As bacterial genomes are completed, new approaches to identify gene expression have been devised. Global approaches reveal patterns of coordinate expression of numerous genes. The major global approaches are DNA microarray analysis of transcription of the genome, and two-dimensional gel electrophoresis (2-D gels) of proteins in the "proteome," the total protein content expressed by a genome. The excitement of genomic and proteomic methods is that they may reveal particular genes and proteins whose expression might not have been tested under prevailing assumptions.

The 2-D gel analysis of proteins was developed by O’Farrell and pursued extensively by Neidhardt, VanBogelen and colleagues to explore stress response in Escherichia coli. Recent advances in gel technology and protein identification, coupled with the availability of genomic sequence, have increased the attractiveness of 2-D gels and made the technique accessible even to small laboratories such as our own. Here we present our current method that works best in our lab.  For a published example, see Kirkpatrick et al (2001) or Stancik et al (J. Bacteriol, in press).

Escherichia coli. E. coli W3110 is cultured in modified Luria Broth with 100 mM KCl replacing NaCl (LBK), to avoid the toxicity of sodium ion in cultures grown at high pH. For 1 L add 10 g of Tryptone, 5 g of Yeast Extract, 7.45 g of KCl, and one or more pH-appropriate sulfonate buffers totaling 100 mM (Table 1), pH adjusted with KOH. Sulfonate buffers are preferred to amine-based buffers such as Tris, whose deprotonated form can cross the cell membrane and penetrate the cell. Bring to volume in double distilled water and filter sterilize.

We filter-sterilize all media, and pipet into sterile glassware just before inoculation.  This avoids possible break-down of the buffers during autoclave.  For aerobic growth, we pipet 20 ml medium into a pre-sterilized 250-ml baffled flask, using rotary shaking.  For anaerobic growth, 20 ml medium per screw-cap tube, leaving small head space; rotate slowly in a vertical rotator.

Minimal Medium (M63): For 1 L add 3 g KH₂PO₄, 7 g K₂HPO₄, 2 g (NH₄)₂SO₄, 0.5 ml 1mg/ml FeSO₄, 2 ml 0.5M MgSO₄, 20 ml 1.5M Glycerol, 20 ml 5mg/ml Thiamine, and pH-appropriate sulfonate buffer at 100 mM total. Bring to volume in double distilled water and filter sterilize. Overnight cultures of 2 ml per test tube are diluted 200-fold into 15 ml per 125 ml flask; to obtain early log-phase growth, 1000-fold dilution is preferred.

Helicobacter pylori. For results, see Helicobacter pH-dependent proteins.  H. pylori 26695 is grown in buffered yeast tryptone medium (HPYT, Ref. 10) or on Brucella agar plates under oxygen regulation (T. Seyler and J. L. Slonczewski, unpublished). The advantage of growth in liquid HPYT medium is that the cell density can be controlled. The advantages of plate growth is that cultures appear healthier, showing cleaner protein content; and that oxygen concentration can be better controlled.

Table 1: Sulfonate Buffers
 

Buffer		pKa	Approximate
Abbreviation	Full name	at 37°C	pH range
HOMOPIPES	Homopiperazine-N,N?-bis-2(ethanesulfonic acid)	4.55	4.0-5.0
MES	2-(N-Morpholino)ethanesulfonic acid	5.96	5.5-6.5
PIPES	Piperazine-N,N?-bis(2-ethanesulfonic acid)	6.66	6.0-7.0
MOPS	3-(N-Morpholino)propanesulfonic acid	7.01	6.5-7.7
TES	N-[Tris(hydroxymethyl)methyl]-2-aminoethanesulfonic acid	7.16	6.8-8.2
TAPS	N-[Tris(hydroxymethyl)methyl]-3-aminopropanesulfonic acid	8.11	7.5-8.5
CAPSO	3-(Cyclohexylamino)-2-hydroxy-1-propanesulfonic acid	9.43	9.0-10.0
CAPS	3-(Cyclohexylamino)-1-propanesulfonic acid	10.08	9.5-10.5

All buffer stock solutions are 0.5M and sterilized by filtration.

