Column storage may be causing trouble?

[ABV]

Hi everybody,
I have a problem with a newly developed SE-HPLC method used for stability testing: During time (0 - 3 -6 -9 months ) the target protein peak is decreasing and moving slightly to the right, also the control, a different batch of the same protein shows this behaviour. Our RP-HPLC analysis of the same protein shows no signs of decreasing signal! So it seems the problem is with the column/method.
Another tricky part is that the 5 proteins from the MW calibration marker stay relatively stabile at a constant signal level througout these different timepoints. So these 5 protein go right through the column.
The target protein is known to be sticky, but when multiple injections are made in the same sequence of samples, there is no loss in recovery! However, when the column has been stored between two timepoints the decrease in signal occurs, but ONLY for the target protein as mentioned.

Column: BioSec-3 Agilent, polymer coated silica (extra hydrophilic and pH stabile), 300 Å, 7,8 x 300mm.
Mobile Phase: PBS buffer.
Storage: after rinsing throughly with water I store the column in 20% EtOH.

Hope, you have ideas of what could be wrong!

[nonagon]

Hi ABV
There is a chance it is not the storgae that causes the problem but equlibration.
Do you condition the column before analysis?
The procedure should be:
Column from storage>>Flush ethanol with water>>Equilibrate with mobile phase>>Load a total of not less than 150ug of your protein over several injections >>perform analysis.
If you do that and the problem persists try adding up to 5% n-propanol (start with 3% IPA) to your mobile phase, this is helpful with "sticky" hydrophobic proteins especialy in your MW range of <40kDa.
From the chromatogram, I gather your flow rate is <=0.6ml/min (Vc=14.3ml). if it is higher than you have secondary analyte-column interactions.
Are you using a biocompatible LC system? some proteins tend to adsorb to teflon and stainless steel surfaces of the LC system's flowpath never to be seen again. I had that experience, it is not pretty.

If I got it right, the highest peak is the 0 timepoint than lowers with time progression. check adsorption to storage containers.
was the analysis performed of the same day or is this an overlay of four different analyses?
cheers

[danko]

I think it’s degradation. And it’s consistent with the fact that in SEC smaller molecules generally elute later than larger ones.

Best Regards

[nonagon]

If this is degradation than we are talking utter destruction in the sense that a lot of the protein is missing.
what is noticeable is that a specie with MW of the native protein is hardly present and if so where is the rest of it?
protein degradation pathways, even under extreme conditions not commonly found in storage, are pretty much modifications to any number of amino acids (Asp deamidation, Met oxidation etc) - this does not change hydrodynamic radius.
it is possible that the degradation here is one that causes decysteinylation, opening disulfide bonds and "flattening" the protein. In such case its steric conformation is changed and may appear smaller in SEC.
Another degradation option is enzymatic cleavage of some sort or chemical digestion, however it must be a very strong reaction since mass balance appears to be low suggesting that the protein has lost chromophores and parts of it are undetectable (very small peptides without chromophores or free amino acids) this can only be caused by a very violent mechanism.
All options discussed here are very bad for stability studies and are, while possible highly unlikely.
I believe the protein is being "snatched" somewhere either by active sites on the column or by the LC system's wetted bits, but who knows - nothings can surprise me with proteins on HPLC.

[haverickm1]

The SE-HPLC is a routing assay but is always filled with problems. I feel that the issue could stem from the fact that your protein has an extremely high affinity for the column you are using, much higher than the standards. When you store the column in 20% ethanol it removes all the proteins that are bound to the column (which you can tell if you monitor the AU as you store the column). What you can do next time you use the column is do several equilibration injections of a standard or even the protein of interest. The equilibration injections will allow your sticky protein to occupy all the available binding sites.

Good luck to you!

[ABV]

Hi again, thanks for a lot for good answers. I guess some of them will be helpful in my coming optimisation of the method.
A general note is that we are now convinced that the degradation-like profile is an artifact of the analysis, primarily because other methods do not see this degradation.
I will try to answer the comments and questions some of you have in the following..

For Nonagon:
- Yes. I do flush the column with water after storage to get EtOH out and basicly follow your outline.
- Each sample sequence starts with a couple of blank runs and then an MW standard, however I guess that I'm not close to loading 150µg before starting to analyse the samples, if that is what you mean? Deliberately I have tried to work with smaller loads (5-10 µg sample per inj.) as I suspected the lifetime of the column would suffer with larger loads!? Maybe it's too little?
- Yes. The working flow is 0.6mL/min
- The LC system is Peek and stainless steel (not that I think that is the problem in this case!?)
- Yes. The highest peak is the 0 timepoint and from here it goes down and to the right.
- The image is an overlay of sample + MW standard run the same day at timepoints: ~0, 3, 6, 9 months.
For Danko:
- No. Not degradation. It's an artifact specific for the protein in these conditions.
For Nanogaon (2):
- No Cys in the protein.
- I support your "snatching" theory, however it ONLY occurs after the storage procedure... (more on that later).
For haverickm1:
- I think you're right only that it seems water does the "trick":



This is how the water rinsing of the column looks after running several samples and MW standards. I had not collected this data when I posted last time, but it seems to me that something, propaply the sample, is more soluble i water than in PBS!!
Hope to get some more good comments! And have I nice weekend!
Best Regards
ABV

[danko]

Hope to get some more good comments! And have I nice weekend!

