Soybean defense responses to the soybean aphid


  • Mention of a trademark or proprietary product does not constitute a guarantee or warranty of the product by the US Department of Agriculture and does not imply its approval to the exclusion of other products that may also be suitable.

Author for correspondence: G. L. Hartman Tel: +1 217 244 3258 Fax: +1 217 244 7703 Email:


  • • Transcript profiles in aphid (Aphis glycines)-resistant (cv. Dowling) and -susceptible (cv. Williams 82) soybean (Glycine max) cultivars using soybean cDNA microarrays were investigated.
  • • Large-scale soybean cDNA microarrays representing approx. 18 000 genes or c. 30% of the soybean genome were compared at 6 and 12 h post-application of aphids. In a separate experiment utilizing clip cages, expression of three defense-related genes were examined at 6, 12, 24, 48, and 72 h in both cultivars by quantitative real-time PCR.
  • • One hundred and forty genes showed specific responses for resistance; these included genes related to cell wall, defense, DNA/RNA, secondary metabolism, signaling and other processes. When an extended time period of sampling was investigated, earlier and greater induction of three defense-related genes was observed in the resistant cultivar; however, the induction declined after 24 or 48 h in the resistant cultivar but continued to increase in the susceptible cultivar after 24 h.
  • • Aphid-challenged resistant plants showed rapid differential gene expression patterns similar to the incompatible response induced by avirulent Pseudomonas syringae. Five genes were identified as differentially expressed between the two genotypes in the absence of aphids.


The soybean aphid, Aphis glycines Matsumura, was first identified in North America in 2000 (Hartman et al., 2001) and has rapidly spread throughout the northern Midwest (Venette & Ragsdale, 2004). Damage to soybean caused by A. glycines includes plant stunting, leaf distortion, and reduced pod set (Sun et al., 1990; Hill et al., 2004). Resistance to A. glycines has been found in soybean germplasm accessions (Hill et al., 2004) and the resistance in soybean cv. Dowling was characterized as antibiosis (Li et al., 2004), but resistance in other genotypes was characterized as antixenosis (Diaz-Montano et al., 2006). Up until 2008, no soybean aphid biotypes were reported (Kim et al., 2008). Other aphid biotypes have been observed in greenbug (Schizaphis graminum) and the Russian wheat aphid (Diuraphis noxia) (Harvey et al., 1997; Porter et al., 1997; Basky, 2003; Smith et al., 2004). A basic understanding of the mechanism of soybean resistance to the aphid may provide an insight into how new biotypes develop and how to develop more durable resistance.

As one of the largest groups of phloem-feeding insects, aphids are a serious problem to many crops. Previous studies of plant response to aphids suggested that jasmonic acid (JA)-, ethylene-, and salicylic acid (SA)-regulated signaling pathways were at least partially activated by aphid feeding. A lipoxygenase, believed to be involved in JA synthesis in plants and induced by wounding (Bell et al., 1995; Leon & Sanchez-Serrano, 1999), was up-regulated in several plant–aphid interactions, including incompatible and compatible interactions in tomato to the potato aphid (Macrosiphum euphorbiae) and green peach aphid (Myzus persicae) (Martinez de Ilarduya et al., 2003), sorghum to the greenbug aphid (Zhu-Salzman et al., 2004), and Arabidopsis to the green peach aphid (Moran & Thompson, 2001). A gene encoding the PR1 protein was reported to be a good marker in the SA-dependent pathway (Uknes et al., 1992; Rogers & Ausubel, 1997) and was induced by aphid feeding in tomato and Arabidopsis (Moran & Thompson, 2001; Martinez de Ilarduya et al., 2003; De Vos et al., 2005). PR1 protein was antifungal (Niderman et al., 1995), and induced by a glucan elicitor, wounding, viruses, SA, and aphids (Hajimorad & Hill, 2001; Graham et al., 2003; Martinez de Ilarduya et al., 2003). Other PR proteins were also differently expressed between resistant (R) and susceptible (S) cultivars in response to aphids. In wheat, the activities of β-1,3-glucanase, peroxidase, and chitinase were induced to higher levels in R than in S cultivars (van der Westhuizen et al., 1998a,b). Russian wheat aphid infestation also induced SA accumulation and peroxidase activity in R compared with S wheat (Mohase & van der Westhuizen, 2002). Greenbug induced PR10 gene expression in sorghum (Zhu-Salzman et al., 2004).

