Introduction

The dominant pathogen of Glycine max (L.) Merr. (soybean) is the parasitic nematode Heterodera glycines Ichinohe (soybean cyst nematode [SCN]), an invasive species first observed in the U.S. in 1954 (Winstead et al. 1955). The SCN reproduces on at least 97 legume and 63 non-legume hosts (Epps and Chambers 1958; Riggs and Hamblen 1962, 1966a, b) with new hosts determined on a regular basis (Creech and Johnson 2006). SCN causes 7–10% reduction in production, worldwide. SCN causes more economic damage than the rest of its pathogens combined (Wrather and Koenning 2006), resulting in about $1.5 billion in losses in the U.S. alone, annually. Approximately 20,000 publically available collections of G. max, classified as plant introductions (PIs), are maintained through the USDA National Plant Germplasm System (USDA-NPGS). This seed bank, including many natural collections, is a resource that has been screened to identify G. max germplasm that can resist H. glycines infection. Through screening studies, two major groups of PIs each composed of a few G. max genotypes have been shown to exhibit specific, but contrasting ways to combat H. glycines (Ross and Brim 1957; Ross 1958; reviewed in Riggs 1992). Defense occurs at the site of infection, a nurse cell known as a syncytium (Fig. 1). The cellular response of G. max [Peking] to SCN has been determined (Ross 1958) and other genotypes including PI 89772, PI 90763 and partially PI 437654 (Mahalingham and Skorupska 1996) have been found to defend against SCN in a similar manner. The G. max [PI 88788] genotype was identified from a second screen (Epps and Hartwig 1972) with PI 209332, PI 548316 and partially PI 437654 (Mahalingham and Skorupska 1996) having similar cytological features occurring during their defense responses.

Fig. 1
figure 1

G. max [Peking/PI 548402] and G. max [PI 88788] resistant reactions. The phase 1 period of the defense response is represented in c, c′ through e, e′. The phase 2 period of the defense response is represented in f, f′ through h′. a Cyst containing eggs. b Pre-infective J2 (pi-J2) nematodes (gray) migrate toward the root. c, c′ the infective J2 (i-J2) nematodes burrow into the root and migrate toward the root stele, d, d′, the parasitic J2 (p-J2) typically selects a pericycle (green cells) or neighboring cell as the feeding site initial (FSi) (yellow cell). This cell is used to initiate the formation of the syncytium. The earlier stages of syncytium development (between 1 and 4 dpi) are similar between G. max [Peking] and G. max [PI 88788]-type of resistant reactions. e′ In G. max [Peking]-type, the neighboring cells (purple) are incorporated into the syncytium at 3 dpi. e′ In G. max [PI 88788], the neighboring cells (purple) are incorporated into the syncytium at 3 dpi. f In G. max [Peking], a rapid and potent resistant reaction occurs by the formation of a necrotic region that surrounds the syncytium (red layer of cells surrounding the yellow FSi) by 4 dpi. f′ In G. max [PI 88788], a prolonged but potent resistant reaction at the syncytium (pink cells) is not yet evident at the cytological level at 4 dpi. g In G. max [Peking], degradation of the syncytium (black cells) is engaged that is accompanied by the mortality of the SCN at the p-j2 stage (purple nematode). g′ In contrast, in G. max [PI 88788], the syncytium (pink cells) continues to develop until 5 dpi. The SCN feeding from the syncytium continues to develop, molting into J3s (red nematode). h′ In G. max [PI 88788], the entire syncytium collapses (black cells) and the SCN dies at the J3 or J4 stage (purple nematode). (The timing of stages is adapted from Endo 1965; Riggs et al. 1973; Lauritis et al. 1983; Kim et al. 1987)

Full size image

Numerous studies have investigated the soybean defense responses to SCN. The G. max [Peking] defense response is potent and rapid because most nematodes die early during parasitism at the parasitic second stage juvenile (p-J2) stage (Colgrove and Niblack 2008). The G. max [Peking]-type of defense response is evident at the cellular level at 4 dpi, involves necrosis of the cells that surround the head of the nematode and separates the syncytium from the cells that surround it (Endo 1964, 1965; Riggs et al. 1973; Kim et al. 1987; Kim and Riggs 1992). Another defining feature of the G. max [Peking]-type of defense response is the presence of cell wall appositions (CWAs), structures defined as physical and chemical barriers to cell penetration (Aist 1976, Schmelzer 2002; Hardham et al. 2008). In contrast, the G. max [PI 88788] defense response is potent but prolonged as the nematodes die at the J3 or J4 stages (Acido et al. 1984; Kim et al. 1987; Colgrove and Niblack 2008). In contrast to the G. max [Peking]-type of defense, the G. max [PI 88788]-type of response lacks the development of a necrotic layer that surrounds the head of the nematode (Kim et al. 1987). The initial stages of the G. max [PI 88788]-type of defense response involves extensive accumulation of cisternae and rough ER that is accompanied by nuclear degeneration within the syncytium by 5 dpi (Kim et al. 1987). The G. max [PI 88788]-type of defense response lacks thickened cell walls or appositions.

