Evaluation of automated cell disruptor methods for oomycetous and ascomycetous model organisms

Takao Kasuga* and Mai Q. N. Bui

USDA-ARS Crop Pathology and Genetics Research Unit, Department of Plant Pathology, University of California Davis, One Shields Avenue, Davis, CA 95616, USA (tkasuga@ucdavis.edu)

Fungal Genetics Reports 58:4-13

Two automated cell disruptor-based methods for RNA extraction, disruption of thawed cells submerged in TRIzol Reagent (method QP), and direct disruption of frozen cells on dry ice (method CP), were optimized for a model oomycete, Phytophthora capsici, and a model filamentous ascomycete, Neurospora crassa.  The results were compared with more conventional methods of manual grinding in a mortar and pestle under liquid nitrogen (method M&P) and those using lyophilized samples.  A chip-based electrophoresis system showed that methods CP and M&P yielded high integrity RNA from both P. capsici and N. crassa.  In contrast, method QP and lyophilized sample-based methods resulted in inconsistent RNA integrity between the two organisms, indicating they are not safe alternatives for method M&P.  Microarray mRNA profiling for P. capsici revealed alterations in global mRNA profiles in those samples that the chip-based electrophoresis detected slight decreases in RNA integrity.  Despite this, RNA integrity of these samples could still be high enough to pass conventional stringent quality control measures.  This demonstrated the necessity of global mRNA profiling for the evaluation of RNA extraction protocols.

Emerging high-throughput transcriptome analysis technologies such as microarray mRNA profiling (DeRisi et al., 1997) and more recently RNA-Seq (Nagalakshmi et al., 2008) enable us to monitor thousands of gene activities simultaneously and greatly facilitate the study of functional genomics.  In responding to the ever-growing demand for transcriptomic analysis for large numbers of samples, automated cell disruptor-based methods for high-throughput RNA extraction are gaining popularity (Van der Vorst et al., 2009).  It should be emphasized that sensitivity and accuracy of high-throughput transcriptome analysis relies heavily upon the quality of input RNA samples.  It is not clear if automated cell disruption is appropriate for high-throughput transcriptome analysis or if it is compatible with only certain tissue types.
Unlike genomic DNA, mRNA transcripts are extremely sensitive to growth conditions of the cell. Transcript levels can change on a minute scale due to both degradation and de novo transcription.  Consequently, alterations in transcript profiles that are not representative for the condition of interest are easily introduced during sample harvesting (Fleige & Pfaffl, 2006; Pieterse et al., 2006) and/or during sample preparation.  Accordingly, in order to correctly interpret the mRNA profiling data, extent of such experimental noise needs to be evaluated and minimized.  As a means to minimize such unwanted changes, biological samples are traditionally snap-frozen in liquid nitrogen, and kept at -80°C.  RNA extraction is then carried out by means of a mortar and pestle with liquid nitrogen (Sambrook & Russel, 2006).  Ground tissues are kept frozen and subsequently transferred to a tube containing extraction buffer and are homogenized immediately.  A chaotropic reagent in the extraction buffer, such as guanidinium thiocyanate (Chomczynski & Sacchi, 1987), denatures and deactivates enzymatic activities, such as nuclease and mRNA transcription activities, thus protecting RNA integrity.  In contrast, in a typical automated cell disruption method, beads such as small glass, ceramic or steel beads and extraction buffer are added to frozen cell cultures resting on an ice bath prior to cell disruption.  Thus, unlike the traditional mortar and pestle-based method, samples are inevitably thawed on ice or in the extraction buffer prior to mechanical disruption of the cell.  Judging from the inherent instability of mRNA transcripts, thawing can perturb global mRNA profiles, thus introducing experimental noise.  To our surprise that reproducibility of global mRNA profiling data for cell disruptor-based methods has not been evaluated.  Furthermore, no standardized protocols are available for cell disruptor-based methods for fungi or oomycetes. 
In this study, a model oomycete Phytophthora capsici (Lamour et al., 2007), and a model filamentous ascomycete, Neurospora crassa (Galagan et al., 2003), were chosen as models, and frozen as well as lyophilized samples were subjected to cell disruption.  There are two main methods for cell disruption; (1) disruption of cells submerged in extraction buffer, and (2) that under liquid nitrogen or dry ice without extraction buffer.  Choice of beads for cell disruption is also crucial.  Four distinctive beads with different sizes and materials; Lysing Matrix A (MP Biomedicals, Solon OH, USA, sold for lysis of hard samples such as cartilage, bone and seed), Lysing Matrix C (MP Biomedicals, for lysis of yeast, algae and fungi), Lysing Matrix D (MP Biomedicals, for lysis of plant and animal tissue) and Lysing Matrix R, which was a home made mixture of two different sized glass beads, were used in combination with the two aforementioned cell disruption methods, and evaluated for RNA yield.  RNA integrity and quality was then evaluated by means of a chip-based electrophoresis system for the two model organisms, and microarray mRNA profiling for P. capsici.  We found that global mRNA profiles were perturbed in response to thawing and lyophilization of samples.  GO ontology enrichment analysis was then used to infer cellular activities which were responsible for the alteration of mRNA profiles.  A means to minimize experimental noise associated with cell disruptor-based protocols are discussed.