There is limited information about the molecular mechanism of plant resistance to aphids (Kaloshian, 2004; Smith & Boyko, 2007). The first cloned insect resistance gene, Mi-1.2, is a member of the leucine zipper, nucleotide binding, leucine-rich repeat (LZ-NBS-LRR) family of plant resistance genes (Milligan et al., 1998), which confers resistance to root-knot nematode (Meloidogyne incognita) and also to a biotype of the potato aphid (Macrosiphum euphorbiae) in tomato (Rossi et al., 1998). Mi-1.2-mediated aphid resistance behaved similarly to the gene-for-gene resistance model in the field (Goggin et al., 2001). Mi-1.2 did not directly confer aphid resistance because Mi-1.2-mediated aphid resistance was not correlated with the transcript abundances of Mi-1.2 (Goggin et al., 2004). Plants with a single recessive mutation in another locus, Rme1 (confers resistance to Meloidogyne spp.), abolished aphid resistance, which indicated that the Mi-1.2-mediated aphid resistance required Rme1 (Martinez de Ilarduya et al., 2001). The Mi-1.2 RNA concentrations in tomato did not change in response to potato aphids or nematodes (Martinez de Ilarduya & Kaloshian, 2001). The only reported differentially induced aphid response between resistant (Mi-1.2/Mi-1.2) and susceptible (mi-1.2/mi-1.2) plants was the earlier systemic induction and accumulation of pathogenesis-related 1 (PR-1) in the incompatible tomato–aphid interaction. More extensive aphid-induced gene expression profiling would be useful to understand the defense responses in resistant plants to aphid infestation.

A recent review (Thompson & Goggin, 2006) on gene expression profiling of plant responses in several plant–aphid interactions indicated the lack of understanding of the early interactions between hosts and aphids, especially for incompatible interactions that may involve race-specific innate resistance. We address that concern here with an investigation of the early responses in soybean to aphids and identification of genes associated with resistance responses by comparing the differential gene expression between R and S soybean with soybean cDNA microarrays representing approx. 18 000 different soybean genes. A. glycines behavior on soybean leaves indicates that resistance in cv. Dowling was actively effective within 8 h after aphid application (Li et al., 2004). Electrical penetration graphing showed clear differences in A. glycines feeding behavior between resistant cultivars, including cv. Dowling, within 9 h after application (Diaz-Montano et al., 2007). Aphids took c. 8 h to reach the first sieve element feeding phase on cv. Dowling versus c. 3.5 h on susceptible checks. Additionally, only 19% of aphids reached sieve elements on cv. Dowling compared with 94% on the control. Once at the sieve elements of cv. Dowling, A. glycine spent c. 2.7 min compared with 18.9 min on the susceptible cultivar. Together, these studies suggest that gene expression studies should focus on time points near 8–9 h post-application (hpa) to potentially identify key genes involved in establishing resistance. Therefore, in this current study, the differential gene expressions in response to A. glycines were compared between soybean resistant cv. Dowling (R) and susceptible cv. Williams 82 (S) at 6 and 12 hpa. In support of the microarray data, quantitative real-time reverse-transcribed PCR (qRT-PCR) was performed on seven genes. A separate experiment followed the expression pattern of three defense genes to further validate the microarray results and to investigate their expression patterns for a longer time. Cross-comparing our list of genes that were differentially expressed in response to aphid in R and S plants to our previous microarray study of soybean response to compatible and incompatible strains of Pseudomonas syringae (Zou et al., 2005) would reveal if soybean activated different or overlapped defense pathways or mechanisms against pathogens and insects.

Additionally, we examined gene expression from R and S plants without aphids to identify genes with constitutive expression differences. Any genes showing significantly higher expression in R than S through three time points were considered candidates for the aphid-resistance gene Rag1 (resistance to A. glycines), which was determined to be a single dominant gene using the F2:3 population of cvs Dowling × Williams 82 (Hill et al., 2006).

Materials and Methods

Soybean cDNA libraries

Two sets of soybean cDNA microarrays, Gm-r1070 and Gm-r1088, were used, representing over 18 000 genes, estimated to be approximately one-third of the soybean transcriptome (Vodkin et al., 2004). Each slide set consisted of PCR product from 9216 cDNA clones and various control clones. Gm-r1070 consisted of mainly developing seed (green tissue) and floral cDNA (approx. 48 and 43%, respectively); Gm-r188 consists of cDNA from a variety of tissue sources, approx. 60% from germinating cotyledons, seed coat and growing shoots, and approx. 40% from stress, pathogen-challenged, or hormone-treated tissues.

