Comparative two-dimensional uorescence gel electrophoresis

Comparative two-dimensional gel electrophoresis (CoFGE) is a special version of two-dimensional polyacrylamide GE (2D-PAGE) and related to difference GE (2D-DIGE). It provides reproducibility and standardisation for 2D-PAGE by introducing a reference to the experiment. CoFGE uses different uorescent labels to distinguish analyte and a marker protein mixture. The method allows in silico correction of the assignment of gel-separated proteins based on the co-run references, which form a grid of landmarks across the entire gel. The variability of spot coordinates is reduced to ~1% error and data can thus be compared to results generated independently with the same method. In this way, searchable repositories of gel-separated proteins become feasible. In addition, the CoFGE experimental principle can be used for protein quantication by applying the proteins of the marker grid in different concentrations. Here we present the protocol for conducting a CoFGE experiment, which takes about 2 days to complete for a technician skilled in GE.


Introduction
Both traditional two-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis (2D-SDS-PAGE) 1 and 2D difference GE (2D-DIGE) 2 are established techniques for the separation of protein lysates with the latter providing improved gel comparability in addition. DIGE was developed to identify changes in the concentration of proteins between different sample states such as normal and stimulated cells. It is based on replicates and uses related uorescent dyes to distinguish the different analytes. DIGE is very valuable in comparative proteomics, but it cannot be applied to singular samples, which only arrive sporadically and need to be compared to archived information, as is, for instance, the case in eld studies of aeroallergens 3 or in rare diseases. In fact, reproducibility has always been the major drawback of GE, because gel-to-gel variation is di cult to overcome 1,2 . In order to ll this gap we introduce a reference, which allows the comparison of gel-separated protein data generated in different laboratories or over long periods of time. To that end, we combine elements of 1D-and 2D-PAGE as well as 2D-DIGE ( Fig. 1) 4 . Thereby, the procedure to separate the analyte proteome is the same as in regular 2D-PAGE. However, in order to co-run reference substances, holes for protein marker solutions are punched across the 2D-gel close to the well for the pI-strip. When running the second dimension of such a gel, both the analyte proteins in the pI-strip, preseparated in the rst dimension by isoelectric focussing (IEF), and the standard proteins in the holes are separated at the same time, the latter de facto undergoing a 1D-PAGE experiment. For the distinction of reference and analyte proteins, they are labelled with different uorescent dyes such as those used in DIGE 2 . In this way, a quasi-internal gel standard grid is created on the gel, which allows the correlation of the analyte spot coordinates to reference landmarks 5 . We named the experiment comparative uorescence GE (CoFGE) 6 . As long as it is performed with the same type of pI-strip and identical marker proteins, in silico matching of independent datasets is possible 4,7 . An exemplary dataset for the separation of Escherichia coli lysate, our test analyte 6,8 , is shown in Fig. 1.
CoFGE was rst developed for vertical 2D-PAGE using a 1D-PAGE comb for the generation of the marker wells 5,6 and it is still performed in this way by independent users 9,10 . However, we found it much more convenient to apply CoFGE to horizontal GE 8 as illustrated in Fig. 1, which has a number of advantages in itself. The resolution is higher and protein spots are sharper, there is no need to handle large buffer volumes, and gel distortions (e.g., "smiling effect") are reduced as a result of better cooling 11,12 . For vertical GE, a stacking gel for the application of the references needs to be casted 5,6 , which is not necessary in horizontal GE, because ready-made gels are now commercially available, which include the wells for the reference grid (Mercator gels, Serva Electrophoresis). Alternatively, the slots for the marker wells can be manually punched out in regular 2D-gels using pipette tips as we did initially 8 . The CoFGE method as described above does not take into account the variability resulting from the rst dimension (IEF). We have, therefore, introduced azo-dyes as marker substances for correction 13 . However, this additional protocol step is optional for most purposes when state-of-the-art high-quality pI-strips are used.
Fluorescence intensity correlates with protein concentration so that it is possible with CoFGE to estimate the protein amount in analyte gel spots. For that purpose, the reference proteins are applied in different concentrations to the marker slots and used for calibration 14,15 .
Depending on the overall goal of a project, be it a comparison of gel images in a short-term study or an investigation carried out over several years, some thought has to be spent on experimental design 4,7 . The analysis of a comparatively small gel set can simply be performed by choosing one of its members as the master gel. All reference grids from this set will then be correlated to this master grid and, subsequently, all analyte gel images will be adjusted using their individually corrected reference grids. In case of a long-term study such as the generation of a special 2D-gel image and protein spot database, the master grid needs to be de ned at the very beginning. To that end, replicate gel runs of the reference proteins are used to calculate a fused image. This "ideal" gel image will serve as the reference grid image for warping of gels throughout the project. Especially in extensive long studies, it is important to observe the protocol and use the same pI-strips and reference proteins throughout the project; otherwise, gel matching will not be possible.
CoFGE takes advantage of the same type of uorescent labels as DIGE and is thus as sensitive (detection limit 0.5 fmol protein) and in the same way linear over a 410,000-fold concentration range 2 . We use two well-matched uorescent dyes such as Cy3 and Cy5 or related paired dye products such as Sci3 and Sci5 or GDyes. Image analysis requires a high-quality uorescence scanner, in our case the Typhoon 9400 imager (GE Healthcare), which provides three-laser illumination of which only two lasers are typically used.
The protocol described here outlines the steps we use in performing a CoFGE experiment 16 .
For method development we use E. coli lysate. CoFGE is otherwise applicable to any sample, which can be separated with traditional 2D-PAGE. CoFGE suffers from the limitations inherent to 2D-PAGE 1 such as di culties with the separation of hydrophobic proteins, but it is otherwise a major improvement towards standardisation of gel electrophoresis. It is a sensitive and robust approach to generate reproducible spot coordinates for proteins separated by 2D-PAGE with the added bonus of possible quanti cation.

