Validation of Reference Genes for Robust qRT-PCR Gene Expression Analysis in the Rice Blast Fungus Magnaporthe oryzae

The rice blast fungus causes significant annual harvest losses. It also serves as a genetically-tractable model to study fungal ingress. Whilst pathogenicity determinants have been unmasked and changes in global gene expression described, we know little about Magnaporthe oryzae cell wall remodelling. Our interests, in wall remodelling genes expressed during infection, vegetative growth and under exogenous wall stress, demand robust choice of reference genes for quantitative Real Time-PCR (qRT-PCR) data normalisation. We describe the expression stability of nine candidate reference genes profiled by qRT-PCR with cDNAs derived during asexual germling development, from sexual stage perithecia and from vegetative mycelium grown under various exogenous stressors. Our Minimum Information for Publication of qRT-PCR Experiments (MIQE) compliant analysis reveals a set of robust reference genes used to track changes in the expression of the cell wall remodelling gene MGG_Crh2 (MGG_00592). We ranked nine candidate reference genes by their expression stability (M) and report the best gene combination needed for reliable gene expression normalisation, when assayed in three tissue groups (Infective, Vegetative, and Global) frequently used in M. oryzae expression studies. We found that MGG_Actin (MGG_03982) and the 40S 27a ribosomal subunit MGG_40s (MGG_02872) proved to be robust reference genes for the Infection group and MGG_40s and MGG_Ef1 (Elongation Factor1-α) for both Vegetative and Global groups. Using the above validated reference genes, M. oryzae MGG_Crh2 expression was found to be significantly (p<0.05) elevated three-fold during vegetative growth as compared with dormant spores and two fold higher under cell wall stress (Congo Red) compared to growth under optimal conditions. We recommend the combinatorial use of two reference genes, belonging to the cytoskeleton and ribosomal synthesis functional groups, MGG_Actin, MGG_40s, MGG_S8 (Ribosomal subunit 40S S8) or MGG_Ef1, which demonstrated low M values across heterogeneous tissues. By contrast, metabolic pathway genes MGG_Fad (FAD binding domain-containing protein) and MGG_Gapdh (Glyceraldehyde-3-phosphate dehydrogenase) performed poorly, due to their lack of expression stability across samples.

Introduction samples and the use of efficient and specific primers [32][33][34]. In addition, the choice and number of reference genes used for data normalisation are critical for robust analysis of gene expression [34,35]. Studies have shown that expression stability of a reference gene varies between species, and that expression could also vary across sample tissue type and experimental conditions [32,33,36,37]. Moreover, using an inappropriate number of reference genes can adversely affect data reliability-as shown with Drosophila melanogaster head cDNA, where the relative expression of genes GSTD1, INR and HSP70 differed significantly when the normalising factor (NF) used either 1, 3 or 13 reference genes [38]. As such, the use of un-validated and/or single reference genes in qRT-PCR assays, such as, for example, MGG_β-Tubulin, MGG_Actin, MGG_Gapdh or MGG_Ubiq in M. oryzae is strongly discouraged. In addition, sample quality control, validation, and assay optimisation prior to running the qRT-PCR in gene expression assays are important for data reliability. Recognising this, recently there has been an increase in the number of reference gene validation studies in plant fungal pathogens such as in Puccinia sp. (wheat rust) [39], Fusarium sp. (wheat head blight) [40], and Aspergillus sp.(Black mould) [41]. However, prior to the work conducted by Park et al. [27], studies of gene expression in M. oryzae used only one, un-validated reference gene such as MGG_β-Tubulin [42,43], MGG_Actin, or MGG_Gapdh [44]. To redress this, Park et al. [27] recently attempted to validate seven candidate genes, identifying MGG_β-Tubulin as the most stablyexpressed gene of their cohort. The work, however, comes short of identifying the appropriate number of reference genes to use.
Through the emphasis for stringent qRT-PCR preparation, this paper was prepared in accordance to the guidelines made for the Minimum Information for Publication of qRT-PCR Experiments (MIQE) [34]. Furthermore, given the need to improve current measurement of gene expression practices in M. oryzae studies, this paper utilises additional candidate reference genes as well as genes commonly used in this field of study. Here, nine candidate reference genes from various functional groups were shortlisted, the stability validated and importantly, the appropriate number of genes needed for normalisation of gene expression under various fungal growth conditions were determined in this study. We then used the chosen reference genes to investigate the expression profile of MGG_Crh2, a cell wall remodelling protein, during various developmental stages and growth under various exogenous stresses. Thus, we have evaluated and determined the optimal number of reference genes to use, but have also appraised and added several further candidate genes, from disparate functional groups that out-perform the expression stability of MGG_β-Tubulin as the ideal reference gene. We hereby propose a robust qRT-PCR data normalisation strategy for transcript analysis in M. oryzae that will improve the confidence and reliability of all future gene expression analysis work related to this fungus.

