1. Gel-based Quantification

2. Sources of variation between gels
Internal variation between gels
Same loading amount?
Same gel condition?
Same staining condition?
External variation after gel developed
Unwanted spots (dye or reagent deposit)
Dirty spots (hair, dust)
Its just like this slide, too much information to deal with
1/3 of all genes have a known function?

3. Image analysis workflow
Manipulation of image/ normalization
-Separation of overlapping spots, removing lines and speckles
Spot detection/ quantification
-Background subtraction, spot segmentation, land-marking ,spot matching
Gel comparison
-Matching of gels (e.g. normal, diseased, treated),alignment
Data analysis
-Defining changes in expression
Data representation
-Annotation of spots, linking of data: spots -intensity - MS data

4. Background subtraction and quantification
Background subtraction method:
No background
Mode of non-spot
Manual background
Lowest boundary
Average boundary

5. Normalization General normalization method: Total spot volume
Single spot
Total volume ratio

6. Expression analysis
Comparison of individual experimental gels to master gels.
Identification of variant spots

7. Expression Analysis Spot detection Spot matching
Normalization of spot intensities
PTM?
Downregulation?

8. Expression comparison
Changes in expression are thought to be significant if they are greater than experimental variation
Its just like this slide, too much information to deal with
1/3 of all genes have a known function?

9. DIGE
DIGE

10. Traditional 2-DE analysis: Sources of variation
Gel to gel variation
Presence - Absences
Volume variation
Resolution Issues
Software
Variation added during spot detection, background subtraction, matching etc
Biological variation
Animal to animal
Culture to culture

11. Replicate gels - problems
At least 3 times as many gels as you should need.
Makes analysis more complex.
Averages out gel to gel variation, instead of removing it.
There is no ‘Gold Standard’ of 2D to see how effective replicate gels are, what part of the system contributes more variation, how many replicates are required to overcome variation

12. Principles of 2-D DIGE methodology
Conventional 2-D
Are spot differences due to induced biological changes or differences in the way the gels have been cast/run/stained?
gel 1
gel 2
Silver stain
control
treated
differences
2-D DIGE
control
treated
differences
Dual scan
Dye 1
Dye 2
Separate two samples on 1 gel
In conventional 1 color 2-D one sample is separated per 2-D gel.
Gels are visualized using a stain e.g. silver.
The spot patterns on each gel are compared to find differences in protein abundance between samples.
However, differences between gels can also be caused by differences in gel casting, run and stained which makes it difficult to tell whether the difference s are really due to changes induced by the treatment e.g. drug/disease state/life cycle.
Using 2-D DIGE, protein samples are covalently labelled using fluorescent dyes. Different samples can be separated on the same gel so any proteins showing change are due to differences between the samples NOT differences due to physical experimental differences between gels.
Achieving accurate quantitative data

13. Overlay of two images
This shows two images obtained from a single gel when two different samples have been separated. The purple spots represent proteins that are present in both samples. Those in blue are only present in the control and those in red in the treated sample. A predominance of one or the other represent a change in expression. A small area has been enlarged to show this more clearly.

14. Fluorescence Emission Filters
488 nm
532 nm
633 nm
Cy2
Cy3
Cy5
520 BP 40
580 BP 30
670 BP 30
Bandpass to distinguish multiple fluors
An example of resolving 2 different fluorochromes, both excited with the green laser. Two band pass filters are used that collect the bulk of emission from the 2 fluors.

15. Volumes expressed as ratios relative to pooled internal standard
Workflow Outline for ‘pooled’ approach
Pooled internal standard
Label with Cy2
Protein extract 2
Label with Cy5
Protein extract 1
Label with Cy3
Mix labelled samples
Separate by 2-D PAGE
Cy2 excitation
wavelength
Cy5 excitation
Cy3 excitation
Volumes expressed as ratios relative to pooled internal standard
Create the standard by pooling aliquots of all biological samples in the experiment
Label standard with one CyDye DIGE fluor
Run standard with individual samples:
Cy2 Cy3 Cy5
Gel 1 Std sample 1 sample 2
Gel 2 Std sample 3 sample N
Samples are in-gel linked to a common standard,
Quantification should be accurate
Separates gel to gel from biological variation
Goes a long way to removing gel to gel variation altogether

16. Linking samples Conventional 1-sample-per-gel 2-D
No intrinsic link between samples
Internal Std
sample 1
sample 2
sample 3
sample 4
sample 5
sample 6
GEL A
GEL B
GEL C
2-D DIGE
Internal standard links samples between gels
Multiplexing links samples within each gel
sample 1
sample 2
sample 3
sample 4
sample 5
sample 6
This slide shows an example of the conventional way a differential 2-D experiment is set up
1-sample-per-gel.
Gel-to-gel differences can arise from many sources including:
Differences in the amount of protein entering into the strip or transferring from strip to 2-D gel.
Differences in IEF running conditions e.g. differences in current between strips run on different IPGphors, or differences in temperature/current between gels run in different positions in the DALT tank.
When protein spots are compared between gels in 1-sample-per-gel experiments, it is impossible to tell whether a spot has changed intensity as a result of an induced biological change or because of this gel-to-gel variation.
Achieving accurate quantitative data

