As the human genome project is completed, microarray technology offers
the potential to study the genomes complexity. This technology facilitates
the direct extraction of functional information from nucleic acids by
measuring the RNA levels of a complete organism or its associated subsets.
After isolation of total RNA, various methods can be applied to prepare
targets for microarray screening (Figure 1). The most common procedure
involves direct cDNA labeling by reverse transcription in which fluorescencelabeled
nucleotides are incorporated. A clear limitation of this technology is
the large amount of RNA required per hybridization. Some of the most important
applications in medical research involve studying very small amounts of
tissues (e.g., microdissected tissue [LCM] or biopsy material from tumors).
Therefore, target amplification methods have been developed to overcome
The linear cRNA amplification procedure, based on
reverse transcription with an oligodT primer connected
to a T7 promoter and in vitro transcription of resulting
DNA with T7 RNA polymerase [1, 2], amplifies a RNA
target approximately 100-fold. This method does not
significantly distort the relative abundance of individual
mRNA sequences within an RNA population [3, 4].
A PCR amplification method based on reverse transcription,
followed by random PCR amplification of the
cDNA and in vitro transcription of the resulting PCR
product with T7 RNA polymerase, is illustrated in
Figure 2. With this procedure as little as 50 ng total RNA
(from 1,000 cells or 0.1 mg of
tissue) are required for
To investigate the bias of these amplification methods, we used yeast
as a model eukaryotic organism due to its low level of splicing variants.
The experiments included two different cell populations grown in rich
and minimal media.
Material and Methods
Target preparation /amplification
Saccharomyces cerevisiae was grown in YPD medium (50 g/l, rich) and SD
base medium (26.7 g/l, minimal). Two percent D+glucose (carbon source)
was added to both media. Cells were grown to an OD of 0.4 in a 1 : 4 dilution
and collected from 40 ml of cell suspension. Total RNA isolation was performed
as described in .
Direct cDNA labeling was performed according to the MWG PAN YEAST Array
Application Guide. One hundred microgram of total RNA was labeled in a
40-l labeling reaction by using moloney murine leukemia virus (MMLV)
reverse transcriptase and Cy3- and Cy5-dCTP (Amersham Biosciences). Linear
amplification was performed with 10 g of starting total yeast RNA according
to the package inserts of the cDNA Synthesis System, Microarray Target
Purification Kit, and Microarray RNA Target Synthesis Kit (T7), using
Cy3- or Cy5-UTP for labeling. Random PCR amplification was performed with
50 ng total yeast RNA according to the package inserts of the Microarray Target
Amplification Kit, Microarray Target Purification Kit and
Microarray RNA Target Synthesis Kit (T7), using Cy3 or Cy5
One microgram of Cy3- and Cy5-labeled cDNA or 10 g
cRNA from the respective yeast population was used for
hybridization. Hybridizations were performed overnight at
42 C using a buffer containing 50% formamide.
Using a contact printing principle, 5-modified 40-bp oligonucleotides
were covalently attached to the solid surface
. Each array was comprised of 6,250 oligonucleotides
that detected different yeast open reading frames (ORFs).
Scanning for signal intensities was performed with a
confocal laser scan microscope. Each slide was scanned
six times per channel with an increasing photomultiplier
setting to expand the dynamic range of the measurement.
Resulting images were analyzed with the Imagene
4.0 Software; the output files were corrected between
both channels via total signal intensities and local background
subtraction using the MWG MAVI Software.
Quantitative RT-PCR was performed on the LightCycler
Instrument using the DNA Hybridization Probe format.
cDNA was produced from yeast total RNA in a 20-l reaction,
including 20 g total RNA and a defined copy number
of neomycin mRNA (spike-in control), oligo dT primer
and avian myeloblastosis virus (AMV) reverse transcriptase.
RNA and primers were first denaturated at 70 C for
10 minutes, then incubated at 42 C for 60 minutes, followed
by a 5-minute denaturation at 94 C. Two hundred picograms and 20 pg of cDNA, primers and hybridization
probes, and LightCycler FastStart DNA Master (as
described in the protocol) were used in a 20-l reaction.
The reaction conditions included: 10 minutes denaturation
at 95 C, 45 cycles at 95 C for 10 seconds, 55 C for 15
seconds, and 72 C for 15 seconds. Fluorescence was
monitored at the end of each 55 C incubation. The fluorescence
detected in channel F2/F1 was analyzed using
the LightCycler Analysis Software. The crossing point for
each reaction was determined using the second derivative
maximum algorithm and the arithmetic baseline adjustment.
For quantification, a titration curve of neomycin
mRNA was used.
Results and Discussion
Correlation of results
obtained with and without amplification
Total RNA from Saccharomyces cerevisiae grown in rich
or minimal medium was labeled with Cy3 and Cy5 by
three different methods (Figure 1):
100 g total RNA by reverse transcription (cDNA
10 g total RNA by linear amplification (cDNA
synthesis followed by in vitro transcription)
50 ng total RNA by random amplification (random
PCR amplification followed by in vitro transcription).
