Your results suggest that the mAb is happy in the buffer you have during the UF step, but not the buffer you are concentrating it into for the DF step.

You haven't posted the buffers you have your mAb in and are trying to DF into.

The mAb may be adhering to the membrane in the TFF cassette, if you do a salt step you may get the mAb back off the membrane (or may not).

First, please confirm that the membrane material such as RC or PES, cut-off values, and used buffer compositions must be identical for different process scales. Since UF membranes are designed to observe low protein binding and your experiment which was completed at the lab scale was satisfactory, we may exclude the filter cake formation and protein loss caused by accumulation and/or adhering.

You should also confirm the aggregation problem and solubility issue in the diafiltration process. SEC analysis with a wide range of cut-off resolution may indicate aggregation. During diafiltration, auxiliary agents for mab solubility are removed. Thereby previously enriched highly concentrated mab is more prone to be aggregated and may also lose its solubility during the DF process. This could be reported as low recovery. If possible, perform SEC-UV/MALLS/MS analysis to check MW shifting to higher rates. You may also use a Native-PAGE to monitor if this is proteolytic activity, aggregation, or simply a protein loss...

İsmail Emir Akyildiz thank you for your detail answer. We used regenerated Cellulose (RC) cassette for UF/DF. As you mentioned above we exclude the filter cake formation and protein loss caused by accumulation or adhering because we did not observe any permeate flow drop and pressure increasement.

In response to your buffer questions; during concentration the protein was in 50 mM Sodium acetate + 80 mM NaCl buffer and during diafiltration we used 20 mM L-Histidine buffer and did not add auxiliary agents.

When we checked the SEC results and compared to reference molecule and our product, it is observed that there is no aggregation.

Stanley Ng thank you for your response. It seems that our mAb does not like formulation buffer :) As an answer for you question: during concentration the protein was in 50 mM Sodium acetate + 80 mM NaCl buffer and during diafiltration we used 20 mM L-Histidine buffer.

Are you able to maintain a reduced concentration of NaCl (~25mM) in your final formulation? If your follow on application will permits that is, Abs have a tendency to aggregate in the absence or in low salt concentrations (you'll need to find out if your will tolerate just a bit).

The 20mM histidine buffer is at physiological pH?

We can consider NaCl as an auxiliary agent too that promotes solubility for its salting-in effect. The final formulation may need sufficient ionic strength to aid solubility. As Stanley Ng indicated, the pI of the mab and pH of the final formulation should also be properly matched to maintain the net charge of the protein ionic and thus soluble.

It would be reasonable to follow proteolytic fragments of mab in the final extract. Other than aggregation, within this approach, you can verify the intact mab structure.

For efficient troubleshooting,

I can strongly recommend testing the buffer exchange efficacy at a lab scale by using a desalting cartridge or dialysis bag/tubing...If the same loss is observed, it definitely caused by buffer incompatibility rather than the pressurized, membrane-integrated approaches such as diafiltration/TFF.

Secondly, I suggest you to adding arginine or urea to the formulation buffer during the diafiltration process and observe for any recovery improvement. Any affirmative rate can reflect the solubility issue.

Good luck

Emir

Stanley Ng and İsmail Emir Akyildiz we removed whole NaCl end of the diafiltration ( we checked the conductivity and pH to be sure to remove NaCl) Before TFF experiment we mimicked buffer exchange exp. with centricon and we did not observe any aggregation. And also we did not observe any aggregation with lab scale (88cm2) cassette. Therefore we did not consider Formulation buffer ingredient and pH as problem. Our molecule nearly pI is 8,3, concentrated sample and formulation pH is 6.

We use Sucrose and Polysorbate 20 as an auxiliary agent in formulation buffer. We are not allowed to add NaCl in final product :( Thank you both for your advices and explanation.

Just a random thing which i've seen before, was the protein concentration in well within the range of detection? i.e. measured and then with a 1:10 dil. measured again to ensure that initial read measurement was correct.

I've seen it before where there was too good a yield and people measured conc., too close of limits of the system and on subsequent steps assumed loss, when the 1:10 dil measurement would have given an indication of overestimation and shown no loss.

Stanley Ng To be sure about exact load amount, we checked protein concentration with different dilution rate included 1:10 ratio.

Hi Eda,

It is common to observe loss of protein at low concentrations, when the protein is transferred/subjected to dialysis etc.

That is the reason why we usually use "carrier proteins" to prevent losses of small amounts of proteins under study. So, by taking the protein solution you had through multiple processes, it is not surprising to see losses. Can you include sovents or detergents to prevent the losses?

Best,

Hediye.

Hi Hediye,

Before diafiltration we concentrated the sample hence it is not at low concentration and we think it is very viscous. When we checked the samples with SEC HPLC, we did not observe any aggregation. So we excluded aggregation as a problem.

• What is the material of your membrane? You need a low protein-binding material like PES, not the high protein-binding ones like NC or PVDF.
• Have you tried to stain the membrane for bound protein with CBB or amido black?
• Is there any chance of protein precipitation during filtration? Changes in protein concentration, salt concentration, pH, ...