Micro simulated moving bed chromatography-mass spectrometry as a continuous on-line process analytical tool

Continuous manufacturing is becoming increasingly important in the (bio-)pharmaceutical industry, as more product can be produced in less time and at lower costs. In this context, there is a need for powerful continuous analytical tools. Many established off-line analytical methods, such as mass spectrometry (MS), are hardly considered for process analytical technology (PAT) applications in biopharmaceutical processes, as they are limited to at-line analysis due to the required sample preparation and the associated complexity, although they would provide a suitable technique for the assessment of a wide range of quality attributes. In this study, we investigated the applicability of a recently developed micro simulated moving bed chromatography system (µSMB) for continuous on-line sample preparation for MS. As a test case, we demonstrate the continuous on-line MS measurement of a protein solution (myoglobin) containing Tris buffer, which interferes with ESI-MS measurements, by continuously exchanging this buffer with a volatile ammonium acetate buffer suitable for MS measurements. The integration of the µSMB significantly increases MS sensitivity by removing over 98% of the buffer substances. Thus, this study demonstrates the feasibility of on-line µSMB-MS, providing a versatile PAT tool by combining the detection power of MS for various product attributes with all the advantages of continuous on-line analytics. Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1007/s00216-023-05023-9.

Table S1: Overview of all model parameters of the SMB process point optimization with CADET-SMB.Table S1 gives an overview of all required input parameters for the process point optimization with CADET-SMB.Fig. S2 shows the single-column desalting experiments of the Mb/Tris system that was used in this study in comparison to the bovine serum albumin (BSA)/ammonium sulfate (AS) system that was applied in a previous study.The peak height of AS was normalized to match that of Tris for better comparability.The retention time difference of the peak maxima is higher for the BSA/AS separation system.In addition, the resolution at the beginning of the peaks is much better.BSA starts to elute before AS, while Mb and Tris almost elute at the same time.The resolution at the beginning of the separation is most important for the raffinate purity.As the desalting performance of the singlecolumn experiment already is much lower for the Mb/Tris system, it is probably not possible to match the µSMB desalting level of the BSA/AS system with further process point optimization.Rather it would be required to change the separation system itself or the µSMB setup instead.Possible approaches are an increased number of columns per zone, longer chromatography columns or the usage of a stationary phase with higher separation performance.Fig. S4 shows the apoMb level and Tris concentration over five consecutive switches in µSMB-MS run 2. The switches were divided into five subsections (24 s per subsection) and the mean detected apoMb percentage was calculated for each subsection.As the interval that was used to calculate the apoMb level is smaller than one switching interval, the influence of the fluctuations in Tris concentration on the detected apoMb level are observable.Despite the differences of the columns, the mean detected apoMb percentage is comparable between the switches for the different subsections.This is favorable for possible on-line analytical applications, as it shows that different time intervals of the SMB process are comparable as long as they are compared to an equivalent time interval of another switch or cycle.
A comparison between different subintervals is not possible, as the detected apoMb percentage differs.Interestingly, there is no direct correlation between the Tris concentration and the detected apoMb level, e.g. the detected apoMb level is highest in the third subsection, while the Tris concentration is lowest in the first one.This again suggests that there are is no proportional relationship between the detected apoMb level and the Tris concentration and further investigation is required to clarify the exact effect.
Figure S1:MS spectra of solutions containing 10 µg/mL Mb and different concentrations of Tris.
Figure S2:Single-column experiments for the separation of Mb and Tris with a Sephadex G10 column and for the separation of bovine serum albumin (BSA) and ammonium sulfate (AS) with a Sephadex G25 fine column.

Figure S3 :
Figure S3: MS signals of Mb, heme and Tris during the third SMB process cycle for each Tris concentration of µSMB run 1.The depicted signals are not normalized with the baseline signal.

Figure S4 :
Figure S4:Detected apoMb level and normalized Tris signal for five consecutive switching intervals of µSMB run 2 (10 mM Tris in feed solution).The apoMb level was calculated over an interval of 24 s, as indicated by the dotted lines.

Fig. S1
Fig. S1 MS spectra of solutions containing 10 µg/mL Mb and different concentrations of Tris

Fig. S2
Fig. S2 Single-column experiments for the separation of Mb and Tris with a Sephadex G10 column and for the separation of bovine serum albumin (BSA) and ammonium sulfate (AS) with a Sephadex G25 fine column.For better comparability, the peak height of the AS peak was normalized to match the Tris peak

Fig. S3
Fig. S3 MS signals of Mb, heme and Tris during the third SMB process cycle for each Tris concentration of µSMB run 1.The depicted signals are not normalized with the baseline signal

Fig. S4
Fig. S4 Detected apoMb level and normalized Tris signal for five consecutive switching intervals of µSMB run 2 (10 mM Tris in feed solution).The apoMb level was calculated over an interval of 24 s, as indicated by the dotted lines

Table S1
Overview of all model parameters of the SMB process point optimization with CADET-SMB.SV: start value; LL: lower limit; UL: upper limit.Fig.S1shows comparably the MS spectra of Mb solutions with different Tris content.For a 1 mM Tris concentration, the highest Tris signal at 593.23 Da is more than ten times higher compared to the Mb signal at 2196.64 Da.