Methods

As with method QP, for method CP, four 2 ml screw-cap microcentrifuge tubes, each containing approximately 100 mg of a frozen mycelium were prepared in triplicate.  However, for method CP, the microcentrifuge tubes were kept on a microcentrifuge tube rack (Subzero Blue IsoFreeze Flipper Rack, GeneMate; BioExpress, Kaysville, UT, USA) on dry ice.  Into each microcentrifuge tube one of each of the four lysing matrices, chilled at -20°C, was added.  The cells were then disrupted at a speed setting of 6 meters/second for 40 seconds, twice, in a CoolPrep adapter, which was filled with crushed dry ice.  One ml of TRIzol Reagent was then added to the pulverized sample, and vigorously vortexed until the sample was completely thawed and homogenized.  Total RNA was subsequently extracted according to the manufacturer’s protocol for TRIzol. 
For the method M&P, up to 200 mg of mycelium was ground with mortar and pestle in liquid nitrogen.  Approximately 100 mg of ground cell tissue was then transferred to each of the 2 ml screwcap microcentrifuge tubes, which were kept on dry ice.  1 ml of TRIzol Reagent was then added to each of the tubes and immediately vortexed to homogenize the samples.  RNA was then extracted according to the manufacturer’s protocol for TRIzol Reagent. 
In addition, two widely used methods for RNA extraction from lyophilized samples were adapted for an automated cell disrupter.  Fresh cell cultures or those stored in -80°C, approximately 100 mg each, were snap-frozen in liquid nitrogen and subjected to lyophilization for 24 hours under 20 to 100 mTorr.  For each model organism three separate lyophilizations were conducted for biological controls.  Lyophilized samples were either pulverized with Lysing Matrices A at room temperature on a QuickPrep adapter (method LQ, treatment 7 in Table 2) or snap-frozen in liquid nitrogen and pulverized on a CoolPrep adapter with dry ice (method LC, treatment 6).  For both methods, cells were disrupted at a speed setting of 6 meters/second for 40 seconds, twice.  One ml of TRIzol Reagent was then added and RNA extraction was carried out as method CP described above.
Up to 100ug of total RNA was further cleaned using the RNeasy mini protocol for RNA cleanup (Qiagen, Valencia, CA, USA).  RNA purity and concentration were determined by NanoDrop 3300 spectrophotometer (Thermo Scientific, Waltham, MA, USA) with RiboGreen fluorescent dye (Invitrogen Life Technologies).  Integrity of RNA was determined by both agarose gel electrophoresis and Experion automated electrophoresis system (Bio-Rad, Hercules, CA, USA).  Experion’s RNA Quality Indicator (RQI), is independent of sample concentration, and is equivalent to RNA Integrity Number (RIN, Agilent Technologies).  Although, different algorithms are employed, RQI and RIN are adjusted to give approximately the same values for identical RNA samples (Denisov et al., 2008).

Microarray design

The P. capsici whole-genome expression oligonucleotide 60-mer arrays (4x72K multiplex format, Roche NimbleGen, Madison, WI, USA) were designed by the manufacturer on the basis of 17,383 gene models derived from the P. capsici database at DOE Joint Genome Institute (PhycaF7_best_transcripts.fasta.gz; http://genome.jgi-psf.org/PhycaF7/PhycaF7.home.html).  On average four probes per open reading frame (ORF), a total of 69,421 probes for 17,363 gene models, were synthesized on each of the four arrays on the multiplex array slide.

cDNA synthesis, hybridization and image acquisition
cDNA synthesis, labeling, hybridization procedure, data acquisition and normalization were carried out according to the manufacturer’s instructions (Roche NimbleGen).  Briefly, 10 ug of total RNA and oligo dT primer were used to synthesize the first strand of cDNA, which was followed by the synthesis of the second strand of cDNA to yield double stranded cDNA.  The Cy3 cyanine dye-labeled random 9-mers were then used to label the cDNA.  The cDNA was then precipitated with isopropanol, vacuumed dried, and afterward used for hybridization.  Hybridization was done at 42°C for 16 to 20 hours on a MAUI Hybridization System (BioMicro Systems, Salt Lake City, UT, USA).  After three steps of washing, microarrays were scanned on an Axon GenePix 4000B (Molecular Devices, Sunnyvale, CA, USA).  Quantile normalization and background correction across arrays were performed using Robust Multi-chip Average (RMA) algorithm (Irizarry et al., 2003) implemented in NimbleScan Version 2.5 software.  A MIAME-compliant microarray dataset (Brazma et al., 2001) has been deposited in Filamentous Fungal Gene Expression Database at Yale University (http://bioinfo.townsend.yale.edu/) (Zhang & Townsend, 2010).  Table S1 lists mRNA profiling results and functional annotations.

