Size standard is 1 cm.

Invertase assay

Sucrose cleavage by acid invertases was assayed by measuring the generation of glucose monomers following a protocol adapted from Arnold & Schultz (2002). Two-hundred milligrams FW of leaf material were ground in liquid nitrogen and extracted in 1 ml of buffer (150 mM Tris-HCl (pH 7.5), 2 mM EDTA, 10 mM ascorbic acid, 5% (w : v) polyvinylpolypyrrolidone (PVPP), 10 mM dithiothreitol (DTT), 2.5 mM benzamidine). Extracts containing soluble acid invertase activity were cleared by centrifugation for 15 min at 18 000 g. Pellet containing cell wall bound acid invertase was washed three times and resuspended with 1 ml extraction buffer without PVPP. A volume of 600 μl of 100 mM sodium acetate (pH 4.5) and 200 μl of 100 mM sucrose were added to 200 μl of each of the two fractions, and incubated for 30 min at 37°C. Reducing sugars formed in the assay were detected with 3,5-dinitrosalicylic acid (DNS) according to Miller (1959), modified with the addition of 15-min incubation at 100°C before cooling to room temperature. Absorbance was measured at 560 nm. Acid invertase activities are reported as μmol sucrose cleaved per gram of tissue FW and minute.

Microarray and quantitative real-time PCR (qPCR) analyses

As described in Philippe et al. (2009), OS treatments consisted of leaves with four tracks of 10 cm-long wounds running parallel to the midvein, made with a fabric wheel, onto which 20 μl of OS was spread with a paintbrush (Fig. 1a). For each tree, the five lowest, fully-expanded, healthy leaves were treated. From each OS-treated and untreated control tree (no wound and no OS treatment) the five lowest healthy leaves (local source leaves; LSo), the five immediately adjacent fully expanded systemic leaves (systemic source leaves; SSo), and the five uppermost juvenile systemic leaves (systemic sink leaves; SSi) were collected at 2, 6 or 24 h after treatment (Fig. 1b), petioles removed, flash frozen in liquid nitrogen, and stored at −80°C. Total RNA was isolated, quantified, and checked for integrity and purity as described in Kolosova et al. (2004). Microarray experiments were designed to comply with MIAME guidelines (Brazma et al., 2001). Details of the 15.5K poplar cDNA microarray platform (NCBI GEO platform number GPL5921) were described in Ralph et al. (2006). Microarray hybridizations, image capture and processing, data normalization and analysis were as previously described (Ralph et al., 2006; Philippe et al., 2009). Scanned microarray TIF images, the gene identification file, and ImaGene quantified data files are available at the NCBI GEO database (series GSE16383). Total RNA from source and sink leaves of OS-treated and untreated control trees was compared using a total of 54 hybridizations (see the Supporting Information, Fig. S1). Details of hybridization design and data analysis are described in Methods S1. The complete set of microarray results is available in Table S1. The qRT-PCR was done as previously described in detail in Ralph et al. (2006), with details of experimental design and analysis described in Methods S1; primers are listed in Table S2.

FLcDNA isolation of galactinol synthase (GOLS) genes

A tblastn search of the Treenomix poplar EST and FLcDNA database (Ralph et al., 2006, 2008) was performed using plant GOLS nucleotide sequences available from GenBank. The CAP3 sequence assembly (Huang & Madan, 1999) was used to group expressed sequence tags (ESTs) into a total of seven different singletons and contigs (40 bp overlap, 95% identity). The corresponding cDNA clones were identified in library glycerol stocks, insert sizes determined and sequenced to high accuracy (GenBank accession numbers EU305718 to EU305724).

Analysis of GOLS sequences and phylogeny

Using blastp analyses of the 41 377 protein-coding gene loci predicted from the poplar genome sequence assembly v2.0 ( we identified GOLS genes in the P. trichocarpa Nisqually-1 genome (Tuskan et al., 2006). As query sequences, we used plant GOLS sequences available in NCBI GenBank and the protein sequences deduced from the seven GOLS cDNAs identified in the Treenomix poplar EST collection. Alignments of multiple amino acid sequences were made with clustalw ( and boxshade (, and manually adjusted before maximum likelihood analysis using phyml, version 2.4.4 (Guindon & Gascuel, 2003) with the JTT (Jones et al., 1992) amino acid substitution matrix. The proportion of invariant sites and the alpha shape parameter were estimated by phyml. Trees were generated using bionj (Gascuel, 1997), a modified neighbour-joining algorithm. SEQBOOT of the phylip v3.66 package (Felsenstein, 1993; was used to generate 100 bootstrap replicates, which were then analysed using phyml and the previously estimated parameters. CONSENSE, also from phylip, was used to create a consensus tree. treeview (Page, 1996) was used to visualize the resultant trees. Bootstrap values above 80% were added to the maximum likelihood tree generated from the original dataset.

Analysis of galactinol and raffinose

Leaves were freeze-dried for 48 h. For each sample 50 mg leaf material was ground with a mortar and pestle in liquid nitrogen, and extracted for 24 h at −20°C with 4 ml of methanol–chloroform–water (12 : 5 : 3). Extracts were centrifuged for 10 min at 5000 g and 4°C, and the supernatant was collected. The pellet was washed with 8 ml of methanol–chloroform–water (12 : 5 : 3), centrifuged for 10 min at 5000 g and 4°C. Combined supernatants were mixed with 5 ml distilled water and, after phase separation, 1 ml of the aqueous phase was removed and dried at 40°C, resuspended in 1 ml distilled, deionized water and filtered through a 4 mm nylon filter (0.45 μm). Soluble carbohydrates were separated and quantified by anion exchange high-pressure liquid chromatography (HPLC) on a DX-600 ion chromatography system equipped with an AS50 autosampler and an ED50 electrochemical detector with gold electrode (Dionex, Sunnyvale, CA, USA). Monomeric sugars were isocratically separated with a 10 μl injection volume on a Carbopac PA-1 (Dionex) anion-exchange column (4 × 250 mm) with distilled, deionized water at room temperature at a flow rate of 1 ml min−1, with a postcolumn addition of 100 mM NaOH before detection. Oligomeric sugars were isocratically separated with a 10 μl injection volume on a Carbopac MA-1 (Dionex) anion-exchange column (4 × 250 mm) with 300 mM NaOH at 0.3 ml min−1. Fucose was used as internal standard for quantitative analysis. Sugar concentrations were determined using regression equations from calibration curves derived from standard solutions of galactinol and raffinose.