Fungal Strains and Growth Conditions
Rice blast fungus Magnaporthe oryzae (M. grisea (T.T. Herbert) M.E. Barr) wild type (WT) strains Guy11 (MAT1-2) and TH3 (MAT1-1) were cultured on complete medium (CM) at 24°C, 14 hour light 10 hour dark cycle. Strain maintenance and media composition were as described by Talbot et al. [45]. Perithecia were produced by crossing Guy 11 with the opposite mating type strain TH3: agar plug inocula of the 2 strains were placed 4 cm apart on oatmeal agar and incubated under constant fluorescent light at 18°C for 29 days. Perithecia were then harvested using a dissecting microscope and snap-frozen in liquid nitrogen. Guy11 spores were harvested from 10-day-old CM plates by scraping the surface with 10 mL deionised water with a glass slide, and poured through triple layered Miracloth (Merck Chemicals Ltd, Padge Rd, Nottingham, NG9 2JR, UK). Spore concentrations were adjusted using a haemocytometer prior to use.

Tissue Preparation for RNA Extraction
Mycelial tissue grown under various stress factors. Two hundred μL of 4 x 10 5 spores/ mL Guy11 spore suspension was added to 500 mL of various liquid cultures, and incubated at 24°C for 3 days in the dark, under continuous shaking (150 rpm). Mycelia were harvested by filtering through double-layered Mira-cloth, dried with paper towels and 0.1 g of fungal mycelia was snap frozen in liquid nitrogen and stored at -80°C. The process was repeated to gather three biological replicate samples for each condition assayed.
For control vegetative growth under optimal conditions, M. oryzae Guy11 was cultured in Complete Medium (CM). Starvation was achieved by culturing in Minimal Medium (MM) made from 1x nitrate salts, 1% D-glucose, 0.1% ml trace elements solution, 0.1% thiamine solution, and 0.05% biotin solution. In extreme starvation, glucose was omitted from the Minimal Medium preparation above (MM-glucose). For cell wall stress, the cell wall perturbant Congo Red (SIGMA-Aldrich) was prepared in a 1% stock solution, filter-sterilised and added to sterile Complete Medium buffered with 50 mM HEPES (pH 7.0) at a concentration of 100 μg/mL. Stress by caffeine was induced by adding 2.5 mM filter-sterilised caffeine (SIGMA-Aldrich) to sterile Complete Medium (pH 6.5). To impose osmotic stress, 1 M Sorbitol was added to Complete Medium (pH 6.5) prior to autoclaving.
Dormant spores at 0 hours post inoculation (hpi). Spores were harvested from 10-dayold WT strain Guy11 grown on Complete Medium agar plates, 25 plates per biological replicate (three biological replicates harvested separately). The combined spores were centrifuged (5 minutes at 13, 000 rpm, 4°C), the supernatant removed and the pellet snap frozen in liquid nitrogen.
Germling development at 2, 8, 24, and 48 hours post inoculation (hpi) on host. Barley seeds (Hordeum vulgare L.) cv. Golden Promise were grown in a 50:50 (w/w) mixture of Erin Multipurpose Compost and John Innes No. 2 soil-based compost, for 7 days at 25°C, 12/12 light, 70% humidity. Seven-day-old barley leaves were cut and placed onto 1.5% (w/v) water agar and inoculated with harvested spores (6-7 x 10 6 spores/mL) re-suspended in 0.4% (w/v) gelatine. Inoculation procedure: for each time point, 2.5 mL of the spore suspension was sprayed uniformly onto 24 leaves, which were then covered to maintain high humidity and incubated at 25°C for 2, 8, 24 or 48 hours post inoculation. The hours post inoculation (hpi) corresponds to the various infective stages: 2 hpi (germ tube emergence), 8 hpi (immature appressoria), 24 hpi (penetration and early invasive growth), and 48 hpi (in planta growth). Prior to harvest, one leaf was selected at random, and its surface viewed by light microscopy to monitor germling development and formation of appressoria. Approximately 0.1 g of leaf material was then snap frozen in liquid nitrogen and stored at -80°C, for RNA extraction. The process was repeated, harvesting three biological replicates.
Perithecia. Guy11 and TH3 were inoculated onto oatmeal (10% (w/v)) agar and incubated as described previously. Perithecia (0.1 g) were harvested, snap frozen in liquid nitrogen and kept stored at -80°C prior to RNA extraction. Three biological replicates were harvested independently.

