The western (note that in this context “western” should be spelt with a lower-case “w”) blot is commonly used to identify, quantify, and determine the size of specific proteins. Western blotting evolved from Southern blotting, which is used to detect specific DNA sequences among DNA fragments separated by gel electrophoresis, and northern blotting, which is used to detect and quantify RNA and to determine its size, and also involves gel electrophoresis to separate RNA. In the late 1970s, Towbin et al. (1979) enabled proteins to be electrophoretically separated using polyacrylamide–urea gels and transferred onto a nitrocellulose membrane. Burnette (1981) later employed the more widely used sodium dodecyl sulfate-polyacrylamide gels (SDS-PAGE), which eventually led to this method being termed western blotting. It is also called protein blotting or immunoblotting and has rapidly become a powerful tool for studying proteins. Basically, gel electrophoresis is used to separate native or denatured proteins. The proteins are then transferred to a membrane for detection using antibodies specific to the target protein. The technique has continued to evolve, and there are many reports on troubleshooting and improving the technique (Kurien and Scofield, 2009).
Protein can be measured from whole tissue or tissue culture extracts. Cells and tissues should be rapidly frozen with liquid nitrogen to avoid protease degradation of proteins or collected and lysed as quickly as possible. It should also be noted that repeated freeze/thaw cycles can have an adverse effect on the quality of protein and should be avoided. Solid tissue is mechanically broken down, usually using a homogenizer or by sonication in a lysis buffer (see below). Tissue preparation can be performed at cold temperatures to avoid denaturation and degradation of protein.
Various detergents, salts, and buffers may be used to enable lysis of cells and to solubilize proteins. There are many different lysis buffers, which can be chosen based on the protein of interest. To avoid degradation of protein (proteases as well as phosphatases can be released during lysis), protease and phosphatase inhibitors can be included in the lysis buffer. Many of these inhibitors come in tablet form and can be simply dissolved in the lysis buffer before use. Lysis buffers used in sample preparation for western blotting should enable efficient protein extraction and maintain antisera recognition of the protein (MacPhee, 2010). Importantly, quantification and comparison with other samples in western blot analysis is dependent on the protein lysates prepared for polyacrylamide gel electrophoresis.
Determination of Protein Concentration
The protein concentrations of the samples to be loaded on a gel need to be determined. Quantification for total protein can be achieved by measuring samples at 280 nm on a spectrophotometer, but the buffer must not contain absorbing materials. When the buffer contains absorbing materials, the Bradford assay (Bradford, 1976) can be used where a standard curve is created to determine unknown sample concentrations.
For optimal separation, it is important to determine the ideal bisacrylamide:acrylamide ratio prior to electrophoresis. Proteins are then separated using gel electrophoresis. Proteins can be separated by isoelectric point, molecular weight, electric charge, or a combination of these. The most common type of electrophoresis uses polyacrylamide gels and buffers loaded with SDS. The procedure of SDS-PAGE is based on the characteristics of SDS, which is a strongly anionic detergent. Because proteins do not all have the same electrical charge, the mixture is treated with SDS, and thus the proteins become denatured and negatively charged. As a result, this allows separation of proteins by molecular weight. Treatment with a reducing agent to remove disulfide bonds and boiling of samples can facilitate denaturing. When voltage is applied to the gel, proteins migrate at different speeds, and these different rates result in separation into bands within each lane. A two-dimensional gel can also be used. This type of gel spreads out proteins from a sample in two dimensions. Proteins are separated by isoelectric point in the first dimension and by molecular weight in the second dimension.
Unlike SDS-PAGE gels, which unfold and denature the native structure of a protein, nondenaturing or native gels can be used to maintain protein complexes for detection after transfer. However, complexes may not separate cleanly or predictably, because they are unable to move through the polyacrylamide gel as rapidly as individual, denatured proteins.
Loading Gels—What Should be Included?
The samples are loaded onto the gel. One lane should include a molecular weight marker that is used to determine the molecular weight of the target protein. Another lane should include an internal control, ideally with a known concentration and molecular weight to determine if the primary antibody is effective (please see websites later and Fig. 1).