Characterization of source–sink relationships

For the characterization of local and systemic responses to simulated insect feeding, we used the lowermost healthy LSo leaves, the immediately adjacent fully expanded SSo leaves and the uppermost juvenile SSi leaves (Fig. 1). As phloem connectivity influences spatial patterns of the systemic defense response in poplar (Davis et al., 1991), we collected groups of five leaves for each leaf type to ensure that orthostichous phloem connections existed between the source and sink leaves (Larson, 1979). We measured soluble invertase (SI, Fig. 2a) and cell wall invertase (CWI, Fig. 2b) activity in leaves of untreated plants to determine the source–sink relationship between leaf groups that correspond to the LSo, SSo and SSi leaves in treated plants. The SI activity did not differ significantly between source and sink leaves (Tukey HSD: LSo vs SSo, P = 0.998; LSo vs SSi, P = 0.998; SSo vs SSi, P = 0.999); the OS treatment did not effect any significant change (Tukey HSD: LSo vs SSo, P = 0.972; LSo vs SSi, P = 0.986; SSo vs SSi, P = 0.982) either. In untreated plants, CWI activity was approximately twofold higher in sink leaves than in source leaves (Tukey HSD: LSo vs SSo, P = 0.999; LSo vs SSi, P = 0.035; SSo vs SSi, P = 0.047). In plants treated with OS, CWI activity increased about twofold relative to undamaged plants after 2 h in LSo, SSo and SSi leaves (Tukey HSD: LSo vs SSo, P = 0.992; LSo vs SSi, P < 0.001; SSo vs SSi, P < 0.001), maintaining the source–sink relationship and potentially increasing phloem loading/unloading capacities with treatment. Two-way ANOVA indicated that CWI activity was influenced by leaf type (P < 0.001) and by OS treatment (P < 0.001), though the interaction term was not significant (P = 0.140).

Figure 2.

 Soluble (SI; a) and insoluble cell wall invertase (CWI; b) activity in source and sink leaves of untreated control and forest tent caterpillar (FTC) oral secretion (OS)-treated poplar trees 2 h after treatment. Open bars, SI or CWI activity in local source (LSo), systemic source (SSo) and systemic sink (SSi) leaves of untreated control trees; tinted bars, SI or CWI activity in LSo, SSo and SSi leaves of OS-treated trees. Values are mean ± SD (n = 5 trees). Data were analysed using two-factor ANOVAs and Tukey multiple comparison tests. Bars with different letters are significantly different at = 0.050; letters are independent such that ‘ac’ is not significantly different from either ‘a’ or ‘c’, while ‘a’ and ‘c’ are significantly different from each other.

Overall spatial and temporal patterns of leaf transcriptomes in response to OS treatment

We used the poplar 15.5K cDNA microarray (Ralph et al., 2006) to examine transcriptome changes in LSo, SSo and SSi leaves in response to OS treatment. Genes that showed changes in transcript abundance (i.e. differentially expressed (DE) genes) were identified using three criteria: at least a 1.5-fold change between the corresponding samples from treated trees and untreated control trees, a Student’s t-test P-value < 0.05 and a Q-value < 0.05. Using these criteria, genes corresponding to approx. 40% of elements on the array were DE in treated trees. Approx. 20% were DE in only one leaf group, indicating leaf type-specific transcriptome responses. The complete set of expression data for all genes represented on the microarray is provided in Table S1.

To accommodate the large number of samples (90 trees for 18 different combinations of time points and leaf types; Fig. S1), the initial transcriptome profiling was done with pooled RNA from five biological replicates for each leaf type and time-point using a total of 54 array hybridizations. To validate these analyses and to assess the variability of the transcriptome responses, we performed additional hybridizations with RNA from each of four independent biological replicates comparing treatment and control of SSi leaves at the 2 h time-point. Results obtained from analysis of pooled and independent replicate samples showed similar variance (Fig. S2), confirming previous validations of microarray analyses with pooled samples from glasshouse-grown clonal poplar (Ralph et al., 2006; Miranda et al., 2007; Philippe et al., 2009).

The overall spatial and temporal patterns of transcriptome changes in response to OS treatment revealed some substantial asymmetry in the three different leaf types (Fig. S3.) The response of LSo leaves to OS was strongest at 2 h (1568 genes upregulated, 938 downregulated) and 6 h (1624 genes upregulated, 900 downregulated) post-treatment with substantially fewer DE genes at 24 h (360 genes upregulated, 58 downregulated). In SSo leaves we detected a slower and weaker response induced by OS treatment with the largest number of upregulated transcript species observed at 24 h (411 genes upregulated, 74 downregulated). In contrast, of all three leaf types, SSi leaves showed the largest number of DE genes at the early time-point 2 h after treatment (1997 genes upregulated, 1632 downregulated). In summary, over the time-course of this analysis, LSo leaves responded rapidly and strongly, the observed response of SSo leaves was slower and weaker, and SSi leaves showed the fastest and strongest induced response.

qPCR reveals differential response of selected genes in source and sink leaves

To validate microarray analyses, we designed gene-specific primers (Table S2) for 10 DE genes and quantified their transcript abundance using qPCR (Fig. 3; Table S3). Genes were selected to cover a relevant range of treatment-induced DE from threefold (histone deacetylase; WS0158_D14; matches Populus v2.0 genome model POPTR_0009s15160; to 24-fold (unknown protein; WS0123_C21; no match in Populus v2.0) change in transcript abundance according to microarray analysis. Expression patterns detected by microarrays were confirmed by qPCR for eight of the ten genes. The remaining two genes were from multigene families, and therefore transcript abundance measured with microarrays may be ambiguous. In general, we observed greater changes in transcript abundance by qPCR than by microarray hybridization. Significant OS-induced changes of transcript abundance ranged from 50-fold downregulation in SSi leaves at 24 h for a putative leucine-rich repeat (LRR) transmembrane protein kinase (WS0205_I02; POPTR_0002s14800) to 2235-fold upregulation for polyphenol oxidase (PPO; POPTR_0001s39660) in SSi leaves at 24 h.

Figure 3.

 Quantitative real-time PCR analysis of local and systemic gene expression of selected genes in poplar leaves in response to simulated forest tent caterpillar (FTC) herbivory. Transcript abundance for each gene was examined in local source (LSo), systemic source (SSo) and systemic sink (SSi) leaves at 2 h, 6 h and 24 h post-treatment. Transcript abundance was normalized to poplar translation initiation factor 5A (TIF5A; WS0116_J23; POPTR_0006s19870) by subtracting the Ct value of each transcript, where ΔCt = Cttranscript − CtTIF5A. Transcript abundance of genes in control (open bars) and oral secretion (OS)-treated (closed bars) samples were obtained from the equation (1 + E)−ΔCt, where E is the PCR efficiency, as described by Ramakers et al. (2003). A transcript with a relative abundance of one is equivalent to the abundance of TIF5A in the same tissue. Error bars show standard error. Statistical significance of expression differences relative to untreated plants was determined using a linear model (see the Materials and Methods and Table S3). Significance thresholds were set at * P < 0.05; **P < 0.01; ***P < 0.001. Black arrows mark statistical differences where bars are small.

The qPCR analysis confirmed that transcriptome changes differ between the three leaf types. For example, the gene with the strongest downregulated transcript abundance in SSi leaves at 2 h (15-fold down; LRR transmembrane protein kinase WS0205_I02; POPTR_0002s14800) showed a 1300-fold upregulation in SSo leaves, highlighting the differences in transcript response of the same gene between systemic source and sink leaves (Fig. 3). Other transcripts also showed upregulation in source leaves and downregulation in sink leaves (e.g. histone deacetylase WS0158_D14; POPTR_0009s15160) or vice versa (e.g. 9-cis-epoxycarotenoid dioxygenase WS0147_P16; POPTR_0019s12320). Some transcripts were most strongly upregulated in sink leaves (e.g. universal stress protein WS0124_D16; POPTR_0014s11710), while others responded most strongly in source leaves (e.g. aminopeptidase M, WS0212_I21; POPTR_0006s24090). We also found genes with upregulation across all three leaf types, with an observed maximum at 24 h (e.g. serine carboxypeptidase S28 WS0214_H20 (POPTR_0001s22060), endochitinase WS0143_A03 (POPTR_0009s14420), Kunitz protease inhibitor (KPI) WS0134_G14 (POPTR_0010s01150), polyphenol oxidase PPO (POPTR_0001s39660), and (−)-germacrene-d synthase TPS1 (POPTR_0001s44080)).