Experimental design

Loop and dye-swap designs were used for the microarrays (Fig. 1). Each loop (Fig. 1a) represented one time point, 6 or 12 hpa, and four samples within one loop could be compared with each other. The cultivar of interest was balanced with respect to dyes since each cultivar was labeled once with both Cy3 and Cy5. The differently expressed genes can be detected with more statistic power in a loop design than a reference design using the same number of arrays (Kerr & Churchill, 2001). Comparisons between aphid and mock treatments on each cultivar were made to identify differentially expressed aphid-response genes. The differences between two cultivars were compared using the mock-treated samples in the loop design, as well as the samples without any treatments at time 0 h in a dye-swap design (Fig. 1b). Independent biological replicates were produced 4 wk apart. For each pair of comparisons, there were at least four replicates (two biological each consisting of two technical).

Figure 1.

Design for microarray experiment. Each arrow represents single microarray hybridization. Tails of arrows represent samples labeled with Cy3 dye while heads are Cy5-labeled. The entire experiment was repeated for two biological repeats. (a) Loop design for microarray analysis of resistant (Dowling, D) and susceptible (Williams 82, W) soybean (Glycine max) cultivars in response to Aphis glycines 6 or 12 h post-application (hpa) after aphid application. (b) Dye-swap design for the comparison between two cultivars at T0 without any treatment.

Soybean plants and aphid clone

Seeds of two soybean (Glycine max (L.) Merrill) cvs Dowling (R to soybean aphid) and Williams 82 (S to soybean aphid) were sown in 12 cm diameter plastic pots filled with soil-less potting medium (Sunshine Mix, LC1, Sun Gro Horticulture Inc., Bellevue, WA, USA), and covered with coarse-grade vermiculite (Hummert International, Earth City, MO, USA) contained in plastic trays without holes (Hummert International). Plants were grown in a growth chamber at 22°C (night) and 26°C (day) under a 14 h photoperiod (approx. 300 µmol m−2 s−1). The seedlings were bottom-watered.

The soybean aphid clone (virus-free) was established from a single first instar nymph isolated from a collection in Urbana, Illinois, in 2000, and maintained on a continuous supply of seedlings of the soybean cv. Williams 82 grown in a plant growth chamber at 22°C under continuous 100 µmol m−2 s−1 PAR irradiation.

Plant treatments for microarray

Ten-day-old plants with two fully expanded unifoliolate leaves of R and S were used for all experimental treatments. Leaves at 0 h without any treatment were collected and used in the dye-swap experiment to compare the constitutive differences at 0 h between the two cultivars. For the loop-design experiment, aphid and mock treatments were completed as follows. In order to keep aphids on the target tissue (one unifoliolate leaf of a plant), a piece of weigh paper of 15 × 15 cm was vertically positioned on the petiole of the leaf to serve as a large lightweight, physical barrier to discourage aphid movement off the leaflet. Forty aphids of various ages (20 wingless adults and 20 nymphs) were gently applied evenly on the abaxial surface of the target unifoliolate leaf of each plant by using a fine tip (1 mm in diameter) paint brush. After 6 and 12 h, infested leaves were cut at the center of the abscission joint connecting leaf to petiole using surgical scissors, and aphids were removed quickly using a soft 5-cm-wide paintbrush. The mock treatment consisted of weigh paper without aphids but the leaves were also brushed when collected. Only one unifoliolate leaf from each plant received aphid or mock treatment. Aphid- or mock-treated unifoliolate leaves for each of three plants were pooled together, respectively, and sealed inside plastic bags and immediately frozen in liquid nitrogen. All of the leaves were stored at –80°C before RNA isolation.

RNA isolation

For the microarray experiment, total RNA was isolated from frozen ground leaf tissue using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) in combination with Phase Lock Gel-Heavy (Brinkmann Instruments, Inc., Westbury, NY, USA) followed by a series of phenol-chloroform extraction, RNA precipitation, and suspension in water. The RNeasy Plant Mini Kit (Qiagen, Valencia, CA) was used to extract RNA from the leaf discs from the clip cage experiment. For quality control, all of the RNA samples were checked with a Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA, USA) to verify lack of RNA degradation. The concentration of total RNA was measured using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA) based on OD260 values.

cDNA microarrays

Reverse transcription of total RNA and fluorescently labeled cDNA probes were prepared from approx. 50 µg total RNA following the indirect labeling protocol described by Zou et al. (2005). Briefly, cDNA probes were synthesized from total RNA. After cDNA was purified and labeled with Cy3 or Cy5, light absorption at 260, 550, and 650 nm was measured to determine the concentration of cDNA, incorporated Cy3 and Cy5, respectively, using a NanoDrop ND1000 spectrophotometer. The hybridization, washing and scanning of the microarray slides were conducted as previously described (Zou et al., 2005). The microarray slides were scanned with a two-laser confocal scanner ScanArrayExpress (Perkin Elmer Life Sciences, Boston, MA, USA), using ScanArrayExpress software.

