Abstract
This thesis used proton transfer reaction mass spectrometry (PTR–MS) complemented with GC–MS for identification purposes to investigate temporal changes in the profile of volatile organic compounds (VOCs) associated with microbial growth. The first model system used consisted of three Pseudomonas strains (P. aeruginosa PAO1, P. aeruginosa ATCC 27853, P. fluorescence MA09) cultivated in Vogel’s broth, supplemented with glucose (0.5 or 1%) and/or protein (egg white powder at 0 or 2%) at 25 °C over 72 h. The glucose or egg white protein influenced the VOCs release profile, particularly for alcohols, esters, aldehydes and ketones. A wider range of metabolites such as alcohols, aldehydes, esters and hydrocarbons were detected in the presence of 1% glucose; in contrast, at 0.5% glucose content, a smaller number of metabolites with different profiles were detected. The lower glucose content (0.5%) also resulted in a different array of compounds as characteristic volatile markers. Such as styrene, benzaldehyde and cresol. The addition of egg white protein resulted in the production of a distinctly different array of VOCs, particularly S-derivatives and cyclic volatiles (i.e. benzene). Esters and fatty acids formation was also influenced as a result of egg white protein supplemented to the medium; e.g., 2-methyl butanoic acid (m/z 103) and m/z 117 (isobutyl acetate, ethyl butyrate), showed increases particularly in the headspace of P. aeruginosa PAO1. In the glucose + egg white protein supplemented media, the production of acetic acid significantly (p < 0.05) increased. Cells near to the end of their growth phase (54, 60 and 72 h), produced a distinctly different array of VOCs compared to those produced in early growth phase as cyclic compounds were detected over early growth phase whereas sulphur derivatives were more common over late growth phase (60 and 72 h). Metabolites produced by the strain ATCC 27853 differed significantly (p < 0.05) from the other tested strains.
In the second model system, temporal changes in the VOCs release profile in the headspace of a psychrotrophic Pseudomonads (Pseudomonas fluorescence MA09) at 4 °C over 38 d were measured . Vogel’s as the basal medium was supplemented with glucose (0 or 0.5 % w/v) and/or sulphur containing amino acids i.e. cysteine (0 or 1% w/v) and/or methionine (0 or 1% w/v). Met and Cys influenced the VOC profile particularly for sulphur volatiles i.e. H2S, methanethiol, dimethyl sulphide and dimethyldisulphide. An alcohol detected in all amino acid-supplemented Vogel’s was 2-ethyl-1-hexanol. Trace level of m/z 59 (acetone) was also detected in the medium containing glucose, Met and Cys. Principal component analysis showed that Vogel’s supplemented with the amino acid, methionine, derived the most variation of the compounds on PC1 plot. The loadings plot also indicated that m/z 35 (H2S), m/z 95 (dimethyl disulfide), m/z 97 (dimethyl furans), m/z 63 (dimethyl sulfide) and m/z 49 (methanethiol) derived the most significant variation along the first principal components. In contrast, the fingerprint of the VOCs from Vogel’s supplemented with 0.5 % glucose and 1 mM cysteine showed a different pattern as plotted on PC2. The addition of methionine to the glucose supplemented medium derived further variation in the volatiles over PC2 plot indicating a further cluster of compounds exist. VOCs emitted from Met and Cys supplemented medium were distributed on PC2 plot whereas VOCs released from glucose and Cys supplemented medium were clustered as a dense cloud on PC2 plot. Results from the above two model systems highlighted how variation in growth phase, carbon and protein content of the medium can influence the fingerprint of bacterial VOCs.
In the third model, ground pork packaged under modified atmosphere (70% CO2, 30% O2) was investigated as a model system. Emphasis was placed on the VOCs profile resulting from microbial and/or biochemical changes using PTR-ToF-MS, with GC-MS/O only being employed in the beginning and at the end point of the trials. Culture dependent methods using designated selective media for targeted species indicated lactic acid bacteria were the dominant microbial species compared to other groups throughout the storage period at 4 °C. Volatiles such as alcohols (ethanol, 1-propanol), ketones (2-propanone, 2-butanone, 2,3-butanedione, 3-hydroxy-2-butanone), acetic acid and aldehydes (acetaldehyde, benzaldehyde) were identified by PTR–ToF–MS and the signal intensity of selected compounds were plotted as well. GS–MS results indicated that aldehydes (i.e., 3-methyl butanal, hexanal and nonanal) were not presented at day 0 but present in 12-day old samples. Alcohols such as ethanol, 1-propanol, 2-ethyl-1-hexanol and 1-hexanol were detected in both fresh (day 0) and 12 day-old samples whereas other alcohols i.e., 1-octen-3-ol, 2-octen-1-ol, 1-pentanol and 2-methylbutan-1-ol were only detected in the 12-day old samples. Cyclic compounds that were detected after 12 day of storage included dimethyl benzenes, benzaldehyde and D-limonene whereas 2-pentyl furan were detected at both day 0 and 12. The identification of flavor-active compounds between fresh (day 0) and stored sample (day 12) indicated a distinction between the VOCs profile by GC-O analysis., Esters and aldehydes were not detected in the fresh samples, however after 12 days of storage at 4°C, fruity, buttery and sweet flavors were the main flavor-active constituents in modified atmosphere packaged raw pork. Such flavor changes were mostly associated with biochemical changes that occur during storage, particularly owing to the growth of LAB rather than to the growth of other microbiota species in pork.
Biochemical changing profile of the sample indicated that the amount of protein hydrazones increased over time. A positive and significant correlation was found between protein carbonyls content and lipid oxidation (R2 = 0.87; p < 0.05). An increase in ferrous myoglobin autoxidation, as well as an increase in MetMb content during extended storage was coincidental with an increase in lipid oxidation and a fall in the redness index. A strong correlation between lipid oxidation (TBARS formation) and the decrease in redness index was also noticeable (R2 = 0.968). The fatty acid profile over the storage period remained unchanged indicating that storage did not show any significant (p < 0.05) effect on the content of each fatty acid. Nevertheless, a concurrent occurrence of myoglobin and lipid oxidations with a progressive formation of metmyoglobin, an increase in TBARS and a decrease in both redness index and the saturation index indicated an association between myoglobin and lipid oxidations. In other words, the autoxidation of myoglobin was associated with an accelerated lipid oxidation and discoloration of the sample as the storage was extended. The VOCs release profile, the proliferation pattern of major spoilage microorganisms during the storage of the meat product along with the biochemical changing profile provided useful information on the mechanism of spoilage in a model system.