 Data analysis
Normalized intensity data for 17,363 genes across 17 arrays were obtained by RMA.  First, between pairwise samples, an intensity ratio of each gene was plotted against the average intensity of gene expression to visualize intensity-dependent ratio (MA plot, (Dudoit et al., 2002)).  Background noise for genes with normalized hybridization intensity of below 64 was disproportionately high.  Therefore, genes with average hybridization intensity below 64 were removed from the dataset.  The remaining 12,504 genes were used for further analysis.  Pearson’s correlation coefficient between global mRNA expression patterns was then use to cluster cDNA samples using hclust function with average linkage option in a statistical software R 2.7.1 (http://bioconductor.org).
Between hierarchical clusters of cDNA samples, two-sample t-test was used to identify differentially expressed genes using the software R 2.7.1.  In order to identify difference in cellular activities between the hierarchical clusters, gene ontology (GO)-based functional enrichment analysis was conducted.  Gene ontology (GO) annotation (Harris et al., 2008) for P. capsici genome was obtained from the DOE Joint Genome Institute (JGI, PhycaF7_GO.tab.gz; http://genome.jgi-psf.org/PhycaF7/PhycaF7.download.ftp.html).  Out of the 17,363 P. capsici gene models that were represented on the microarray, a total of 1,677 GO terms were assigned to 7,767 gene models according to the annotation scheme conducted by JGI.  Over- or under- representation of GO terms in differentially expressed gene groups in relation to the genome was evaluated against an expected hypergeometric distribution using Fisher’s exact test in the software R 2.7.1.  A significant level of 0.05 was used with multiple testing corrections according to Benjamini and Hochberg (Benjamini & Hochberg, 1995). 

Results

Representatives of oomycetes and filamentous ascomycetes, Phytophthora capsici and Neurospora crassa, respectively, were used for optimization of RNA isolation using a bench-top cell disruptor, FastPrep-24 Instrument (MP Biomedicals).  Integrity and global mRNA profiles of obtained RNA specimens were then cross-examined with total RNA isolated by mortar and pestle under liquid nitrogen (M&P) and two lyophilization-based methods.

Optimization of RNA extraction using a cell disruptor
Cell disruptors are widely used for DNA and RNA extraction as high-throughput alternatives for mortar and pestle.  Several different methods of cell disruption and various lysing matrix beads, with which tissues are pulverized, are available.  First, we searched for a method that gives high quality RNA at high and consistent yields.  Two distinctive cell disruption methods, each with a specialized adaptor, were tested.  For the first cell disruptor method, 4°C TRIzol Reagent was added to frozen tissues, and homogenized immediately on a QuickPrep adaptor (method QP).  For the second method, frozen tissues were pulverized at a frozen state on a dry ice-filled CoolPrep Cryogenic Adapter, and subsequently 4°C TRIzol Reagent was added and vortexed (method CP).  For both methods, four distinctive lysing matrix beads were evaluated for RNA yield: i.e. Lysing Matrix A, Lysing Matrix C,  Lysing Matrix D, and Lysing Matrix R (details in Introduction and Materials and methods).  Additionally, two lyophilized sample-based methods, samples disrupted on a dry ice-filled CoolPrep adapter with Lysing Matrix A (method LC) and samples disrupted on a QuickPrep adapter at room temperature with Lysing Matrix A without extraction buffer (method LQ), were conducted.
Similar trends for performance of extraction methods for P. capsici and N. crassa were recognized.  For method QP, the yield difference due to lysing matrix beads were small (Table 1).  However, for N. crassa Matrix R significantly underperformed any other lysing matrices (p<0.001).  For method CP, difference in performance of matrices was more prominent than that of method QP.  Matrices A and R, which were comprised of fine and large beads, outperformed Matrices C and D, both of which were comprised of small uniform beads, for P. capsici (t test, p<0.01) and for N. crassa (p<0.001).  RNA yields from lyophilized samples (methods LC and LQ) and those from a mortar and pestle under liquid nitrogen (method M&P) were comparable to the highest values for method CP.  The quality of RNA obtained by the five methods, QP, CP, LC, LQ and M&P, judged by intensity of 28S and 18S ribosomal RNA bands on agarose gel, was indistinguishable (data not shown).

