(W1030-11-65) Vicious Variability: Capillary Isoelectric Focusing Method Performance Issues with Non-Detergent Sulfobetaine 211 as an Additive

Purpose: Inclusion of solubilization additives, such as non-detergent sulfobetaines (NDSB), is a common approach to optimize charge variant separation in imaged capillary isoelectric focusing (cIEF) methods. We evaluated NDSB-211as a cIEF assay additive to prevent protein aggregation without imparting significant changes to the viscosity or pH of buffered solutions for certain bi-specific antibody programs during development. Over time, challenges were encountered with this cIEF method and variability in peak resolution as well as unpredictable noise peaks were observed during routine release and stability testing. The root cause of this variability was investigated to better understand how to develop robust cIEF assays for commercial programs.

Methods: All cIEF methods utilized 1% MC, 2M NDSB-211, a ratio of Pharmalytes (either 3-10, 8-10.5, or 5-8), and relevant bracketing pI markers. Internal mAb samples were diluted to 10 mg/mL and treated with carboxypeptidase B (CPB) before analysis. cIEF analysis was performed on a Protein Simple iCE3 whole-capillary imaging system. All detection occurred at 280 nm and data were imported into Empower for further analysis. ^>1H NMR data were collected on a Magritek SpinSolve 80 instrument. Samples were dissolved in D₂O at 0.2M concentration without pH adjustment. Spectra were collected in ¹H decoupled mode, spectra delay was set to 3 seconds, and 4096 scans were collected for each sample at a spectral resolution of 0.1Hz. All NMR data were analyzed using Mestrelab Mnova software and the spectra chemical shift was normalized to the HDO signal in the data using a value of 4.80 ppm. Integration areas were normalized to the resolved NDSB-211 signal at 4.1 ppm and adjusted to 2 protons.

Results: When using NDSB-211 as an additive, great differences in profile resolution and precision across vendors of this molecule were observed. Figure 1A shows the typical charge variant profile of an IgG mAb, and Figure 1B/C demonstrates variability in the charge heterogeneity profile when NDSB-211 is sourced from five different vendors. Sigma Aldrich lot SLBV5379 resolved the two basic peaks in the profile very well in comparison to the expected profile, while the other two Sigma lots resolved Basic Peak 2 as a weak shoulder. Hampton Research Labs, Soltec Bio Science, and Anatrace all demonstrated a poor resolution of the basic peaks to different degrees. It is relevant to note that all other peaks in the profiles resolved consistently across vendors and lots with only the basic peaks affected. The resolution differences among different vendors and lots persisted despite filtering the NDSB-211 solution before adding to the parent solution, using different instruments and capillaries, and mitigating the hydroscopic nature of the additive by desiccation. ¹H NMR was employed to determine if there was complexing of the NDSB-211 additive once solvated or to detect any interfering impurities. Figure 2A displays the hydrogen structure of NDSB‑211 and the resulting NMR spectra for the various vendors. A small peak (not shown) was seen in Soltec, Sigma SLBX6523, and Hampton at 8.0 ppm. The peaks appearing at 3.1 ppm show different splitting between vendors, and such strong splitting in multiplets is not usually a result of proton coupling. The shoulder in Soltec (B) and SLBX6523 (C) and the significant splitting in Hampton (D) indicate an impurity present with hydrogens at a similar chemical shift. Vendor communication revealed the possibility of residual dimethylformamide (DMF) from the purification process, and when tested, chemical shifts were observed in the regions of interest (3.1 ppm and 8.0 ppm) (Figure 3A). To confirm that the residual DMF was impacting the NDSB-211 region, an NMR spiking experiment was performed (Figure 3B). The spectra showed significant overlap and interference of the NDSB-211 profile with the growing DMF peaks; the control NDSB-211 used was determined to have an underlying 2% of DMF present. A spiking experiment was also performed on cIEF (Figure 3C), which resulted in profiles with worsening resolution of basic peaks as the overall percentage of DMF in the parent solution increased. All other peaks remained the same, but the two basic peaks quickly merged into a peak and a shoulder by the 5% DMF level.

Conclusion: DMF is an impurity of the manufacturing process that is not assessed on the Certificate of Analysis for any NDSB. Although not anticipated, a low amount of DMF in the NDSB-211 caused variability and unpredictable resolution issue of mAb charge heterogeneity profiles in cIEF. As this impurity is not controlled by the vendor, inconsistent across vendors and among lots of the same vendor and developing a screening tool to determine DMF concentration for each received NDSB-211 lot before use would increase overall time taken to test sample at manufacturing sites, NDSB-211 was removed from the cIEF method. Methods had to be re-developed under tight timelines using urea as an additive in order to effectively solubilize the same mAbs for cIEF analysis.

Acknowledgements: Internally funded by Johnson & Johnson Innovative Medicine. Thanks to Matthew Myers and Steven LaBrenz.

cIEF of IgG mAb. A) Typical profile of specific IgG mAb; B) Full spectrum of IgG mAb tested with 6 different lots of NDSB-211; C) Zoomed view of panel B, showing differences in Basic Peak resolution of the IgG mAb profile when different lots and vendors of NDSB-211 are used.

1H NMR of different NDSB-211 vendors and lots. A) Structure of NDSB-211; all other panels show resolution of key peaks in 1H NMR spectra for NDSB-211, with highlighted region showing distinct differences between the lots/vendors. Differences in this region indicate an overlapping interference from an impurity.

Confirmation that DMF is present in NDSB-211 to interfere with cIEF profile. A) 1H NMR of DMF; B) 1H NMR experiment spiking increasing levels of DMF into NDSB-211 sample, showing growth of interference peaks in expected region; C) cIEF confirmation experiment spiking increasing levels of DMF into parent solution, showing greater DMF leads to decrease in Basic Peak resolution, reflecting profiles in Graphic 1.