Transfer of Proteins
Once electrophoresis is complete, the separated proteins can be transferred from within the gel onto a membrane (a western blot) made of nitrocellulose, polyvinylidene difluoride, activated paper, or activated nylon (Towbin et al., 1979; Kurien and Scofield, 2006). Nitrocellulose is the most commonly used membrane. Electroblotting is the most popular procedure for transferring proteins from a gel to a membrane. Its main advantages are speed and completeness of transfer. This process uses an electric current to pull proteins from the gel onto the membrane. It can be achieved by immersion of a gel-membrane sandwich (wet transfer) or by putting the gel-membrane sandwich between absorbent paper that has been soaked in transfer buffer (semidry transfer). The effectiveness of protein transfer is dependent on the type of gel used, the molecular mass of the protein, and the type of membrane. Some limitations associated with protein transfer include a lower molecular weight limit of ∼10 kDa, the use of specialized transfer buffers (e.g., 3-(cyclohexylamino)-L-propanesulfonic acid) to facilitate transfer of proteins with a high isolelectric point, and problems associated with using a transfer buffer with a lower pH than the protein's isoelectric point (i.e., the protein will run backward).
Blocking and Antibodies
It is important to prevent interactions between the membrane and the antibody chosen to detect the target protein. To block nonspecific binding, the membrane is placed in a dilute solution of protein such as bovine serum albumin and nonfat dry milk. Researchers should ensure that the blocking buffer that is appropriate for the specific antiserum is also appropriate for the type of membrane. Blocking helps mask any potential nonspecific binding sites on the membrane, thus reducing background “noise” in the final product of the western blot, eliminating false positives and providing a clear result.
After blocking, the most popular method is to incubate the membrane with primary antibody, wash, reblock, and then incubate with secondary antibody and wash again. It is important to determine the optimal concentration of antibodies before running all the samples as optimization is a prime determinant of the sensitivity of the assay (Burnette, 1981). The antibody concentration should be optimized to provide the best signal to noise ratio. Both monoclonal and polyclonal antibodies can be used for western analyses, with advantages and disadvantages in using either type (MacPhee, 2010).
The probes that are labeled and bound to the protein of interest need to be detected on the western blot. For detection methods, colorimetric, radioactive, and fluorescent methods can be used. However, chemiluminescent detection is used most often and therefore, will be briefly described. Enhanced chemiluminescence (ECL) is a sensitive method and can be used for relative quantitation of the protein of interest (Kurien and Scofield, 2006; MacPhee, 2010). The primary antibody binds to the protein of interest and the secondary antibody, usually linked to horseradish peroxidase, is used to cleave a chemiluminescent agent. The reaction product produces luminescence, which is related to the amount of protein. Only a single light detector is required, and the light is detected by photographic film or by a charged-couple device camera (more sensitive, greater resolution, and a larger range of exposures than film). It is helpful that many manufacturers produce a variety of ECL-based western blot detection kits according to the researchers' specific needs.
Once exposures have been captured, blots can be washed in a buffer and then “stripped,” which involves removing bound antisera to enable reuse of the blot. Blots can then be stored for future reprobing several more times. However, subsequent reprobing can interfere with protein antigens, resulting in a decreased signal (Alegria-Schaffer et al., 2009).
RELATIVE QUANTIFICATION AND ANALYSIS
Once an image (Fig. 1) is achieved from the blot, it can be analyzed by densitometry to measure the relative amount of a specific protein on the blot by comparing it with a control or specific time point. This quantification is necessary, if a researcher wants to compare samples (e.g., a treatment or time effect). It is important to attempt to achieve an exposure of the image where the bands are sharp (not fuzzy or indistinct edges). If samples are being compared, the ideal situation is to run all samples on one blot, because there can be variation between blots. There are commercial software programs for image analysis of bands on film, and charged-couple device cameras usually have their own software designed for analysis. Imaging software allows the user to select lanes, bands, or regions of interest for analysis either automatically or manually. Relative levels of protein expression can then be obtained by comparing ratios of intensities of a reference band (e.g., β-actin or glyceraldehyde-3-phosphate dehydrogenase [GAPDH]) or a band of known protein concentration. Relative optical density units can be plotted in a graph, and the appropriate statistical analysis can be performed on the samples that have been converted to optical density units (Fig. 2; for a published example see: Jensen et al., 2007).
The author thanks Sue McGlashan for generously providing the data for examples of a western blot (Figs. 1 and 2).