Growth of Culture

1. Overnight medium is inoculated by using a toothpick to transfer single colonies.
2. Overnights are rotated overnight, incubating at 37ºC, using a Lab-Line Cel-Gro tissue culture rotator placed inside of a Lab-Line Imperial III incubator.
3.  A 1:200 dilution is made into growth media for all LBK and M63 overnight cultures. A separate overnight is used for the replicates of each condition. Four different conditions are run, each with four independent cultures.
4. Each flask is rotated gently, between 125-175 RPM, at 37ºC, using a Lab-Line heated orbital shaking bath with a platform containing 16-125 ml flask clamps.
5. LBK cultures are grown to an OD₆₀₀ of 0.400 to 0.500, M63 cultures are grown to an OD₆₀₀ of 0.200 to 0.300. OD₆₀₀ is measured using a Molecular Devices Spectra Max Plus spectrophotometer with a 96 well plate reader.
6. The final OD₆₀₀ is recorded and each culture is carefully transferred into a 15 ml sterile conical centrifuge tube. The culture volume is recorded.
7. The cultures are placed on ice for at least 10 min to stop cell growth.  Each culture is pelleted by spinning 10 min at 7000 RPM in an IEC table top centrifuge.
8. The supernatant is discarded and the pellet is washed by resuspending in either 1 ml of plain LBK, or 1 ml of plain M63.  The resuspended pellet is transferred into a lock-lid 1.5 ml microcentrifuge tube and pelleted by centrifugation for 10 min at 10,000 RPM in a Savant high speed centrifuge.  Samples are stored at -80ºC, or immediately used in the protein preparation.

The following procedure is based on Genomic Solutions, bacterial protein sample preparation, developed by Ruth VanBogelen. Make up the following three Sample Buffers and freeze away in aliquots for use. Note: urea needs to be exceptionally fresh during buffer preparation, in order to avoid deleterious reactions with protein.

Sample Buffer 1. 0.3% (w/v) Sodium Dodecyl Sulfate (SDS), 200 mM Dithiothreitol (DTT), 28 mM Tris-HCl, 22 mM Tris base, and 18 megohm-cm water. Aliquot the mixture to 1 ml volumes and store at -80ºC.

Sample Buffer 2. 24 mM of a 1.5 M Tris base stock, 476 mM of a 1.5 M Tris-HCl stock, 50mM of a 1.0 M MgCl₂, 1 mg/ml DNAse I, 0.25 mg/ml RNAse A, and 18 megohm-cm water. Mix the water and first three reagents and chill on ice before adding the DNAse I and RNAse A. Aliquot the mixture to 100 ml volumes and store at -80ºC.

Rehydration Solution.  (For IPGphor Immobiline gels; replaces Genomic Solutions Sample Buffer 3)   8.0 M Urea, 2.0% (w/v) CHAPS, trace amounts of Bromophenol blue, and 18 megohm-cm water.  (Note: Urea must be ultrapure electrophoresis grade.)  Aliquot the mixture to 2.5 ml and store at -20ºC.   Just before use for a gel run, add 7 mg of DTT and 12.5ml IPG Buffer (same pH range of the IPG strip) per 2.5 ml aliquot of rehydration stock solution.