The genetic basis underlying defense to SCN resulted in the identification of the major recessive (rhg1, rhg2 and rhg3) (Caldwell et al. 1960), and dominant (Rhg4) (Matson and Williams 1965) and Rhg5 (Rao-Arelli 1994) loci. Of these, the rhg1 locus is currently the best understood since it has been defined in a region spanning approximately 611,794 nucleotides on chromosome 18 (Concibido et al. 1994; Mudge et al. 1997; Cregan et al. 1999; Hyten et al. 2010). Allelic variants are known to exist between different soybean genotypes harboring rhg1 (Brucker et al. 2005; Kim et al. 2010). Furthermore, fine mapping efforts in the G. max [PI 88788] background has allowed the locus to be narrowed down to within a region of approximately 67 kb (Kim et al. 2010).

While mapping efforts have made large contributions to understanding resistance, recent evidence has shown the value in applying gene expression to compliment mapping efforts in plants with complex duplicated genomes (Bancroft et al. 2011). The availability of the G. max genome (Schmutz et al. 2010) allows for similar expression mapping to be performed. However, expression studies typically examine differentially expressed genes whereby expression is measured in both a control and experimental sample and relative levels of expression are compared under various statistical parameters. The problem with the differential expression approach is that genes that have expression in one sample type and lack expression in a second sample type are discarded because statistical analyses cannot be done when expression is lacking in one of the two samples (Fig. 2). Detection call methodology (DCM) makes possible the cross-comparison of gene activity measured in one sample type to a second sample type where activity is not measured. Therefore, it is possible to identify and analyze genes with expression that is limited to one cell type. The important concept to recognize is that this pool of expressed genes could represent gene activity that defines a specialized cell type such as a syncytium proceeding through a series developmental events that culminates in a terminal phenotype such as cell death. Therefore, it could be imagined that an undifferentiated cell type like pericycle would lack expression of genes involved in programmed cell death where they would be found to be expressed in a syncytium undergoing the terminal steps of resistance. Thus, DCM in concert with differential expression analyses could provide a broader and more comprehensive analysis of gene expression in specialized cell types such as syncytia undergoing defense.

Fig. 2
figure 2

Detection call methodology. A and B represent genes with measured detection from two different cell types. The center gray region, which represents the union of A and B having the same pool of genes, including the green pool (induced genes) and red pool (suppressed genes), are genes that can be analyzed by differential expression analyses because they are expressed in both sample types. The gray region, lacking the red and green pools are the genes expressed in both sample types but do not exhibit statistically significant differences in expression between A and B. The genes of the white region of the A pool do not meet the statistical criteria of differential expression studies and would be discarded because they are expressed in only one sample type. Likewise, the genes in the white region of pool B pool are those that are discarded in differential expression studies. The white pools of genes are the focus of the detection call methodology

Full size image

The analysis presented here compares gene expression occurring during the potent and rapid defense response found in G. max [Peking/PI 548402] to the potent but prolonged process found in G. max [PI 88788]. Expression is presented graphically using a custom-developed KEGG application called Pathway Analysis and Integrated Coloring of Experiments (PAICE) (Hosseini et al. unpublished). Comparative analyses of transcriptional activity in these cytologically and developmentally distinct defense responses are used to determine gene expression in relation to the sequenced genome of G. max [Williams 82/PI 518671], identifying the chromosomal coordinates of the expressed genes. Further analyses place the expressed genes in relationship to an important resistance locus, rhg1, defined within a 67 kb region of DNA on chromosome 18 between the markers BARCSOYSSR_18_0090 and BARCSOYSSR_18_0094 (Kim et al. 2010). Gene expression of a subset of 1,000 genes is confirmed by Illumina® deep sequencing.