 
Experimental design for evaluation of cell disruptor protocols by means of a chip-based capillary electrophoresis and microarray mRNA profiling
With appropriate combinations of cell lysing matrices and adaptors, the high throughput cell disruptor was shown to yield a comparable quantity of RNA in comparison to method M&P.  The next objective was to examine whether RNA samples derived from the cell disruptor, fresh or lyophilized samples, had high integrity and were able to reproduce global mRNA profiles of RNA samples derived from method M&P.  Especially, unlike M&P, in method QP, TRIzol Reagent is needed to add to samples prior to cell disruption, which could potentially impact global mRNA profiles.  Also, effect of lyophilization on transcriptome has not been investigated.
For each of the five RNA isolation methods, the number of biological replicates used for the measurement of RQI values and sample names of biological replicates used for microarray profiling were shown in Table 2.  For the QP method, no more than four samples were handled simultaneously, and no longer than 2 minutes had elapsed between unscrewing caps for the addition of TRIzol Reagent to frozen samples and the initiation of mechanical disruption. 

 
Additionally, we simulated a situation where a large number of samples are handled together.  Commercially available cell disruptor such as Mini-Beadbeater (Biospec, Bartlesville, OK) and Precellys Homogenizer (Bertin Technologies, Montigny-le-Bretonneux, France) are able to hold 45 and 24 microcentrifuge tubes, respectively.  The maximum sample load for a QuickPrep adaptor on FastPrep 24 is 24 microcentrifuge tubes.  When 24 samples are handled, 5 to 10 minutes is needed between unscrewing the first sample and initiation of a homogenizer.  In such a scenario, if samples are kept in an ice bath, they are exposed to 0°C or a higher temperature, and thus there is an opportunity for thawing.  Upon the addition of 4°C TRIzol Reagent, samples will inevitably be thawed in a short period of time.  It is possible that prolonged duration in a thawed state could introduce noise to mRNA profiles.  In order to account for high-throughput RNA extraction, two additional treatments were designed for method QP (Table 2).  First, approximately 100 mg each of frozen P. capsici tissues were incubated with 4°C TRIzol Reagent in a 4°C water bath for 10 minutes to evaluate the effects of TRIzol and thawing (Table 2, treatment 1).  Frozen samples of P. capsici were also incubated in microcentrifuge tubes in a 4°C water bath to assess the thawing effect alone (treatment 2).  In addition, RNAlater (Ambion), an aqueous reagent widely used to preserve RNA in fresh animal tissues (treatment 4) (Florell et al., 2001; Mutter et al., 2004), and RNALater-ICE (treatment 3) used to preserve RNA in previously frozen animal tissues (Li et al., 2004), were tested for the hope that they were capable of preserving mRNA profiles in P. capsici after harvest.  In order to assess mRNA response to cold-shock alone, cultures were harvested and transferred to microcentrifuge tubes, without liquid nitrogen snap-freeze, and incubated at 4°C for 10 minutes (treatment 5).  The two lyophilized sample-based methods (treatments 6 and 7), were also included for evaluation of their mRNA quality.

Effect of post-harvest treatments on mRNA profiles of P. capsici
Except for the RNAlater ICE-treated samples, all the P. capsici samples produced comparable quantities of total RNA.  Several attempts were made to extract RNA from RNALater ICE-treated samples, however, the yielded RNA was always degraded.  RNA quality was then assessed by RNA quality indicator (RQI) values presented by Experion microfluidic capillary electrophoresis system (Bio-Rad).  For those samples without post-harvest treatment, average RQI values ranged from 9.5 to 9.8, indicating that RNA integrity was high and comparable among the three extraction methods (Table 2).
For those samples subjected to various post-harvest treatments, a diverse range of responses was observed.  Thawed samples in the absence of TRIzol Reagent (method 2) showed slightly decreased RQI values (9.3).  It was our surprise that lyophilized samples showed the lowest RQI values accompanied with elevated baseline signals.  Note that method LQ yielded lower RQI (6.8) than method LC (RQI 8.1),  We repeated lyophilization a total three times including six biological replicates for each method but decrease in RQI value was consistently detected.  Remaining samples with post-harvest treatment showed RQI values comparable to samples without post-harvest treatment.  Integrity measures of RNA for these thawed samples were in the range of “intact RNA” (Denisov et al., 2008), and those for lyophilized samples with method LC were in the range of “less intact” but reliable for microarray mRNA profiling (Copois et al., 2007).  RQI values for lyophilized samples with method LQ were below the reliable range for microarray but in the reliable range for quantitative reverse transcription PCR (Fleige & Pfaffl, 2006; Fleige et al., 2006).  All the RNA samples except for that treated with RNAlater ICE were, thus, processed for microarray mRNA profiling.
In order to visualize effects of sample treatments on mRNA, seventeen P. capsici samples were hierarchically clustered according to pairwise distance (1 - correlation coefficients) between global mRNA profiles (Fig. 1).  Three distinctive clusters, designated groups A, B and C, were observed.  The group A includes all the samples prepared by methods M&P (MP1), QP (QP1, QP2, QP3) and CP (CP1, CP2, CP3), all without post-harvest treatment, as well as the sample treated with 4°C RNAlater (RL2) and two samples incubated in 4°C TRIzol Reagent for 10 minutes (TR1, TR2).  One sample, treated at 4°C for 10 minutes without snap-freeze in liquid nitrogen (NF1) was outgrouped to all the above-mentioned samples in group A.  Group B includes all the four lyophilized samples (LC1, LC2, LQ1, LQ2), and group C includes two snap-frozen samples, which were thawed and incubated at 4°C for 10 min (FT2 and FT3).

