Dear Matthias,

your question touches many aspects of Western blot analysis of which some (using other control proteins, using the antibodies against the non-phosphoform of your protein, using kinase inhibitors, etc. have been adressed).

(At least) one point remains unanswered and this point is the difference between statistical significance and functional relevance. If 30% increase in a phosphoprotein signal is significant is a matter of how many times you repeated the experiment, how many "n"s you have and if all results point into the same direction.

The functional relevance is far more difficult to estimate and depends on what your phosphoprotein is supposed to do what the non-phosphoprotein does not. In the case of cyclicAMP/calcium regulated element binding protein (CREB) I doubt that 30% elevation is relevant, in other cases it may.

A technical comment: many proteins display a phospho-group-based size shift in SDS-Page which can be seen on the blot membrane. If you do DAB staining instead of chemiluminescence you may see both phospho- and non-phospho with the total protein antibody or do two blots with phospho- and total protein. From the shifted versus the non-shifted signal you can estimate how much more of the protein is transformed into the phospho-form (one example is Fricke et al., Endocrinology 2005 for Hormone-sensitive lipase but there are many more). Good luck, Erik

Well, I cant understand this fully, some more details may be helpful and if you think your results are convincing then I would repeat it once or twice to compare them.

Normalizing the protein expression by actin will tell you the amount of your protein of interest that you have on your blot. Actin is very abundant compared to other proteins so you may have more actin expressed than your protein of interest and that would bring down the level of your protein when you do the normalization. In this case, you can directly compare your total protein expression level to the level of phosphorylated protein without the actin calculation since the total protein would be used to normalize the phosphorylated protein expression level.

@Ananda: I repeated the experiment 4 times and got the same result each time

@Kristen: we use actin for normalization to correct for possible pipetting errors. reanalyzing the values without prior actin normalization raises the phospho-protein/protein to 1.5 fold

if your loading is fairly equal according to Bradford this means that when you check actin in your reference blot, it should also be decreased. Accordingly, since your phosphoprotein reduction is half that of your total protein, then it sounds reasonable to assume that you have increase in phosphorylation. However, it should be better: 1. to use another reference protein and 2. see what happens with the addition of kinase inhibitors (if you suspect or know of a certain kinase species, then look for the appropriate specific inhibitor: this should be critical experiment)

Does your protein has got other isoforms?

@George: I resolved 30 microgram of protein on the gels, as determined by bradford assay. actin is not really changed due to the treatment... the first reference protein I used was tubulin... which was downregulated by 60% :-)

If there is an antibody available for the non-phosphoryled form of your protein you can use it to compare as well.

So there seems to be an interesting protein that you study. If along with actin also ponceau staining shows invariable protein bands, then you may look at the possibility that your protein is selectively degraded due to the treatment. I cannot tell whether this may be induced upon protein phosphorylation, but you may look into its primary structure for degradation signals, like polyubiquitination. Again I would rather insist with kinase inhibitor treatments and see if degradation is somehow depending on phosphorylation. I just happened to observe similar "problems" when comparing the amount and phosphorylation status of a certain protein between wild type Arabidopsis and a MAPK mutant. Unfortunately I put those results behind and I hope that I will be able to study them sometime

One likely explanation is that the unphosphoryalted form of the protein is being preferentially degraded over the phosphorylated form.

@ Matthias, I would go ahead and use the values you have. You can separately analyze the actin levels of your treatment groups and see if you systematically pipetted more protein for one treatment group. If the answer is no, then protein loading variation is not the source of your variance and you have a true experimental effect. It sounds like this is what you have.

Another way of stating it is that, the phosophorylated form is stable over the non-phosphorylated form.

With the normalization with actin you can assure the fold in phosphorylation and the decrease in the amount of target protein, the most interesting thing is the interpretation of that, an increase in the activation of the protein along the decrease in the amount of protein

Hi Matthias,

have you controlled different time points after treatment? I mean, if the protein is degradated and if (like Amitavo said) the phosphorylated form is more stable, than you possible see only one form of your protein after a specific period of time.

@Manuel and Dayana: and that is exactly my problem... how to interprete that....

@Amitavo: also an interesting idea.... i have to check that...

I was thinking that, if it's known, you should consider also the function of this protein and how it works. A kinase, a transcription factor, a protease, a protein working as a dimer, alone or as part of a big complex are quite good informations to interpret your data...what do you think?

Dear Matthias,

your question touches many aspects of Western blot analysis of which some (using other control proteins, using the antibodies against the non-phosphoform of your protein, using kinase inhibitors, etc. have been adressed).