Cluster analysis reveals large-scale differences of OS-induced transcriptome responses in source vs sink and local vs systemic leaves

Next we comprehensively assessed the complete transcriptome data for possible differences in the response across the three different leaf types. Despite the large number of genes showing OS-induced changes in transcript abundance, there was relatively little overlap in the change of specific transcript species across all leaf types and time-points (Table S1), indicating that distinct sets of genes are affected by OS in different leaf types and that these gene sets are activated with temporally distinct profiles. We used the divisive DIANA algorithm (Bryan, 2004) for cluster analysis, to identify global patterns of co-expressed genes in response to OS treatment (Fig. 4; Table S4). Transcripts corresponding to 7231 microarray elements with a fold-change > 1.5, P < 0.05 and Q < 0.05 in at least one leaf type and time-point in response to OS treatment fell into eight unique clusters showing distinct patterns of expression.

Figure 4.

 Cluster analysis of expression profiles of differentially expressed (DE) genes. A set of 7231 genes were identified as DE (fold-change > 1.5 × for treated/control leaves; < 0.05; < 0.05) for at least one time-point and leaf type and then clustered using the divisive DIANA algorithm (Bryan, 2004). For each panel, fold-change expression ratios are plotted for 2 h, 6 h and 24 h post-treatment for local source (LSo), systemic source (SSo) and systemic sink (SSi) leaves. Solid red lines represent the median expression ratio for a given cluster of gray lines, where each gray line represents the expression profile detected with an individual microarray element. For each leaf group, boxplot representations of the expression profile for the 7231 array elements included in this analysis are provided. Each boxplot shows the median value as a line dissecting the box, upper (75%) and lower (25%) quartiles at the top and bottom box edges, the nonoutlier minimum and maximum values as whiskers outside the box (1.5 × the interquartile range), and outlier values (beyond the whiskers) as open circles.

Four large clusters (clusters 1–4) identified in this analysis contain a total of 6841 genes, or 95% of all DE genes (Fig. 4). Cluster 1 contains genes that were upregulated early (2–6 h) mainly in LSo leaves. By contrast, cluster 3 contains genes that were upregulated early (2 h) mainly in SSi leaves, and cluster 4 contains genes that were downregulated early (2 h) mainly in SSi leaves. These three clusters identified genes with predominantly leaf type-specific responses. Cluster 2 contains genes that showed early (2–6 h) upregulation in SSi leaves and downregulation in LSo leaves. Many of the genes upregulated in LSo leaves identified in cluster 1 (1631, or 23%) have annotations associated with signaling, general stress response, primary metabolism or unknown functions (Table S4). The large number of genes in clusters 2, 3 and 4 (5210, or 72%) revealed that changes in SSi leaves comprise the major portion of the total OS-induced transcriptome response observed. These three clusters contain mainly genes annotated as functioning in primary metabolism or general stress responses (Table S4).

Substantially fewer genes are represented in the four additional clusters, 5–8. Cluster 5 contains the small number of genes whose transcripts were upregulated across all three leaf types and most rapidly and transiently increased with their observed peak at 2 h in the treated LSo leaves. This cluster contains genes annotated in primary metabolism, transport, signaling, redox reactions, flavonoid metabolism and volatile organic compound synthesis (Table S4). Cluster 6 also contains genes that responded throughout the plant, and more quickly in LSo leaves than in SSo or SSi leaves, but with overall slower response than genes in Cluster 5. This cluster contains a relatively high number of genes with putative functions in calcium binding or calcium signaling (Table S4). Cluster 7 contains genes with late or sustained response whose upregulation was greatest at 24 h throughout the plant. Genes of this cluster are annotated with functions in defense against insect herbivores, such as polyphenolic oxidase, Kunitz protease inhibitors, endochitinases or octadecanoid signaling, along with several apyrases (Table S4). The very small cluster 8 contains genes that respond in all three leaf types, were rapidly upregulated and were back to basal levels by 24 h in both LSo and SSo leaves, while peaking later at 6 h in SSi leaves. Thus, this cluster highlights differences between source and sink, as opposed to treated and systemic leaves.

The cluster analysis supports the distinct spatial and temporal patterns of transcriptome changes noted in different leaf types with strong responses in LSo leaves, a strong early response in SSi leaves and the generally weaker and later response in SSo leaves.

SSi leaves show a unique and dynamic response to FTC OS

The contrast in expression profiles between LSo and SSi leaves shown by cluster analysis (Fig. 4) highlights the differences of temporal patterns of transcriptome changes between treated and systemic leaves, and between source and sink leaves. The strong early response in SSi leaves was particularly striking. For a more detailed investigation of the SSi leaf response, we identified the 20 most strongly upregulated and downregulated genes in these leaves at each time-point and compared their transcript abundance across LSo, SSo and SSi leaves (Tables 1, 2). For DE genes that are members of gene families where several members were among the most strongly responding genes, only a single representative is listed in Tables 1 and 2. For example, WS0133_I11 (POPTR_0010s01150) (Table 1) is shown to represent seven different KPIs that were found among the most strongly upregulated genes in SSi leaves at 24 h.

Table 1.   Top 20 microarray elements revealing strongest upregulated genes in systemic sink (SSi) leaves in response to mechanical wounding plus forest tent caterpillar (FTC, Malacosoma disstria) OS at 2 h, 6 h and 24 h. For comparison corresponding expression levels are also shown for treated local source (LSo) and systemic source (SSo) leaves1j
 Clone IDMatch Genome2Match AGI#AnnotationE-valueLSoSSoSSi
2 h6 h24 h2 h6 h24 h2 h6 h24 h
  1. 1Microarray elements ranked by fold-change (FC) induction of response in SSi leaves to mechanical wounding plus FTC OS vs untreated control at 2 h, 6 h and 24 h, with FC values for those elements from LSo and SSo leaves also included for comparison. Subsequent redundant examples of a gene family at each time-point are removed to improve the diversity of different families included. Only FC values with statistical significance (P < 0.05, Q < 0.05) are coloured as different from ‘–’ (‘no change’) according to the colour scale shown at right, where dark green to dark red correlates with the listed fold-change in expression. Abbreviations: AGI, Arabidopsis genome initiative; E, E-value; OS, oral secretions.

  2. 2Genome model names have the ‘POPTR_00’ prefix removed to fit the table.

