Hi all,

I am developing a SEC-HPLC method for analyzing the size homogeneity (%monomer, %aggregate) of a purified antibody-drug conjugate. This conjugate is made up of an antibody decorated with a few hydrophobic small molecule drugs.

I've traditionally used Tosoh's G3000SWXL (7.8 x 300mm) column. In our hands, this column gives the best resolution between monomer and aggregate as compared to other columns (Phenomenex, Bio-Rad, Agilent).

Previously, a Tosoh column has given us near 100% recovery (as determined by the area-under-the-curve and extinction coefficient of the conjugate). However, 2 new Tosoh columns we've tried has given us poor recovery. It appears that the conjugate is being irreversibly adsorbed in the column. The first few injections into the Tosoh column gives very bad recovery; the recovery increases with subsequent injections. What's even more disturbing is that the %monomer and %aggregate is a function of the recovery: the higher the recovery, the higher the %monomer.

The mobile phase I've traditionally used is PBS (154mM NaCl).

I've tried the following:

1. Pre-conditioning the column by multiple injections of large amounts of the conjugate.

2. Including 10-20% of MeCN, MeOH, IPA, propylene glycol, and DMSO.

3. Using other silica-based SEC-HPLC columns from Phenomenex, Agilent, and Bio-Rad.

Nothing seems to help. I would like to see 90% or greater recovery.

Any suggestions would be greatly appreciated. I was thinking about trying:

1. Tosoh's "non-hydrophobic" SuperAW polymer SEC-HPLC column. Anyone have any experience with this column?

2. Including some other additives.

3. Decreasing the temperature of the method.

As mentioned previously, TSKgel Super SW 3000 columns have done a tremendous job (do not seem to have any competitors on this) of separating Mab (MW~150000) from its Fab (~100000), except that TFA, etc., caused a permanent deterioration (still no satisfying explanation of this available). Recovery has never changed, but then I wouldnīt consider a 10% lowering in area significant (qualitatively, not mathematically)in this system.
On this subject, I would certainly appreciate to be informed of your calc. method to arrive at the area from the absorption coefficient. Also, how did you get at the absorption coefficient?
Modifyers other than TFA also interfered, partially severely, though they did not permanently change the column. If you want to know what the antibodies do in their natural setting you should stay away from any modifyers, their presence can change the protein severely. What is your phosphate conc. and pH of your PBS?

I am always sceptical when I see chromatography of aggregates, antigen complexes, etc., of antibodies. The energies of associations are generally quite low, in practical terms this means that it would be surprising if they survive the chromatography unchanged. How do you prove that no changes take place? (You mentioned a case of a change taking place). Your small molecule drugs stick with the antibodies?? Do you know how they chromatograph on your column?
On temp.: One may suspect that adhesion to the column increases at lower temp., but proteins show unfamilier entropy effects....you just have to try, but are you really sure that you get adhesion? Also, temp. will have an effect on your equilibria.

Thanks for your reply HW Mueller.
1. I should first clarify something: the small molecule drugs are covanlently conjugated to the antibody. Since the drugs are hydrophobic, they make the entire conjugate more hydrophobic.

2. Determination of size homogeneity (% monomer and % aggregate) by SEC-HPLC is always employed to characterize antibodies during development and manufacturing. As I understand it, the aggregates detected during SEC-HPLC of antibodies are typically irreversible (i.e., energies of associations are high). That is, they are probably formed as a result of denaturation (unfolding), followed by aggregation of the partially-unfolded antibodies. However, I still agree with you that one must be careful when including organic modifiers in the running buffer because it could potentially alter the equilibrium of some reversible aggregates.

3. As to the calculation of recovery by the area-under-the-curve (AUC). First, convert the AUC from units of absorbance*time (e.g., mAU*sec) to AU*mL. Then, by the integrated form of Beers' Law (you can prove this for yourself by integrating the absorbance over time for the entire peak of interest; i.e., obtain the AUC). The integrated form of Beers' Law: AUC = e*m*l, where:

AUC has units of AU*mL
e (extinction coeff.) has units of AU*M-1*cm-1
l has units of cm
m has units of mmoles

In order to solve the above equation for m, you need to estimate the extinction coeff. of your protein of interest. Since I know the amino acid sequence, I simply use the equation found in the excellent paper by Pace et al. (Protein Science, 4:2411-2423, 1995):

e(280nm)=(#Trp)(5,500)+(#Tyr)(1,490)+(#Cystine)(125)

where e(280nm) is the extinc. coeff. at 280nm with units of M-1 cm-1.

Mike

On your point 2.: Did you ever reinject the different fractions, separatly to see whether there are any changes? Do you do light scattering etc., analyses on the proteins? Itīs just that I wonder what MAB we get from industry for very good money.
On your point 3.: Your equation relating the absorption coefficient and peak area seems to be the same as the one we are using (there was a discussion on this), though I am not sure of your reasoning. I asked, because I am also not 100% sure whether we "fudge" the injection amt. correctly. The examples we did worked, but ....
On the estimation of the absorption (extinction) coeff.: We have used an equation with slightly different constants for Trp and Tyr and no Cysteine. Thanks for the ref. Using the peptide bond absorption at 230 nm should give an even better estimation, unfortunatly for 230 nm.

Hi there Dr. Mueller,

Sorry I haven't checked this forum for a long time, so I didn't see your reply.

I haven't collected fractions and reinjected them yet. I think this is a good idea. In parallel, I plan to perform ultracentrifugation sedimentation analysis as an orthogonal method for detection of aggregates. Unfortunately, the mAb I'm working on is VERY far from commercialization; so, not to worry.

I derived the equation for estimating the recovery based on simple calculus. I'm pretty confident in my derivation. I don't understand what you mean by "fudge" the injection amt. correctly. Maybe you can elaborate your concern about the validity of this equation. I know it's tough to explain without writing it down on a piece of paper.

I still haven't solved the problem of recovery yet. However, I've got a few new things to try (e.g., surfactant in the running buffer).

Mike

Mike,

What do your recovery numbers look like if you base them on a flow injection instead of your calculated expected area.

I normally just put a piece of restriction capillary tubing in one position of a column selection valve, then base recovery on the area I obtain from an injection without the column.

Have you done a pH study to determine how pH is affecting aggregation. As you probably already know the aggregation and dissociation kinetics of MAb can be quite complex. You can also have more than one mechanism leading to a single dimer species for example.

Hi Tom,

Recovery as calculated by peak area = 74%
Recovery as calculated by a on-column/off-column method (i.e., the method you mentioned) = 73%

Therefore, the low recovery does not appear to be an artifact due to the way I calculate recovery.

I have not varied the pH of my running buffer. However, I'm fairly certain that the low recovery is directly related to the hydrophobic drugs that I'm conjugating onto my antibody. Therefore, I would not have guessed that pH should make a difference.

Why do you mention pH in particular?

Thanks,

Mike

Hi Mike

I've just recently started to tune into this forum. I've had similar issues with differential recovery of HMW and monomeric species. See my posting dated October 24th. I was wondering if you've had any success solving your recovery issues or did you make any helpful contacts on this topic?

Thanks, Rich