To prepare proteins:

1. The pellets are removed from the -80ºC freezer and placed on ice.
2. Sample buffer 1 is added to each of the samples, then mixed by pipetting up and down until the pellet has completely dissolved. Add a volume (in  ml ) of sample buffer 1 equal to the product of 6.15 X the culture volume (ml) times the final OD₆₀₀ reading after growth of the culture.
3. Pellets are boiled 3.5 min in a water bath. Locking lid microcentrifuge tubes keep the lids from opening under pressure.
4. Pellets are placed on ice for 10 min.
5. Sample buffer 2 is added (1/10 the added volume of sample buffer 1), then mixed by pipetting up and down.
6. Pellets are placed on ice for 10 min.
7. Rehydration solution is placed in a water bath, no higher than 37ºC, to allow it go into solution.
8. Rehydration solution is added (4x the added volume of sample buffer 1), then mixed by pipetting. In place of sample buffer 3, rehydration solution can be used. This is particularly useful when protein blotting. IPG Buffer is not added during this step, because it limits different IPG strip ranges that can be used during 1-D gel isoelectric focusing.
9. Samples are stored at -80ºC, or immediately used for 1-D gel isoelectric focusing.  Extracted samples can be reused several times, if refrozen immediately after aliquots are removed for gels.

Reagents. Rehydration Solution: 8.0 M Urea, 2.0% (w/v) CHAPS, trace amounts of Bromophenol blue, and 18 megohm-cm water. (Note: Urea must be ultrapure electrophoresis grade.)  Aliquot to 2.5 ml and store at -20ºC. Just prior to use add 7 mg of DTT and 12.5ml IPG Buffer (same pH range of the IPG strip) per 2.5 ml aliquot of rehydration stock solution.

For the IPGphor immobiline strip, there are a range of choices. We find that the range of pH 4-7 covers a majority of the proteins of interest in E. coli, and is the best choice for beginning investigations. For H. pylori, however, a greater proportion of alkaline proteins are present; we therefore use the nonlinear pH 3-10 range. The nonlinear gradient spreads out proteins in the range of pH 4-7, where the majority of proteins still appear, but also enables visualization of proteins up to pI = 9 or 10 (Fig. 2).

For best results, pour 2-D gel sandwiches the same day as starting the 1-D; let sit overnight, covered with 0.5X Running Buffer.
Start the gel coolers to chill lower buffer solution to 14ºC.

Procedure.

1. Strip holders are prepared by washing each with detergent, to remove residual protein, and rinsed thoroughly with 18 megohm-cm water. The strip holders are air dried, or wiped off using a cotton swab or lint free tissue.
2. For 10 protein samples, two aliquots of rehydration solution are removed and thawed from the freezer. After the DTT and IPG Buffer is added and dissolved, 325 ml of rehydration solution is added to 10 microcentrifuge tubes. To ensure complete sample uptake, no excess rehydration solution is added.
3. 25 ml of each protein sample is added to the rehydration solution, for a total of 350 ml. These numbers may need to be adjusted depending on the staining technique and sample preparation. If blotting, use 150ml of protein sample and 200 ml rehydration solution.
4. Sample/rehydration solution is slowly applied to the center of each strip holder. The solution is spread throughout the length of the holder by tipping it back and forth.
5. The protective plastic cover is removed from the IPG strip and, with the gel side down, the anodic (pointed) end of the gel strip is lowered onto the anodic (pointed) end of the strip holder. Anodic end first, the gel is lowered into the solution, to coat the entire gel, then lifted and lowered to ensure complete wetting. The cathodic end is then lowered and the gel strip pushed toward the anodic end of the strip holder to ensure both electrodes are covered with the gel.
6. To each gel strip holder, 0.75-1.0 ml of PlusOne IPG DryStrip cover fluid (Amersham Pharmacia Biotech) is applied, dropwise, until the entire IPG strip has been covered. Strip holder covers are then placed on each strip holder, with the protrusions facing the gel, to provide complete contact between the gel and electrode.
7. The gels are loaded and run on the IPGphor Isoelectric Focusing System. The program is set with a 14 hr rehydration, 1 hr at 500 V, 1 hr at 2000 V, and 5 hr at 8000 V. Optimal length of time for each step will depend on the nature of each sample and method of application.

Buffers and Reagents.