Not so much a comment - but more of a question. What exactly does the following mean?

- No. Not degradation. It's an artifact specific for the protein in these conditions.

What happens?

Best Regards

[nonagon]

Hi ABV
Three quick questions:
1. Is you polypeptide His-tagged?
2. Are there any dteregents/surfectants in your formulation (even highly diluted)?
3. Are metal interactions involved in your polypetide's activity?

And to clarify - I mean 150ug over several injections e.g. 15 injections of 10ug or something of sorts.
cheers

[ABV]

For Nonagon:
1) No. There is no His-tag.
2) The protein is formulated in PBS, pH 7,4 + 0,1% Tween20. For different reasons i chose to leave out the Tween20 from the mobile phase, and at first it seemed alright! (Adding Tween20 to the mobile phase of cause will be on my list of "thing to try")
3) It is not known what the exact role of this protein is. However, from Mass Spec there has been indications that it likes to bind several Na and K atoms (not confirmed).

[nonagon]

OK!
Tween will elute after about two column volumes as seen in your water run.
You can avoid this by injecting two samples followed by a 90min flush run with water.
Tween is more soluble in Ethanol so you should consider flushing with about 5% of the stuff.
The best solution, however is to adjust your mobile phase to contain the same %tween as the sample. This, of course, renders your column strictly tween dedicated.
As for myself, I use detergent/surfactant removal columns or dialysis since I cannot be bothered with flush runs (I validate both procedures).
As for the separation, try adjusting ionic strength:
Remove the "s" from pbs in your mobile phase as in prepare a phosphate buffer at the same concentration and pH without NaCl.
Run your samples and check for improvement or worsening in separation - see if there is any significant difference from pbs.
remember to condition your column with your protein as discussed.
If separation gets worse, you can do the opposite - prepare phosphate buffer at the same concentration and pH with 0.3M NaCl, equilibrate and run your sample.
If this does not help, there are other pathways for solution.

cheers

[danko]

I don’t seem to be able to find the detection technique/mode in the descriptions above, but assuming it’s UV, I don’t think you’re seeing Tween20 peak – no UV absorption.
Another potential issue: Can’t find the mobile phase pH either. But if low pH – NaCl is a bad, bad idea in relation to the steel parts in the flow- path – i.e. pumps, capillaries, valvs etc.

Best Regards

[nonagon]

danko
Take a look at ABV's posts and chromatograms, you will find the answers.
We are talking UV at 214nm, pH=7.4 with ionic strength of about 0.15 NaCl.
It is true that stainless steel parts are bad for protein analyais and may very well be the problem with this particular one, most non-bio HPLC's I've worked with had no problem with 0.3M NaCl at the discussed pH.
As for tween not having UV absorbance, you're right - there is, however an effect of refrective index change where Tween changes the amount of light getting to the detector due to angular shift thus detectable in UV.
Furtermore tween may be bound to the protein so when it finally elutes, the bound protein shines beautifuly in UV.
This is what we call "tough sample" sticky metal chelating protein with surfactant.
In a perfect world we would be using a bio-inert HPLC with PEEK column and no surfactants or detergents but we have to be practical with what we got.
Moderate toying with ionic strength has far more chance of benefit than damage in this case. A good advice though would be to check for precipitation prior to testing with 0.3M NaCl.

[danko]

I still can’t find the mobile phase pH but I can go along with the assumption that when the protein is formulated in PBS, pH 7.4 then the mobile phase is at the same pH – if I try hard anyway.
I wasn’t talking about potential protein/steel interactions which often are hugely exaggerated, but rather the NaCl effect on the steel material – very degrading in deed, but at lower pH so 7.4 wouldn’t worry me too much.
Finally, refractive index artifacts are usually observed as less absorbance or negative peeks if you will, due to the fact that the light beam is refracted away from the detection unit – PMT or whatever.

[danko]

Sorry the last sentence was a mess. I meant negative peaks due to the changes in the transparency of the mobile phase while the non absorbent compound elutes, but the other affect is observed too so one typically observes rapid changes in the absorbance/baseline.

best Regards

[nonagon]

I just wrote a comment but failed to submit before I was automaticaly logged out. BUMMER!

Refractive index shift can indeed go either way as seen in many gradient applications - less light=absorbance.
Surfactants have a way of misbehaving in chromatography:
Tweens become yellow in water, impurities jump and prance at 214nm, they may elute within column volume and more.
I do not think this is the problem here though.
I think playing with ionic strength may improve things here.
My bad, failed to notice no mobile phase composition ABV please post.