Fig. 1. Hierarchical clustering of cDNA samples with various post-harvest treatments. 17 samples were clustered based on their expression patterns of 12,504 genes. The seven samples used for the optimization of cell disruption methods, CP, QP and M&P were indicated with filled circle, filled square and filled star, respectively. Group B is comprised of lyophilized samples (indicated with filled inverted triangles), and group C is comprised of samples which were thawed and incubated without TRIzol Reagent (indicated with filled triangles). All the remaining samples were found in group A.

Hierarchical clustering showed mRNA profiles to be perturbed by both lyophilization and freeze and thaw.  It should be noted that the culture (NF1) incubated at 4°C for 10 minutes without snap-freeze in liquid nitrogen, was found in group A, which was comprised mainly of samples without prolonged post-thaw incubations.  This indicates that the observed deviation of mRNA profiles in group C were largely attributable to the freeze-thaw process rather than a response to environmental changes, such as deprivation of nutrient due to transfer of cells from culture medium to a microcentrifuge tube, injury incurred during transfer, or the temperature shift from 23°C to 4°C.  The sample TR1 and TR2 incubated at 4°C for 10 minutes with TRIzol Reagent, were closely associated with samples which were not subjected to post-harvest treatments prior to cell disruption.  These data indicate that the change in the global mRNA profiles due to freeze and thaw was effectively suppressed when samples were immersed in TRIzol Reagent.  In addition, the mRNA profile of the fresh culture treated with RNAlater overnight at 4°C (RL2) was found within group A.  This indicates that RNAlater is able to immediately stabilize mRNA in P. capsici, and perturbation by the treatment is minimum, if there is any.  Four lyophilized samples were found in group B, which is distant from the sample prepared by mortar and pestle (MP1), indicating that lyophilization affects mRNA profiles.  Furthermore, samples pulverized at room temperature (LQ1 and LQ2) and those on dry ice (LC1 and LC2) formed discrete subclusters, indicating that sample temperature during pulverization also affects mRNA profiles.

Genes associated with certain biological processes were enriched in lyophilized and frozen-thawed samples.
Although repeated attempts were made, P. capsici cultures have never survived the process of snap-freeze in liquid nitrogen and rapid thawing or lyophilization, in the conditions used in this research (data not shown).  Inviable cells, however, displayed alteration in global mRNA profiles in response to freeze and thaw.  It is not clear if the observed alteration is of biological implication or is a part of the natural process of cell decomposition.  Elucidation of the mechanism of the alteration in mRNA profiles could yield clues to minimize mRNA noises that occur during sample preparation.  We hypothesized that if alteration of mRNA profiles is a consequence of a biological response, then activation/deactivation of groups of genes involved in specific biological processes or metabolic pathways will be observed.  Enrichment analysis of functional categories can be used to detect the activation/deactivation of such specific biological processes or metabolic pathways.  To test this hypothesis, we conducted gene ontology (GO) enrichment analysis (Subramanian et al., 2005).  GO terms describe gene products in terms of their associated biological processes, cellular components and molecular functions in a species-independent manner (Harris et al., 2008).  Out of 17,363 P. capsici genes on the microarray, at least one GO term was assigned to 7,767 gene models.  From microarray mRNA profiling data, we first identified apparently differentially displayed genes with statistical support (p>0.05) between groups (Fig. 1 and Table S1).  Student’s T-test showed that transcript abundances of 7,310 genes were significantly different between groups.  Note that because multiple hypothesis correction has not been conducted, these genes include truly significant genes as well as false positives. 
GO enrichment analysis revealed that 22 GO terms were enriched in the comparison between group A and group B (Table S2).  Memberships of these GO terms are partially overlapped, and they can be summarized into five main terms (Table 3).  In response to lyophilization, mRNA transcripts for 1) “alcohol catabolic processes” and 2) “glucose catabolic process” became less abundant, whereas genes for 3) “phosphorylation”, 4) “microtubule-based process” and 5) “cell communication” became abundant.  Likewise in comparison between group A and group C, fourteen GO terms were enriched.  Again, a large part of them has redundant descriptions having overlapped gene sets, and they can be summarized into two main terms.  In response to freeze and thaw, mRNA transcripts for 1) “cellular amino acid and derivative metabolic process” became less abundant, whereas those for 2) “microtubule-based process” became abundant. 

