Buffer Optimization (SEC), odd results

[ScottHorn]

We've recently been evaluating the Tosoh SuperSW3000 column (4.6x30) to replace the Superdex 200 10/300 GL columns that we've been using to analyze antibodies and antibody conjugates. The primary reason for the change was the speed, 60 min per injection for the superdex vs 15 min for the SuperSW. The resolution is noticeably better as well. I have noticed the tailing that others have mentioned experiencing with this column, but only with some antibodies. I noticed that Tosoh literature recommends using increased salt concentrations as well as <20% organic solvents to minimize undesired interactions with the stationary phase. To that end I decided to try a different buffer to see if I could improve my results. Previously, we've been using PBS w/ 0.1% sodium azide, pH 6.8. I made the same phosphate buffer, but increased the NaCl conc. from 0.15M to 0.4M, and added 10% ethanol (before pH adjustment). I injected the same antibody (a monoclonal mouse IgG) both before switching buffers, and three times after (I got impatient before the column was fully equilibrated). I'm attaching the relevant chromatograms, but in short what I saw was an increase in retention time, decrease in plate count, increase in tailing, and decrease in area, for the IgG peak. All peaks had better resolution. What is going on here? Are the increased salt and the ethanol having opposite effects? I like the increase in resolution, but the tailing on the IgG peak is getting to be ridiculous. I'd appreciate any theories as to what I've done, and perhaps some suggestions as to how to keep this resolution increase without the increased tailing.

The chromatograms are as follows:
1. Old buffer, PBS w/ azide, 150mM NaCl, pH 6.8
2. New buffer (after about 2 column volumes), PBS w/ azide, 0.4M NaCl, 10% EtOH, pH 6.8
3. Same buffer, after 4 column volumes
5. Same buffer, after many column volumes

Flow rate: 0.35 ml/min
Sample Conc: 2 mg/ml
Injection vol: 5 ul
Detector: UV @ 280 nm









Any ideas?

Scott Horn

[Uwe Neue]

You got a better separation for the dimer, trimer (or whatever these are) from the IgG, so you could be happy.

You varied two parameters at the same time. In my opinion, there was no need to add the ethanol. The primary concern is ionic interaction with silanols, and that could be fixed with the addition of the NaCl.

Are you injecting the sample dissolved in mobile phase? If not, this could be a problem, if yes, it could be a problem if your purpose is to determine dimer formation.

I would take out the ethanol.

[ScottHorn]

Thanks for the quick reply. I'm going to try it again without the ethanol. Hopefully I can keep the good separation of the aggregates and lose the tailing. I have a few more questions to pose. Why did this change increase the tailing on the monomer peak? Also, when I went to the higher salt/ethanol containing buffer, the area of the aggregate peaks/smears went from around 20 percent to around 10 percent, in addition to separating better. Each sample was from the same vial, so I don't think the aggregate percentage would have changed. So I'm wondering which result is "correct"?
Also, I'm new to this whole high pressure, silica based column thing. With our old superdex columns, we just left the pump running all day, so that we didn't have to wait for the baseline to stabilize each time we wanted to make an injection. Is doing that with my new column going to shorten it's lifespan?

Scott Horn

[HW Mueller]

This is a beautiful example of how cutting corners to save time doesn´t work. As Uwe mentioned one should not change two things at a time when one wants to optimize. Now you have to do another pH series to get the optimum pH without the EtOH (this is also one reason for why I always have shunned adjusting the pH with organic in solution).
If you check the archives you will see that I mentioned an aspect of my experience with Mab, Fab separation on a Super SW. I tried to find out why TFA permanently deteriorated performance, . . . never did find out. The separation of Mab from its Fab was also extremely sensitive toward changes in pH, organic content, ionic strength, and sample solvent (reversibly, though). The changes involved peak broadening more than just tailing. It was so complex that I never had a chance to interpret it all.

[Uwe Neue]

Hi ScottHorn,

I don't have all the answers...
With respect to the simultaneous loss of aggregates as well as good peaks that do not smear: it is possible that in your original analysis you had loosely associated, i.e. physical, aggregates that fell appart during the dilution in LC, or just by exposure to the LC buffer. Now, with the new method, you are only seeing the chemical aggregates.

The tailing of the IgG could either be a version of overload, or a structural chage due to the exposure to the ethanol, assuming that your sample is not dissolved in the ethanol containing mobile phase (which could be counterproductive for the assay that you are trying to do). Removing the ethanol might help. We will see...

[danko]

Another example – in addition to HW Mueller’s poin - of loosing control over the effects is, the fact that the “oldâ€

[ScottHorn]

I understand the perils of varying two parameters at once. This was my first attempt at using a different mobile phase w/ this column. The only difference between the buffer used in the first chromatogram and the buffer used with the "old" superdex column is the pH (lowered to 6.8 from 7.4). This was done because I've read that silica based columns are vulnerable to stationary phase dissolution at pHs higher than 7. The manufacturer advises that buffers up to pH 7.5 are ok, but I wanted to ensure maximum column lifetime. I've not had a chance to do any "fine" optimizations of pH yet.
Based on Uwe Neue's advice I removed the ethanol. I saw the retention times for all peaks decrease by about 0.5 min. The enhanced resolution for the aggregate peaks remained, and the peak symmetry for all peaks was slightly improved.
Does anyone have any response to my earlier question about the safety of leaving the pump running constantly in order to avoid having to wait for the baseline to stabilize before each injection?

[Uwe Neue]

I do not think that running the column around the clock will shorten column lifetime, unless you have a lot of dust etc. in your mobile phase. On the other hand, the mobile phase is rather simple, and I do not see a reason for it to take a long time to give a flat baseline.