(T1530-06-34) Ion-pairing Agent-Free SFC Separation of Polypeptides Using Automated Method Development Tools

Purpose: Method development for reverse-phase LC has long enjoyed the support of coordinated software and hardware, that provides end users assistance with experimental design, method execution, and data analysis. Combined with predictive modeling and QbD robustness software, much effort has been made to streamline this task for HPLC. SFC has long been viewed as a tool of necessity for chiral separations, leading to ongoing real and perceived barriers to wider acceptance of the technique. While predictive modeling for chiral SFC separations is still a future dream, the tools for automated method development and separation design space analysis are now available for analytical SFC. In that context, the growth of biologic therapeutics, including polypeptide and oligonucleotide therapies, presents a new application that can benefit from the use of SFC.  Of particular interest for these classes of biopolymers is the reduction or elimination of the use of ion-pairing agents to provide adequate separation.  Initial investigations have shown that for many polypeptides of various length and composition, TEA and TEAA can be omitted under SFC conditions.  This has the potential to eliminate the need to desalt following traditional reverse-phase purification.

Methods: To take advantage of automated solvent preparation, four modifier solvents were prepared:  100% MeOH (A); 90%/10% MeCN/H₂O (B); 90%/10% MeOH/H₂O (C); and 79.7%/10%/10%/0.3% MeOH/IPA/H₂O/AmOH (D).  By proportioning solvents B, C, and D against solvent A, blends from 1-5% H2O were created for each solvent system.  Blended modifier solvents and neat MeOH were then used in a binary gradient against CO₂, creating 16 defined modifier solvents.  Polypeptide samples were diluted in 50%/50% H₂O/MeCN. The first stage of method development was conducted as a column screening, testing 4 columns (Shim-pack UC PBr, Shim-pack UC Py, Cosmosil Cholester, and Cosmosil Pi-Naphthyl) against 98%/2% MeOH/H2O in a 5-95% linear gradient to look for initial hits. Only one column in this regime, the Shim-pack UC Py (silica-based pyridinyl functional group), indicated some retention and separation for a 4-component polypeptide mix.  The next stage of method development tested the Shim-pack UC Py column against the 16 defined modifier solvents to closely observe the effects of "small percentage" additives, as H2O and AmOH can have a profound effect on separation, even with concentration changes of less than 1%.  Automated method development software prepared batch runs including all necessary equilibration and solvent exchanges for the system and columns. Upon screening the 16 modifiers, the best separation was chosen and promoted to further optimization varying column oven temperature, gradient start and end concentration, and gradient length.  A Box-Behnken model was used in the DOE process, and data collected and evaluated using LabSolutions MD software.

Results: Screening of modifiers showed distinct results, with blends containing acetonitrile providing little or no separation.  Blends including water at 2% or greater exhibited the best overall performance, with increasing water content leading to greater retention.  Likewise, for blends including ammonium hydroxide, retention increased as additive concentration increased. As initial screening was conducted using a 5-95% gradient, idle time at the beginning of the elution was reduced by increasing the initial concentration to 30%.  None of the screened polypeptide samples eluted later than 60% modifier, so the final concentration of the gradient was reduced to 70%.  However, when a second round of polypeptide samples were added, at least one compound was observed to elute during the 70% hold, necessitating an extension of the hold, and a clean-up step to 90% added to avoid on-column carryover.

Conclusion: The use of automated method development tools with SFC simplifies method development, allowing for rapid screening of multiple stationary phases, mobile phase blends, and method conditions.  This approach, particularly the use of automated solvent preparation, can inform separations where small amounts of additives, e.g. +/-1% H2O or +/-0.06% AmOH, have a large effect on observed retention and selectivity.  Creating and testing 16 mobile phases manually would involve multiple batch runs, and hours of additional preparation and use of the instrument software to execute.  Using automated method development tools, allows that workload to be reduced to preparing 4 modifier solvents and configuring a single unattended batch run to acquire the same result. The obtained separation shows that a wide variety of polypeptides of differing length and polarity can be separated without the use of ion-pairing agents under SFC conditions.  This may open the door to SFC purification of these biopolymers that reduces fraction volume, reducing energy and time expense, and may completely eliminate post-purification desalting.  Further research aims to determine the limits of sample loading and fraction collection under similar conditions.

Initial non-optimized column screening hit

SFC separation of polypeptides without ion-pairing agents