1.5 M Tris Blend: 190.8 g TRIZMA Pre-set Crystals (Tris[hydroxymethyl]amino-methane and Tris hydrochloride) pH 8.8 (Sigma), in 1 L 18 megohm-cm water.  Filter sterilize, then store at room temperature.

SDS equilibration buffer: 50 Tris (from Tris blend stock, above), 6 M Urea, 30.0% (v/v) Glycerol, 2.0% (w/v) SDS, trace amounts of Bromophenol blue, and 18 megohm-cm water. Aliquot to 25.0 ml and store at -20ºC. Just prior to use add 0.25 g DTT.

10x Tris/Gly/SDS Running Buffer: 10 g SDS, 30.28 g Tris base, 144.13 g Glycine, and 18 megohm-cm water to 1 L. Genomic Solutions sells buffer pre-made.

Slab Solution (11.5% acrylamide): 465 ml Duracryl (Genomic Solutions), 300 ml 1.5 M Tris blend, 436 ml 18 megohm-cm water, 12.3 ml 10% SDS, 0.618 ml TEMED (Genomic Solutions), 3.04 ml of fresh 10%Ammonium Persulfate (APS) (Genomic Solutions). Note: 10% Ammonium persulfate is, for example, 0.5 g in a total volume of 5 ml.  It looks like too much, but it goes in fast.

Top Chamber Running Buffer: 2 L 18 megohm-cm water and 250 ml 10x Tris/Gly/SDS running buffer

Bottom Chamber Running Buffer: 10 L 18 megohm-cm water and 1.1 L 10x Tris/Gly/SDS running buffer

Procedure. (Pour the night before running)

1. 2-D plates are cleaned and assembled, separated by blue spacer sheets. To begin assembly, a folded blue spacer sheet is placed on the bottom of the casting chamber. The beveled edge side of two plates are cleaned using 70% ethanol, then sandwiched together with spacers (and mushroom caps) containing a small line of glue stick. The 2-D casting chamber is then covered and leveled. Once the filling chamber has been connected to the casting chamber, with the clamp shut, the slab solution is poured into the filling chamber. When most of the bubbles have risen, the clamp is opened and the plates are filled to about 0.5 cm from the top. Shutting the clamp stops the slab solution from filling above the desired height. If too much is poured, the filling chamber can be lowered below the casting chamber to let some of the slab solution out. 0.65 ml 18 megohm-cm water is applied from the middle outward toward each end of the gel. To do this, the water is applied slowly with the pipette tip pointed at a slant directly inside the spacer.  It is important to load slowly, along the length of the glass edge, rather than in one spot.  This avoids a "bowed up" gel top.
   
2. After polymerization (at least 30 min), tilt the gel box slightly to let all the water run out.  Check that the surface of each gel is completely smooth; no bowing, no tiny tooth-like bulges.  The slightest imperfection in the top edge will mess up the gel patterns.   If they look bad, take them apart and pour them over again!  If they look good, overlay them with 0.5x Tris/Gly/SDS running buffer. Plastic wrap is place on top of the chamber and left overnight. The gel running chamber is set to 14° C and allowed to equilibrate on the coolers.
   
3. The next day,  the sandwiches are taken out of the casting chamber, with a metal spatula, and washed with warm water to remove excess acrylamide. The slot is rinsed with 1x Tris/Gly/SDS running buffer and set the gel on a rack to dry. Once dry, a gasket is placed around the gel sandwich and fitted it in the running chamber. The gaskets are fitted beginning at the ends and working toward the middle. Obtaining a watertight fit is easier if the gaskets and running chamber slots are dry. When the gel sandwiches are loaded, the bottom chamber running buffer is drained just above the electrodes and the top running buffer is added to just above the top of the plates.
   