(At least) one point remains unanswered and this point is the difference between statistical significance and functional relevance. If 30% increase in a phosphoprotein signal is significant is a matter of how many times you repeated the experiment, how many "n"s you have and if all results point into the same direction.

The functional relevance is far more difficult to estimate and depends on what your phosphoprotein is supposed to do what the non-phosphoprotein does not. In the case of cyclicAMP/calcium regulated element binding protein (CREB) I doubt that 30% elevation is relevant, in other cases it may.

A technical comment: many proteins display a phospho-group-based size shift in SDS-Page which can be seen on the blot membrane. If you do DAB staining instead of chemiluminescence you may see both phospho- and non-phospho with the total protein antibody or do two blots with phospho- and total protein. From the shifted versus the non-shifted signal you can estimate how much more of the protein is transformed into the phospho-form (one example is Fricke et al., Endocrinology 2005 for Hormone-sensitive lipase but there are many more). Good luck, Erik

Can you determine protein and degree of phosphorylation (before and after) by different techniques? Using radioactive label? MS or so. I feel that Western Blot is just a preliminary step to evaluate protein phosphorylation. I personally work with bacterial proteins glycosylated and phosphorylated in the same fragments of proteins. I assume, the two processes act as regulators of protein function.

If you're obtaining the total protein and the phosphorylated protein levels from the same blot, you need no normalization to actin levels to calculate the phosphorylated/total protein ratio because this ratio is independent of the amount of actin: phospho/total = (phospho/actin) / (total/actin).

However, you said that your ratio after actin normalization was 1.3 whereas it was 1.5 without actin normalization. So, I wonder if the problem (and a possible way to understand your results) is that you're obtaining the total and the phosphorylated protein levels from different blots.

Hi Matthias,

Interestingly, in my hands, actin levels can be affected by whatever treatment the cells are undergoing. This is not true in response to all treatments but with certain ones I sometimes see a doubling of the amount of actin. If this is the case then obviously actin would not serve as a good control. Therefore I have resorted to staining my membranes with Coomassie Blue R250 after I have probed for my target protein (strip the membrane first). This stain usually shows an even protein load in all the lanes, i.e. the difference in actin levels was not due to pipetting error. I then scan the stained membrane and use a representative band as a control.

Good luck!

Gisela

One way to be sure that you have the same total protein in both conditions is to perform phosphatases treatment. Phosphatases have very good activity and I'm sure you are going to obtain almost all protein completely dephosphorylated.

Good luck!

Western blot is a semi-quantitative technique to evaluate protein phosphorylation or any other PTM. I would say your observation is correct with some limitations. I would suggest to reconfirm your observation using other quantitative techniques such as ELISA or Mass-spectrometry.

I assume that you are treating a cell culture with some drug or compound and as a result you observe the protein reduction and changes in phosphorylation. I would say that it depends on what protein you are studying. Is it possible that your treatment causes changes in cell cycle, which also changes phosphorylation state of many proteins? What about the total number or cells that you lyse? Someone mentioned before about the antibodies for your protein of interest, which I think is a great idea. If you are interested in the ration of phosphorylation state of your protein regardless of total protein concentration I would do WB with antibodies against your protein that recognizes both forms, and then do WB with phospho-specific antibodies. Densitometry and comparison between treated and untreated cells should give you the answer whether the compound that you are testing influences phosphorylation state of your protein of interest. I hope that any of it answers your question... Good Luck!

Matthias,

I’m not familiar with the nature of your protein of interest and whether it becomes active or inactive upon phosphorylation. But when using beta actin as a control and your loading looks identical everywhere, than your densitometry quantitation should be accurate. This would be even more accurate if there is an antibody available against either the non-phosphorylated form, or the phosphorylated amino acid residue.

If your loading is correct and you are still trying to see whether your treatment causes alterations in the protein levels of both phosphorylated and unphosphorylated form, you should probably try a 2-D Western. By using ReadyStrip IPG strips, with pH 3-10 for example. Based on the protein charge, you could easily distinguish between the isoforms, and you can proceed with your current antibody that you have tried already. Good luck!

If I understood correctly, you got a result that, phospho-protein/protein ratio of treatment group is 1.3 fold increased, compared to control group. In spite of any technical problems, the result at least indicates that the treatment group has a higher phosphorylation ratio, no matter how much the total protein is increased or decreased.

So I would like to conclude that 'protein phosphorylation is enhanced' instead of 'phospho-protein is enhanced', judging from the ratio part.

I think that's a significant result.