RNA Extraction, Quantity, and Quality Test
Tissues were extracted with TRIzol1 Reagent (Ambion1) using the manufacturers' instructions, followed by the addition of 50% volume isopropanol, with 5 minutes incubation at room temperature. Samples were loaded into an RNA Quick Spin column (Qiagen RNeasy Minikit) for on-column DNAse digestion and subsequent elution in RNase-free water. RNA sample concentration was determined using the Thermo Scientific 'NanoDrop1 ND-1000' spectrophotometer. Samples were normalized to the same concentration of 1000 ng/μL (±100 ng). RNA quality and RIN number for each sample were tested using the 'Agilent 2100 Bioanalyzer Instrument' as per the manufacturers' instructions to ensure minimal RNA degradation. Sample RNA was discarded if the RIN value was below 7.5 for samples containing only fungal RNA, and if the RIN value was below 6.0 for samples containing a mix of plant leaf and fungal pathogen RNA.

cDNA Preparation
Qiagen 'Maxima First Strand cDNA Synthesis Kit for RT-qPCR' kit was used as per manufacturers' instructions. The same amount of template RNA (1.0 μg), was added per 20 μl reactions for each tissue sample. Three controls were prepared: Non-template control (NTC), minus Reverse Transcriptase (-RT) and RNA from Barley leaf only (BL). The reaction mix was incubated for 10 min at 25°C, followed by 30 min at 50°C, and the reaction terminated by heating at 85°C for 5 min. Products were retained at -20°C and used within 2 weeks.

