Microarray and Metabolite Profiling Methods
Total RNA for each sample was extracted and prepared for hybridization according to
the protocols outlined in the GeneChip Expression Analysis Technical Manual
(Affymetrix Santa Clara, Ca). The quality of all RNA samples was evaluated using the
Agilent’s 2100 bioanalyzer before target labeling. The IVT reaction is also run through
the bioanalyzer. This served as a QC for RNA quality and quality of the IVT product.
Briefly, 8 g of total RNA was used as template for first strand cDNA synthesis
(Superscript, Invitrogen) which was primed with a T7-(dT)24 primer containing a T7
RNA polymerase promoter sequence (Genset Oligos). In vitro transcription was
performed on second strand product using biotinylated UTP and CTP (Bioarray High
Yield RNA Transcript labeling Kit, Enzo Diagnostics). Biotinylated cRNA was heated in
Mg buffer resulting in 35-200 base fragments. Arabidopsis arrays (Affymetrix) were
hybridized for 16 hr at 45oC with 15 g of fragmented cRNA. Arrays were stained with a
streptavidin-phycoerythrin conjugate (Molecular Probes) and scanned with an Agilent
argon-ion laser with a 488 nm emission and a 570 nm detection (GeneArray™ Scanner).
MAS5.0 software was used for image acquisition. Microarray Suite version 5
(Affymetrix, Santa Clara, CA) was used to generate *.cel files. MAS5.0 was used to
perform an absolute analysis of each chip. Several metrics associated with the image
were examined to determine if the chip is providing "good" information. The metrics
included percent present, background, raw Q, scaling factor, 3':5' ratios for actin and
gapdh. Probe ProfilerTM version 1.3.11 software (Corimbia Inc, Berkeley, CA) was used
to convert *.cel file intensity data into quantitative estimates of gene expression for each
probe set. The signal values for genes with a significant treatment effect (p< .05) were
normalized by performing a Z-transformation thereby generating a distribution with mean
0 and standard deviation of 1 for each gene (Draghici S. 2003. Data analysis tools for
DNA microarrays. Chapman & Hill/CRC Mathematical biology and medicine series. p
466, 2003). After acquisition, all metric values must fall into acceptable ranges and
within those acceptable ranges be similar. These ranges are empirical and based on
previous experience. The metrics associated with the software indicate if the chips in the
data set are comparable. The metrics include brightness, outliers, saturation, background,
cel outliers, signal from hybridization controls and 3':5' ratios of actin and gapdh. The
Probe profiler software identifies probe pairs which to not provide informative
information and either eliminates the probe pair or assigns a realistic values in its place.
The probe Profiler software generates a signal value for a probe set, a SD for this score
assuming additive variance, an SDlog which is an estimate of the natural log, loge of the
multiplicative standard deviation assuming multiplicative variance and the p-value testing
the null hypothesis that the gene in this scan has an expression score greater than zero.
The average intensity of the array signal values determine and then scaled to an arbitrary
value (100). The data is then filtered to eliminate genes whose p-value for presence is >
.05 on all chips in the experiment. The data is filtered to remove genes for which there
are any missing values. The signal values for genes significantly effected by the treatment
a normalized by subtracting the genes mean signal value by the SD of the genes signal
value. This generates a standard normal distribution for each gene. The normalized values
are run through various clustering programs (PCA & 2-way hierarchical) to determine if
the treatment replication are responding similarly.
Metabolite Methods
Time points for metabolite profiling during cold acclimation were selected based
on freeze-tolerance time-course experiments (Kaplan et al., 2004). Cold acclimation was
initiated 2 h after the onset of the light period, which allowed for the harvest of all
samples within the light period. Three-week-old 20°C grown plants were placed at 4°C
and sampled at 1, 4, 12, 24, 48 and 96 h. Additionally, untreated controls were taken at 0
time of the experiment and 4 h after the experiment began. All samples were rapidly
harvested, flash-frozen in liquid nitrogen (<30 sec), and stored at -80°C until metabolite
extraction. Two sets of temperature stress experiments were performed, each comprising
2-4 replicate measurements per time point. Aerial tissues were ground in liquid nitrogen
with pestle and mortar.
Polar metabolites were prepared as previously described (Kaplan et al., 2004;
Wagner et al., 2003). Briefly, aerial tissues were ground in liquid nitrogen with pestle
and mortar. Aliquots of 60 mg frozen powder were extracted with hot MeOH/CHCl3 and
the fraction of polar metabolites processed as described (Wagner et al., 2003). Ribitol,
isoascorbic acid, and deuterated alanine were added as internal standards. Carbonyl
moieties were protected either by methoximation or by ethoximation, using alkoxyamine
hydrochloride in pyridine. Afterward, acidic protons were derivatized with N-methyl-Ntert-butyldimethylsilyltrifluoroacetamide (MTBSTFA) at 70 C or N-methyl-Ntrimethylsilyltrifluoroacetamide (MSTFA). One-microliter aliquots of these solutions
were injected at a split ratio of 1:25 into a GC/MS system consisting of an AS 2000
autosampler, a GC 8000 gas chromatograph, and a Voyager quadrupole mass
spectrometer including a dynode/phosphor/photomultiplier detector (all ThermoQuest,
Manchester, U.K.). Mass spectra were recorded from m/z 50 to 600 at 0.5 s scan-1 for
trimethylsilylated samples (TMS) and from m/z 50 to 800 at 0.7 s scan-1 for tertbutyldimethylsilylated samples (TBS). Accurate mass measurements were made using a
Finnigan MAT magnetic sector field instrument (Finnigan, Bremen, Germany).
Chromatography was performed using a 30 m × 250 m SPB 50 column (Supelco,
Bellefonte, PA) or a 30 m × 250 m DB 5-MS column (J&W Scientific, Folsom, CA).
Injection temperature was 230 C, the interface was set to 250 C, and the ion source was
adjusted to 200 C. Helium flow was 1 mL min-1. After a 5-min solvent delay time at 70
C, the oven temperature was increased at 5
C min-1 to 310 C, 1 min isocratic, cool-
down to 70 C, followed by an additional 5-min delay. Plant samples were derivatized
and then used to confirm peak identities by comparison to reference compounds and use
of the NIST, Wiley, and internally compiled spectra libraries. Retention time correction
was done by internal reference compounds in order to minimize run-to-run errors. All
chemicals were purchased at Sigma-Aldrich-Fluka (SAF, Deisenhofen, Germany). Stable
isotope reference metabolites ([13C12]sucrose, [13C6]glucose, glycerol-d8, ethanolamined4, ethylene-d6 glycol, aspartate-d3, [13C5]glutamate, alanine-d4, valine-d8, leucine-d3 and
benzoic-d5 acid) were obtained from Campro Scientific (Emmerich, Germany).
Metabolite Identification
GC-MS based metabolite profiling detects and quantifies specific mass spectral
fragments in defined retention time windows. Identification of these fragments was
performed through standard addition experiments using pure authenticated compounds to
confirm identity by retention time index and mass spectrum. Compounds were designated
as metabolites if they were identified with a match >750 on a scale of 0-1000 and RI
deviation < 3.0. Other unidentified compounds are designated as mass spectral tags
(MSTs) designated by the code NA and a unique number. In cases of high mass spectral
similarity of MSTs to available commercial or custom mass spectral libraries, MSTs were
named in square brackets by a preceding match value and a compound name taken from
these libraries. Representative mass spectra and RI, which serve for metabolite
identification in Arabidopsis thaliana, and novel identifications post publication, will be
available through CSBDB (http://csbdb.mpimp-golm.mpg.de/csbdb/dbma/msri.html).