4. Each gel strip is equilibrated for 10 min. Two aliquots of equilibration buffer are thawed, and 0.25 g DTT is added to each. 10 ml equilibration buffer is pipetted into 5 extrusion trays. Two 1-D gels are placed into each tray and, moved periodically to completely wet the each gel in the buffer. After 10 min, tweezers are used to place the 1-D gel into the beveled edges of the 2-D sandwiches. With the plastic back of the gel toward the back of the chamber, the gelstrip is firmly positioned into place with a gel installation tool. There should be no bubbles between the gel and gel strip.  Make sure the middle of the strip is pushed down firmly, and that both ends of the strip fully contact the surface of the slab gel.
   
5. The remaining top running buffer is carefully poured into the corner of the top chamber.
   
6. The 2-D gels are run at 500 V for 4-4.5 hr, or until the dye has reached roughly 0.5 cm to the bottom of the gel.  To program the power supply:
• Select Function Setup.
• Process 3 (for 2-D slabs)
• Slab output: 3 [for both units].  If only one gel box (5 gels) is run: 1 [for left unit] or 2 [for right unit].
• Number of gels: 10 [or whatever number are run]
• Max voltage: 500V
• Duration (h 0-99): 10  (But never run them that long.)
• Duration: (m 0-59): 0
• Max power: 14,000 for 9-10 gels (or 16,000 for 1-8 gels)
• Press Start
• Press 3 [for Process 3; otherwise, 1 or 2, for Process 1 or 2]
  
7. The 2-D sandwich is lifted from the running chamber, and the top plate is removed with a metal spatula. With the spatula, crop each gel (slash corner or square corner) in order to mark #1-5 according to their position in the 2-D chamber.
   
   
   
8. For Silver Stain the next day, an equal amount of Fix (recipe under Silver Stain ) is poured into two 7.8 L storage containers, followed by the 2-D gels. Each container is labeled 1 or 2 to indicate what chamber the gels are from, then rotated gently, for 30 min, and placed in a cold room, overnight.
   
9. To clean the glass plates: Place overnight in a tank of Micro detergent solution, in racks made out of test-tube holders to keep the glass plates apart.  The next day, rinse plates for two hours in a bucket with a line from the sink directing water from the bottom upward.  Place in test-tube racks to dry.

The following procedure is modified from Pharmacia Biotech’s Silver Staining Kit, and from Yan et al (2000).  Make up all solutions fresh.

Reagents. Fix: 800 ml 95% ETOH, 200 ml Glacial Acetic Acid, and 1 L 18 megohm-cm water to a 2 L flask.

Sensitizing Solution: 600 ml 95% ETOH, 80 ml 5%(w/v) Sodium Thiosulfate, 136 g Sodium Acetate, and 18 megohm-cm water to 2 L.

Silver Stain: 0.75 g Silver Nitrate to 300 ml 18 megohm-cm water. Repeat this for as many gels to be stained.

Developing Solution: Add 7.5 g Sodium Carbonate to 300 ml 18 megohm-cm water. Repeat this for as many gels to be stained. Add 120 ml 37% (v/v) Formaldehyde just prior to use

Stop Solution: 29.2 g Disodium Ethylenediamine Tetraacetate (EDTA) to 2 L 18 megohm-cm water.

Preserve Solution: 450 ml 95% ETOH, 69 ml 87% (v/v) Glycerol, and 981 ml 18 megohm-cm water.

Procedure.