Candidate Reference Gene Selection and Primer Design
Candidate reference genes were selected on the basis of being (a) previously used in M. oryzae gene expression studies [42,43,[46][47][48], and (b) from several essential functional groups (protein translation, glycolysis pathway, cytoskeleton). These genes were then cross-checked with the publicly available COGEME high throughput-super SAGE transcript profile M. oryzae database [49], for an estimation of stability. Stability ratios were calculated using the given base mean values between samples of the same growth stage (example: Complete Medium against Minimal Medium, or 4 hpi against 16 hpi). Genes with a ratio less than 10 from the transcriptomic database were then selected for further analysis. Sequences for the respective genes were obtained from the 8 th annotation (MG8) Magnaporthe oryzae strain 70-15 sequenced by Broad Institute [30]. All primer pairs (Table 1) were synthesized by SIGMA-Aldrich and designed according to the following specifications: annealing Tm 60°C (±2°C); G-C content 50-60%; no non-specific product amplification; no primer-dimer or secondary structures; amplicon of 60-110 bp; primer efficiency 80-110%, r 2 value 0.98-0.99; either Forward or Reverse primer to span exon/exon boundary (transcript-specific). Primer specificity: primers were deemed acceptable if the targeted gene showed an e value less of than 10 −3 in NCBI Blast search. In addition, primer pairs were shown to be pathogen-specific by conducting qRT-PCR using only host barley leaf cDNA as a negative control (BL). Primers were tested for non-specific product/s by amplicon separation on 2% (w/v) agarose gel electrophoresis and 10-300 bp GeneRuler Ultra Low Range DNA Ladder (Thermo Scientific) at the end of a qRT-PCR run. [50] Quantitative Real Time PCR (qRT-PCR) Real time PCR for the Infective group was conducted using an Mx3000P qPCR System (Agilent Technologies), which was replaced with an ABI 7000 apparatus (Applied Biosystems) for the Vegetative and Global group. All reactions were conducted in 96-well plates (MicroAmp1 Optical 96-Well Reaction Plate cat N8010560) utilising a sample maximisation arrangement and the addition of inter-run calibration (IRC) wells for each gene in each 96-well plate [51]. 'Power SYBR Green PCR Master Mix' assay (Applied Biosystems) was used in a total volume of 25 μL per well/reaction. Each 25 μL reaction contained 2.0 μL cDNA, 8.5 μL water, 1.0 μL (10 μM) forward primer, 1.0 μL (10 μM) reverse primer and 12.5 μL 'Power SYBR Green PCR

Evaluation of Reference Gene Expression Stability Using GeNorm Analysis
Analysis of raw Ct data for all candidate reference genes was performed using 'qBase+ For each group (Infective, Vegetative or Global), raw ct values were converted to relative expression values using the Pfaffl Method [52] and inter-run calibration was performed to correct for run-to-run variation [51]. The GeNorm algorithm [53] in 'qBase+ Basic' licence software version 2.5 (Biogazelle) was used to calculate the average expression stability score (M) for each reference gene, and the pairwise variability score (V) for sequential normalisation factors, resulting from stepwise inclusion of additional reference genes. Expression stability values were used to identify the most stably-expressed reference genes at each stage, whilst pairwise variability scores were used to determine the most appropriate number of reference genes to use as the normalisation factor.
Gene of Interest (GOI): Orthologue of S. cerevisiae Crh1 gene The Crh1 gene was first identified in the model yeast S. cerevisiae [19]. The protein sequence was obtained from the Saccharomyces Genome Database [54] and the corresponding orthologue in M. oryzae was obtained from "Magnaporthe comparative Sequencing Project" by the Broad Institute of Harvard and MIT [30] using the BLASTp search tool [55]. Here, the yeast Crh1 protein sequence was queried against proteins predicted to be encoded by the M. oryzae 70-15 (MG8) sequenced genome. The best hit genes with a cut-off value of e -10 , were then reciprocal BLASTp [55] against S. cerevisiae genome in the NCBI database (www.ncbi.nlm.nih.gov).