Materials and methods

Plant and nematode procurement

The materials and methods pertaining to H. glycines populations, G. max genotypes, experimental procedures and data analysis methods are published (Klink et al. 2005, 2007, 2009; Alkharouf et al. 2006). The G. max [Peking/PI 548402] and G. max [PI 88788] stocks were originally obtained from the USDA-NPGS (http://www.ars-grin.gov/npgs/acc/acc_queries.html). The H. glycines NL1-RHg population used in the studies is race 3, HG-type 7 (H. glycines [NL1-RHg/HG-type 7]) (Klink et al. 2009, 2010a). The G. max [Peking/PI 548402] and G. max [PI 88788] genotypes were used in the experiments to obtain defense responses by the use of H. glycines [NL1-RHg/HG-type 7]. The H. glycines [TN8/HG-type 1.3.6.7] (race 14) population was used to obtain susceptible reactions (Klink et al. 2007, 2009, 2010a). Seedlings were grown according to Klink et al. (2007, 2009). Prior to infection, the nematodes were diluted to a final concentration of 2,000 pi-J2/ml and one ml of nematode stock was added to each root of each plant. The roots, including the mock-infected control samples, were washed after 1 day to remove nematodes that had not penetrated the roots. Infected roots were grown for 3, 6 or 9 dpi. Maximally infected lateral roots were harvested for analyses. The process was subsequently repeated twice, providing three independent sets of samples for each genotype.

LCM and microarray hybridization

Slides were prepared according to Klink et al. (2005, 2007, 2009, 2010a). LCM was performed on a Leica® ASLMD microscope® (Leica®). Serial sections of approximately 100 syncytia were used to obtain the RNA for the studies for each replicate. Over 100 ng of RNA per replicate was obtained for the studies. Work to obtain RNA was done with the PicoPure RNA Isolation kit, (Molecular Devices®). A DNAse treatment was added, just before the second column wash, using DNAfree® (Ambion®). RNA quality and yield were determined using the RNA 6000 Pico Assay® (Agilent Technologies®) using the Agilent 2100 Bioanalyzer® according to the manufacturer’s instructions. Both probe preparation and hybridization procedures on the GeneChip® Soybean Genome Array (Affymetrix®) were performed according to their guidelines.

Data analysis

All microarray hybridizations were performed at the Laboratory of Molecular Technology, SAIC-Frederick, National Cancer Institute at Frederick, Frederick, MD 21701, USA. Local normalization was used. The measurement of the presence or absence of transcripts by particular probe set on a single array was determined using the Bioconductor implementation of the standard Affymetrix® DCM according to Klink et al. (2010b). In summary, the DCM consists of four steps: (1) removal of saturated probes, (2) calculation of discrimination scores, (3) P-value calculation using the Wilcoxon’s rank test, and (4) the detection (present/marginal/absent). Ultimately, the algorithm determines if the presence of a probe set’s transcript is provably different from zero (present [P]), uncertain (marginal [M]), or not provably different from zero (absent [A]). A probe set was considered present only if it measured expression on all three replicate microarrays corresponding to that condition. To be considered absent, the probe set had to lack detection on all three replicates for a given condition. A description of the supplemental files is provided. Microarray gene expression has been confirmed using the Illumina® Genome Analyzer II® (Illumina®) at the USDA-ARS Beltsville, MD according to the manufacturer’s protocols. Data is maintained at the Soybean Genomics and Microarray Database (Alkharouf and Matthews 2004).

Gene pathway analyses

The PAICE software (Paice_v2_90.jar) http://sourceforge.net/projects/paice/ (Hosseini et al. unpublished) was developed for the pathway analyses. The PAICE software visualizes pathways according to Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg/catalog/org_list.html) from Affymetrix® gene expression data. There are 38,099 probe sets on The Affymetrix® soybean GeneChip®. As of June, 2011, 9,717 probe sets (29%) have reference pathway enzyme commission (E.C.) numbers. There are 23,583 probe sets with matches to Arabidopsis thaliana accessions (62%). The number of probe sets matching both A. thaliana accessions and having E.C. numbers is 4,156 (11%). The PAICE pathway analysis was performed according to Klink et al. (2011) using a modified version for data obtained through the DCM. Data supplemental to each table and figure and GO terms (Harris et al. 2004) are provided. The seven supplemental datasets (Supplemental Datasets 1–7) can be found at the website: http://dl.dropbox.com/u/428435/DCM%20PATHWAYS.zip.

Chromosomal map coordinates

The Genbank accessions of probe sets on the Affymetrix® soybean GeneChip® (Supplemental table 1) were queried against the sequences G. max [Williams 82/PI 518671] (Supplemental table 2) (Schmutz et al. 2010) genome at http://www.phytozome.org/. The queries were performed in the Glycine max database using the Blast option. Once the chromosomal map coordinates were obtained, the coordinates were queried into http://www.soybase.org, allowing for the identification of the microarray-identified genes and their chromosomal map coordinates to physical map positions in relation to the genetic positions of the resistance loci.