(a)2 h
WS0123_C21n/an/aNo significant hitn/a1.160.930.951.101.010.9124.631.721.37
WS0124_D1614s11710At3g62550Universal stress protein9e-370.930.510.751.130.830.6822.201.350.97
WS01213_L1008s19370At1g60470Galactinol synthase3e-280.540.570.750.971.000.9122.111.321.25
WS0131_I2014s03080n/aNo significant hitn/a0.920.570.801.141.200.7517.742.181.17
WS0162_F0915s06040At4g23740Leucine-rich repeat transmembrane protein kinase2e-790.940.580.751.540.900.7814.721.941.03
WS01210_A0739s00330At5g53550Oligopeptide transporter2e-790.750.600.710.680.920.6312.811.931.21
WS0132_B0817s06920At2g24210[Isoprene synthase] Terpene synthase TPS101e-80.581.060.770.810.940.7511.791.311.03
WS01213_O1413s12390At5g04530KCS1 fatty acid elongase (3-ketoacyl-CoA synthase 1)1e-380.821.081.340.870.750.5811.291.531.15
WS0152_B1302s12840n/aNo significant hitn/a0.370.400.890.850.740.7910.901.810.97
WS0162_D0906s08780At5g60020Laccase/diphenol oxidase3e-80.710.441.041.280.850.8010.091.261.36
WS0133_F0202s08920At1g24620Polcalcin/calcium-binding pollen allergen2e-220.620.440.750.790.810.699.841.491.04
WS01117_C0405s26930At1g76180Dehydrin (ERD14)6e-100.700.500.801.010.780.768.591.021.06
WS0134_L0902s12850n/aNo significant hitn/a0.460.600.601.140.780.658.312.081.05
WS0178_N2204s18200At5g42190E3 ubiquitin ligase SCF complex subunit SKP1/ASK11e-640.880.580.601.131.070.638.261.711.42
WS0147_P1619s12320At4g191709-cis-epoxycarotenoid dioxygenase1e-280.540.350.571.141.220.907.801.791.44
WS0162_A2416s08770At2g37090Glycosyl transferase family 43 protein5e-681.050.790.990.841.
WS0143_H2017s06670At3g21890Zinc finger (B-box type) family protein5e-
WS0143_D1113s06850At2g30410Tubulin folding cofactor A (KIESEL)9e-180.850.560.870.911.210.846.671.241.18
WS0162_F0822s00830At1g71900Expressed protein7e-151.011.370.720.901.210.806.031.162.11
(b)6 h
WS0191_J0309s15010At2g2950017.6 kDa class I small heat shock protein2e-537.502.701.143.621.201.721.6326.521.07
WS02010_D0304s07190At5g52640Heat shock protein 81-11e-313.742.041.012.800.990.951.1412.811.20
WS0202_P2203s10860At4g2520023.6 kDa mitochondrial small heat shock protein1e-514.452.471.073.190.981.460.9312.401.49
WS0231_D0811s07550At2g34070Expressed protein2e-431.751.311.061.631.020.942.4912.301.08
WS0172_K2105s11910At3g51130Expressed protein1e-845.892.421.212.930.881.281.4410.490.85
WS0192_G0514s14690At5g48570Peptidyl-prolyl cis-trans isomerase8e-
WS0178_C2113s06780At2g14880SWIB complex BAF60b domain-containing protein3e-232.401.850.902.
WS0162_E1105s23040At4g02830Expressed protein2e-141.562.
WS0211_C1710s12370At1g56300DNAJ heat shock protein1e-192.051.770.901.991.331.102.796.330.85
WS0131_A16256s00200At1g17180Glutathione S-transferase1e-582.142.481.042.212.781.241.805.111.16
WS01117_M0903s18280At1g16040Phosphatidylinositol-glycan biosynthesis3e-680.991.280.991.240.910.962.225.080.58
WS0222_K1809s16050At2g27080Harpin-induced protein-related2e-601.711.371.092.532.651.411.974.931.21
WS0208_H1817s02810At5g3767015.7 kDa class I-related small heat shock protein5e-441.701.431.192.210.911.210.534.491.04
WS0132_I1002s24310At3g07090Expressed protein1e-322.141.860.891.831.991.671.283.901.19
PX0015_F0205s12410At4g36600LEA domain-containing protein1e-101.681.271.012.582.820.921.113.721.57
WS0195_F1805s28110At3g50770Calmodulin-related protein7e-172.851.631.121.221.411.532.563.490.87
WS0204_K0615s12060At5g07330Expressed protein1e-
WS0132_E1210s21350At3g12580Heat shock protein 703e-782.072.580.801.292.621.070.703.470.98
WS0163_C2307s01500At2g30700Expressed protein6e-231.081.321.
WS0145_K1608s22840At5g18600Glutaredoxin family protein2e-381.411.451.031.301.251.260.723.360.71
(c)24 h
WS0146_J0204s18880At3g12500Basic endochitinase6e-250.701.2620.740.920.4732.333.122.3527.91
WS0212_O0506s28990At2g24520ATPase, plasma-membrane-type3e-972.062.4711.210.790.846.882.321.3410.71
WS0212_I2106s24090At4g33090Aminopeptidase M3e-662.092.7710.281.160.7011.791.881.679.27
WS0133_I1110s01150At1g17860Kunitz protease inhibitor3e-62.422.387.820.970.6110.291.711.648.65
PPO01s39660n/aPolyphenolic oxidasen/a1.666.173.380.941.171.941.401.388.15
WS0212_O0107s11000At4g36980Expressed protein2e-191.943.6910.101.200.719.381.781.507.34
TPS101s44080n/a(−)-germacrene-d synthasen/a1.949.643.
WS0214_H2001s22060At5g22860Serine carboxypeptidase S281e-382.303.991.721.211.351.481.611.186.65
WS0141_I1906s11980n/aNo significant hitn/a1.802.259.470.950.717.281.471.176.39
WS0152_K2308s00830At4g29905Expressed protein3e-102.182.3311.471.150.596.561.101.055.42
WS01120_O2402s11490At4g07960Glycosyl transferase family 2 protein4e-131.882.395.951.190.767.
WS01120_K1603s14550n/aNo significant hitn/a3.427.5212.111.192.1318.770.881.764.79
WS0141_A0304s12470At5g39410Expressed protein3e-422.512.4211.931.050.753.651.311.474.60
WS0212_C1501s05560At5g12950Secreted protein SCF41.30c2e-381.812.3510.821.060.6213.870.820.834.43
WS0144_M1510s16070At3g17210Stable protein 16e-362.092.763.671.161.333.730.791.303.83
WS0156_A0909s07810At2g30080Metal transporter ZIP69e-511.482.396.180.860.822.101.401.093.45
WS01211_J2006s28210At5g10780Expressed protein HSPC1847e-131.351.978.370.800.842.740.910.953.44
WS01118_E0113s03860At1g04240Auxin-responsive protein3e-352.843.609.521.170.713.031.001.813.42
WS0178_N2416s12600At2g29420Glutathione S-transferase4e-273.274.767.601.321.9111.660.791.953.24
Table 2.   Top 20 microarray elements revealing strongest downregulated genes in systemic sink (SSi) leaves in response to mechanical wounding plus forest tent caterpillar (FTC, Malacosoma disstria) OS at 2 h, 6 h and 24 h. For comparison corresponding expression levels are also shown for treated local source (LSo) and systemic source (SSo) leaves1
 Clone IDMatch Genome2Match AGI#AnnotationE-valueLSoSSoSSi
2 h6 h24 h2 h6 h24 h2 h6 h24 h
  1. 1Microarray elements ranked by fold-change (FC) induction of response in SSi leaves to mechanical wounding plus FTC OS vs untreated control at 2 h, 6 h and 24 h, with FC values for those elements from LSo and SSo leaves also included for comparison. Subsequent redundant examples of a gene family at each time point are removed to improve the diversity of different families included. Only FC values with statistical significance (P < 0.05, Q < 0.05) are coloured as different from ‘–’ (‘no change’) according to the colour scale shown at right, where dark green to dark red correlates with the listed fold-change in expression. Abbreviations: AGI, Arabidopsis genome initiative; E, E-value; OS, oral secretions.