Western blotting is a powerful technique for quantifying protein levels; however it is often not well optimized and relies greatly on antibodies which are poorly validated. As our study suggests, antibodies differ in regards to the optimal blotting conditions and results they yield; even different antibodies to the same target peptide can give different results. Currently more polyclonal antibodies are used for Western blotting than monoclonal antibodies, mainly due to the ease and lower up-front cost of making polyclonal antibodies. However, polyclonal antibodies vary from lot to lot due to different animals, improper storage, and different bleeds from an individual animal. Since the previously mentioned studies in S1 Table as well as other studies have shown that some popular antibodies to specific proteins show artifactual signals [3–5], we investigated two sets of antibodies to common PTM epitopes associated with ISGylated and ubiquitinated proteins.
Comparison of Ubiquitin Antibodies
Ubiquitin antibodies have been developed to target free ubiquitin, ubiquitin chains linked in a specific manner, or ubiquitin in any form. Depending on the quality and specificity of the ubiquitin antibody used, different researchers may obtain different results when examining ubiquitination or free ubiquitin levels. Our objective was to compare ubiquitin blots using five different antibodies to see if they gave similar results. It was expected that most anti-ubiquitin antibodies would detect high molecular weight polyubiquitinated proteins in the heart and liver samples as well as the polyubiquitinated protein-enriched lysate. It was also expected that the main polyubiquinated proteins detected would be similar to other anti-ubiquitin antibodies.
Comparison of the mouse heart and liver cytosolic fractions (20 μg each) using five commercially available antibodies showed that three antibodies identified consistent major bands at approximately 26 and 60 kDa (Fig 1). These antibodies were utilized under the same conditions (1:1000 dilution) except for the last two lanes on the right which were at 1:100 and 1:2000 dilutions (Fig 1). Two antibodies (VU101 and P4G7-H11) detected a large number of high-molecular weight ubiquitinated proteins while another antibody (U5379) detected some high molecular weight bands. However, only one antibody (VU101) detected free ubiquitin in the liver samples under the conditions investigated (the location of free ubiquitin is shown by an arrow in Fig 1). Four antibodies showed higher levels of ubiquinated proteins in heart than liver for the same amount of total protein. The antibody that did not show more ubiquitination in heart (AP1228a) detected no proteins in the heart sample and only one protein in the liver sample. Although two antibodies, VU101 and FK1, gave similar results, the FK1 antibody did not detect high molecular weight proteins. Since the manufacturer recommends the FK1 antibody be utilized in BSA instead of nonfat milk (NFM), Western blot analysis using FK1 was also carried out using 1% BSA in TBST (lane FK1* in Fig 1). All the other blots were carried out using NFM. When BSA was used instead of NFM a few additional bands were detected.
Fig 1. Comparison of anti-ubiquitin antibodies.
Heart and liver lysates (20 μg each) were investigated by Western blotting using five commercially available anti-ubiquitin antibodies (VU101, U5379, AP1228a, P4G7-H11, FK1). Arrow shows location of free unbound ubiquitin. Stain-free staining of total proteins loaded was used as the normalization control. H, heart; L, liver. BSA was used as the blocking reagent for the blot labeled FK1* while non-fat milk was used as the blocking reagent in all the other blots shown. All antibodies were used at a dilution of 1:1000 except for blots labeled U5379* and U5379^ which were used at dilutions of 1:100 and 1:2000 respectively.
Since antibody concentration is also important, all antibodies were utilized at the dilution that was recommended by the manufacturer. The antibodies VU101, AP1228a, P4G-H11, and FK1 are all recommended for use at 1:1000 while U5379 is recommended at a concentration of 1:100. Varying the concentration of U5379 from 1:100 to 1:2000 showed the importance of the concentration of antibody used, as the 1:2000 dilution only faintly detected one high molecular weight band in the liver sample. However, independent of blocking reagent used or concentration of antibody, the results suggest that different anti-ubiquitin antibodies give distinctly different banding patterns when using Western blot analysis.