Results

Intergenotype analyses identify genes that are expressed in a genotype-dependent manner

Detection call methodology (DCM) was used to compare the G. max [Peking/PI 548402] and G. max [PI 88788] defense responses using the Affymetrix® GeneChip fabricated with 38,099 soybean probe sets. Comparisons were made at the 3 dpi time point syncytia (Fig. 3a; Supplemental table 3), 6 dpi (Fig. 3b; Supplemental table 4) and 9 dpi time points (Fig. 3c; Supplemental table 5). The analyses of the combined data from the 3, 6 and 9 dpi time points demonstrate that G. max [Peking/PI 548402] and G. max [PI 88788] are undergoing gene expression that is different from syncytia undergoing the susceptible reaction (Fig. 3d; Supplemental table 6). Expression was confirmed using Illumina® deep sequencing platform (Table 1; Supplemental table 7).

Fig. 3
figure 3

Intergenotype analyses represented by Venn diagrams depicting the four comparative analyses made between the G. max [Peking/PI 548402] resistant syncytium (blue circle), G. max [PI 88788] resistant syncytium (red circle) and G. max [Peking/PI 548402] susceptible time course (black circle). The blue ring represents the G. max [Peking/PI 548402] resistant syncytium pools of genes. The red ring represents the G. max [PI 88788] resistant syncytium pool of genes. The black ring represents the G. max [Peking/PI 548402] susceptible syncytium time course pool of genes. a 3 dpi G. max [Peking/PI 548402] and G. max [PI 88788] resistant syncytium samples compared to G. max [Peking/PI 548402] susceptible syncytium time course pool of genes. b 6 dpi G. max [Peking/PI 548402] and G. max [PI 88788] resistant syncytium samples compared to G. max [Peking/PI 548402] susceptible syncytium time course pool of genes. c 9 dpi G. max [Peking/PI 548402] and G. max [PI 88788] resistant syncytium sample pool compared to G. max [Peking/PI 548402] susceptible syncytium pool of genes. d Combined 3, 6 and 9 dpi G. max [Peking/PI 548402] and G. max [PI 88788] resistant syncytium sample pool compared to G. max [Peking/PI 548402] susceptible syncytium time course pool of genes

Full size image
Table 1 Illumina® deep sequencing experiment for the G. max [PI 88788] 9 dpi resistant reaction time point
Full size table

Time point analyses identify genes that pertain to defense

Experiments were then designed to determine gene expression that is common to G. max [Peking/PI 548402] and G. max [PI 88788] syncytia as they undergo their respective defense responses. Experimental data are presented here only from genes that are present in all replicates for a particular cell type of both genotypes (G. max [Peking/PI 548402+PI 88788]) undergoing their defense responses (Fig. 4). The analyses of the 3 dpi time point demonstrate that G. max [Peking/PI 548402+PI 88788] syncytia undergo gene expression that is different from both pericycle and the surrounding cells as well as syncytia undergoing the susceptible reaction (Fig. 4a; Supplemental table 8). The analyses were followed by examining expression occurring at 6 dpi (Fig. 4b; Supplemental table 9) and 9 dpi (Fig. 4c; Supplemental table 10). Combining data from the 3, 6 and 9 dpi time points demonstrate a core set of constitutively and perhaps uniquely active genes in G. max [Peking/PI 548402+PI 88788] syncytia undergoing their defense responses (Fig. 4d; Supplemental table 11). Analyses presented in Fig. 4d show that 1,787 probe sets, representing ~5% of the total array, measure expression specifically in G. max [Peking/PI 548402+PI 88788] syncytia throughout defense.

Fig. 4
figure 4

Time point analyses represented by Venn diagrams depicting the four comparative analyses made between the G. max [Peking/PI 548402+PI 88788] pericycle and surrounding cells, the G. max [Peking/PI 548402+PI 88788] defense response and the G. max [Peking/PI 548402] susceptible reaction. The green ring represents the G. max [Peking/PI 548402+PI 88788] pericycle and surrounding cells pool of genes. The light blue ring represents the G. max [Peking/PI 548402+PI 88788] syncytium defense response pool of genes. The black ring represents the G. max [Peking/PI 548402] susceptible syncytium pool of genes during a time course of infection. a Comparative analysis of 3 dpi G. max [Peking/PI 548402+PI 88788] resistant reaction to G. max [Peking/PI 548402+PI 88788] pericycle and the G. max [Peking/PI 548402+PI 88788] and G. max [Peking/PI 548402] susceptible reaction sample pool of genes. b Comparative analysis of 6 dpi G. max [Peking/PI 548402+PI 88788] defense response to G. max [Peking/PI 548402+PI 88788] pericycle and the G. max [Peking/PI 548402+PI 88788] and G. max [Peking/PI 548402] susceptible reaction sample pool of genes. c Comparative analysis of 9 dpi G. max [Peking/PI 548402+PI 88788] defense response to G. max [Peking/PI 548402+PI 88788] pericycle and the G. max [Peking/PI 548402+PI 88788] and G. max [Peking/PI 548402] susceptible reaction sample pool. d Comparative analysis of 3, 6 and 9 dpi G. max [Peking/PI 548402+PI 88788] defense response to G. max [Peking/PI 548402+PI 88788] pericycle and the G. max [Peking/PI 548402+PI 88788] and G. max [Peking/PI 548402] susceptible reaction sample pool