Microarray data analysis

Blocks and spots on the array images were defined and the intensities of spots and background were quantified using GenePix Pro (Axon Instruments, Union City, CA, USA). The quantified raw data files (including light fluorescence intensities of Cy3, Cy5, and background) as well as the images were uploaded into GeneTraffic (Iobion, La Jolla, CA, USA) for data storage, visualization, and analysis. The background-subtracted Cy3 and Cy5 signal intensity values were normalized using a linear-logarithmic transformation and analyzed by analysis of variance (ANOVA) in a linear mixed model with array as random effect and dye and sample as fixed effects, using R/MAANOVA statistical package (Wu et al., 2003). Genes that showed significant (P < 0.05 from contrast t-tests) differences in expression in the desired comparisons were saved as a gene list and uploaded to GeneTraffic. Fold-change calculations of significant genes were performed in GeneTraffic after local background subtraction and locally weighted scatter plot smoother (lowess)-subgrid normalization. Our gene selection cutoff required that a gene meet the statistical P-value cutoff of 0.05 in R/MAANOVA and, as an added stringency measure, that the average of the two biological replicates have a 1.5-fold increase or decrease before being considered ‘significantly’ differentially expressed.

Quantitative real-time PCR (qRT-PCR)

The qRT-PCR was performed according to the procedure described by Zou et al. (2005). The expression of a soybean β-actin (AW350943) was used as an internal standard to normalize the possible differences in template amounts (Zou et al., 2005). The primers (Supplementary material, Table S1) were designed based on the tentative consensus sequences with emphasis on the unique oligomer region or singleton EST sequence from TIGR, using Primer Express 2.0 software (Applied Biosystems, Foster City, CA, USA). The specificity of the primers was validated by the presence of a single peak in the dissociation curve analysis run after the qRT-PCR. The qRT-PCR data were analyzed using the relative quantification 2−ΔΔCT method (Livak & Schmittgen, 2001).

Leaf clip cage experiment

In order to add further verification to the microarray results and to study the expression patterns of the selected genes using qRT-PCR over a longer time period, a new set of samples from a separate experiment was collected at additional time points from leaf clip cages (12 mm outside diameter, 1-mm-thick walls and 12 mm high) as described in our previous study (Li et al., 2004). Ten-day-old plants with two fully expanded unifoliolate leaves were used. Three plants were used for each sample. Five wingless adults and five nymphs of various ages were put inside the leaf cage and clipped on the abaxial surface of one unifoliolate leaf of each plant. Empty cages were used as mock treatment. At 6, 12, 24, 48, and 72 hpa, clip cages were removed and aphids were brushed off. Leaf areas under the cages were cut using a 12-mm-diameter core borer and used for RNA isolation. Three leaf discs from three plants per treatment were pooled together and stored at −80°C before RNA isolation.


Responses in resistant vs susceptible soybean to aphids

Gene expression was compared between two cvs, resistant Dowling and susceptible Williams 82, under mock and aphid treatments at 6 and 12 hpa. Comparison combinations focused on aphid vs mock in R, aphid vs mock in S, R vs S under mock treatment, and R vs S under aphid treatment, following a loop design (Fig. 1a). Differential gene expression of R and S plants in response to aphid vs mock was used to select aphid-response genes within each time point (Figs 1a,  2). The differentially expressed genes were selected based on significance (P < 0.05) and a fold-change ratio (> 1.5 or < –1.5). Only four genes overlapped between R and S in response to aphids (Fig. 2).

Figure 2.

The number of genes significantly (P < 0.05) above 1.5-fold change up-regulated at 6 h post application (hpa) (a); down-regulated at 6 hpa (b); up-regulated at 12 hpa (c); and down-regulated at 12 hpa (d) in resistant (Dowling, D) and susceptible (Williams 82, W) soybean (Glycine max) cultivars.