As Shubhankar said, WB is semi-quantitative technique. In my opinion it is incorrect to do the mathematical calculations you are doing. You must be sure that you are working in the range of concentrations that gives a linear labeling with the antibody concentrations you are using for detection. Assuming that you are in the correct range, you must take into consideration that the signal is not directly proportional (even if signal vs. antigenconcentration is linear) to the amount of antigen in the membrane. I mean that a band of an intensity that doubles another’s band intensity does not mean that you have two times the amount of the protein of interest: it simple means that you have more protein, maybe more than twice, maybe exactly twice or maybe less than twice.

Also you should state if actin is a good housekeeping protein for your samples. Otherwise it is advisable to control protein load on your membrane (as Gisela suggested). Remember that nitrocellulose membranes cannot be stained with Coomasie.

I am not sure about the nature of your protein-of-interest and your treatment, but it is possible that phosphorylation decrease the turnover of your protein-of-interest, so I think it is not appropriate to conclude that the phosphorylation is enhanced with your current data.

it is a question of stoichiometry assuming you don't have artefacts. Your phosphorylated protein is a substrate of kinases and phosphatases, and because phosphorylation may modify protein stability one may find situations that both the phospho and total forms are regulated by a particular treatment or condition. The interpretation in your case maybe : you have less phosphorylated but much less total protein so the stoichometry is higher (phospho/total) than before treatment. The kinase that phosphorylates the substrate is putting more phosphate per total protein left or the phosphatase is taken the phosphate from it. If still have less phospho (actived or inactivated from of your protein function). Just look at enzymology with the equilibrium between enzymes and substrates it should help you. In the case of mTOR, for example, activation is associated with more stability so you have more phospho and more total. The stoichiometry may remains the same but you have more phospho-mTOR. And so on, so for...

This is a very complex situation, and the answer will depend on how you measure your protein abundance. If your results of the total protein measurement are obtained by normalizing them to actin, and your phosphorylated protein is also normalized to actin. The data of phosphoprotein/protein from the normalized data will be difficult to interprete. However, if you measure the abundance of protein and phosphorylated protein against the respective standard curves. The expression of phosphoprotein/protein data will be trust worthy.

Yes, the levels of phosphoryolation is NOT proportional to the leves of protein expression...and sure you can state that the level of phsphorylation of your protein is enhanced

If your experimental design (especially negative and positive control) is appropriate, you can say that the phosphorylation was induced. Phosphorylation levels should be simply evaluated by the ratio of phosphoprotein to total protein.

As to your observation that total and phospho proteins are reduced with increased ratio of phospho to total protein, I have some comments.

The phosphorylation levels are regulated by kinase, phosphatase and the protein expression levels regulated by de novo synthesis and degradation. In your case, total protein levels were reduced presumably via degradation or loss of synthesis. Normally, this reduction of total protein will also result in reduction of phosphoprotein (the ratio of phospho to total will be equilibrated soon to the original levels before).

Therefore, if you observe increased in the ratio of phospho to total, it means there is a change in the regulation, e.g. increase in kinases or decrease in phosphatases. There are a number of feedback regulatory systems in phosphorylation signals. Some important phosphoproteins are kept at certain levels. In your case, I assume that the phosphorylation was induced as a compensation to the loss of total protein.

Actually, this kind of observation is not rare in the phosphorylation signals.

In this scenario it will be better to perform a proteoasomal block along with the treatment in both the control as wells treated samples and then look in to the phosphorylation status which should answer your questions at this moment.

This is how I would look at the results:

Your experiment addresses two separate issues:

(i) What happens to the expression level of the protein in response to treatment?

(ii) How much of the expressed protein is phosphorylated after treatment ?

The answer you have is:

60% of the protein is expressed in response to treatment and of that 80% is phosphorylated.

You can compute phosphorylation under control conditions relative to protein level (in control), if that is 100% then the protein expression as well as phosphorylation is compromised after treatment.

I suppose it depends on what you are studying really - the degredation or the downstream activity and the pathway.

In the labs I have worked in, we have always analysed phospho-protein (e.g. p-ERK) in relation to actin. The reason for this is that it is well known that levels of ERK and other MAPK will change in response to stimuli, however, downstream changes in the pathway are reliant on a net increase in the active p-ERK, in relation to its downstream target, not to unphosphorylated ERK.

So if you have decreased ERK but increased phosphorylation, the net effect on the throughput of the pathway might be zero. That is not to say though, that the change itself might not be interesting in other ways, just that I would not describe this (as I understand it) as an increase in phosphorylation. Particularly if I was trying to claim it were indicative of increased downstream effects such as activation of transcription factors.