Full size image

Intergenotype PAICE analyses reveal genotype-specific metabolic pathway activity

PAICE (Hosseini et al. unpublished) was developed to place the expressed genes into their metabolic context (Table 2). The pathways that are identified in the resistant reactions of G. max [Peking/PI 548402], G. max [PI 88788], the susceptible reaction and those found in all three cell types for pericycle (Supplemental dataset 1), 3 dpi (Supplemental dataset 2) 6 dpi (Supplemental dataset 3) and 9 dpi comparisons (Supplemental dataset 4) are provided. The analyses reveal that while commonalities exist in gene activity for hundreds of pathways between G. max [Peking/PI 548402] and G. max [PI 88788], specific alterations in expression exist that accompany the distinct forms of their genotype-defined defense responses (Fig. 5).

Table 2 PAICE pathway count for the intergenotype analyses accompanying Fig. 3
Full size table
Fig. 5
figure 5

Intergenotype PAICE analysis of sphingolipid metabolism, map 00600. Probe sets detecting expression for Peking resistant reaction at 9 dpi (green); PI 88788 resistant reaction at 9 dpi (red); Peking resistant reaction at 9 dpi, PI 88788 resistant reaction at 9 dpi and the susceptible time course (blue). Note No expression was found for sphingolipid metabolism that was limited only to the susceptible time course

Full size image

Time point PAICE pathway analyses identify syncytium-specific expression

PAICE was then used to analyze the pooled transcript data obtained from G. max [Peking/PI 548402+PI 88788] syncytia undergoing their defense responses at 3, 6 and 9 dpi to G. max [Peking/PI 548402+PI 88788] to both pericycle and their surrounding cells and syncytia undergoing the susceptible reaction. The time point analyses resulted in the generation of a total of 658 PAICE pathways (Table 3). Removing duplicate pathways occurring in the different time points and cell types resulted in the identification of 119 pathways with gene expression activity (Fig. 6).

Table 3 PAICE pathway count for the time point analyses accompanying Fig. 4
Full size table
Fig. 6
figure 6

PAICE pathway analyses. Pathway activity is shown for 3, 6 and 9 dpi time points. Red Resistant syncytia only, green expressed in resistant and susceptible syncytia, gray susceptible syncytia only, yellow expressed in all cell types, blue expressed in pericycle and resistant syncytia, purple pericycle only, pink expressed in pericycle and susceptible syncytia, black expression not detected in sample

Full size image

Time point analyses identify genes found at the rhg1 locus that are expressed in the syncytium during defense

The time point PAICE analyses are designed to visualize metabolic activity for all genes exhibiting expression. Gene activity at the rhg1 locus can be measured by the 112 probe sets fabricated onto the array that represent the locus (Supplemental table 12). Expression studies show that 18 of the 112 probe sets spanning the rhg1 locus are measuring expression in at least one sample at the studied time points (Fig. 7). The gene lists for the 3 dpi (Supplemental dataset 5) 6 dpi (Supplemental dataset 6) and 9 dpi time points (Supplemental dataset 7) are provided. In sum, two adjacent genes within the 67 kb rhg1 region had expression only in syncytia undergoing defense and at all time points as revealed by the experimental conditions (Fig. 7).