  2. 2Genome model names have the ‘POPTR_00’ prefix removed to fit the table.

(a)2 h
WS01214_O2113s13990At1g51990O-methyltransferase family 2 protein3e-431.7811.2081.1031.2091.1661.3920.0850.5030.711
WS0161_F0503s01440At5g64080Protease inhibitor/seed storage/lipid transfer protein3e-120.7831.5531.4841.2310.9850.8960.1040.4190.495
WS01217_E2201s47550n/aNo significant hitn/a2.0301.4260.7811.2801.0411.0630.1540.7231.398
WS01224_K2309s10800At1g47480Expressed protein similar to PrMC3 (Pinus radiata)5e-231.0691.3530.6901.2291.1151.2390.1720.6000.763
WS0162_B1602s12780At5g42890Sterol carrier protein 2 (SCP-2) family2e-460.6961.8932.2731.3401.0671.6020.1910.9152.377
WS0119_H1709s03410At2g28790Osmotin-like protein6e-700.5300.7431.2551.6470.6991.4590.1990.8620.729
WS01210_K0906s10860At2g38540Nonspecific lipid transfer protein 1 (LTP1)1e-300.6721.5862.2551.5190.9971.2810.2220.8430.372
WS0111_L1413s02990At1g54690Histone H2A2e-481.2651.1381.1321.2920.9851.2200.2220.9840.758
WS01121_O2101s30920At4g34160Cyclin delta-3 (CYCD3)2e-650.9610.7890.8511.6190.9030.8570.2260.9641.237
WS0122_I2308s23110n/aNo significant hitn/a1.5201.2551.0741.0471.0161.2460.2270.5410.606
WS0175_M2216s00270n/aNo significant hitn/a1.1450.9461.0801.3341.1401.1660.2360.6100.615
WS0224_G0805s25620At1g05150Calcium-binding EF hand family2e-650.7110.7720.9830.5690.8920.7700.2420.6660.849
WS0221_M1801s25490At2g29570Proliferating cell nuclear antigen 2 (PCNA2)2e-600.7331.0190.8621.4871.0191.2280.2520.7280.784
WS0113_K0703s22240At5g65360Histone H33e-610.7420.8080.7831.2000.8200.9460.2521.0361.012
WS02011_J1406s11500At3g54220Scarecrow transcription factor9e-630.9190.4770.7021.0830.8911.1270.2530.8361.158
WS0116_O0801s31790n/aNo significant hitn/a1.0090.9861.2211.6670.9261.7520.2550.8680.969
WS0195_A0801s00240At2g45190Axial regulator (YABBY1)1e-591.0240.8700.8280.9851.4321.0650.2570.9511.005
WS02011_L1907s08070At5g60990DEAD/DEAH box helicase3e-301.4731.0610.8741.0340.8920.8260.2590.9080.878
WS0166_C2319s00770At5g63660Plant defensin-fusion protein, putative (PDF2.5)4e-131.1551.0780.9651.2500.7631.1190.2630.4640.623
WS0181_G1312s11570At5g61670Expressed protein7e-950.7900.7320.6190.8241.1641.0030.2681.0781.388
(b)6 h
WS0158_J1619s11560At1g71880Sucrose transporter (SUC1)1e-060.4040.5360.6050.7970.6910.9191.3540.3241.398
WS0185_D1918s03570At5g10695Expressed protein1e-230.3120.3401.1301.1220.4330.8831.2080.3910.613
WS0161_F0503s01440At5g64080Protease inhibitor/lipid transfer protein (LTP)3e-120.7831.5531.4841.2310.9850.8960.1040.4190.495
WS0125_C2013s03890At3g04720Hevein-like protein (HEL)7e-460.9341.0781.1081.1891.1301.0990.3170.4210.626
WS01222_I2101s39660n/aNo significant hitn/a0.6460.8301.0050.9491.0771.5980.3660.4270.982
WS0143_J0701s03470At1g80920DNAJ heat shock protein3e-330.3840.4251.0050.7350.3680.7420.8050.4420.455
WS0166_C2319s00770At5g63660Plant defensin-fusion protein, putative (PDF2.5)4e-131.1551.0780.9651.2500.7631.1190.2630.4640.623
WS0174_G2005s23220At1g03140Splicing factor Prp18 family protein7e-411.3621.3581.1220.9320.8801.2460.8520.4781.138
WS01215_E1603s10270At1g31812Acyl-CoA binding protein1e-310.9841.1501.5501.5331.1041.1240.8480.4790.799
WS0221_N1807s10350At2g23170Auxin-responsive GH3 family protein3e-721.1410.7520.9471.0441.2291.3060.4210.4850.773
WS01218_M1207s07110n/aNo significant hitn/a0.6720.7200.8941.1210.6820.7770.6940.5000.880
WS0113_N0104s18240At2g16850Plasma membrane intrinsic protein, putative (SIMIP)9e-640.6870.9201.0911.5011.3941.7560.6470.5010.435
WS01214_O2113s13990At1g51990O-methyltransferase family 2 protein3e-431.7811.2081.1031.2091.1661.3920.0850.5030.711
WS0199_D1216s05020At3g12150Expressed protein3e-600.7260.8800.9891.0510.9731.1810.7510.5050.869
WS0113_J1210s21590n/aNo significant hitn/a0.9241.2391.3061.3441.0040.9550.7730.5090.908
WS01127_I15n/aAt3g10390Amine oxidase family protein5e-081.1770.7511.0580.7420.9420.8880.4950.5100.839
WS01212_F1406s01020At3g22120Protease inhibitor/lipid transfer protein (LTP)2e-181.1911.2231.0671.1030.9950.8270.2800.5130.728
WS01218_H1310s19490At3g55990Expressed protein contains7e-091.0950.8711.3011.0240.7210.4980.6930.5131.423
WS0193_L2015s08290At5g14920Gibberellin-regulated family protein9e-230.7951.0960.8500.7510.7540.6940.3450.5160.779
WS02012_E0404s17490n/aNo significant hitn/a1.0531.2391.0731.0141.1270.9700.5050.5180.939
(c)24 h
WS0188_A1405s02270At5g27860Expressed protein1e-101.7281.9881.2300.9201.5100.9021.3111.2280.261
WS0151_F0204s17810At2g17880DNAJ heat shock protein2e-270.2430.1920.8720.7200.2900.4400.8260.6830.343
WS0151_H1308s06260At5g5972018.1 kDa class I heat shock protein2e-493.0071.5671.1543.2770.8971.4781.2371.6020.365
WS01210_K0906s10860At2g38540Nonspecific lipid transfer protein 1 (LTP1)1e-300.6721.5862.2551.5190.9971.2810.2220.8430.372
WS0224_E0505s17200At4g08950Phosphate-responsive protein, putative1e-871.5231.6391.2560.6100.9011.2011.0030.8150.388
WS01217_C2104s02340n/aNo significant hitn/a1.4420.9731.9781.2011.1451.2340.5630.8150.397
WS0141_D1906s24030At5g1202017.6 kDa class II heat shock protein1e-363.0371.3911.3362.1660.6851.1500.8610.9930.402
PX0011_G0510s20310n/aNo significant hitn/a0.9970.5801.4281.7700.7461.4020.2900.6380.408
WS0154_I0315s06000n/aNo significant hitn/a1.5903.0222.6941.1301.0451.1240.4330.6470.419
WS0121_D1708s06940n/aNo significant hitn/a1.4991.5941.4290.7741.3900.9570.8131.2500.428
WS0113_N0104s18240At2g16850Plasma membrane intrinsic protein, putative9e-640.6870.9201.0911.5011.3941.7560.6470.5010.435
WS0113_E0202s17100At5g08610DEAD box RNA helicase (RH26)2e-301.4483.4521.2800.8341.1201.3660.4140.8990.438
WS01210_O1812s12760At2g40140Zinc finger (CCCH-type) family protein4e-071.1461.3470.9721.0291.4281.3180.8401.3540.438
WS0232_K1816s14490n/aNo significant hitn/a0.7600.4971.1611.1070.8580.7380.6911.4540.442
WS0196_L2106s04820At2g41710Ovule development protein, putative4e-150.9170.6240.8441.0350.7771.2100.7120.6920.443
WS01121_M0809s13470At2g21050Amino acid permease, putative4e-480.7290.8561.3201.0730.8881.0400.9140.6980.446
WS0234_G1712s00760n/aNo significant hitn/a0.8091.0281.3400.8740.9691.0530.8680.8400.453
WS0231_C1106s03090At2g4101065 VQ motif-containing protein2e-220.5320.4300.6980.9750.9430.8500.9501.0890.453
WS0188_G0712s03800At1g48430Dihydroxyacetone kinase family8e-771.0901.3611.2760.8051.1040.8040.7730.7420.459
WS0123_D1207s03520At2g22500Mitochondrial substrate carrier family protein4e-541.3971.6401.0790.6471.2961.1760.9311.4150.467

