209: A Complementary Chemical and Biological Fingerprinting Approach to Monitor Food Safety, Quality, and Remaining Shelf-Life of Vegetable Juices

Introduction

As the consumption and demand of vegetables and their ready products increases globally, there is a need to develop strategies and tools to report on and improve their safety, quality and stability. Approaches such as chemical fingerprinting (excitation emission matrices or EEMs) and microbial fingerprinting (RNA sequencing) may provide complementary comprehensive tools to scout progressive quality and safety risks in vegetable products.

Methods

Fresh organic baby spinach was blended with deionized water (1:1 w/w) to prepare juice-like products. Each sample was divided into three aliquots (1- untreated (UJ), 2-pasteurized at 72°C for 1 min (PJ), 3- addition of gallic acid (8 mg/mL) (GJ) and stored at 4 and 15°C for 7 days. Samples were collected daily for EEMs (λex=250-530nm, λem=270-750nm, slits=2 and 3 nm) using a fluorescence spectrophotometer. DNA was also extracted from the juices using the Qiagen DNeasy Kit. 16S rRNA sequencing was performed on the Illumina MiSeq and analyzed using bioinformatic pipelines.

Results

Three main EEM regions exhibited the most changes during storage for treated (n=28) and untreated (n=14) samples, whose photophysical properties differed depending on treatment type and testing time. Culture-independent 16S rRNA amplicon sequencing approach yielded more bacterial genera than the culture-dependent approach. Higher abundance of Pseudomonas, Exiguobacterium and Erwinia was observed in untreated samples. The two treatments reduced and/or inhibited Aeromonas, Chryseobacterium, Delftia, Enterobacteriaceae, Exiguobacterium, Paenibacillus, and Sphingobacterium, which contain pathogenic and/or spoilage microbial species. Higher abundance of Bacillus was seen in PJ, while Pseudomonas was more prevalent in GJ. Using an ANN model to analyze the changes in intensity collected from the EEMs (λem 325-700nm @ λex 325nm & λem 420-700nm @ λex 420nm) and microbial growth allowed for the classification of 90% of samples as “fresh” vs. “spoiled.”

Significance

Vegetable spoilage results in extensive food waste and foodborne disease outbreaks have increased lately. PJ and GJ were effective in inhibiting microbial communities and had unique EEMs. Analyzing these chemical and biological fingerprinting data using a machine learning approach (ANN) allowed identifying early spoilage and quality changes in both treated and untreated samples. This may improve food safety, quality and shelf-life reporting.

Authors: Maleeka Singh, Xiaoli Liu, Valeria R. Parreira, Opeyemi U. Lawal, Maia Zhang, Xue Jun, John Shi, Lawrence Goodridge, Maria G. Corradini