17. Differential Analysis using the pool
GEL A
Sample 3
Sample 4
Internal Std
GEL B
Sample 5
Sample 6
Internal Std
GEL C
DIA
In-gel
co-detection
DIA
DIA
Sample 2
Sample 1
Master
BVA
Cross-gel matching
Internal Std
This slide summarizes the whole analytical process in DeCyder.
IN DIA:
Each image is co-detected with its internal standard, producing two image pairs per gel.
The ratio of standard:sample is calculated for each protein in each image.
IN BVA:
One of the internal standard images is selected as the master image and all internal standard images matched to this.
Sample:standard spot ratios for each protein in each sample are then compared, giving T-test values, fold changes and ANOVA values for each protein.
Proteins of interest are filtered out to generate a pick-list.
Protein difference ratios
Achieving accurate quantitative data

18. DeCyder Software Structure
Differential In-gel Analysis (DIA)
Co-detection of image pairs and triplets
Automated detection, background subtraction,
Automatic quantification, normalization and first level (In-gel) matching
Low user interaction, high throughput
Biological Variation Analysis (BVA)
Automated gel to gel matching
Statistical/Trend analysis
Batch Processor
Processing of 500 gel image pairs and triplets without user interaction
XML Toolbox
For the purposes of explanation I will divide the software into four parts.
Firstly the batch processor; up to 500 image pairs can be selected for analysis at one time. This reduces user hands-on time significantly.
Secondly the Differential In-Gel analysis. This part of the software automatically co-detects the image pairs, subtracts background, quantitates, normalises and matches the detected spots.
It has been designed to minimise user interaction which in turn ensures that subjective editing is removed reducing user to user variation. The only parameters required to input is approximate spot number and a filter number set to automatically remove dust particles that might otherwise thought to be proteins spots. This is very simple and the user becomes familiar very quickly. It is important to remember that ratio measurements are always from within gel and never obtained from two different gels. A standard should be used in each gel increasing the accuracy of the result. This makes the process analogous to an assay.
The third part is the biological variation analysis which matches together the spots from different gels. The matching process is made easier because the internal standard can be used to match which ensures that the same complement of proteins are represented in each gel. Again this is only possible with our novel and proprietary DIGE multiplexing technology. I will come back to that in a moment
The fourth part is the XML toolbox, designed for easy extraction of data from your analysis in two output forms, web-based or table-based.

19. DIA (Difference In-gel Analysis) Results
Number of Spots Detected:
About 1500 in each gel image

20. BVA (Biological Variation Analysis) Results

21. Using an Internal Standard Sample
Pooled standard has many benefits:
Every protein in the population should appear on each gel
Decreases gel to gel variation
Each sample is compared internally to the same standard
Enables fully automated accurate spot statistics
Makes gel to gel matching easier

22. Traditional 2DE analysis: Types of variation
Present only in (a)
Stronger in (b) than (a) or (c)
Sample 1 a
Sample 1 b
Sample 1 c
Present in (a) and (c) not (b)
Is this gel to gel variation or biological variation?
29-Apr-04

23. Internal standard - theoretical benefits
Cy5
Cy3
Gel 1
Gel 2
Cy2
Standard
Sample 3
Sample 4
Gel to gel variation – since spot absent from standard in gel 2

24. Biological variation: In both standards but absent in sample 1
Internal standard - theoretical benefits
Standard
Cy2
Cy5
Cy3
Gel 1
Gel 2
Sample 2
Sample 1
Sample 3
Sample 4
Biological variation: In both standards but absent in sample 1

25. Biological decrease, not increase, also gel to gel variation
Internal standard - theoretical benefits
Cy5
Cy3
Gel 1
Standard
Cy2
Gel 2
Sample 2
Sample 1
Sample 3
Sample 4
Biological decrease, not increase, also gel to gel variation

26. Normalisation of spots between gels
Normalised, but not to standard
Normalised to standard
This slide highlights the importance of normalisation. The major source of error occuring when obtaining ratio measurements from two gels is gel to gel variation.
In this example after normalisation it is clear that these two data groups are showing a significant difference. In this example it is an average change of 1.79-fold with greater than 99.9% t-test confidence.
Now going to the non-normalised information if you view the paired results of the standard and group 1 or standard and group 2 joined by the broken lines it is obvious that there is an upward trend for group 1 compared to the standard and a downward trend for group 2 compared to the standard. Now consider the ringed group in blue and the ringed group in red only, this is what you would have if single gels had been prepared using conventional post-labelling.It would be very difficult to make the same conclusion that DIGE allows you to make.
Key message DIGE allows accurate assessment of differences that are not possible when ratio measurements are obtained from different gels