Effect of post-harvest treatments on mRNA profiles of N. crassa
In P. capsici, we found that RNALater and TRIzol post-harvest treatments preserved RNA integrity and expression profiles, whereas, lyophilization and thawing of samples at 4°C reduced RNA integrity and altered mRNA profiles.  In N. crassa, methods CP and M&P also yielded RNA with high RQI values (Table 2, RQI 9.7 for both methods).  However, unlike P. capsici, method QP slightly reduced RNA integrity in N. crassa in comparison to methods CP and M&P (RQI 9.4, t-test, p=0.04).  Furthermore, when samples were incubated for a longer period of time in TRIzol (treatment 1), a decrease in RNA integrity became more noticeable for method QP (RQI 8.9).  As has been observed for P. capsici, thawing at 4°C slightly reduced RNA integrity (RQI 9.2, treatment 2).  Lyophilization again reduced RQI values in N. crassa.  The lyophilized sample pulverized at room temperature showed smaller RQI (8.8, treatment 7) than that pulverized on dry ice (RQI 9.2, treatment 6).  However, compared to P. capsici, reduction of RQI values was less severe.  In N. crassa both RNALater ICE and RNALater gave heavily degraded RNA (treatment 3 and 4).  

Discussion

There is a growing demand for high-throughput RNA isolation, and automated cell disruptors have been gaining popularity (Van der Vorst et al., 2009).  Validation of the technology is, however, limited to 1) RNA integrity measured by capillary electrophoresis and 2) mRNA profiles of a handful of genes measured by real time quantitative PCR (Van der Vorst et al., 2009).  In this research we first optimized automated cell disruptor protocols for an oomycete plant pathogen P. capsici and a filamentous ascomycete N. crassa and then inspected RNA quality by capillary electrophoresis for both N. crassa and P. capsici and by global mRNA profiling for only P. capsici.
We found that lyophilization as well as thawing of samples prior to the addition of extraction buffer resulted in significant alteration of mRNA profiles.  This finding is notable becase thawing is inevitable for the majority of protocols devised with cell disruptors, (e. g. (Kasuga & Glass, 2008; Monteiro et al., 2009; Rautio, 2010)) as well as the hot phenol protocol for frozen yeast cells (e.g. (Castillo et al., 2002; Eelderink-Chen et al., 2010)).  It should be emphasized that the effect of thawing, which is evident from microarray mRNA profiling, cannot be explicitly detected by conventional quality control measures.
For method QP for P. capsici, it was found that adding TRIzol Reagent to frozen samples could minimize the change of mRNA profiles due to thawing.  This indicates that TRIzol Reagent percolates into P. capsici cells and ceases enzymatic activity; as a consequence this prevents the alteration of mRNA profiles.  In contrast, a 4°C TRIzol treatment was found to compromise RNA integrity in N. crassa.  This implies that the cell wall of N. crassa is less permeable to TRIzol, allowing RNA to degrade.  We do not have mRNA profiling data for N. crassa, however, it is likely that addition of TRIzol also affects its transcriptome.  We, therefore do not recommend the use of method QP unless effects of TRIzol on the preservation of mRNA of particular species or tissue types being investigated has properly been evaluated by means of global mRNA profiling. 
RNAlater (Ambion), is an aqueous reagent, and is widely used to stabilize and protect RNA in animal tissue (Mutter et al., 2004).  The global mRNA profile of the RNAlater-treated P. capsici sample closely resembles that prepared by M&P, indicating that the reagent immediately permeated the tissue and stabilized mRNA.  On the other hand, an RNALater treatment resulted in severely degraded RNA from N. crassa.  This is consistent with our finding that while a 4°C TRIzol treatment successfully preserved RNA integrity in P. capsici, the same treatment on N. crassa yielded slightly degraded RNA.  We again do not recommend RNALater unless its effectiveness has been proved for interrogated species or tissue types. 