Data analysis: Expression Profile of MGG_Crh2 Across the M. oryzae life-cycle
The validated combination of reference genes according to tissue types were used as Normalising Factor (NF) to analyse the expression profile of a cell wall remodelling enzyme, MGG_Crh2 (MGG_00592) under various stages of the life cycle (Infective and Global assay) and growth conditions (Vegetative assay). In the latter, the expression of MGG_Crh2 under starvation (MM), osmotic stress (Sorbitol), extreme starvation (MM-glucose), and cell wall stress imposed by Congo Red (CR) or caffeine was measured relative to vegetative growth in nutrient-rich medium (CM). In the Infective assay, expression profile was measured during germling development for tissues harvested from 2 hpi (germ tube), 8 hpi (immature appressoria), 24 hpi (penetration and early invasive growth) and 48 hpi (in planta growth), relative to 0 hpi (dormant spores). Finally, the expression profile of MGG_Crh2 was captured across all stages of the life cycle in the Global assay by comparing MGG_Crh2 expression in representative tissues. These were 8 hpi (immature appressoria), 48 hpi (in planta growth), vegetative growth (CM) and perithecia, relative to 0 hpi (dormant spores).
Two technical replicates were used per reaction. When possible, the plate design carried all three biological replicates for target and control sample respectively, on the same plate; and, for each sample analysed, MGG_Crh2 and two validated reference genes were deposited onto the same plate. For expression analysis involving more than one 96-well plate, comparison of samples across different plates was achieved by incorporating inter-run calibration (IRC) wells for each gene in equivalent wells within each 96-well plate. The generated Ct values for samples from multiple runs were imported into MxPro QPCR Software or 'qBase-Plus Basic Version 2.5' for IRC calibration [51].
The calibrated values from multiple runs and raw Ct values from one well, were exported into a Microsoft Excel file and analysed separately according to the recommendations by Willems et al. (56). In particular, following their protocol we log transformed, mean centered, and autoscaled the data from three biological replicates, before evaluating the significance of up/ down regulation at the 0.05 level by observing whether or not the 95% confidence interval for fold-change increase/decrease in expression contained a fold-change value of 1 (no change). Significant up/down regulated results are indicated by an asterisk on all graphs.

Results & Discussion
Nine candidate reference genes (Table 1) were selected on the basis of being (a) previously used in M. oryzae gene expression studies [42,46,47], (b) from several essential functional groups (protein translation, glycolysis pathway, cytoskeleton) and (c) expression data in the publicly available high throughput-superSAGE transcript profiling COGEME database of M. oryzae [49]. A base mean ratio above 10 in the COGEME database was interpreted as indicating that gene expression was unstable, and the gene was unsuitable to be a candidate reference gene. Genes with a base mean ratio below 10 were selected for subsequent sensitive analysis (qRT-PCR using GeNorm calculations) to identify the most stable gene.

Primer Specificity and Efficiency Check
Each primer specificity assay yielded a single amplicon of the expected size for the primer sets tested (Table 1 and S1 Fig-S2 Fig). There was no amplification, or high Ct values (>35 Ct), for both non-template controls, minus reverse transcriptase (-RT) and the host-only control (BL). This showed that the reagents were free from contamination as there was no amplification of gDNA and no non-specific amplification of plant cDNA. Calculation of primer efficiencies using five-fold dilution of pooled complementary DNA (cDNA) for all nine HKG primers and MGG_Crh2 gave r 2 >0.99 and 80-110% efficiency (E) values (S3 Fig).

RNA and cDNA Quality and Quantity Check
The RNA Integrity Number (RIN) values derived from pure fungal RNA samples (dormant spores, perithecia, and vegetatively-grown tissues) were >7.5 (S4 Fig), whilst those containing a mix of pathogen and host plant tissues had RIN values >6.0 (S5 Fig). The lowered RIN values from plant-derived RNA were attributable to the presence of both pathogen and host leaf RNA (contributing cytosolic and chloroplastic ribosomal RNA), giving rise to additional multiple peaks during electrophoretic separation. Consequently, the slightly lowered RIN number is not due to actual nucleic acid degradation (RIN <6.0), but due to the presence of additional peaks, which increases the calculated total area below the electrophoretic graph, a measurement required for the RIN algorithm [56,57].

Expression Stability of Candidate Reference Genes
The gene stability (M) values of the nine candidate reference genes, assayed separately for each group of tissues (Vegetative, Infective, and Global), are shown in Fig 1. Candidate genes with the lowest M value were the most stably expressed, whilst genes with the highest M value were least stable. The minimum number of reference genes required for reliable and accurate normalisation for each tissue group was determined as two genes, with a "cut-off" value was set at 0.15 [53] as shown in Fig 2. The gene stability (M) and pairwise variation (V) measurement data were used to identify the two best-performing reference genes for each group; MGG_Ef1 and MGG_40s for the Vegetative Group; MGG_40s and MGG_Actin for the Infection Group, and MGG_Ef1 with MGG_S8 or MGG_40s (same M value) for the Global Group.