Further validation of these antibodies showed that one of these antibodies (AP1228a) did not recognize either free ubiquitin or polyubiquitinated proteins (Fig 2). FK1 only recognized one of the polyubiquitin chains and did not recognize free ubiquitin under the conditions utilized (Fig 2). Purified ubiquitin was used as the positive control for free ubiquitin, while commercially obtained polyubiquitin chains (tri-ubiquitin, penta-ubiquitin and octa-ubiquitin chains) were used as a positive control for polyubiquitin chains. Purified polyubiquitinated proteins and lysate depleted of polyubiquitinated proteins were made in our laboratory using TUBEs. TUBEs has been shown to be highly efficient at removing polyubiquinated proteins from lysates . Two steps were taken to ensure the lysate was depleted of ubiquitinated proteins: significantly more bait (TUBEs) was used than required, and the lysate remaining after the ubiquinated proteins were removed was further processed with TUBEs to remove any trace amounts of ubiquinated proteins. Use of the different controls showed that the VU101 antibody was the best antibody for detection of free ubiquitin and polyubiquitinated proteins.
Fig 2. Validation of anti-ubiquitin antibodies.
VU101 in the presence and absence of 0.5% glutaraldehyde pre-treatment, U5379, AP1228a, or P4G7-H11 were used to detect ubiquitin and ubiquitinated proteins. A) Western blot of polyubiquitin chains (Ub3, Ub5, Ub8) (lane A), purified ubiquitin (lane B), polyubiquitinated proteins from H9c2 cells treated with 10μM MG-132 for 36 h obtained from affinity purification using TUBEs (lane C), and unbound fraction from H9c2 cells after removal of polyubiquitinated proteins (lane D). B) Upper figure, Western blot of free ubiquitin (lane A) and polyubiquitin chains (lane B) with U5379 antibody diluted at 1:100 and 1:2000. Lower figure, Western blot of free ubiquitin (lane A) and polyubiquitin chains (lane B) with FK1 antibody diluted at 1:1000 in BSA. Even when the blots were imaged for long time periods no additional bands were seen.
Due to the relatively low amounts of polyubiquitin chains (300 ng) and purified ubiquitin (1 μg) used, no protein bands were detectable by total protein staining methods such as Stain-free (Fig 2). Stain-free works by cross-linking a fluorescent adduct to tryptophan residues, so the Stain-free method will not detect the tryptophan-less ubiquitin . The VU101 antibody was able to detect all three polyubiquitin chains, the purified ubiquitin, and numerous proteins in the purified polyubiquitinated samples, while detecting only free ubiquitin in the polyubiquinated protein depleted samples. TUBEs do not bind free ubiquitin with high affinity so free ubiquitin was expected to be present in the polyubiquinated protein-depleted fractions. In the experiments shown in Fig 1 the VU101 was used without glutaraldehyde pretreatment. The manufacturer’s protocol for VU101 suggests for optimal results the membrane should be pre-treated with 0.5% glutaraldehyde (Fig 2). Although VU101 recognized all the positive controls without pre-treatment, inclusion of the 0.5% glutaraldehyde pre-treatment increased the signal intensity of free ubiquitin and polyubiquitinated proteins. Inclusion of 0.5% glutaraldehyde pre-treatment in the protocol for Western blotting using other anti-ubiquitin antibodies resulted in no bands being detected, suggesting that the pre-treatment affects the antibody-antigen interactions (data not shown).
For both VU101 and U5379, ubiquitin dimers were also detected. Non-covalent dimerization of free ubiquitin has been previously described . Of concern is that the AP1228a antibody detected a major band in the lysates from which polyubiquitinated proteins were removed. While this antibody detected the octa-ubiquitin chain it did not detect the tri- and penta-ubiquitin chains, the polyubiquitinated proteins in the polyubiquitinated enriched lysate, or free ubiquitin. It is possible that the protein detected by AP1228a is a monoubiquitinated protein still present in the polyubiquitinated depleted lysate. P4G7 detected two of the polyubiquitinated chains, free ubiquitin and polyubiquitinated proteins in the polyubiquitinated enriched lysate, suggesting that it is also a good antibody with the limitation that it does not recognize all polyubiquinated chains. The U5379 showed reactivity to two of the polyubiquinated chains, but also detected some proteins in the polyubiquinated depleted lysate. The VU101, which was the best anti-ubiquitin antibody according to our results, had fewer citations than four of the other anti-ubiquitin antibodies (Table 1). These results suggest that using different antibodies to examine ubiquitination may give contradicting results, as each antibody recognized different subsets of proteins, with some anti-ubiquitin antibodies not recognizing polyubiquitinated standards and at least one antibody recognizing a potentially non-ubiquitinated target. Interestingly, the most commonly cited antibody in CiteAb is the anti-ubiquitin antibody U5379 from Sigma Chemical Company, while the best performing antibody VU101 had no references on CiteAb. The results from four of the antibodies tested suggest that 20 μg of mouse heart contains more polyubiquinated proteins than 20 μg of mouse liver. The anti-ubiquitin antibody U5379 showed distinct target binding properties compared to any of the other anti-ubiquitin antibodies. VU101, P4G7-H11 and FK1 antibodies all detected two major polyubiquitinated proteins at 25 and 58 kDa while the U5379 antibody did not detect either of these bands (Fig 1). While the U5379 antibody clearly detects polyubiquinated proteins (Fig 2), this antibody also detected proteins in the polyubiquinated protein depleted lysate suggesting that this antibody could potentially be recognizing non-ubiquitinated proteins. It is also possible that U5379 may be detecting monoubiquinated proteins in the polyubiquinated protein depleted lysate. The potential detection of some non-ubiquitinated proteins may account for the significantly different target protein identification obtained with this antibody compared to the other antibodies. A possibility also exists that the tri-, penta, and octa-polyubiquinated chains may be forming dimers which would further complicate the analysis of the Western blots. These results emphasize that the most popular antibody is not necessarily the best antibody for the target protein(s).