Lyophilization is another widely used method to cease cellular activity prior to RNA extraction (Leary et al., 1969; Sanchez-Rodriguez et al., 2008).  Lyophilized samples can be effectively pulverized on a mortar and pestle or an automated cell disruptor.  We found that lyophilization damaged RNA integrity and altered mRNA profiles in P. capsici.  The adverse effect of lyophilization was also observed for N. crassa but to a lesser extent.  In lyophilization, samples frozen in liquid nitrogen are placed in a vacuum chamber at room temperature, where gas pressure is typically at 100 mTorr.  Because the chamber pressure is below the saturation pressure of water vapor, sublimation will progress, while removal of heat by sublimation keeps the sample frozen.  It is possible that a rigid cell wall and increasing concentrations of solutes in the cytosol effectively prevent the cell from complete desiccation, by which the cell maintains residual cellular activity such as de novo transcription and mRNA degradation.  GO enrichment analysis revealed that in the lyophilized samples 105 genes with a GO term “phosphorylation”, of which 53 were annotated as “kinases”, and also 41 genes with a GO term “cell communication” appeared to be increased.  It seems that P. capsici is capable of activating signal transduction cascades in response to lyophilization.  Reactivation of transcriptional machinery likely happens when samples in the vacuum chamber are thawed during lyophilization.   Again, the effect of lyophilization, which is evident from microarray mRNA profiling, cannot be detected by conventional quality control measures with certainty. 
Thawed samples also showed a large alteration in the transcriptome.  A cryoinjury of the cell is likely responsible for the shift of mRNA profiles.  During freezing, formation of intracellular ice and shrinkage of cells occur in various ascomycete and oomycete species (Morris et al., 1988), which lead to loss of integrity of cell membrane, organelles, and thus viability.  GO enrichment analysis showed that in response to freeze and thaw, genes for microtubule motors were activated.  Lyophilized samples, which were also likely to have incurred cryoinjury, showed activation of genes for microtubule motors.  Taken together, activation of microtubule motor might be associated with cell repair. 
In summary, we demonstrated that a transcriptome is extremely sensitive to RNA extraction protocols.  A brief thawing of samples on ice or in contact with extraction buffer or lyophilization process can significantly alter mRNA profiles.  We have analyzed mRNA profiles for only the oomycete P. capsici, however, such alteration in transcriptomes is likely to occur for the ascomycete N. crassa as well as diverse organisms e.g. bacteria, plants and animals.  We recommend the use of a mortar and pestle or frozen-phase cell-disruption method unless application of RNAlater or TRIzol-submerged cell disruption method is proven safe by means of global mRNA profiling.

References

Benjamini, Y., Y. Hochberg, 1995. Controlling the false discovery rate - a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B-Methodological 57: 289-300.

Brazma, A., P. Hingamp, J. Quackenbush & other authors, 2001. Minimum information about a microarray experiment (MIAME) - toward standards for microarray data. Nat Genet 29: 365-371.

Castillo, E. A., J. Ayte, C. Chiva, A. Moldon, M. Carrascal, J. Abian, N. Jones, E. Hidalgo, 2002. Diethylmaleate activates the transcription factor Pap1 by covalent modification of critical cysteine residues. Mol Microbiol 45: 243-254.

Chomczynski, P., N. Sacchi, 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate phenol chloroform extraction. Anal Biochem 162: 156-159.

Copois, V., F. Bibeau, C. Bascoul-Mollevi & other authors, 2007. Impact of RNA degradation on gene expression profiles: Assessment of different methods to reliably determine RNA quality. J Biotechnol 127: 549-559.

Denisov, V., W. Strong, M. Walder, J. Gingrich, H. Wintz, 2008. Development and validation of RQI: an RNA quality indicator for the Experion automated electrophoresis system. Bio-Rad Bulletin #5761.

DeRisi, J., V. R. Iyer, P. O. Brown, 1997. Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278: 680-686.

Dudoit, S., Y. H. Yang, M. J. Callow, T. P. Speed, 2002. Statistical methods for identifying differentially expressed genes in replicated cDNA microarray experiments. Statistica Sinica 12: 111-139.

Eelderink-Chen, Z., G. Mazzotta, M. Sturre, J. Bosman, T. Roenneberg, M. Merrow, 2010. A circadian clock in Saccharomyces cerevisiae. Proc Natl Acad Sci U S A 107: 2043-2047.

Englander, L., L. F. Roth, 1980. Interaction of light and sterol on sporangium and chlamydospore production by Phytophthora lateralis. Phytopathology 70: 650-654.