Reference Gene Validation
Our recommendations for use of the named reference genes for the given tissues are robust but we advocate caution in extrapolation to untested experimental conditions, such as in the use of different exogenous stresses (eg. temperature, pH or heavy metal stress). Moreover, inferred co-regulation is an important consideration in the choice of reference genes and should be avoided, if possible. Here, we selected at least two candidate genes from key functional classes, that is, the glycolysis pathway (MGG_Fad & MGG_Gapdh), protein synthesis (MGG_Ef1, MGG_S8 & MGG_40s), cytoskeleton assembly (MGG_β-Tubulin & MGG_Actin) and protein degradation (MGG_Ubiq). For the Infective Group, the two most stable genes, MGG_40s, and MGG_Actin belong to two distinct functional groups. However, for the Vegetative and Global Groups, genes belonging to the protein synthesis pathway proved most robust in stability, being unaffected by changes to environmental stressors or fungal growth phases. In this case, it is more important to use a set of validated stable genes instead of prioritising genes from different biological pathways. Furthermore, for the Vegetative Group, the two most stable genes belong to protein synthesis pathways with exclusive functions: MGG_Ef1 and MGG_40s, therefore have little risk of co-regulation [58,59]. In the Global Group, the most stable genes lie in the protein synthesis pathway. Here, as MGG_40s and MGG_S8 both form one 40S functional subunit, we recommend instead MGG_Ef1 (M = 0.38), used in conjunction with MGG_40s (M = 0.37) or MGG_S8 (M = 0.37). Here, the M value differences are very small, suggesting that all three genes are similarly stable and thus this will not significantly affect normalisation reliability.
Thus far, a single study has attempted identification and validation of M. oryzae reference genes for qRT-PCR [27]. In that study, Park et al. (2013) chose seven candidate genes, three of which are used herein and can therefore be compared directly, namely MGG_β-Tubulin, MGG_Actin and MGG_Gapdh. Based on GeNorm analysis, Park et al. (2013) ranked their seven genes across tissue sample types that concur with our Global Group, whereby MGG_Gapdh has lower stability (high M value), with MGG_Actin and MGG_β-Tubulin being more stably expressed (low M value). Our combined findings also correspond for the Infective Group, whereby MGG_Actin is more stable than MGG_Gapdh and MGG_β-tubulin. This consistency between studies confirms the reproducibility and reliability of qRT-PCR when quality control practices were adhered to. In addition to the three genes shared by both studies, here we introduce a validated set of more robust genes, particularly those belonging to the protein synthesis pathway (MGG_Ef1, MGG_40s, and MGG_S8) that can be used for gene expression studies in M. oryzae. This study also demonstrates that for accurate and reliable normalisation of M. oryzae genes across all tissue groups (Infective, Vegetative, and Global), the optimal  Determination of the optimal number of reference genes for accurate normalisation. Graphs show calculated pairwise variation, V n/(n+1) , between normalisation factor NF n and NF n+1 for a total of nine candidate reference genes. The optimal number of reference genes for accurate normalisation was taken to be the smallest number for which V was less than a cut-off of 0.15 [53]. The analysis was repeated for the three designated groups: A) Infective; B) Vegetative, and; C) Global. doi:10.1371/journal.pone.0160637.g002 Robust Reference Genes for Gene Expression Analysis in the Rice Blast Fungus number of reference genes to use is two and not one as used in many M. oryzae gene expression analysis studies hitherto.