1. The morning after running gels, transfer to fresh Fix for 15 min.  Gels are carefully transferred into two separate storage containers, each with half of the Sensitizing Solution, then rotated vigorously for 30 min.
2. The gels are washed four times by transferring them into new containers, filled with enough 18 megohm-cm water to cover the gels, and rotated vigorously. The first three washes are done for 5 min and the last one for 10 min. This step is important in removing excess Sensitizing Solution from the gel, which can result in background noise on the gel.
3. Using only one set of gels, each is individually transferred to developing trays containing Silver Stain, then rotated gently for 20 min.
4. The gels are transferred into a wash container for two washes: first 2 min, then 1 min.   Follow with a quick rinse to ensure excess Silver Stain has been removed.
5. The gels transferred to developing trays containing Developing Solution, then rotated back and forth until each gel has developed.
6. Once the gel has developed it is transferred to a container containing half of the Stop solution. Let the gels shake gently for 10 min.
7. The second set of gels is started on step 3.
8. The gels are washed three times for 5 min each.
9. The gels are transferred into a container with half of the Preserve Solution and rotated gently for 1 hr.
10. Each gel is bagged into 1 gal plastic bags labeled with the strain information, protein prep date, and the run date.
11. The gels are scanned and saved. Bubbles are rolled out of the gel using a test tube. To provide contrast, for better image quality, a white background is placed behind the gel during the scan.

The protein spot must be cut from the gel using completely protein-free implements and gloves, repeatedly changed.  The cut protein is placed into a microfuge tube and 2 ml filtered water is added to maintain moisture.  The proteins can be stored frozen at -80 deg, or they can be shipped overnight to a sequencing lab at room temperature.  An experienced lab is necessary to obtain good results; we recommend the U. Mass Protein Microsequencing Lab.

To compare different growth conditions, the size of a given protein spot can be compared between patterns of protein spots from gels performed on samples grown under different conditions. In the older literature, a protein was considered to be "induced" if it appeared in two out of three gels from independently grown cultures of an experimental condition, compared with gels from a control growth condition. More recently, computer image analysis has been used to subtract background from the protein pattern, and normalize the spot pixel densities against the total protein density. This approach suffers from high errors associated with the background count, and from the skewing of total protein count by a few proteins whose high concentration overloads the relatively narrow range of image intensity available for current scanners; see for example the protein pattern of H. pylori, dominated by the greatly overloaded spots of UreB and GroEL.  A complication often observed is the appearance of a given protein in several different spots, most commonly in a "train" of spots over a range of pI.   The reasons for multiple spots remain unclear, although they probably arise from a combination of posttranslational modifications in vivo as well as chemical modifications that occur during sample preparation.

Currently, computer software such as Compugen Z-3 replaces background subtraction with a spot quantitation algorithm that is relatively insensitive to background and overexposure, as well as a normalization algorithm that compares the histogram of all proteins visualized between two gel patterns. By quantifying protein spots using pixel density, it is now possible to detect smaller differences that may have gone unnoticed otherwise. Furthermore, a large number of pairwise comparisons can be performed, enabling a level of statistical analysis not possible previously.

The gel images are loaded into Z3 to be analyzed and made into layered views. We use three gels from independent cultures at each of two growth conditions, an experimental and a control; for example, cultures grown in the absence or presence of 50 mM D-lactic acid (Fig. 3). It is best to select gel images that exhibit comparable amounts of overall protein and intensity of stain, although Z-3 does compensate for modest differences and effectively normalizes the comparison between gels. Relative spot densities are computed by comparing the non-saturated pixels of the spot on each of two gels and fitting by linear regression. Normalization of overall protein content is performed by comparing the overall histograms of spot ratios across the entire gel.

The Z3 program is a bit finicky, but we find reliable results with this order of analysis:
1.  Open all six images.
2.  Layer the first pair, reference to comparative.  Check entire layered view for spots that need fixing by hand.
3.  Match all spots in the first pair.  Cut out large trains of "spots" that are actually smears.
4.  Now layer the reference gel to a second comparative gel
5.  Match all spots in this second pair.
6.  Now layer the reference gel to the third comparative gel, etc.
7.  After all nine pairs are done, save and backup a copy.
8.  Export to Excel.

Table 2 shows examples of typical differential expression ratios (DE values) for proteins from pairwise comparisons of gel images from two sets of three replicate cultures. Control gels are designated as gel 1 and experimental gels are designated as gel 2. We consider a protein to exhibit significant differential expression if its DE values are greater than 1.5 (50% induced) or less than 0.67 (30% repressed) for at least 7 of 9 control-to-experimental layered views. In Table 2, the protein in the first row is "unmatched," appearing only in the experimental condition; the second row shows a protein induced nearly four-fold; and the third row shows a protein repressed three-fold. The lower two rows show typical DE values of proteins with no consistent pattern of induction or repression.