Fig. 7
figure 7

Time point analyses of the rhg1 locus. Expression is presented in relation to map positions of Affymetrix®probe sets and genes at the locus with gene activity as demonstrated. The list represents probe sets that have chromosomal coordinates on chromosome 18 in the region of rhg1 and also have expression data. P, pericycle and surrounding cells; C, common; S, syncytium. Purple Pericycle; red expressed in pericycle and syncytium; green syncytium; black uniquely expressed in syncytia undergoing a defense response as compared to pericycle and syncytia undergoing a susceptible reaction. Duplicate probe sets having identical gene expression were consolidated. The red box represents the only genes within the 67 kb region (Kim et al. 2010) between the markers BARCSOYSSR_18_0090 and BARCSOYSSR_18_0094 that had any expression

Full size image

Discussion

An analysis of gene expression of soybean germplasm obtained originally from ecological collections was used to show how natural genetic variation is a useful tool in understanding defense at cellular resolution. The study generated a map of cell fate decisions as soybean was undergoing infection leading to either a susceptible reaction or a successful defense response to SCN infection. The analysis presented here was accomplished by examining gene expression occurring at the site of infection, the syncytium. Notably, the genes focused in on here do not meet the statistical cut-off parameters in differential expression studies and are therefore discarded from further analysis (Klink et al. 2007, 2009). This outcome occurred because the probe sets measured expression in samples isolated from one cell type but lacked the measurement of expression in other cell types, making statistical comparisons impossible for differential expression studies. The work presented here was built off the premise that cell-type specific expression is a hallmark of cellular identity, especially during specialized processes such as defense to pathogens. The analysis presented here demonstrates that there is a basic conserved expression program in place that would likely be common to all soybean genotypes undergoing defense to SCN (Klink et al. 2011). It is on this conserved gene expression platform that genotype-specific expression is organized and orchestrated during defense to SCN (Klink et al. 2011). This expression is what governs the different cellular features that are present during the G. max [Peking/PI 548402] and G. max [PI 88788] forms of the resistant reaction. Identifying differences in gene expression at the cellular level is not unexpected and is consistent with single cell type gene expression studies done in other experimental systems undergoing a developmental process (Benfey and Mitchell-Olds 2008; Chiang and Melton 2003; Guo et al. 2011; Tang et al. 2011).

Detection calls confirmed by Illumina® deep sequencing

While the DCM provides a statistical output, it is difficult to determine what the relative quantities of a transcript are. It was revealed through Illumina® deep sequencing that some of these genes can be represented by a fairly large percentage of the transcripts in a sample type. For example, the Affymetrix probe set Gma.3940.1.S1_at, whose sequence is Glyma13g06450.1, is an unknown gene that does not appear to be conserved with other organisms. However, it represented over 17% of the transcripts in the 9 dpi G. max [PI 88788] sample isolated from cells undergoing resistance. The XYLOGLUCAN ENDOTRANSGLYCOSYLASE 6 (XTR6) gene was represented by over 11% of the transcripts. The XTRs modify the plant cell wall xyloglucan-cellulose framework and by doing so, modulate strength and expansion. This activity appears to occur at the point of formation of secondary cell walls through restructuring of the primary cell walls (Bourquin et al. 2002). Very little is understood about XTRs and defense. However, the reinforcement of cell walls is a component of defense of soybean to SCN (Mahalingham and Skorupska 1996). Another gene, BOTRYTIS-INDUCED KINASE1 (BIK1) that constituted almost 10% of the Illumina®-identified transcripts was originally identified as a defense gene (Veronese et al. 2006). BIK1 is activated within minutes after infection of Arabidopsis thaliana by Botrytis cinerea (Veronese et al. 2006; Laluk et al. in press). BIK1 has been shown to play essential roles in plant growth, ethylene signaling and pathogen activated molecular patterning (PAMP) during defense. The BIK1 protein localizes to cell membranes and it was suggested that it may act early during the interaction between the plant and pathogen (Veronese et al. 2006). Of note, the defense requirement of BIK1 functions in relation to salicylic acid levels. The addition of the Illumina® sequencing thus places the Affymetrix® detection data into an expression context since relative amounts of transcripts that are present within a sample are obtained. What is clear from the Illumina® experiments is the accuracy of the Affymetrix® detection calls.

