with Chris Gripton Principle Scientist GSK
In his presentation at the TIDES 2021 event, Chris Gripton, principal scientist at GSK started by explaining that in the synthesis of 20-mer strands of anti-sense oligonucleotides (ASO), the impurities generated fall into two categories: those of differing chain lengths (which can easily be separated chromatographically), and those with the same chain length as the API and which contain modifications. The modifications include missing or altered nucleobases, phosphodiester, or phosphorothioate linkages, sugars or protecting groups. These impurities are not as easily resolved and are typically separated using a mass selective detector.
Gripton emphasized that the methodology employed for ASO impurity analysis should be simple and robust, and while a single process is the goal, sometimes multiple simple processes are preferred over a single complex one.
In the case study presented by Gripton, the technology for anti-sense oligonucleotide impurity analysis was preset and used ultraviolet and mass spectrometry liquid chromatography (UV/MS LC) comprising a reverse phase ion pair separation mechanism on a standard c18 column. For screens of alkylamine modifiers a series of acetate buffers with hydrofluoride-isopropanol acid modifier (HFIP) was prepared. Amine choice proved critical for the method, affecting both selectivity and resolution. Use of HFIP in solutions resulted in poor liquid chromatography (LC) resolution of phosphorothioate modified ASOs but provided excellent mass spectrophometry (MS) responses. However, in acetate buffers the opposite response was observed, which mean that use of a single analytical method was not found to be desirable. Therefore UV-LC methodology was developed in buffers containing tributylamine and acetate modifiers while MS methodology initially contained dibutylamine/HFIP. Problems with quantification of abasic impurities and their behaviour on the column were encountered using dibutylamine/HFIP, so ultimately a switch was made to acetate buffers containing tertiary amines at a neutral pH.
In the experiment, HFIP/MS responses were found superior for certain types of amines, even when allowing for charge state distribution effects. Use of 5 mM dibutylamine and 10 mM HFIP (DBA/HFIP) in acetonitrile was initially identified as potential system with excellent MS response and few spectral artefacts. However during pre-validation, when a genuine M-G abasic impurity was used, an unexpected peak profile was observed; the iso-retentive peak was unexpectedly small and two larger peaks appeared that could be attributed to shorter impurities. An older method using UV conditions with tributyl ammonium acetate (TBAA) demonstrated the expected single iso-retentive. This inferred that on-column ASO degradation was occurring in DBA/HFIP buffer.
The abasic impurities are unstable at high pH, monitoring of the eluent revealed the pH of DBA/HFIP buffer was high (9.5) while use the TBAA conditions yielded eluents that were slightly acidic (pH 6.2). Degradation of M-G abasic full length product (FLP) impurity was monitored by measuring UV %area of corresponding abasic shortmer of FLP. In experiments where the pH of the DBA buffer was adjusted by addition of HFIP or Acetic acid, significant degradation was observed at all pH values 7.7 and above. Degradation was found to be temperature sensitive, again indicative of the reaction occurring on the column.
Effect of pH and temperature on Abasic Mass Spectra
Evidence of temperature sensitive degradation was still observed in DBA Acetate buffers but at much lower levels. MS analysis of the degradation products from each the two conditions, DBA/HFIP or TBAA buffer, revealed that in high pH conditions, elimination of the nuclear base and hydration of the sugar occurred while in basic conditions M-G hydrolysis was observed. Because it was
shown that elimination abasics can be formed from hydrolysis abasics they should be quantified as API-related impurities. The observed degradation was not solely due to pH as degradation, and was also observed in acetate conditions even though levels were low. Switching to tributylamine over dibutylamine reduced the occurrence of shorter ASO impurities, thus nucleophilicity and sterics are involved it is not just the basicity of the amine that plays a role. The structural choice of the amine in the buffer impacts the behavior of the abasic impurity on the column. MS results in TBA/HFIP were poor most likely due to wide charge state distributions. A number of different tertiary amines were examined, but very little degradation was observed for any tertiary amine compared to DBA/HFIP. Dimethylhexylamine (DHMA/HFIP) was found to have an excellent MS and UV response but it ultimately was rejected as analytical buffer over for long term sourcing concerns and robustness. Other tertiary amines were found to be more readily available and in high purity.
Major considerations for abasic analytical measurements include: choice of LC eluents -- avoid high pH and primary and secondary amines; and MS source settings -- may elevate elimination impurities, avoid high temperatures and fragmentor settings.
Speaker Chris Gripton, PhD, is Investigator, Drug Substance and Product Analysis UK, CMC Analytical, GSK Medicines Research Centre, United Kingdom