Two hybridizations for each set of probes were performed
on the MWG PAN YEAST Array, including a dye swap to
minimize normalization and hybridization artifacts . A threshold for low signal intensities was calculated based
on the average signal intensities of 15 Arabidopsis-specific
negative control spots. The average value of those 15
spots, plus the standard deviation multiplied by 1.96, gives
a 95 % confidence that signals above this value are based
on specific hybridization. Among a total of 6,250 spots, 75% (4,690) displayed signal
intensities above the threshold. For those yeast ORFs, the
average ratio of hybridization signals obtained from yeast
grown in rich or minimal medium was determined. Within
the 4,690 yeast ORFs, 562 ORFs displayed a ratio of ≥2,
and 496 ORFs displayed a ratio o
f ≤0.5 in the direct-labeling
approach. These expression ratios were very reproducible
for each target-labeling method applied. This is shown in Table 1 for the cDNA labeling procedure. The
vast majority of the genes (97%) showed a less than twofold
variation from the average expression ratio. Similar
results can be obtained by comparing linear amplification
with linear amplification, and random amplification with
random amplification (data not shown).
The most important question is whether the three applied
methods deliver similiar expression differences. Therefore
we compared the cDNA labeling method with the amplification
methods. Table 1 shows that 92% of the yeast
genes show comparable expression alterations (≤ 2-fold
variation) when cDNA labeling is compared to linear
amplification. When cDNA labeling is compared with random
PCR amplification, 89% of the yeast genes show
similar expression rates. These results indicate that different
amplification methods might not generate the same
number of labeled target molecules from each template.
Rather, the degree of amplification within a method is
reproducible from one reaction to another. In addition, the
different methods deliver similar expression alterations.
Validation of results with
quantitative PCR on the LightCycler Instrument
To find out which target preparation methods would reflect
the biological situation best, an independent method
(quantitative RT-PCR) was applied to a number of genes.
We randomly selected two genes from a group of upregulated
genes and two genes from a group of downregulated
genes in rich vs. minimal medium. We also
selected three housekeeping genes that should show no
alterations, and four genes from the group in which different
methods show differences in the expression alterations.
The group of u
p- and down-regulated genes, as well
as the houskeeping genes, show similar ratios (Figure 3).
The yol 086c and yjl153c ORFs indicate that cDNA labeling
and LightCycler quantification (to a higher degree)
have a broader dynamic range.
For the genes varying in their expression levels, depending
on the target preparation method used (Figure 3D), Light-
Cycler analysis confirmed this result. Surprisingly, yor 091w
seems to be down-regulated when using the cDNA labeling
method, but is up-regulated when applying other
methods. This was not expected since cDNA labeling is
considered to introduce less bias. Nevertheless, this result
was confirmed in an independent experiment (Figure 3E).
Besides potential distortion of the relative abundance of
individual mRNA by the method used, differences
between amplified and non-amplified targets may also
be due to different structures (RNA vs. DNA) and
length. Targets produced by direct labeling of cDNA varied
from 400bp to several kb in length. Amplified targets
(cRNA) require fragmentation prior to hybridization.
Therefore those fragments vary from 50bp to several
100bp in length. Whereas 1 g of labeled cDNA could be
applied on the array, 10 g cRNA is needed for sufficient
signal intensities. Differences in target length, structure
(RNA vs. DNA) and concentration can lead to differences
based on the formation of secondary structures
among different targets during hybridization as well as
the variability in hybridization patterns.
The analysis of expression profiles from different organisms
and cell types by microarrays is now being implemented
in many laboratories. Based on the variable
amount of starting material, we have shown that cDNA
labeling, linear amplifica
tion, and PCR-based amplification
methods provide consistent and comparable results.
Source:Page: All 1 2 3 4 5 6 7 Related biology technology :1
. An Epitope Tagging Vector for Gene Expression in Mammalian Cells2
. Antibodies for Studying NMDA Receptor Protein Expression and
. High-Level Expression of Peanut Allergens Affected by Rare Codon Usage4
. Versatile Vectors for Ponasterone A- Inducible Control of Gene Expression in
. Innovative Tissue Array Technology for High-Throughput Screening of Gene
. New Mammalian Expression Vectors Employ Stable, High-Level Fluorescence
Humanized Renilla GFP Reporter7
. Functional Cloning Using
ViraPort Retroviral cDNA Expression Libraries8
. High-Level Protein Expression, One-Column Purification, and FLAG
Epitope Tagging in E. coli9
. A New Lambda Vector for Mammalian Expression10
. Expression and Purification of Recombinant Proteins That Have Native Amino
. Codon Bias-Adjusted BL21 Derivatives for Protein Expression