Selectivity and Resolving Power of Hydrophobic Interaction Chromatography Targeting the Separation of Monoclonal Antibody Variants

This study presents a comprehensive investigation of the mechanistic understanding of retention and selectivity in hydrophobic interaction chromatography. It provides valuable insights into crucial method-development parameters involved in achieving chromatographic resolution for profiling molecular variants of trastuzumab. Retention characteristics have been assessed for three column chemistries, i.e., butyl, alkylamide, and long-stranded multialkylamide ligands, while distinguishing column hydrophobicity and surface area. Salt type and specifically chloride ions proved to be the key driver for improving chromatographic selectivity, and this was attributed to the spatial distribution of ions at the protein surface, which is ion-specific. The effect was notably more pronounced on the multialkylamide column, as proteins intercalated between the multiamide polymer strands, enabling steric effects. Column coupling proved to be an effective approach for maximizing resolution between molecular variants present in the trastuzumab reference sample and trastuzumab variants induced by forced oxidation. Liquid chromatography–mass spectrometry (LC–MS)/MS peptide mapping experiments after fraction collection indicate that the presence of chloride in the mobile phase enables the selectivity of site-specific deamidation (N30) situated at the heavy chain. Moreover, site-specific oxidation of peptides (M255, W420, and M431) was observed for peptides situated at the Fc region close to the CH2–CH3 interface, previously reported to activate unfolding of trastuzumab, increasing the accessible surface area and hence resulting in an increase in chromatographic retention.


Estimating the Retention Equilibrium Constant (K)
To compare the hydrophobicity of the different columns, we define here the retention equilibrium constant, K. Adopting an adsorption-based retention mechanism and assuming the analyte retention is proportional to the surface area available for retention, we can write: where k' is the retention factor of the analyte, S' the area of the surface upon which retention is occurring, Vm the volume of the mobile phase, and K is the corresponding equilibrium constant given by: where n is number of moles in equilibrium with the concentration in the mobile phase and Cm,e is equilibrium mobile phase concentration.The subscript "e" in each term represents equilibrium.K can be used as an index for hydrophobicity as it defines the number of moles of the analyte retained on the surface area S' in equilibrium with the mobile-phase concentration.The higher its value the more hydrophobic the stationary phase.
To obtain K, we need to determine the k' of an analyte and the S'/Vm ratio of the stationary phase.The k' for phenanthrene was measured in isocratic mode for each column applying a mobile phase of 30:70 (v/v)% acetonitrile: 0.1M ammonium acetate.The column hold-up time (t0) was determined by injecting a water plug.The column was operated at 1 mL/min and the column was thermostatted at 30°C.The S'/Vm ratio can be derived from the relationship mass (m) = density (ρ) × volume (V).For the porous alkylamide and polyalkylimide particles, the S' values were determined by the manufacturer to be 20 m 2 /g, which corresponds to 20 10 -3 m 2 /kgsilica  msilica.Consequently, where the term   (1 −   ) represents the volume of the silica particles excluding the mesopores, Vp is the volume of the particles which is equivalent to   (1 −   ),   is the column volume,  is the porosity inside the particle (~30%), and   is the external porosity (~40%).Porosity values have been obtained Cabooter et al. 1 Considering the density of silica (  = 2.2  10 3 kg/m 3 ), S' becomes: The volume of the mobile phase ( m ) can be expressed as: where (  + (1 −   )    ) corresponds to the total porosity (  ).Therefore, the S'/Vm ratio becomes: S-3 For the nonporous particle (butyl),   = 0, and the ratio S'/Vm is estimated by: By substituting the values of  and   in Eq. ( 6) and ( 7) and then S'/Vm and k' in Eq. ( 1), the estimated values of K were obtained.Values have been reported in Table S1.

Determination of Retention Parameters
The logarithmic retention factor of analytes in HIC can be approximated by the following linear relation based on the LSS model mode: where [M] is the molar salt concentration, ln k'b the extrapolated value of ln k' to aqueous buffer solution.The ln k'b (intercept) and S (slope) values listed in Table S2 were determined by injecting trastuzumab reference sample on to the different columns (butyl, alkylamide, polyalkylimide) while operating the columns in isocratic mode and applying (NH4)2SO4 or NaCl as modifiers, respectively.

RP-HPLC-MS/MS Peptide Profiling of Major HIC Fractions
Initially the glycosylation patterns of fraction 1 (pre peak) and 2 (main peak), in trastuzumab reference sample, were compared, considering the glycopeptide, EEQYN297STYR, which carries this oligosaccharide, see Fig. S1.The standard deviations are low, and the glycosylation pattern is homogenous between both fractions.The dominant glycoforms are of complex type with one or two terminal N-acetylglucosamine (A), zero terminal galactose (G) and core-fucose (F); A2G0F and A1G0F, accounting for approximately 57% and 20% fractional abundance, respectively.The remaining glycoforms were below 10%.In the next step, all possible tryptic peptides of trastuzumab with potential modification sites were screened.The protein sequence of trastuzumab from DrugBank (Accession number: DB00072) is presented below in FASTA format and peptides with modifications potentially are listed in Table S3.
Some peptides contained more than one modification site, all of which were considered in the analysis.( a ) The different peptides considered in this study with potential modification sites highlighted in bold and the position of the amino acid written as subscript.
( c ) No oxidation was detected on the tryptophane residues in these peptides which corresponds to 100% fractional abundance (unoxidized) in all fractions.
Trastuzumab contains the PENNY-motif in the Fc subdomain, which is susceptible to deamidation. 2,3However, we did not observe deamidation at these positions GFYPSDIAVEWESN387GQPEN392N393YK in any of the fractions.For the light-chain peptide ASQDVN30TAVAWYQQKPGK, the fractional abundance of deamidated asparagine in the main peak was 1%, whereas the pre-peak reached 45%, see Fig. S2.

Figure S1 .
Figure S1.Comparison of glycovariants in fractions 1 (black) and 2 (grey) from trastuzumab reference material.LC-MS conditions: peptides were separated on a 75 µm i.d. 150 mm column packed with 2 µm C18 particles (100 Å) utilizing a two-step linear gradient from 1-30% in 35 min and then 30-60% in 10 min (with 0.1% formic acid in acetonitrile).The column was operated at a flow rate of 300 nL min -1 applying 50°C.MS conditions are described in the 'Experimental Section' in the main manuscript.

Figure S2 .
Figure S2.Deamidation level of peptide ASQDVN30TAVAWYQQKPGK in trastuzumab reference material (A) and trastuzumab after oxidation (B).

Table S1 .
Estimated retention equilibrium constant (K) for the different HIC stationary phases.

Table S2 .
k'b and S data determined for the trastuzumab reference sample.

Table S3 .
Screening of post-translational modifications in trastuzumab reference sample and trastuzumab after forced oxidation.