Most of the 18 000 genes represented on the microarrays did not show differential expression between the two cultivars under mock treatment. However, we identified 140 aphid-response genes that showed specific response in R but not S (Table S2) at 6 and 12 h (the shaded area in Fig. 2 without the overlapped four genes; the sum of the numbers of genes in the shaded area is > 140 because of overlapped genes between T6 and T12). If a gene was up-regulated in R but down-regulated in S, it was considered as an R-specific response since its differential expression in R was different from that in S. As selected, these genes showed significantly differential expression after aphid feeding in R but not the same trend in S. On the other hand, 95 genes (nonshaded area in Fig. 2) showed responses to aphids that were specific to S (Table S3). In subsequent text, the gene annotations given are the summarized call based on analysis of both 5′ and 3′ EST comparisons to NCBI nr protein (BLASTx), TIGR TC, and Aridopsis (MIPS) databases (detailed information provided in Table S2).

Resistance-associated responses to aphids in soybean

The 140 R-specific aphid-response genes were classified into 16 functional groups based on the annotation from multiple databases (Table S2). These functional categories included groups of genes related to cell development, cell wall, cytoskeleton, defense, DNA/RNA, membrane, oxidation, primary metabolism, protein, secondary metabolism, senescence, signaling, stress, miscellaneous, no hits, and unknown. We focused on five groups, including cell wall, defense, DNA/RNA, secondary metabolism, and signaling.

Six of seven significant cell wall-related genes were down-regulated in R plants to aphids. Three homologs of pectate lyases (Gm-r1070-3571, Gm-r1070-4002, and Gm-r1070-7069) were down-regulated.

Significant defense-related genes included gene homologs to R genes, PR proteins, HR-associated proteins, and antimicrobial proteins. All genes were significantly up-regulated in R in response to aphids at 6 and/or 12 hpa, except for the snakin homolog (Gm-r1070-7107) which was down-regulated. Two hsr23J (HR-induced protein) homologs (Gm-r1070-628 and Gm-r1070-4998) were induced at 6 and 12 hpa. Three R gene homologs (Gm-r1088-3067, Gm-r1070-3058, and Gm-r1070-2852) and two R gene-related proteins (X21-binding protein Gm-r1088-3841 and NPR1 like-protein Gm-r1088-1044) were all up-regulated in R at 6 and/or 12 hpa. In addition to PR1a precursor (Gm-1088-8829), another two PR protein genes, soybean P21 (Gm-r1070-5189) and thaumatin (Gm-r1088-8018), which belongs to PR-5 family, also were induced.

The DNA/RNA functional group included 14 differentially expressed homologs of transcription factor (TF) and genes related to DNA or RNA processing. Eleven genes in this group were induced by aphids in R, and three genes were homologs of WRKY TFs (Gm-r1070-4784, Gm-r1070-4040, and Gm-r1070-2914).

All significant genes in the secondary metabolism (most genes related to phenylpropanoid pathway) were up-regulated in R, including homologs of chalcone synthase, isoflavone synthase, flavanone 3-hydroxylase-like protein, and cytochrome P450s. Secondary metabolites play an essential role in many plant defense responses, and several genes associated with synthesis of these chemicals that may serve as both antimicrobial and antioxidants were differentially expressed in response to aphids.

For the significant genes of the signaling group, two homologs of the gibberellin-interacting protein GIP1 (Gm-r1070-1066 and Gm-r1070-4900) and a homolog of calmodulin1 (Gm-b10BB-40) were down-regulated. Four calcium-associated genes (Gm-r1070-2651, Gm-r1088-1400, Gm-r1088-8402, and Gm-r1088-6724), two kinase homologs (Gm-r1070-6315 and Gm-r1070-8303), and a homolog to the jasmonate biosynthetic gene allene oxidase (Gm-r1070-2203) were induced in R in response to aphid at 6 hpa.

Array confirmation and expression patterns of defense-related genes by qRT-PCR

To validate the microarray results, seven genes were analyzed by qRT-PCR using the same RNA samples used for the microarrays. The results were consistent with the microarray results, as the genes assayed showed similar differential gene expression in response to aphids (Fig. 3), except that the fold change from qRT-PCR was generally higher than that observed for microarrays. Three genes related to defense, including soybean P21 (Gm-r1070–5189), MMP2 homolog (Gm-r1070-749), and PR1a precursor homolog (Gm-r1088-8829), increased in abundance from 6 to 12 hpa based on qRT-PCR (Fig. 3).

Figure 3.