Genes with the strongest upregulation in SSi leaves at 2 h were not changed at the same time-point in SSo leaves, and many were substantially downregulated in LSo leaves at 2 h (Table 1, part a). Galactinol synthase was one of the most strongly upregulated at 2 h in SSi leaves (Table 1). A distinct fingerprint of strongly upregulated genes in SSi leaves was also seen at 6 h, although some of the same genes were also upregulated, albeit with lower fold-change, at 2 h in LSo and SSo leaves (Table 1, part b). The strongest upregulated transcripts in SSi leaves at 6 h grouped into a variety of heat-shock proteins and expressed protein of unknown function. At 24 h, the strongest-responding SSi genes were also highly upregulated in LSo and SSo leaves (Table 1, part c).

Results shown in Table 2 reveal that genes with the strongest downregulation in SSi leaves were practically unchanged in SSo leaves and possessed a relatively sparse and weak response in LSo leaves. Genes annotated as functioning mainly in transport and signalling are the most downregulated in SSi leaves at 2 h and 6 h, with heat-shock proteins appearing at 24 h. The strongest downregulated genes in SSi leaves at 2 h were not changed at the same time-point in SSo leaves, while the small proportion of responding genes in LSo leaves at 2 h (Table 2, part a) responded with weak upregulation and downregulation. An O-methyltransferase was one of the most strongly downregulated genes at 2 h in SSi leaves, with downregulation sustained at 6 h and lessening through 24 h (Table 2), mirrored by transient upregulation in LSo leaves at 2 h (Table 2, parts a and b). A distinct fingerprint of strongly downregulated genes in SSi leaves was also seen at 6 h, again with the near-absence of response in these genes in SSo leaves and weak response in a few genes in LSo (Table 2, part b). At 24 h, the strongest responding SSi genes were again mirrored by weak upregulation or downregulation in a few of the genes in LSo and SSo leaves (Table 2, part c). Interestingly, a quarter of the strongest downregulated genes in SSi leaves at 24 h were significantly downregulated earlier, at 2 h, with the majority of these being unknown genes.

The galactinol synthase (PtGOLS) gene family

Given the rapid and strong induction of transcripts for several GOLS genes in SSi leaves in response to OS-treatment, we characterized this gene family in poplar as a basis to explore the involvement of carbon metabolism in insect-induced defense responses. Nine unique gene predictions with sequence relatedness to functionally characterized plant GOLS were identified in the most recent assembly of the P. trichocarpa genome sequence. An additional gene model annotated as a putative galactinol synthase (POPTR_0010s24860.1) is more similar to glycosyl transferases when tested by blast against the NCBI nr database (data not shown), and was thus omitted from this analysis. The location of the P. trichocarpa (Pt)GOLS gene models on chromosome scaffolds is shown in Fig. 5. Based on amino acid sequence identity (Table S5), the nine predicted PtGOLS genes appear to have evolved from four ancestral genes through genome duplication. PtGOLS6g and PtGOLS7g are duplicated genes on chromosome scaffolds 2 and 14, respectively (93% amino acid sequence identity), as are the PtGOLS1g/PtGOLS8g and PtGOLS2g/PtGOLS9g pairs (89% shared identity within each pair) on chromosome scaffolds 8 and 10, although there has been an inversion in one of these last two PtGOLS gene pairs. PtGOLS5g is likely a duplicate of PtGOLS3g and PtGOLS4g (92% identity), where the last two genes appear to have arisen by tandem duplication within chromosome scaffold 13 (92% identity between PtGOLS3g and PtGOLS4g). These patterns of duplicated GOLS on different scaffolds are in agreement with the large-scale genome duplication and chromosome rearrangement patterns identified by Tuskan et al. (2006).

Figure 5.

 Genome organization of the Populus trichocarpa galactinol synthase (PtGOLS) gene family. The ‘g’ at end of gene name indicates these are gene models predicted from the most recent assembly of the poplar genome sequence v2.0 ( Chromosomes are indicated by their scaffold number to the left of the chromosome representation. Total length of chromosomes is shown to the right in megabases (Mb). Gene orientation is indicated by the arrowhead. Arrowheads of the same colour indicate gene pairs that arose apparently by duplication from a common ancestor. The asterisk indicates tandem duplication.