Based upon the results obtained it is recommended that positive controls should be included when Western blot analysis is carried out using anti-ubiquitin antibodies. It is in the interest of the scientists working in this field to establish what the best optimal controls would be.
Comparison of ISG15 Antibodies
ISG15 is another small protein modifier that can be conjugated to proteins to regulate their activity. Proteins which are covalently linked to ISG15 are referred to as ISGylated proteins. The effect of aging or skeletal muscle disuse on ISGylated protein levels in hearts has not been previously reported. To investigate this, we initially utilized two antibodies against ISG15 and expected to find that one antibody would detect more ISGylated proteins than the other antibody but that both antibodies would detect the same main ISGylated proteins. However, we obtained significantly different results for the two antibodies by Western blot analysis. Further investigation of five anti-ISG15 antibodies from Santa Cruz and one from eBioscience showed that only two of these antibodies gave similar results (Fig 3). E9 and ISG15 antibodies from Santa Cruz and eBioscience respectively (both monoclonal) gave similar results. The samples that were investigated were young and old hearts from normal (control) and hind-limb suspended (HLS) rats. The most common major bands recognized in these samples were 25 and 50 kDa bands which were identified by three antibodies tested. The other most common bands were 37, 42, and 100 kDa which were recognized by two antibodies each. The H150 antibody was the only antibody that recognized a 260kDa protein band. This H150 antibody which gave different results from every other anti-ISG antibody investigated is currently the most cited anti-ISG15 antibody (Table 1). Validation of these ISG15 antibodies was not carried out, because we were unable to enrich for ISGylated proteins without using antibodies, and ISG15 siRNA was only able to reduce ISG15 levels by 80% in C2C12 skeletal muscle cells. Quantification of the results to determine if old rat hearts have similar, lower, or higher levels of ISGylated proteins than young hearts showed that three antibodies detected higher levels of ISGylated proteins in old hearts when compared to young hearts, while four other antibodies showed no statistically significant change (Fig 3B). One antibody (H150) showed increased ISGylated protein levels in young HLS hearts when compared to young hearts while other antibodies showed similar ISGylated protein levels in these hearts. The eBioscience anti-ISG15 antibody showed higher ISGylated protein levels in old HLS hearts when compared to young HLS hearts, while the other six antibodies showed similar protein levels in HLS hearts. These results strongly suggest that different ISG15 antibodies recognize different epitopes and give different results.
Fig 3. Comparison of anti-ISG15 antibodies.
(A) Seven anti-ISG-15 antibodies were used to detect the levels of ISGylated proteins in four different types of samples. (B) Quantification of ISG15 Western blots. Young, 10 month old hearts; Young HLS, high-limb suspended 10 month old hearts; old, 30 month old hearts; Old HLS, high-limb suspended 30 month old hearts. * p < 0.05, ** p < 0.01 by 1-way ANOVA.