Confirmation of microarray results using quantitative real-time reverse-transcribed PCR (qRT-PCR). Transcriptional responses of resistant (Dowling, D) and susceptible (Williams 82, W) soybean (Glycine max) cultivars in response to soybean aphids (Aphis glycines) at 6 and 12 h post application (hpa) were evaluated by both microarray and qRT-PCR. The changes in transcript abundance after aphid feeding were expressed as log2-transformed fold change compared with the mock treatment. qRT-PCR data showed the mean ratios from three replicates and the error bars represent the standard error (SE) of the mean. The mean ratios and SE of the mean for microarray data were obtained from GeneTraffic. The annotations for the genes are listed as the following: A, calreticulin; B, HR-associated Ca2+-binding protein; C, ferritin; D, DNA-binding protein; E, soybean P21 protein; F, matrix metalloproteinase GmMMP2; G, PR1a precursor.

Three defense-related genes that were up-regulated in the microarray analysis (P21, MMP2, and PR1a) were further verified at 6 and 12 hpa and were analyzed for their expression patterns over a longer time course (6, 12, 24, 48, and 72 hpa) using a leaf clip cage. All three genes had a higher induction (aphid vs mock) before 48 hpa in R than in S (Fig. 4). The induction of soybean P21 and the MMP2 homolog reached a peak at 24 h in R, and the PR1a precursor homolog peaked in expression at 48 h in R. Defense gene expression in the S genotype responded to the aphid infestation more slowly and less robustly as transcription abundances of these genes were induced only after 24 hpa.

Figure 4.

The expression patterns of three genes in soybean (Glycine max) in response to Aphis glycines from 6 to 72 h post-application (hpa) using qRT-PCR. Induction ratio (aphid vs mock) was expressed as mean log2 fold change, and the error bars represent the standard error of the mean. RNA samples were collected from pooled leaf discs in the leaf clip cage experiment, in which five wingless aphid adults and five nymphs were caged in the area where leaf samples were collected. Solid lines, resistant; dotted lines, susceptible.

Comparison between aphid and P. syringae induced responses

In a previous microarray study (Zou et al., 2005), soybean responses to P. syringae lacking (Virulent, Vir) or carrying (Avirulent, Avr) the avirulence gene avrB were compared. Soybean cv. Williams 82 showed susceptibility to Vir (compatible interaction) but resistance to Avr (incompatible interaction). Among the 140 R-specific aphid-response genes, 100 showed significant (P < 0.05) differential expression between incompatible and compatible induced responses when comparing the incompatible vs compatible at 8 h (Table S4). Hierarchical clustering of these 100 genes across the different experiments revealed that R responses to aphids at 6 hpa (T6-Dowling) were most similar to incompatible induced responses (T8-Avr_vs_MgCl2) (Fig. 5a). On the other hand, only 22 of the 95 S-specific aphid-response genes were identified as differentially expressed (P < 0.05) between incompatible and compatible (Table S5). The cluster analysis did not show a close relationship between S responses to aphids and incompatible or compatible (Fig. 5b). Combining R responses to aphids at 6 and 12 hpa shows a clear overlap in expression behavior of the 100 genes that were both significant in this aphid study (P < 0.05) and significant in the Avr P. syringae-induced incompatible response (P < 0.05) at 8 hpi, revealing shared rapid responses for these genes in both R interactions.

Figure 5.

Hierarchical clustering of soybean (Glycine max) responses to Aphis glycines and Pseudomonas syringae (Zou et al., 2005) using Pearson correlation. (a) Using 100 genes showed both R-specific aphid response and differential expression (P < 0.05) between Avr and Vir P. syringae-induced responses. (b) Using 22 genes showed both S-specific aphid response and differential expression (P < 0.05) between Avr and Vir P. syringae-induced responses. T6-Dowling, aphid vs mock in resistant Dowling at 6 h post-application (hpa); T6-W82, aphid vs mock in susceptible Williams 82 at 6 hpa; T6-mock, Dowling vs Williams 82 under mock treatment at 6 hpa; T6-aphid, Dowling vs Willams 82 under aphid treatment; T12 is as T6; T8-Avr vs MgCl2, Avr P. syringae-induced responses at 8 h; T8-Vir vs MgCl2, Vir P. syringae-induced responses at 8 h; T8-Avr vs Vir, differential expression between Avr and Vir P. syringae-induced responses at 8 h.

