Rosario-Rosario, Arnaldo J.

Profile Picture

Filters

20081
Reset filters

Settings

Citations

Publication Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Evaluation of NIR, FTIR and Raman spectroscopies as monitoring techniques for recombinant GFP protein production in fed-batch cultures of Escherichia coli
    (2008) Rosario-Rosario, Arnaldo J.; Saliceti-Piazza, Lorenzo; College of Engineering; Romañach, Rodolfo; Carlos Sáez, Juan; Department of Chemical Engineering; Orellana, Lynette E.
    The expression of green fluorescent protein (GFP) was induced and monitored in fed-batch cultures of a genetically-engineered Escherichia coli strain. This recombinant E. coli is commercially available (Bio-Rad pGLO kit), and it synthesizes GFP when induced by addition of L-arabinose. The E. coli fed-batch culture considered in this work utilizes glycerol as carbon and energy source and ammonium ion as nitrogen source. Acetic acid (acetate) is produced as a metabolic by-product because of incomplete oxidation of glycerol (Hall et al., 1996). Given that higher concentrations of acetate, ammonium and glycerol can inhibit E. coli growth (can decrease the biomass bench scale concentration), all fed batch fermentations were inoculated to start at high initial cell mass concentrations in close analogy to an industrial setting. After inoculation, all fermentations were monitored and samples collected regularly to be analyzed at line using NIR, IR and Raman spectroscopies. Acetate and glycerol concentrations were measured with an offline HPLC during each batch with very good accuracy (R2 > 92%). To contribute to the monitoring and eventually to bioprocess automation and industrial applications improvements (such as PAT), Raman spectroscopy was used along with chemometric techniques to model and predict analyte concentrations. During the fermentation, samples were collected and analyzed with at-line Raman and the off-line HPLC to establish a correlation. Specifically, to accomplish the correlation, a design of experiments (DOE) was programmed with Minitab (software version 14). A mixture design was used to correlate the analytes concentrations and the Raman spectra using a chemometric techniques, partial least squares (PLS). The correlation between Raman spectroscopy and the analyte mixture showed excellent model results (R2 > 99 % and Q2 > 99 %) with three factors. The correlation between the HPLC and the Raman with fermentation data was determined and validated. Different modes were used, such as crossvalidation (after model was obtained, it was tested by predicting the results with the set of data using in the model) and test set validation (reserve a representative data and predict it with the model). Both validation procedures presented a good model fit (linear y = x relation). The biomass concentration is another critical variable to ensure rigorous fed-batch fermentation. Specifically, the biomass concentrations were measured using two methods: 1) an at-line absorbance method (ultraviolet spectrophotometry) and 2) dry cell mass concentration, or gravimetric determination. Both methods were then correlated with near infrared spectroscopy (NIR). PLS regression presented a good model fit (R2 = 99%, two factors) as well as the validation analysis (y = x). After obtaining a good correlation between different analytical methods used for estimating dry cell mass and by-products concentrations, a GFP protein fluorescence correlation was investigated using flourometry as the reference method and Fourier transform infrared (FTIR) spectroscopy as the primary method. Analysis and PLS modeling resulted in a good fit for GFP concentration (protein concentration is proportional to the fluorescence). FT-IR was used with chemometric techniques to model and predict the recombinant protein concentration. Following available literature, a partial least squares (PLS) regression analysis was used to establish the correlation between fluorescence and the FT-IR spectra (R2 = 99 %, two factors). Also a cross validation and test-set validation were achieved with good linear relationships (y = x). Overall, a substantially better understanding on how we can measure the critical parameters of a recombinant protein fed-batch culture system using spectroscopic techniques and chemometrics was accomplished. Specifically, the ability to accurately determine glycerol, acetate, biomass and recombinant protein concentrations was achieved. More research is needed to test the techniques “in line” to determine if the bench scale experimental approach can be scaled-up to industrial and commercial scale working volumes.