Reagents
Instead of the dyes, proteins or kits mentioned below, similar products from other vendors will also work.
CRITICAL Use puri ed deionized water such as MilliQ water and analytical grade reagents throughout the protocol. Sample purity is critical, especially in labelling and IEF. CRITICAL STEP Use a higher concentration of lysine than recommended by the manufacturer to avoid mislabeling due to quenching failure 18 .
PAUSE POINT Aliquots can be stored at -80°C until further use.

6| Protein extraction
Add 100 µl Tris buffer to the lter unit and vortex (45 min, 1,000 rpm). Transfer the protein solution to a fresh tube.
CRITICAL STEP If you wish to evaluate the protein concentration in addition to analyte spot correction, you need to determine the concentration of the grid mix proteins again after above procedure, because protein loss may occur 4,14,15 .
PAUSE POINT Store the reference proteins at -32°C until further use. Alternatively, if cup-loading is used, place the IPG strip gel side up into the manifold and pipet the sample (13.2 µl, total protein amount 50 µg) followed by mineral oil (10 µl) into a cup located on the anodic side of the strip.
CRITICAL STEP In case of multiple parallel experiments, only load identical protein amounts per strip, e.g. 50 µg, for reliable results. Scanning Timing ~1 h 15| Scan gels immediately using the green laser for Sci 3 (532 nm, emission lter 580, band pass (BP) 30, photomultiplier tube (PMT) 525) and the red laser for G-Dye300 (633 nm, emission lter 670, BP 30, PMT 490). Set the resolution for the main scans to 100 µm per pixel. Adapt the PMT response in such a way that the gel image shows the most intense protein spot slightly below saturation.
16| Visualize gel images using ImageQuant software and store them.

PAUSE POINT
Image analysis Timing ~1 to several h depending on project 17| Delta 2D software has been speci cally modi ed to allow user-friendly analysis of CoFGE projects. However, any image analysis software capable of warping will be suitable.
First, chose a reference gel as discussed above ( nd examples in ref. [6][7][8]. The x-coordinates of the nodes are determined by the well distances; the y-coordinates by the position of the marker protein spot. Match the individual marker grid for each gel to the chosen master grid by assigning every experimental grid spot to the corresponding spot of the theoretical grid. This process determines the match vectors for the respective gel.
18| Apply the determined gel match vectors to the analyte proteome on this gel. Control the mapping of the match vectors manually.

Troubleshooting
Reasons for poor results are the same as in 2D-PAGE 1 and DIGE 2 . Speci cally, insu cient protein labeling may occur when the pH of the lysate is not correct (< 8) or in the presence of primary amines competing with the label (30 mM TRIS does not interfere signi cantly). The labelling modi es ~5% of all proteins so that a slight molecular weight-shift between labelled and unlabelled forms for proteins < 25 kDa is expected 2 . IEF may be disturbed by salts and other contaminants, which can be avoided by prior protein precipitation or the use of a commercial clean-up kit. The placement of the IPG-strips in the well for 2D-PAGE requires practice to avoid a horizontal shift of the analyte proteins with respect to the grid proteins.
In case the method is to be used for quantitative purposes, reproducible sample preparation should be observed to avoid variation of the results. Typical mistakes include different labelling times, di culties during sample loading for IEF and errors in the protein concentration determination.

Anticipated Results
The described protocol generates a 2D-gel containing two samples: the reference protein mixture grid seen in green in the Fig. 1 false-colour image and the analyte protein (here E. coli in red). In a CoFGE experiment, several of these gels are matched with the goal of describing the spot location of the analyte proteins as reproducibly as possible using software such as Delta 2D, which has been updated by the manufacturer to accommodate this task. The accurate protein coordinates allow archiving of 2D-gel spot information in browser-based searchable databases. In addition, with the use of different concentrations of the marker proteins in the reference grid analyte protein concentrations can be deduced. CoFGE thus provides a technical solution to both standardization and quanti cation in 2D-PAGE.

Figure 1
Schematic of a horizontal CoFGE experiment4. Both sample and reference proteins are pre-labelled using different uorescent dyes. The analyte proteome is separated by IEF as in regular 2D-PAGE. The gel for the second dimension, however, contains marker slots close to the channel for the pI-strip, which are lled with a reference protein mix (R1-R8). During GE, both the analyte and the marker proteins are separated and the latter experience the same gel distortions as the former. A uorescence scanner separates the images for analyte and reference grid, respectively. The inset shows an exemplary gel for the separation of E. coli lysate (red, marker grid in green) as is achieved with the protocol below. Subsequently, analyte spots (A1-A5) are associated with reference landmarks in silico with 2D-PAGE analysis software capable of warping (see, for example, spots A2 and A4).