To verify the PtGOLS gene models predicted in the P. trichocarpa genome sequence (Fig. 5), we cloned seven different full-length (FL)cDNAs for GOLS genes (three from P. trichocarpa, four from P. trichocarpa × deltoides). The FLcDNAs nomenclature corresponds to the PtGOLS gene models based on sharing > 95% amino acid identity, with the .1/.2 designating putative alleles of the same gene (Fig. S4; Table S5). The predicted PtGOLSs share between 66% and 93% amino acid identity (PtGOLS2g and PtGOLS3g, and PtGOLS6g and PtGOLS7g, respectively) (Table S5). A carboxyl-terminal pentapeptide motif APSAA (Sprenger & Keller, 2000) was only partly conserved in poplar GOLSs, with PtGOLS1g possessing an APTAA sequence (on three independent FLcDNA clones) and PtGOLS8g and PtGOLS9g possessing an LPSAA sequence (Fig. S4). A putative manganese-binding motif DXD (Breton et al., 1998; Busch et al., 1998; Wiggins & Munro, 1998) was found in all PtGOLS gene models. A serine phosphorylation site was absent from PtGOLS8g and PtGOLS9g, as is the case with Arabidopsis thaliana AtGolS2 and AtGolS3. Phylogenetic comparison of 26 GOLS from other plant species with the nine PtGOLS genes and the seven FLcDNAs showed that the level of sequence divergence among poplar GOLS is similar to the overall divergence across the plant GOLS family (Fig. 6).

Figure 6.

 Phylogeny of the galactinol synthase (GOLS) gene family of poplar and other plant species. Amino acid sequences of 42 proteins were analysed by maximum likelihood using phyml. Bootstrap values are indicated with an asterisk only for nodes with > 80% support. Genes shown with names in bold represent full-length poplar cDNAs. Genes shown with names in italics and a ‘g’ are predicted gene models from the Populus trichocarpa v2.0 genome sequence assembly ( Details for GOLS nomenclature, species names and accession numbers are listed in the Supporting Information, Table S5.

GOLS transcripts show source- or sink-specific induction patterns

Expression analysis in source and sink leaves of hybrid poplar in response to OS was done by qPCR to quantify transcripts for four poplar GOLS for which gene-specific primers could be identified and verified by amplicon sequencing (Fig. 7). Before OS treatment GOLS transcripts were present at very low levels in sink leaves, but were expressed at higher levels in source leaves (data not shown). Treatment with OS caused leaf type-specific GOLS induction (Fig. 7). Two GOLS genes, PtdGOLS1.2 and PtdGOLS2.1, which have high constitutive transcript levels in source leaves (not shown) exhibited the greatest increase (c. 250-fold) in transcript abundance at 2 h in SSi leaves. By contrast, PtGOLS3.1 expression was upregulated only in SSo leaves but was downregulated in LSo and SSi leaves (c. 400-fold up vs 15 000-fold and 10-fold down, respectively). The upregulation of PtGOLS3.1 in SSo was highest at 2 h, whereas the suppression in LSo and SSi was strongest at 24 h. Transcripts of the fourth pGOLS tested (PtdGOLS6.1) were increased c. 10-fold in all three leaf types with highest abundance at 2 h in source (LSo and SSo) leaves and at 6 h in distant sink (SSi) leaves.

Figure 7.

 Quantitative real-time PCR analysis of gene expression of galactinol synthase (GOLS) in poplar leaves in response to simulated forest tent caterpillar (FTC) herbivory. Values represent fold-change differences between leaves from untreated control trees and leaves from trees treated with mechanical wounding plus FTC oral secretion (OS). Transcript abundance was examined in local source (LSo), systemic source (SSo) and systemic sink (SSi) leaves at 2 h, 6 h and 24 h post-treatment. Data were normalized to poplar translation initiation factor 5A (TIF5A; WS0116_J23; POPTR_0006s19870) by subtracting the Ct value of each transcript, where ΔCt = Cttranscript − CtTIF5A. Transcript abundance for each gene was obtained from the equation (1 + E)−ΔCt, where E is the PCR efficiency, as described by Ramakers et al. (2003). Fold-change of GOLS transcript abundance was calculated as a treatment/control ratio of relative expression levels. Statistical significance of fold-change differences was determined using a linear model (see the Materials and Methods section and Table S6). Significance thresholds were set at *P < 0.05; **P < 0.01; ***P < 0.001.

Rapid increase in galactinol and raffinose in poplar leaves

Next we tested if levels of the sugar alcohol galactinol changed during the OS-induced systemic defense responses in poplar leaves (Fig. 8). Galactinol levels were significantly increased in systemic leaves at 2 h after OS treatment (SSi 2-fold, P = 0.034; SSo 1.2-fold, P = 0.048). Raffinose, a trisaccharide formed from galactinol and sucrose, also increased significantly in concentration throughout the plant, and again more so in SSi leaves (LSo 1.3-fold, P = 0.030; SSo 1.3-fold, P = 0.016; SSi 2-fold, P < 0.001).

Figure 8.

 Changes of levels of galactinol (a) and raffinose (b) in poplar leaves after simulated herbivory. Soluble sugars were isolated from local source (LSo), systemic source (SSo) and systemic sink (SSi) leaves of oral secretion (OS)-treated trees (2 h post-treatment) (tinted bars) and untreated control trees (open bars) and analysed by anion-exchange high pressure liquid chromatography. Values are represented as mean ± SD (n = 5 individual trees). Data were analysed separately using two-way ANOVA and Tukey multiple comparison tests. Bars with different letters above them are significantly different at = 0.050; letters are independent such that ‘ac’ is not significantly different from either ‘a’ or ‘c’, while ‘a’ and ‘c’ are significantly different from each other.


Simulated FTC attack

In previous work on the response of poplar to FTC feeding, Major & Constabel (2006) identified FTC OS as a reliable mimic of insect herbivory when OS was added to mechanically wounded leaves. Major & Constabel (2006) identified N-hydroxylinolenoyl-l-glutamine, commonly known as volicitin (Alborn et al., 1997), as a potential elicitor in the FTC OS. Because insect feeding involves mechanical damage and contact of wound sites with OS we used a combination of wounding plus OS application to simulate insect attack. Given the complexity of the experimental design of the microarray study with multiple time-points and leaf types, application of OS to wounded leaves served as a practical alternative to actual FTC feeding. Given the large number of samples required for replicated microarray analysis of the temporal and spatial patterns of the response, we did not attempt to identify the effect of wounding alone. It is therefore important to note that the effects described in this paper are in response to a combined wounding and OS treatment. Furthermore, FTC OS is a complex mixture of compounds (Major & Constabel, 2006). The effect of OS on poplar leaves may depend on variables that were beyond those controlled for in this study. For example, we cannot exclude that the pH of the OS may have had an effect on the response observed in treated leaves. By keeping OS frozen before application we reduced the possible effect of enzymatic degradation of elicitor active compounds in the OS.

Source–sink relationships of systemically responding poplar leaves

Simulated FTC herbivory via mechanical wounding plus FTC OS resulted in a significant increase in CWI activity in both SSo and SSi leaves, similar to the induction of CWI activity in response to jasmonic acid treatment or gypsy moth (Lymantria dispar) feeding in P. trichocarpa × nigra (Arnold & Schultz, 2002). As CWI activity increased in both systemic source and sink leaves, the source–sink relationship was maintained. Insect herbivory has previously been shown to influence carbon partitioning in poplar (Babst et al., 2008).

