Case study: Fermentation monitoring for a biopharmaceutical startup working to treat depression

Background

Volatile AI combines purpose built chemical analysis equipment with AI algorithms to achieve 10x faster turnaround in chemical analytics.

The team has recently conducted a trial in monitoring fermenting yeast with a goal to isolate and partially quantify target analytes.

Psylink is an innovative startup developing a process to produce pharmaceutical ingredients in a scalable way to treat depression. Their target compounds include psylocibin and benzothiophenes. The company uses yeast to run a fermentation cycle to synthesize these target compounds. As the company develops the optimal approach for the production of the ingredients, it is crucial to monitor the process and quantify the molecules in the fermentation tank as frequently as possible.

Process control in biopharmaceutical fermentation today

As is typical in the biopharmaceutical industry, the process analytics today involve HPLC or LC-MS. Liquid based chromatographic techniques can be time consuming both from the sample preparation point of view and also during the analysis itself, which takes up to 30 minutes. It can also mean shipping samples to third party labs to run the full testing routine. The turnaround times at external labs can extend into weeks significantly delaying the results.  Such reality is not optimal for production environments as the length of the process feedback loop is directly correlated to how fast a new ingredient can be offered to the market. If that feedback loop extends into days or even weeks, there is a significant delay to the iteration cycle of the fermentation process.

Volatile AI experiment to monitor the fermentation process using a 3-minute test replacing HPLC

Volatile AI ran an experiment to monitor the concentrations of benzothiophene family compounds, used for the treatment of depression. The measurements were taken directly from the fermentation broth at Psylink facilities using three gas phase measurement systems in parallel: gas chromatography – mass spectrometry (GC-MS), ion mobility spectrometry (IMS) and the Scout2 electronic nose (eNose). The results were also validated using conventional LC-MS measurements.

The approach involved taking a sample of the fermenting broth, including the yeast sediment, heating to a set temperature and scraping the headspace off the sample into the gas analysis instrumentation. The samples were collected at different timestamps in the process and transferred frozen to preserve the production environment conditions. The limited sample preparation was intentional in order to simulate production conditions and to test an at-line measurement approach not requiring any specific sample preparation.

The data was collected using a Volatile AI adapted ion mobility spectrometry device, which includes an easy to use headspace sampling cup and a Volatile AI built eNose with PID and MOX sensors. GCMS and LCMS analyses were conducted in partner laboratories.

Result: Ion mobility spectrometry could see the increasing concentrations of the target analyte

The results showed that IMS performed best and could identify the relative intensity of one of the emerging benzothiophene compounds in the fermentation tanks. The GCMS could identify benzothiophene family compounds, but since the analysis was not calibrated, the stability of the relative peaks was not sufficient enough to indicate the evolution in the concentration of the compound. The eNose achieved a reasonable correlation to one of the compound intensities, however during the test suffered in terms of the stability of the results. Finally, LCMS confirmed the emergence of 2 different molecules during the fermentation process and could provide clear relative quantification for them.

Below displays the already processed results of the best performing systems in the test, the ion mobility spectrometry measurements (x axis) compared to the LCMS analysis (y axis), during the progression of the fermenting process:

Conclusions

The results provide a glimpse of new analytical methodologies which can be applied in the field of biopharmaceutical production controls, potentially accelerating time spent on liquid chromatography analyses conducted today. The analysis using an ion mobility spectrometer takes anywhere between 30s to 90s and did not require any sample preparation, therefore significantly accelerating time required vs. HPLC approaches of 15-30min.