Fleige, S., M. W. Pfaffl, 2006. RNA integrity and the effect on the real-time qRT-PCR performance. Mol Asp Med 27: 126-139.

Fleige, S., V. Walf, S. Huch, C. Prgomet, J. Sehm, M. W. Pfaffl, 2006. Comparison of relative mRNA quantification models and the impact of RNA integrity in quantitative real-time RT-PCR. Biotechnol Lett 28: 1601-1613.

Florell, S. R., C. M. Coffin, J. A. Holden, J. W. Zimmermann, J. W. Gerwels, B. K. Summers, D. A. Jones, S. A. Leachman, 2001. Preservation of RNA for functional genomic studies: A multidisciplinary tumor bank protocol. Mod Pathol 14: 116-128.

Galagan, J. E., S. E. Calvo, K. A. Borkovich & other authors, 2003. The genome sequence of the filamentous fungus Neurospora crassa. Nature 422: 859-868.

Harris, M. A., J. I. Deegan, J. Lomax & other authors, 2008. The Gene Ontology project in 2008. Nucleic Acids Res 36: D440-D444.

Irizarry, R. A., B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, U. Scherf, T. P. Speed, 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4: 249-264.

Kasuga, T., N. L. Glass, 2008. Dissecting colony development of Neurospora crassa using mRNA profiling and comparative genomics approaches. Eukaryot Cell 7: 1549-1564.

Lamour, K. H., J. Win, S. Kamoun, 2007. Oomycete genomics: new insights and future directions. FEMS Microbiol Lett 274: 1-8.

Leary, J. V., A. J. Morris, A. H. Ellingbo, 1969. Isolation of functional ribosomes and polysomes from lyophilized fungi. Biochim Biophys Acta 182: 113-120.

Li, M., W. Li, H. J. Min, Q. Z. Yao, C. Y. Chen, W. E. Fisher, 2004. Characterization of somatostatin receptor expression in human pancreatic cancer using real-time RT-PCR. J Surg Res 119: 130-137.

Monteiro, J. P., K. V. Clemons, L. F. Mirels, J. A. Coller, T. D. Wu, J. Shankar, C. R. Lopes, D. A. Stevens, 2009. Genomic DNA microarray comparison of gene expression patterns in Paracoccidioides brasiliensis mycelia and yeasts in vitro. Microbiology-Sgm 155: 2795-2808.

Morris, G. J., D. Smith, G. E. Coulson, 1988. A comparative-study of the changes in the morphology of hyphae during freezing and viability upon thawing for 19 Species of fungi. Cryobiology 25: 517-517.

Mutter, G. L., D. Zahrieh, C. M. Liu, D. Neuberg, D. Finkelstein, H. E. Baker, J. A. Warrington, 2004. Comparison of frozen and RNALater solid tissue storage methods for use in RNA expression microarrays. BMC Genomics 5: 88.

Nagalakshmi, U., Z. Wang, K. Waern, C. Shou, D. Raha, M. Gerstein, M. Snyder, 2008. The transcriptional landscape of the yeast genome defined by RNA sequencing. Science 320: 1344-1349.

Pieterse, B., R. H. Jellema, M. J. van der Werf, 2006. Quenching of microbial samples for increased reliability of microarray data. J Microbiol Methods 64: 207-216.

Rautio, J. J., 2010. Multiplex gene expression analysis by TRAC in fungal cultures. Methods Mol Biol 638: 165-173.

Sambrook, J., D. W. Russel, 2006. The Condensed Protocols from Molecular Cloning: A Laboratory Manual. New York: Cold Spring Harbor Laboratory Press.

Sanchez-Rodriguez, A., O. Portal, L. E. Rojas, B. Ocana, M. Mendoza, M. Acosta, E. Jimenez, M. Hofte, 2008. An efficient method for the extraction of high-quality fungal total RNA to study the Mycosphaerella fijiensis-Musa spp. Interaction. Mol Biotechnol 40: 299-305.

Subramanian, A., P. Tamayo, V. K. Mootha & other authors, 2005. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102: 15545-15550.

Van der Vorst, S., A. F. Dekairelle, L. Irenge, M. Hamoir, A. Robert, J. L. Gala, 2009. Automated cell disruption is a reliable and effective method of isolating RNA from fresh snap-frozen normal and malignant oral mucosa samples. Clin Chem Lab Med 47: 294-301.

Vogel, H. J., 1956. A convenient growth medium for Neurospora (Medium N). Microbial Genet Bull 13: 42-43.

Zhang, Z., J. P. Townsend, 2010. The filamentous fungal gene expression database (FFGED). Fungal Genet Biol 47: 199-204.