Time point PAICE analyses

The time point PAICE analyses revealed gene activity occurring specifically during the resistant reaction. These experiments demonstrated the activity of many pathways at 3 dpi, a period when the soybean is altering its gene expression that leads to visible signs of the defense response (Ross 1958; Endo 1964, 1965; Endo and Veech 1970; Gipson et al. 1971; Jones and Northcote 1972; Riggs et al. 1973; Kim et al. 1987; Kim and Riggs 1992; Mahalingham and Skorupska 1996; Klink et al. 2007, 2009, 2010a, b). However, the expression of a pathway in cells that will undergo defense was never limited only to the 3 dpi syncytium samples. In some cases gene expression was measured in syncytia undergoing defense for the earlier 3 and 6 dpi time points while lacking any expression at the 9 dpi time point. In these cases, the analyses revealed the expression of lipoic acid metabolism that is part of a nonenzymatic antioxidant system in plant cells (Pérez-López et al. 2010) and processes leading to lipid production (Baud et al. 2007). The pathways for indole alkaloid biosynthesis (Onkokesung et al. 2010; Hanssen et al. 2011) and isoquinoline alkaloid biosynthesis (Facchini et al. 1996; Holková et al. 2010), having roles in defense, were also observed to be active. The pathway involved in the biosynthesis of glycosphingolipids was observed to be active. Glycosphingolipids are a class of sphingolipids of which there are greater than 168 in A. thaliana (Markham and Jaworski 2007) and perform roles in cell death. Other active pathways include the biosynthesis of alkaloids from the shikimate pathway (Zulak et al. 2008). The shikimate pathway provides metabolites for the production of phenylpropanoids that are metabolized into substances that perform defense roles. The pathway involved in the biosynthesis of alkaloids from terpenoid and polykeytides that are jasmonate regulated (Menke et al. 1999; Montiel et al. 2011) and fatty acid biosynthesis (Savchenko et al. 2010; Gao et al. 2011; Meldau et al. 2011) are also active.

Some pathways appeared to be active at all points during the defense response, including increased activity of components of the arachidonic acid pathway (Klink et al. 2011). The direct link of arachidonic acid metabolism to α-linoleic acid metabolism, shown to be differentially expressed during defense to SCN (Klink et al. 2011) provides additional support to the metabolic pathway leading to the synthesis of methyl jasmonate possibly being involved in G. max defense to H. glycines (Klink et al. 2007, 2009, 2010a). Notably, genetic data in Zea mays has already linked the involvement of JA signaling during its defense of the plant parasitic nematode Meloidogyne incognita (Gao et al. 2008). Arachidonic acid functions in defense by triggering programmed cell death (Bostock et al. 1981, 1986). Arachdonic acid metabolism is active through genes leading to the synthesis of hydroxyepoxyeicosadienoic acid and tetrahydrofuran diols. This observation is important because experiments in other plant-pathogen systems demonstrate that furans are an important component of plant defense responses, working efficiently on inhibiting larva development of insects (Rodriguez-Saona et al. 2000). Other gene pathways that are active at the 3, 6 and 9 dpi time points include N-glycan biosynthesis (reviewed in Pattison and Amtmann, 2009) and nicotinate and nicotinamide metabolism (Steppuhn et al. 2004) perform roles in defense. The analyses also identified glycerolipid metabolism (Kachroo et al. 2004; Chaturvedi et al. 2008; Xia et al. 2010), glyoxylate and dicarboxylate metabolism (Emmerlich et al. 2003) and zeatin biosynthesis (Smigocki et al. 1993; Gális et al. 2004), pathways known to perform roles in defense.

Gene activity at the rhg1 locus

An expression mapping analysis focused in on rhg1. In the time point analyses, 18 probe sets representing 18 different genes, detected expression in one or more cell types. Of those 18 probe sets, only two map within the 67 kb rhg1 region between the markers BARCSOYSSR_18_0090 and BARCSOYSSR_18_0094 (Kim et al. 2010). The identified amino acid transporter and the α-SNAP probe sets measured expression at all of the time points (3, 6 and 9 dpi) throughout the defense response. α-SNAP, through vacuolar sorting would be considered to play a role in defense involving autophagy (Liu et al. 2005; Hofius et al. 2009). In contrast, no clear role has been determined for the amino acid transporter and defense. The lack of detection for CBL-interacting protein kinase, a conserved unknown gene, a speckle-type POZ protein-related gene, a conserved unknown gene, a cys-rich domain protein gene, an elicitor inducible protein gene and an unknown gene in pericycle and their surrounding cell samples and samples isolated from syncytia undergoing defense or the susceptible reactions are noted. However, the expression presented here only reflects what was observed under our experimental conditions.

PAICE analyses link the cytological events pertaining to resistance to genes present at the rhg1 locus