Because expression values represent ratios between conditions, we perform logarithmic conversion and represent their distribution as the mean log₁₀ (DE), or  LDE, with standard deviation (see explanation below). Proteins induced in the comparative gel show a positive LDE; proteins repressed show a negative LDE.

The pixel densities are calibrated by setting the overall histogram of the proteins in the total image such that a maximum of DE values are close to 1 (equality), on the assumption that 90% of proteins have the same concentration in both cell samples.  This method of calibration is a big improvement over "summing the whole gel," which is what most methods use.  The DE is actually calculated by matching pixel by pixel between the spots on the two gels, plotting Gel 2 over Gel 1, then taking the  slope of a fitted line. The benefits of using this method are that (1) it is more robust than summing each spot and taking a simple ratio, and (2) no background subtraction is needed.

Reporting protein expression ratios.  Because DE values are ratios, the appropriate average measure is the geometric mean; that is, exponentiation of the mean logaritm values.  However, we believe the induction and repression of proteins is best indicated by reportingpositive and negative mean log₁₀ (DE) which we call LDE.  Thus, a positive LDE = 1.00 represents a ten-fold increase of a spot in the comparative gel over the reference gel: for example, LDE=2.00, hundred-fold increase; -1.00, ten-fold decrease.  [Note: The natural logarithm works as well, but log₁₀ has the advantage of producing values in which each unit of 1 means a ten-fold increase.]

To determine the mean log₁₀ DE:
Assume one has two groups of three gels (reference group 1 -- Gels A, B, C; and comparative group 2 -- Gels A, B, C).  One would like to quantitate a protein's expression ratio between Gel 1(A, B, or C) and Gel 2(A, B, or C), based on a set of pairwise DE ratios (values obtained from pairwise comparisons of group 2 over group 1).  There would be a total of nine DE ratios.

Since ratios are not additive, one should use the geometric mean and not the arithmetic mean. Note, however, that the geometric mean is related to the arithmetic mean of the log-ratios. Specifically, let DE₁, DE₂, ... DE₉ be the 9 ratios, then the geometric mean of them can be expressed as:
[geometric mean of DE]  =  n-th root of the product  =  10 ^ [mean log₁₀(DE )] = 10 ^ LDE
where "10^X" is ten raised to the power X, and  "log₁₀" is the logarithm to the base ten.
LDE = Mean log₁₀ DE =  [ log₁₀(DE₁) + log₁₀ (DE₂) + .... + log₁₀ (DE₉) ]  /  9

By taking the log₁₀ of the ratios we can average them arithmetically.  The inverse function (10^) of the LDE would generate the geometric mean.

Estimating Error (Standard Deviation).   The standard deviation (SD) of log₁₀(DE) can be taken simply based on the mean log₁₀ (DE) and the individual log₁₀(DE) values.  We represent the distribution of the mean log₁₀ (DE) as LDE +/- SD.  (See Table 2 for typical results.)

The values can however be converted back to geometric mean by exponentiation:
Geometric mean = 10 ^ [mean log₁₀(DE)]
Range of DE is between:
10^[LDE + SD] = 10^[mean LDE] * 10^[SD]
10^[LDE - SD] = 10^[mean LDE] * 10^[-SD]

Dealing with the limits to sensitivity of silver stain.  Silver stain is great for showing tight dark spots; but this very advantage is linked to its disadvantage, a narrow range of sensitivity.  In practice, we find that typical LDE values rarely go above 1.00 (ten-fold difference).  Therefore, we take ten-fold as the maximum observable log expression ratio.  If a protein appears unmatched for one or two of the nine pairs, but gives an expression ratio for most of the pairs, we designate "unmatched spots" as 10-fold increase or decrease, depending on the direction of difference, then average it in.  If a protein is unmatched for all or most gel pairs, it is designated + or -  respectively.  We welcome any comments on this analysis procedure, or suggestions for alternatives.