Transcriptome profiling reveals unique responses in systemic sink tissues

Forest tent caterpillar OS contains active elicitors and can be used as a faithful mimic of insect herbivory in poplar (Major & Constabel, 2006). The OS induces much of the same transcriptome responses as FTC herbivory (Ralph et al., 2006), although the timing of the response varies, with OS treatment causing faster transcriptional responses than FTC feeding, probably owing to a stronger initial stimulus caused by application of OS to a wounded leaf surface compared with the initially small but increasing feeding damage caused by FTC. Previous transcript profiling of poplar responses to wounding or herbivory have identified genes responding locally in treated leaves (Lawrence et al., 2006; Major & Constabel, 2006; Ralph et al., 2006; Babst et al., 2009). Here we went beyond confirming the involvement of many of these previously identified genes in the local defense response of poplar, providing a time-course profile of the transcriptome response in treated leaves. Babst et al. (2009) also established transcriptome responses in both local source leaves and in systemic sink leaves in response to herbivory and jasmonic acid (JA) at 22 h after onset of treatment. The later time-points (22 h) of the induced transcriptomes studied by Babst et al. (2009) and those identified in this study at 24 h share many similarities. However, the unique earlier response in SSi leaves was not captured in previous work. The results of the present time-course analysis also showed differential timing (i.e. separation in time) of the responses in LSo leaves with early induction of transcripts of oxidative stress response and octadecanoid signalling (e.g. allene oxide cyclase WS0155_D02; POPTR_0004s10240, peaking at 2 h mainly in LSo leaves) and later induction of known or putative defense genes (e.g. Kunitz protease inhibitor WS0151_M13; POPTR_0010s01160, maximum at 24 h in LSo and in SSo and SSi leaves). It remains to be investigated whether the observed changes in gene expression result in altered protein profiles and signalling activities.

In the systemic response, the induced transcriptome change in SSo leaves is weaker and slower than that of LSo leaves; however, a rapid, strong and distinct transcriptome response was activated in distant, juvenile SSi leaves. The early SSi response may allow for increased resource allocation and import into sink leaves for the production of a systemically induced defense at the growing shoot apex. The developing SSi leaves (Fig. 1) may lack constitutive resources to produce these defenses at the levels required (Jones et al., 1993), and resources from source leaves may be necessary to provide substrate for the suite of induced defense genes seen throughout the plant over the 24 h time-course. The strong induction of carbon metabolism genes in the transcriptome response of SSi leaves at 2 h could be a signature of induced increase of carbon resource allocation along source–sink gradients. By contrast, the later transcriptome signatures of sink leaves at 6 h and at 24 h show prominent features of induced defense genes.

GOLS and galactinol in the systemically induced response to biotic stress

The putative role(s) of the raffinose family of oligosaccharides (RFO) in plants include transport and storage of carbon resources and osmoprotectants in response to abiotic stress (Dey, 1985; Bachmann et al., 1994; Haritatos et al., 1996; Sprenger & Keller, 2000; Taji et al., 2002). They are produced by the sequential addition of galactinol units to sucrose. Galactinol is synthesized from UDP-galactose and myo-inositol by GOLS (inositol 3-α-galactosyltransferase; Keller & Pharr, 1996). The addition of one, two or three galactinol units to sucrose yields the trisaccharide raffinose, the tetrasaccharide stachyose, or the pentasaccharide verbascose, respectively (Peterbauer & Richter, 2001). Isoforms of GOLS are differentially expressed during drought, heat, and cold stress in Arabidopsis thaliana (Liu et al., 1998; Taji et al., 2002; Cunningham et al., 2003; Panikulangara et al., 2004), and overexpression of GOLS in A. thaliana leads to enhanced drought tolerance (Taji et al., 2002). Here we showed differential induction of some of the poplar GOLS genes with gene-specific patterns in source and sink leaves in response to simulated herbivore feeding. The involvement of GOLS in both abiotic and biotic stress responses perhaps points to a general role for GOLS and galactinol in stress-induced changes of carbon metabolism and reallocation of carbon resources. Indeed, while salt stress strongly induces GOLS isoforms in Populus euphratica, the galactinol produced does not itself play a direct role as compatible solute in osmoregulation in this species (Ottow et al., 2005). It is also possible that galactinol could be a component of systemic defense signalling.

Possible signals involved in activation of systemic defense response in poplar

Simulated FTC herbivory elicited a cascade of transcriptome responses in local and systemic and in source and sink leaves. The responses of the different leaf types vary in the genes involved, as well as in their temporal patterns of expression and overall magnitude of change. Transcripts from different signalling pathways changed in abundance throughout the plant, though some were leaf-type specific. For example, genes involved in JA signalling were upregulated throughout the plant, but more strongly and rapidly in LSo than in SSo and SSi leaves. Similarly, genes involved in ethylene signalling were upregulated most strongly in LSo leaves. On the other hand 9-cis-epoxycarotenoid dioxygenase, involved in ABA synthesis, was strongly and transiently upregulated in SSi leaves at the 2 h time-point (Fig. 3; Table S3). A possible role for ABA in the defense response of SSi leaves would be supported by the upregulation of ABA-sensing proteins in sink leaves in response to herbivory (Babst et al., 2009). Several genes for calcium signalling responded strongly at 2 h in SSi leaves, supporting findings of OS-induced calcium signalling in the early response to insect herbivory (Maffei et al., 2004; Lippert et al., 2009). Given the changes in soluble sugars, it is also noteworthy that transcription of a number of genes responding in SSi leaves is known or proposed to be controlled by soluble sugars (Rolland et al., 2002).

The rapid response in SSi leaves requires a signal that moves at a rate similar to that measured for phloem transport (50–100 cm h−1; Canny, 1975) in order to elicit transcriptional changes of the magnitude observed by 2 h in SSi leaves that are c. 150 cm away in an acropetal direction from the LSo leaves. Such a signal could involve jasmonates (Li et al., 2002; Howe, 2004), sugar sensing (Ehness et al., 1997; Rolland et al., 2002) or other signals (Lautner et al., 2005; Frost et al., 2007; Maffei et al., 2007; Pandey et al., 2008). The nature of the systemic signal(s) in poplar remains to be identified in future work. It is possible that the dynamic pattern of transcriptome responses observed in SSi leaves at 2 h, 6 h and 24 h results from a combination of signals acting in sequential cascades to produce an early reallocation of resource and culminating in the systemically induced response of defense genes.


Distinct spatial and temporal transcriptome patterns are induced by simulated insect attack in local and systemic, source and sink leaves of poplar. Induced transcriptome cascades are associated with induction of cell wall invertase activity, enhancement of source–sink relationships, leaf type specific changes of galactinol synthase gene expression and concomitant increase in galactinol and raffinose levels. Systemic transcriptome changes of SSi leaves are rapid and strong with initial signatures of metabolism and signalling, followed by induction of defense genes.


We thank Rick White for help with experimental design and statistical analysis, Sharon Janscik for help with cDNA preparation, David Kaplan for greenhouse support, Sarah Martz for OS collection, Elizabeth Chun and Mack Yuen for bioinformatics assistance, and Bob McCron for supply of FTC. This work was supported by the Natural Science and Engineering Research Council of Canada (NSERC) (grant to J.B., fellowship to R.N.P.), Genome British Columbia, and Genome Canada (grant to J.B.). Salary support for J.B. was provided in part by the University of British Columbia Distinguished Scholar Program.