The results of our investigation of anti-ISG15 antibodies were more complicated than the ubiquination results, as five of the six antibodies detected different proteins. The most cited anti-ISG15 antibody on CiteAb was H150, and the second most cited was anti-ISG15 from eBioscience. Both of these antibodies give distinctively different banding patterns. Bands at 37 and 100 kDa detected by at least two other antibodies were not detected by either H150 or eBioscience anti-ISG15. The H150 antibody showed a >2.5 fold increase in ISGylated protein levels (mainly due to a 260kDa protein) in young HLS hearts when compared to young hearts while the anti-ISG15 from eBioscience showed no such change. These results show that depending on the ISG15 antibody used, different results can be obtained. Hence it is critical to have the catalog number of the antibody used in all publications. The ratio of sub-standard antibodies to quality antibodies is likely to increase, as the number of post-translational modification (PTM) specific antibodies is increasing at an exponential rate. Antibodies are now available for many PTM sites including acetylation, methylation, and the more common phosphorylation sites, but many of these PTM specific antibodies do not work well. An additional complication is that the antibody specificity may change under different experimental conditions and in different tissues . An antibody may work well for one cell type or species but not for other cell types.
Even if the scientific community is able to generate quality monoclonal antibodies or recombinant antibodies for Western blotting, other factors important for the Western blotting technique also need to be taken into account. One example is the primary antibody concentration. The amount of antibody that should be utilized for Western blotting depends on many factors including the concentration of the enzyme, the abundance of the target protein and the detection system used. The specificity and high affinity of antibodies for their targets allows antibodies to be used in low concentrations (ranging from 1:100 to 1:500,000 for a 1mg/ml starting concentration). The optimal dilution of the primary antibody has to be determined experimentally. In general lower amounts of antibody result in increased specificity for the target protein. Using lower amounts of antibody also reduces cost.
Overall, similar to the anti-ubiquitin antibodies, proper Western blot analysis using anti-ISG15 requires proper controls. An ISG15 knock-out mouse is available and tissues from this mouse model would be an excellent negative control for anti-ISG15 antibodies. It may be time for NIH to start a resource center with tissues from knock-out animal models that could be used as negative controls for Western blot analysis. Including information in manuscripts about controls or validations done using antibodies not previously validated would give reviewers greater confidence in the results and would also be beneficial to other researchers [2, 28].
Effect of Primary Antibody Dilution on Western Blotting Results
Since other factors are also important for Western blotting accuracy and reproducibility, we investigated two of these important factors which are often overlooked: 1) the effect of primary antibody dilution on signal detected and 2) the effect of the main buffer used in Western blotting on signal detection. Different concentrations of rat liver lysates were probed with different dilutions of anti-β-actin and anti-PSMA6 to determine the effect of antibody dilution on signal detection. β-actin is a commonly used Western blotting normalization control and PSMA6 is a subunit of the proteasome. While it was expected that lower concentrations of antibody (higher dilutions) would result in lower signal intensity, we observed that anti-β-actin dilution did not significantly affect the signal intensity detected (Fig 4E). This was surprising since a 1:25000 dilution of the anti-β-actin gave similar results to a 1:1000 dilution. Using this 25 fold dilution of anti-β-actin would significantly reduce the cost of the Western blot without compromising the sensitivity of detection.
Fig 4. Effect of antibody concentration on linearity of target proteins detected by Western blotting.
(A) Western blot of rat liver samples (3–12 μg) using anti-PSMA6 at four different concentrations (1:5000, 1:10000, 1:20000, and 1:50000). (B) Quantification of anti-PSMA6 Western blots without including any normalization. (C) Quantification of anti-PSMA6 Western blots using total protein normalization. (D) Western blot of rat liver samples (3–12μg) using anti-β-actin at four different concentrations (1:1000, 1:2500, 1:10000, and 1:25000). (E) Quantification of anti-β-actin Western blots without including any normalization. (F) Quantification of anti-β-actin Western blots using total protein normalization. * p < 0.05 by 1-way ANOVA.
