Synchronous Fluorescence Spectroscopy to Monitor Relevant Biochemicals over Zika-Virus-Like Particles' Production.

This work evaluates using synchronous fluorescence spectroscopy with chemometric modeling techniques, Partial Least Squares (PLS), and Artificial Neural Network (ANN), to monitor key biochemical parameters during Zika virus-like particle (Zika-VLP) production. The study aimed to predict concentrations of lactate (Lac), glutamine (Gln), glutamate (Glu), ammonium (NH₄⁺), total protein (Tp), viable cell density (Xv), cell viability (Cv), and viral titer (VT) from offline synchronous fluorescence spectra at various wavelength differences (Δλ) captured throughout bioreactor assays. Biochemical parameters were initially monitored using standard methods. Both PLS and ANN models demonstrated good predictive capabilities for several parameters, with ANN generally outperforming PLS in accuracy and lower error rates. The study found that Δλ = 80 nm spectra, both processed and raw, yielded the best results, although the optimal preprocessing method varied by parameter. The model's predictive accuracy was assessed using Mean Relative Error (MRE) in percentage, showing promising results, especially for parameters like glutamate (1.1%), glutamine (1.6%), ammonium (4.0%), and viral titer (5.8%) when ANN was used as a modeling technique. Transmission electron microscopy confirmed the successful production of Zika-VLP, with sizes consistent with previous reports. Synchronous fluorescence spectroscopy, coupled with appropriate spectral preprocessing and chemometric modeling, is a viable technique for monitoring multiple parameters during Zika-VLP production.