Layered Composite Gels
 

Visual Analysis Generate a composite gel image for each experimental condition, then to overlay the two composite images. Proteins induced in gel 2 appear pink, whereas proteins repressed appear green. Quantitative Analysis The individual gel pairs still need to be checked, because a protein appearing in only one of three replicate cultures can look induced or repressed in the layered composite.

1. Stancik, L. M.,* D. M. Stancik,* B. Schmidt, D. M. Barnhart,* Y. N. Yoncheva,* and J. L. Slonczewski.  2002.  pH-Dependent Expression of Periplasmic Proteins and Amino Acid Catabolism in Escherichia coli.  J. Bacteriol. 184: 4246-4258.
2. Kirkpatrick, C., L. M. Maurer,* N. E. Oyelakin,* Y. Yontcheva,* R. Maurer, and J. L. Slonczewski. 2001.  Acetate and formate stress: opposite responses in the proteome of Escherichia coli.  J. Bacteriol. 183:6466-6477.
3. J. L. Slonczewski and C. Kirkpatrick.  2002.  Proteomic analysis of pH-Dependent stress responses in Escherichia coli and Helicobacter pylori, using two-dimensional gel electrophoresis.  Methods in Enzymology, Bacterial Pathogenesis Part C, Academic Press.
4. VanBogelen, R. A., K. Z. Abshire, A. Pertsemlidis, R. L. Clark, and F. C. Neidhardt. 1996. Gene-protein database of Escherichia coli K-12, Edition 6, Ch. 115, pp. 2067-2117. In F. C. Neidhardt, R. I. Curtiss, C. C. Gross, J. L. Ingraham, and M. Riley(ed.), Escherichia coli and Salmonella typhimurium Cellular and Molecular Biology, 2nd ed. ASM Press, Washington, DC.
5. VanBogelen, R. A., K. Z. Abshire, A. Pertsemlidis, R. L. Clark, and F. C. Neidhardt. 1996. Gene-protein database of Escherichia coli K-12, Edition 6, Ch. 115, pp. 2067-2117. In F. C. Neidhardt, R. I. Curtiss, C. C. Gross, J. L. Ingraham, and M. Riley(ed.), Escherichia coli and Salmonella typhimurium Cellular and Molecular Biology, 2nd ed. ASM Press, Washington, DC.
6. M. R. Wilkins, K. L. Williams, R. D. Appel, and D. F. Hochstrasser, eds. (1997) Proteome Research: New Frontiers in Functional Genomics. Berlin: Springer.
7. Lambert, L., K. Abshire, D. Blankenhorn, and J. L. Slonczewski. 1997. Proteins induced in Escherichia coli by benzoic acid. J. Bacteriol. 179:7595-7599.
8. D. Blankenhorn, Phillips, J., Slonczewski, J. L. 1999. Acid- and base-induced proteins during aerobic and anaerobic growth of Escherichia coli revealed by 2-d gel electrophoresis. J. Bacteriol. 181:2209-2216.
9. Slonczewski, J. L., T. N. Gonzalez, F. M. Bartholomew, and N. J. Holt. 1987. Mu d-directed lacZ fusions regulated by low pH in Escherichia coli. J. Bacteriol. 169:3001-3006.
10. J. L. Slonczewski, D. J. McGee, J. Phillips, C. Kirkpatrick, and H. L. T. Mobley. (2000) pH dependent protein profiles of Helicobacter pylori analyzed by two-dimensional gels. Helicobacter 5:240-247.

Yan (2000)

This material is based upon work supported by the National Science Foundation under Grant No. 9982437.