Genes with constitutively higher expression in resistant than susceptible soybean

In addition to comparing mock-treated controls, we also directly compared nontreated R with S at the start of the experiment (T0). In this T0 comparison, 68 genes, or less than 0.4% of the spots on the array, showed significantly higher expression (P < 0.05, and fold change > 1.5) in R than in S (Table S6). Only five genes showed constitutively higher expression in R than in S throughout T0, T6, and T12. These five genes were as follows: Gm-r1070-2257, no homolog; Gm-r1088-3786, a homolog of an unknown protein; Gm-r1088-7714, a homolog of a soybean NBS-LRR type resistance gene RPG1-b; Gm-r1070-4664, similar to a Myb family transcription factor, one of which has been shown to regulate the phenylpropanoid biosynthesis pathway and plant resistance to insect (Johnson & Dowd, 2004); and Gm-r1088–4309, a homolog of LTCOR11 or snakin2, which is potentially antimicrobial.


The soybean aphid is an invasive insect pest that is new to North America. The soybean aphid was not reported in North America before July 2000 (Hartman et al., 2001) and has rapidly spread throughout the Midwestern USA and southern Canada causing economic losses (Venette & Ragsdale, 2004). Other than the discovery of finding sources of resistance, and determining the inheritance and map location of the gene for resistance (Hill et al., 2004, 2006; Li et al., 2007), there has been very little research done on the molecular mechanism of soybean resistance to the aphid.

This is the first report to show early regulation of global gene expression in soybean in response to A. glycines, and to identify different functional groups of aphid-response genes. It is also the first to show a correlation of gene expression in R plants to aphids and the incompatible response induced by avirulent P. syringae. Another study compared the transcriptome changes in A. thaliana with those in P. syringae and M. persicae (De Vos et al., 2005), but was not R-specific. Studies on global transcriptional regulation of plant response to aphids include the sorghum–greenbug interaction (Zhu-Salzman et al., 2004; Park et al., 2006), and the compatible interaction between A. thaliana and M. persicae (Moran et al., 2002; De Vos et al., 2005; Couldridge et al., 2007). The earliest time point examined by microarray analysis in these studies was at 24 hpa, and comparisons between resistant and susceptible plants were made at 72 hpa, when susceptible plants were already badly damaged.

Although cDNA microarrays have their limits, many of their initially reported shortfalls (Li et al., 2002) have been diminished by the use of improved experimental designs and appropriate statistical analyses (Yauk & Berndt, 2007). Properly used, cDNA microarrays are powerful tools for the identification of candidate genes for further studies and can reveal interesting underlying physiologies when one examines the data across functional categories and/or by cross-comparison to other expression studies. Our microarray data showed that the differential gene expression pattern in aphid resistance had many similarities to Avr P. syringae-induced expression patterns (Zou et al., 2005), suggesting that the aphid-resistance gene Rag1 in Dowling might be mediated by an incompatible-like interaction. However, we were unable to detect cell death at 24 or 48 hpa in R or S (data not shown). This lack of detection might be because the aphid-induced cell death in R was only limited to a few cells, making it difficult to detect, or because cell death is not induced in response to aphids in R. Although incompatible reactions are often characterized as a hypersensitive response (HR) that induces cell death, there are examples that an HR can occur without cell death (Yu et al., 1998; Cole et al., 2001). It is possible that the aphid resistance in Dowling was induced by HR-like responses (including ROS, PR proteins) without causing cell death. Additionally, the probing/feeding behavior of aphids would presumably stimulate fewer cells compared with the leaf spongy mesophyll infiltration used for P. syringae inoculation (Zou et al. 2005) and would explain our limited ability to detect equally high numbers of differentially expressed genes.

In other plant–aphid interactions, an HR-like reaction was only reported in barley resistance to Russian wheat aphid, where the resistant barley had more collapsed, autofluorescent cells (because of the polymerization of phenolics) which was usually observed in cells undergoing the HR (Belefantmiller et al., 1994). No cell death was observed in the resistant tomato to aphid at 24 h using trypan blue stain (Martinez de Ilarduya et al., 2003).

Our microarray results showed that many genes were selectively activated in R plants as early as 6 hpa in addition to PR1a precursor, which is suggestive that the R gene Rag1 might be a regulatory gene associated with triggering resistance in the early stage of the soybean–aphid interaction. Although cvs Dowling and Williams 82 used in this study were not isogenic lines, the comparison between these two cultivars without aphids at three time points provided several R gene candidates. Five genes with constitutively higher expression in Dowling throughout all time points were identified, and the NBS-LRR type of R gene (Gm-r1088-7714) would be the top candidate because two other aphid-resistance genes were mapped in a NBS-LRR cluster region (Brotman et al., 2002; Klingler et al., 2005) in addition to the Mi-1.2 gene. By BLASTx search, the closest match to Gm-r10880-7714 was Rpg1-b, the soybean disease-resistant gene against P. savastonoi pv. glycinea (avrB) at 83% identity. Rpg1-b maps to soybean linkage group F (Ashfield et al., 2003), whereas Rag-1 maps to linkage group M (Li et al., 2007). It is possible that Rag-1 is a homolog of Rpg1-b if it is a NBS-LRR type of R gene. Furthermore, Rag-1 could be absent from the array, or there may be no differential expression between two cultivars. The sequence of selected candidate genes in this study will be used to design primers for use in more detailed mapping.