In contrast, higher concentrations of anti-PSMA6 (1:5000) gave higher signal intensities of PSMA6 when compared to lower concentrations of the anti-PSMA6 at all liver lysate amounts investigated (Fig 4B, not normalized data). When normalized to total protein loaded it is expected that all the protein concentrations and antibody dilutions would give a normalized value of 1 as observed for β-actin (Fig 4F). However lower concentrations of anti-PSMA6 resulted in lower normalized signal intensities at 3 and 6 μg of liver lysate (Fig 4C). Also of significance is that the normalized values for 12μg of lysate was statistically significantly higher than for 3μg of lysate when the anti-PSMA6 was used at its highest concentration (1:5000 dilution, Fig 4C). This suggests that to achieve optimal results with antibodies such as the anti-PSMA6 antibody used in these studies (as opposed to the β-actin antibody used in these studies), it is important that the amounts of proteins loaded on a gel be similar in each lane. In contrast to the PSMA6 antibody, using higher amounts of the β-actin antibody did not increase the signal intensity, suggesting that for certain antibodies it is not advantageous to use higher amounts of antibody (Fig 4E). These results show that the amount of protein loaded on a gel is important, and each gel lane should have very similar total protein amounts to avoid artefacts due to protein loading.
Effect of buffer on western blotting results.
A variety of Western blotting buffers are currently used and range from TBS or PBS without any additives to TBS and PBS with several additives including Tween-20 and low amounts of blocking reagent such as non-fat milk. TBS, however, a commonly used buffer for Western blots, is not always the best buffer for certain antibodies. Commercial preparations of TBS can also be different in the concentration of Tris present and the pH of the solution. We carried out a Western blot on rat liver lysates with the PSMA6 and β-actin antibodies using TBST or PBST to see if varying the buffer affected our results (Fig 5). Although some researchers utilize PBS while other utilize TBS we expected that these buffers would result in different target protein signal intensities. While anti-PSMA6 antibody showed similar normalized relative amounts of detected targets in different amounts of rat liver lysates (3–12μg), anti–β-actin showed a greater than 10 fold increase in intensity when PBS was used instead of TBS. When low abundance targets are being detected, 10 fold differences in the detection signal could be the difference between detecting the protein of interest and not detecting the protein. These results show that the choice of buffer is very important for signal intensity when certain antibodies are used. It is recommended that Western blot analysis using uncharacterized antibodies be carried out with both PBS and TBS to determine if one buffer is significantly superior to the other buffer.
Fig 5. Effect of buffer reagent on Western blotting linearity.
(A) Western blot of rat liver samples (3–12 μg) using anti-PSMA6 and different buffers (TBST and PBST). (B) PSMA6 quantification, not normalized to total protein. (C) PSMA6 quantification, normalized to total protein. (D) Western blot of rat liver samples (3–12 μg) using anti-β-actin and different buffers. (E) β-actin quantification, not normalized to total protein. (F) β-actin quantification, normalized to total protein. * p < 0.05, ** p < 0.01 by 1-way ANOVA.
Other Western Blotting Concerns
Several other common Western blotting problems exist; however since experimental data from different laboratories including our laboratory are already available on this topic these problems are only briefly mentioned [29–35]. A common problem in Western blotting is the total amount of protein in each well is too high. When a protein of interest is expressed at very low amounts, some labs use very high protein loads (>60μg of total protein) to be able to detect the protein and use a housekeeping protein (which is usually expressed at relatively high levels) as a loading control. While the protein of interest may be in the dynamic range of the tissue or sample being investigated, housekeeping proteins because of their abundance have a limited dynamic range and are not linear at high protein concentrations . Another problem with blotting high amounts of proteins is the increased chance of artifacts due to non-specific binding of the antibody utilized. Using lower amounts of protein for Western blotting has been shown to improve the detection of some poorly expressed proteins [30, 31]. The secondary antibody used also affects the signal intensity of the target protein. IgG subclass specific secondary antibodies were found to be superior to anti-mouse IgG (H+L) antibodies in immunohistochemistry and Western blotting . A recent publication also showed results which demonstrated that the current lack of established procedures in densitometry results in inaccurate quantification of many Westerns .
All of these reports suggest that the current standards for reporting Western blots are inadequate. A requirement by journals for researchers to deposit detailed antibody information into one of the online databases would be very beneficial, but the best solution would be an NIH mandated depositing of detailed Western blotting data into one site for all NIH funded research. A repository with the tissue used and optimized buffer for antibodies would be especially useful to all research labs using this technique. Like polymerase chain reaction and mass spectrometry data, reporting minimal standards are needed for Western blotting. A 10 point requirement referred to as the Western blotting minimal reporting standard (WBMRS) is suggested (Table 2 and S1 text). With limited research funding, identifying poor quality antibodies and poor Western blotting techniques will save money, save researchers time and improve the quality of the results.