WB is a semi-quantitative method. For variations of less than 2 fold, realize the experiment 3-5 times and then we can talk

Well, if you relate phosphoprotein/protein (reduced value after treatment) of each single sample, with phosphoprotein/protein before the tratment, the result may be acceptable. It's possible that the treatment induce a negative reguation of protein expression probably via a phosphorilation mechanism.

Sorry, but I can response your question because I haven got experience with westhern blot.

Bioq. Paola Ferrero

Strictly speaking about WB technique, what I’ve tried to say before is that as each protein (actin and your protein of interest) will have its own linear curve of signal vs. amount of antigen, that are probably not going to be parallel, so relationships between them are not going to be proportional. Normalization in WB has this extra-problem, that is why in publications you usually see that the actin band has the same intensity in each lane. This is the only way to be sure that the fold changes in your protein signal are well determined. It is different, for example, in real time PCR for which you can calculate efficiency of reaction for each pair of primers and then you can mathematically take into account slight differences among housekeeping levels of different samples.

Maria that's a very good point. It is really important to load equal amounts of protein.

I found using densitomety that the signal to protein ratio is not even linear in many cases. I think there are two components to the densitometric measurement 1. The density/opacity & 2. the size/area. If you load too much protein, the density proportion hits its maximum and then the only factor increasing the signal is the band area; so the line levels out.

To my mind you can´t quantify on the same blot change of the protein expression AND change of post modification. The main reason is that following subtle change of two parameters in a single experiment will provide you almost data statistically impossible to interpret.

I would suggest:

1/ Also use normalization of the loading by adjusting level of the protein of interest, really interesting to check if a post modification really occur or not.

2/ Use more mild protein extraction protocol and do not freeze your sample (if needed aliquot them)

3/ It would be of big help to include other methods to confirm your results, make your story more trustable (http://www.rndsystems.com/mini_review_detail_objectname_MR08_ProteinPhosphorylation.aspx)

To complete the nice answer of Maria and Peter, I used to check difference of signal without integrating loading control, and then with integrating it. If the normalization change my tendency, I just throw away results and redo the blot using a better loading.

Let us know what are your conclusions!!!

Best

if you are lucky and your protein is bandshifted by the phosphorylation events, you can estimate the ratio of phospho/non phospho form. I think this is more accurate thant comparing the signal obtained with two differents ab

The real question to answer here is why are you studying phosphorylation of this protein-specifically what is the biological consequence of its phosphorylation? Peter makes a good point point in answer, but ultimately you need to relate it to the role of protein phosphorylation. I have worked on quite a few proteins and their regulation by phosphorylation, and have always tried to relate the specific changes in the amounts of phospho- to total protein. This works well generally when the levels of total protein do not change, for example IGF stimulation of serum starved cells will cause a massive increase in phospho ERK1/2 or AKT within 15 minutes. However there are cases when the total amount of protein changes, and the interpretation is tricky. Firstly you can use the vaues for total and phospho protein to re-run your gels so that you load the same amount of total protein of interest (i.e. normalising to the target, not total protein from the extract) and then check the levels of phosphorylation, and similarly load the same amount of phospho target and compare the levels of total protein. This will give you a better idea of how the levels of phosphorylation have actually changed.

There is also a caveat here which cell signalling technologies do talk about and that is sometimes if you re-probe a blot you have used for phospho proteins with the corresponding total protein, sometimes if the epitopes are too close the phospho-protein antibody can inhibit the total antibody binding to the membrane. Usually this is not a problem, but is worth considering.

However I would suggest that you look at a biological readout of the proteins functionality-does phosphorylation change the activity or stability of a protein? I have seen massive increases in the expression and phosphorylation of a transcription factor that is inhibited by phosphorylation, but the the absolute increase in the levels of that protein meant that there was sufficient un-phosphorylated protein to drive the transcription of targets genes as studied by reporter assays and RNA expression. I have also seen cases where phosphorylation results in decreased stability of the protein, and you would have to study that differently by measuring the effects on half life and compare that to biological activity etc.

If you blot phospho- and total-protein on the same membrane, blot actin (to monitoring samples loading ) and phospho protein first, respectively, and then strip the phospho-signal to blot total protein. To calculate phospho ratio, (phospho signal/total signal).

If you run two gels, one is for phospho and another is for total, so blot them and their actins, respectively. To calculate phospho-ratio, (phospho-signal/actin in phospho gel)/(total-signal/actin in total signal).

Hope help. Good luck.