Gene of Interest (GOI): MGG_Crh2 Expression Profile
Reciprocal blast search using the S. cerevisiae gene Crh1 protein sequence revealed MGG_00592 as the best-hit gene and most probable orthologue, with a cut-off value of e -10 . It is hereby named MGG_Crh2. We exploited the stable pairs of recommended reference genes as normalising factors to trace the expression profile of the cell wall remodelling gene MGG_Crh2, an orthologue of S. cerevisiae Crh1 gene (S1 Dataset). In the vegetative assay, we used MGG_40s and MGG_Ef1 as reference genes and observed that MGG_Crh2 expression responded to cell wall stress imposed by Congo Red, being elevated some 2-fold (p<0.05), and was downregulated under extreme starvation (MM-glucose) (Fig 3, Graph A and B) but was not affected by growth under cell wall stress imposed by caffeine, hyperosmotic stress (Sorbitol), or by moderate starvation (MM).
In S. cerevisiae, Crh1 has previously been reported to be upregulated 2-4 fold in the presence of Congo Red [26], and cell wall stress caused by Congo Red was shown to activate the cell wall integrity pathway directly via the phosphorylation of Rlm1 [26]. Additionally, the lack of response to cell wall stress induced by caffeine is an interesting observation. Caffeine is a compound known to activate the cell wall integrity pathway leading to phosphorylation of the S. cerevisiae Mpk1 and shown to alter yeast cell wall architecture upon transient exposure to caffeine [60]. Further studies need to be conducted to investigate whether MoMig1p [61], an orthologue of S. cerevisiae Rlm1p, indeed regulates the expression of MGG_Crh2 under cell wall stress and if so, whether it is selective to certain types of stressors.
In the infective assay, MGG_40s and MGG_Actin were used as the normalising factor with expressions measured relative to 0 hpi (dormant spores). Results in Fig 3 (Graph C) showed that MGG_Crh2 is not regulated during germling development as its expression in each tissue type (germ tube, immature appressorium, penetration and early invasive growth as well as in planta growth) was not significantly different from dormant spores. This is an interesting outcome, given that in other M. oryzae cell wall protein studies cell wall biosynthesis and remodelling enzymes such as the chitin synthase Chs7 gene [62] and two Gel orthologues (MGG_11861 and MGG_08370) [63] were found to be upregulated during appressorium development and for the former, mutants were non-pathogenic. Interestingly, studies of the human fungal pathogens Candida albicans and Aspergillus fumigatus showed that the orthologues of S. cerevisiae Crh1p (Crh11p and Crf1p respectively) are specifically detected by the mammalian host immune response system [64][65][66][67]. It is therefore conceivable that the lack of upregulation of MGG_Crh2 in infective structures aids M. oryzae in evading detection. The literature is currently devoid of information on plant immune responses elicited by recognition of secreted fungal cell wall remodelling enzymes. There is however, evidence showing that the cell wall of M. oryzae is remodelled during infection to 'mask' the fungi with α-1,3 glucan [68], suggesting that the fungi is indeed under selection to evade detection. This unexplored area of plant-pathogen interactions is worth exploring in future work.
The theory is further supported by results in the Global assay (Fig 3 graph D). In this assay, MGG_Crh2 gene expression in tissues representative of each stage of the life cycle was each measured relative to dormant spores (0 hpi). Data analysis showed that this gene is most highly expressed during vegetative growth by up to 3-fold higher (p<0.05), followed by 2.4-fold higher in perithecia and lowest expression during the infective stages, as shown in Fig 3, during appressorium development (8 hpi) and in planta growth (48 hpi).
In summary, this expression study has shed light into the regulatory profile of the transglycosylase cell wall remodelling gene, MGG_Crh2, which was found to be most highly expressed during vegetative growth with down-regulation during the development of infective structures and in planta growth. We also have identified an optimum set of control genes for studying gene expression across the life-cycle of M. oryzae using qRT-PCR. Further work is required to expand this gene-set to cover a broader range of growth conditions, such as varied exogenous  stresses, but this robust approach sets a standard of practice for qRT-PCR gene analysis which can only increase our understanding of this devastating rice pathogen.