|
/ Index / Introduction / Principles
DNA microarray principle
DNA microarrays allow for rapid measurement and visualisation of differential
expression between genes at the whole genome scale. If technique implementation
is quite complicated, it’s principle is very easy. Here are described the
major steps involved in this process:
Microarray production process:
DNA fragments amplified by PCR technique are spotted on a microscopic glass
slide coated with polylysine prior to spotting process. The polylysine coating
goal is to ensure DNA fixation through electrostatic interactions. PCR fragments
are in our case the expressed part (ORF) of the 6200 Saccharomyces cerevisae
genes (baker yeast). Slide preparation is achieved by blocking the polylysine
not fixed to DNA in order to avoid target binding. Prior to hybridisation, DNA
is denatured to obtained a single strand DNA on the microarray, this will allow
the probe to bind to the complementary strand from the target. Apart from glass
slide microarray other types of chips exist:
|
High density filters(macroarrays) |
Glass slides (microarrays) |
Oligonucleotides chips |
 |
 |
 |
Detail:
 |
Detail:
 |
Detail:
 |
| Size: 12cm x 8cm |
Size: 5,4cm x 0,9cm |
Size: 1,28cm x 1,28cm |
- 2400 clones by membrane
- radioactive labelling
- 1 experimental condition by membrane
|
- 10000 clones by slide
- fluorescent labelling
- 2 experimental conditions by slide
|
- 300000 oligonucleotides by slide
- fluorescent labelling
- 1 experimental condition by slide
|
Target preparation:
RNA are extracted from two yeast cultures from which we want to compare expression
level. Messengers RNA are then transformed in cDNA by reverse transcription.
On this stage, DNA from the first culture with a green dye, whereas DNA from
the second culture is labelled with a red dye.
Hybridisation:
Green labelled cDNA and red labelled ones are mixed together (call the target)
and put on the matrix of spotted single strand DNA (call the probe). The chip
is then incubated one night at 60 degrees. At this temperature, a DNA strand
that encounter the complementary strand and match together to create a double
strand DNA. The fluorescent DNA will then hybridise on the spotted ones.
Slide scanning:
A laser excites each spot and the fluorescent emission gather through a photo-multiplicator
(PMT) coupled to a confocal microscope. We obtained two images where grey scales
represent fluorescent intensities read. If we replace grey scales by green scales
for the first image and red scales for the second one, we obtained by superimposing
the two images one image composed of spots going from green ones (where only
DNA from the first condition is fixed) to red (where only DNA from the second
condition is fixed) passing through the yellow colour (where DNA from the two
conditions are fixed on equal amount).
Data analysis:
We have now two microarray images from which we have to calculate the number
of DNA molecules in each experimental condition. To dos o, we measure the signal
amount in the green dye emission wavelength and the signal amount in the red
dye emission wavelength. Then we normalise these amount according to various
parameters (yeast amount in each culture condition, emission power of each dye,
…). We suppose that the amount of fluorescent DNA fixed is proportional
to the mRNA amount present in each cell at the beginning and we calculate the
red/green fluorescence ratio. If this ratio is greater than 1 (red on the image),
the gene expression is greater in the second experimental condition, if this
ration is smaller than 1 (green on the image), the gene expression is greater
in the first condition. We can visualize these differences in expression using
software as the one developed in the laboratory call ArrayPlot (cf below image).
This software allows from the intensities list of spot to display the red intensities
of each spot as a function of the green intensities.
Expression profile clustering:
Then we can try to gather genes that share the same expression profile on several
experiments. This clustering can be done gradually as for phylogenetic analysis,
which consist in calculating similarity criteria between expression profiles
and gather the most similar ones. We can also use more complex techniques as
principal component analysis or neuronal networks.
At the end hierarchical clustering is usually displayed as
a matrix where each column represent one experiment and each row a gene. Ratios
are displayed thanks to a colour scale going from green (repressed genes) to
red (induced genes).
Stéphane LE CROM and Philippe MARC
|