A model was proposed to summarize the resistance-associated defense responses in R in this study (Fig. 6). Before aphids successfully locate phloem cells to feed on phloem sap, they repeatedly probe by inserting their mouth stylets into plant tissue (Mclean & Kinsey, 1965). This probing activity releases discharged aphid saliva containing peroxidases, β-glucosidases, and other potential signal-generating enzymes, which produces potential chemical signals (such as turonides released from plant cell wall polysaccharides) as a result of digestion by aphid salivary enzymes (Miles, 1999). Additionally, the cells in the stylet path release wound signals as they are punctured (Morgham et al., 1994). When aphids locate phloem cells to feed, additional signals are released (Tjallingii, 2006). Wounding signals can induce synthesis of JA (Bell et al., 1995), which is an essential signal in the defenses against insects, including aphids, in Arabidopsis (McConn et al., 1997; Moran & Thompson, 2001; De Vos et al., 2005). The aphid-generated signals could act as elicitors to induce calcium- and reactive oxygen species (ROS)-related signaling. ROS, including H2O2, might be directly toxic to insects, leading to decreased herbivory (Bi & Felton, 1995) and contributing to cell wall strengthening (Bradley et al., 1992; Brisson et al., 1994), and may signal the incompatible reaction in challenged cells as well as function as a diffusible signal to induce the genes encoding protectants in adjacent cells (Levine et al., 1994). H2O2 can induce SA accumulation (Wu et al., 1997) and SA is a key signaling molecule in local and systemic defenses against pathogens (Dempsey et al., 1999). The SA- and JA-dependent signaling pathways may also overlap.

Figure 6.

Proposed model to summarize the defense responses in resistant soybean (Glycine max) cv. Dowling at 6 and 12 h post-application (hpa). The genes labeled by upward vertical arrows were up-regulated and those labeled by downward vertical arrows were down-regulated after aphid feeding in this study. HR, hypersensitive reaction; JA, jasmonic acid; SA, salicylic acid; WRKY, WRKY transcription factors.

The defense responses we observed at the RNA concentration in soybean generally coincided with previously observed aphid behavior. In a previous nonchoice test, it was shown that soybean aphids started to avoid feeding on R leaves after 4 hpa and some aphids died as early as 48 hpa, while aphids stayed and colonized leaves of the S genotype to the end of the 72 hpa experiment (Li et al., 2004). Electrical penetration graphing studies likewise showed clear differential feeding behavior between R and S genotypes, with A. glycines finding sieve elements within c. 4 hpa on a susceptible host, but taking c. 8 h on resistant genotypes (Diaz-Montano et al., 2007). In our study, more genes were induced in R than in S at 6 hpa, supporting these observations from behavioral studies that defense-related regulation of gene transcription leading to aphid resistance occurs early. Assuming that the expression of three defense genes we studied by qRT-PCR represents defense in general, one would conclude that the response to A. glycines before 48 hpa was induced earlier and more strongly in R than in S. In R, the induction of defense responses declined after 24 or 48 hpa, probably because of the lack of continual feeding or probing by the aphids, caused by avoidance of feeding or cessation of feeding when aphids died after 48 hpa. On the other hand, the induction of defense gene expression in S plants increased after 24 hpa and reached the higher amount of induction after 72 hpa, suggesting timing of the defense induction was important to the aphid resistance in R. Our results are consistent with resistance to A. glycines being controlled in a gene-for-gene model involving the early reorganization of aphid-generated signals by a resistance gene triggering rapid downstream defenses to prevent further damage.


We thank Dr Kris Lambert for use of the ABI PRISM 7900HT machine for qRT-PCR analysis, Beth Weldon and Robert Trieu for assistance with GenePix image analysis, and Andrea Pabon and Curt Hill for assistance with aphid propagation and use of leaf clip cages. This research was in part supported by the USDA-Agricultural Research Service, the United Soybean Board, the Illinois Soybean Check-off, and the University of Illinois Soybean Disease Biotechnology Center.