From the studies that combined the expression data from the two genotypes, it appears that methyl jasmonate activity may be a part of a pathway that leads to transcriptional activation of genes involved in defense of soybean to the SCN. Previous reports have shown JA activity to be important in the resistance of plants to parasitic nematodes (Gao et al. 2008), supporting the transcriptomic work in G. max (Klink et al. 2007, 2009, 2010a). While JA activates defense pathways to many pathogens, upstream events including arachidonic acid metabolism supply metabolic products through the α-linoleic acid metabolic pathway that leads to the synthesis of 12-oxo-phytodienoic acid (OPDA) and methyl jasmonate. As already discussed, many pathways involved in lipid metabolism are observed to be expressed only in samples isolated from syncytia undergoing defense. Some of the downstream events, obvious through cytological examination of roots undergoing the defense response, include lignification and suberinization that stain readily with safranin as demonstrated by the original studies of Ross (1958) and re-examined later (Klink et al. 2009, 2010a, 2011). The synthesis of lignin and suberin is mediated through the activity of the phenylpropanoid pathway and shown to be induced in syncytia undergoing defense (Klink et al. 2007, 2009). Phenylpropanoid metabolites are involved in defense, providing a physical barrier to infection. One of the earliest structural features identified as providing a barrier against infection by some pathogens are CWAs. CWAs are present during the defense of plants to fungi (Aist 1976) and the plant parasitic nematode H. glycines in G. max [Peking] and G. max [PI 437654], but not G. max [PI 88788] (Kim et al. 1987; Mahalingham and Skorupska 1996). The formation of CWAs is linked to the aggregation of subcellular components at the infection site, a process that is dependent on the polarization of actin at the site of infection. The induced transcriptional activity of actin is observed in syncytia undergoing defense in G. max [Peking/PI 548402] (Klink et al. 2007, 2009), a soybean genotype known to have CWAs. The work of Böhlenius et al. (2010) implicates vesicular transport in CWA formation during infection of barley by Blumeria graminis. The experiments Böhlenius et al. (2010) relate increased production of phenylpropanoids to their delivery at localized sites in the cell at the cell wall. In related experiments, RNAi of the phenylpropanoid pathway components for monolignol biosynthesis that includes phenylalanine ammonia lyase (PAL), caffeic acid O-methyltransferase (CAOMT), caffeoyl-CoA methyltransferase (CCoAMT), and cinnamyl alcohol dehydrogenase (CAD), key components of lignin synthesis, results in super-susceptibility of wheat leaf tissues to an appropriate pathogen, B. graminis f. sp. tritici (Bgt) (Bhuiyan et al. 2009). All of these components have been shown induced during the defense response (Klink et al. 2007, 2009). The RNAi treatment also resulted in compromised penetration defense to a non-appropriate pathogen, B. graminis f. sp. Hordei (Bhuiyan et al. 2009). These observations are not surprising since CWAs are composed of materials such as lignin, pectin, suberin and chitin that are synthesized through the phenylpropanoid pathway. It is suggested that the methyl units synthesized through S-adenosylmethionine synthetase activity in the epidermal cells at the site of infection are metabolized into CWAs (Bhuiyan et al. 2007). A link between the synthesis of the CWA component lignin and methyl units has been made. Arabidopsis thaliana S-methionine synthetase mutants, although appearing identical to wild type, have a 22% decrease in lignin (Shen et al. 2002). The enzyme S-methionine synthetase is found to be highly induced in syncytia undergoing a defense response in G. max [Peking/PI 548402] (Klink et al. 2009). Other proteins known to compose CWAs include hydroxyproline-rich glycoproteins (HRGPs) and peroxidases. The analysis of the rhg1 region identified an extensin protein that is expressed specifically at all stages of the defense response. However, the extensin lies outside of the 67 kb rhg1 region as defined by Kim et al. (2010). The synthesis, deposition and assembly of extensin appear to be accompanied by localized release of reactive oxygen species (ROS) including H2O2. The release of H2O2 can drive the cross-linking of proteins like extensin, directly intoxicate the pathogen and/or drive the defense response in neighboring cells (Aist 1976; Bradley et al. 1992; Levine et al. 1994; McLusky et al. 1999; Hueckelhoven et al. 1999; Mellersh et al. 2002). The activity of genes involved in H2O2 production is observed in syncytia undergoing a defense response (Klink et al. 2007). Many of these observations suggest altered cell wall composition being important in defense.

Conclusion

The experiments show that a vast amount of relevant gene expression data, typically representing between 5 and 20% of the genes in the soybean genome, is discarded from cell-type specific differential expression studies. Analyzing biological processes occurring in homogeneous cell types will require a modified approach that can examine the unique gene expression profiles of the different cell types and different genotypes. As shown by the Illumina® deep sequencing data, the Affymetrix® DCM profiling is accurate and should be used as an additional complimentary measure of gene activity during any biological process under study. The advantage of the quantitative Illumina® methodology, in the absence of differential expression knowledge, is that relative expression of these genes can be obtained, providing a measure of the activity of the gene in a specific cell type. The ability to map the expression to resistance loci in a genotype-dependent manner should allow for a better understanding of expression nuances that define the cellular strategies employed by the different genotypes as well as generalized expression features as they combat SCN.