Student Number: 201477310

o Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

o NonCommercial — You may not use the material for commercial purposes.

o ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.

How to cite this thesis

Surname, Initial(s). (2012) Title of the thesis or dissertation. PhD. (Chemistry)/ M.Sc. (Physics)/ M.A. (Philosophy)/M.Com. (Finance) etc. [Unpublished]: University of Johannesburg. Retrieved from: https://ujcontent.uj.ac.za/vital/access/manager/Index?site_name=Research%20Output (Accessed: Date).

Metabolomics, Physicochemical Properties and Mycotoxin Reduction of

Whole Grain Ting (a Southern African fermented food) Produced via Natural

and Lactic acid bacteria (LAB) fermentation

A Thesis submitted to the Faculty of Science,

University of Johannesburg, South Africa

In partial fulfilment of the requirement for the award of a

Doctoral Degree in Food Technology

OLUWAFEMI AYODEJI ADEBO

STUDENT NUMBER: 201477310

Supervisor : Dr. E. Kayitesi

Co-supervisor: Prof. P. B. Njobeh

October 2018 EXECUTIVE SUMMARY Drought and challenges related to climate change are some of the issues facing sub-Saharan Africa countries, with dire consequences on agriculture and food security. Due to this prevailing situation, drought and climate resistant crops like sorghum (Sorghum bicolor (L) Moench) can adequately contribute to food security. The versatility and importance of sorghum is well reflected in its use as a major food source for millions of people in sub-Saharan Africa. Recent interest in gluten free food products has equally positioned this crop as a potential substitute for wheat, with numerous studies attesting to this possibility. Although other food processing techniques for the transformation of these sorghum grain exists, fermentation is an important and dominant technique for processing sorghum into other food products particularly in the developing world. Porridge and gruels are commonly and usually consumed in developing countries. A form of this sorghum fermented food is ting, referred to as mabele, bogobe (when processed to hard porridge) or motogo (when processed to soft porridge). It is commonly consumed in South Africa, Botswana and other neighbouring Southern African countries. Although ting is conventionally processed using refined grains, the role of whole grain meals with regards to the provision of desirable health benefits is emphasized in recent literatures. Their composition yields better contents of bioactive, health promoting and beneficial compounds in subsequent products as compared to products from refined grains.

The composition of fermented foods depends on such factors as fermentation time and temperature, meaning that changes and choice of these conditions would influence the quality of the final product. Such differences in fermentation conditions would thus affect ting quality, necessitating an optimization of these conditions. Accordingly, a response surface methodology (RSM) optimization technique was employed in this study to optimize fermentation conditions of ting from two whole grain sorghum types: high tannin sorghum (Avenger) and low tannin sorghum (Titan). The effect of fermentation time and temperature on whole grain-ting parameters such as pH, titratable acidity (TTA), total viable bacteria count (TBC), total lactic acid bacteria count (TLABC), total fungal and yeast count (TFYC), tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC) and antioxidant activities (AA) was investigated. Scanning electron microscopy (SEM) was also used to investigate possible morphological changes in the samples during fermentation.

Results obtained showed that fermentation significantly (p  0.05) influenced all the parameters investigated in both high tannin (HT) and low tannin (LT) ting samples. Higher levels of pH, TTA, TBC, TLABC, TFYC were observed in the LT-ting samples, with lower amounts of TPC, TFC, TNC and AA when compared to HT-ting samples. More visible modifications in morphology were observed for the LT-ting with no considerable morphological changes noted in HT-ting samples. Statistical models generated and validated using different statistical indices [absolute average 2 deviation (AAD), bias factor (퐵푓), accuracy factor (퐴푓) and coefficient of determination (R )] suggested the validity of the models obtained. A multi response numerical optimization (MRNO) technique for production of whole grain (WG) ting with better quality indicated optimal conditions for WG-ting from the HT- and LT-sorghum type as 28 oC for 72 h and 34 oC for 24 h, respectively. Subsequent predicted values confirmed via experiments (in triplicates) showed relatively close experimental values with the predicted ones, further indicating the strength of the models obtained.

The study further investigated the effect of natural (spontaneous) fermentation as compared with a controlled fermentation using two LAB strains namely Lactobacillus fermentum FUA 3165 and L. fermentum FUA 3321 on different ting quality parameters. Optimal fermentation conditions earlier obtained with the multi response numerical optimization were adopted for both the HT- and LT-sorghum types. Both pH and TTA values of naturally fermented WG-ting were low, while accelerated fermentation led to a rapid drop in pH and significantly (p  0.05) higher TTA values, after fermentation with LAB strains. Likewise, a significant (p  0.05) decrease in the TNC, TPC and TFC was observed in LAB-fermented WG-ting samples, suggesting an increased metabolism of phenolic compounds and possible production of compounds that contributed to higher AA. Quantification of some bioactive compounds using liquid chromatography tandem mass spectrometry (LC-MS/MS) showed that WG-ting from the HT-sorghum had higher composition of catechin, quercetin and gallic acid. A combination of the two strains used gave relatively significant (p  0.05) lower values for WG-ting parameters investigated, indicating a negative competitive action between the LAB strains.

Mycotoxin contamination of food crops and persistence in derived food products have been of immense concern due to associated health and economic implications. An effective, safe and practicable way of reducing these mycotoxins in food through fermentation was explored in this study. Although the initial mycotoxin content of the whole grain sorghum types (HT- and LT-)

ii was well below levels regulated in South Africa, fermentation was observed to significantly (p  0.05) reduce these contents. This was particularly more pronounced in LAB fermented WG-ting samples, with over 80% reduction in fumonisin B2 (FB2), T-2 toxin (T-2) and alpha-zearalenol (α-

ZEA). A 98% reduction in fumonisin B1 (FB1) was also recorded with other mycotoxins equally reduced at varying rates. Observations made in this phase of this study correlate with those of the preceding paragraph, indicating the effective potential of these L. fermentum strains as starter cultures for fermentation.

Adopting a metabolomics approach, an untargeted analysis of the metabolites in WG-ting was done using a gas chromatography high resolution time of flight mass spectrometry (GC-HRTOF- MS) system. The raw HT and LT-ting samples were included to understand changes in sorghum metabolome before the formation of WG-ting. Multivariate data analysis using principal component analysis (PCA) effectively separated the tested samples into different clusters, relative to fermentation types (natural and control) and sorghum types. Subsequent orthogonal partial least square discriminant analysis (OPLS-DA) on the dataset revealed significant metabolites contributing to differences observed in both HT- and LT-sorghum samples as groups of acids (methylene cyclopropanecarboxylic acid and 3,4-difluorobenzoic acid, 2,2,2-trichloroethyl), pesticide, phenol and fatty acid ester (FAEs), some which might contribute to reduced fungal load and attendant mycotoxin reduction in both samples as noted. Although OPLS-DA equally revealed variations (majorly reduction of metabolites) as a result of fermentation (natural fermentation, use of L. fermentum FUA 3165, L. fermentum FUA 3321 singly and in combination), different significant metabolites were identified as being responsible for these differences. Differences in raw HT-sorghum and derived ting were due to a phenol, ketone, pesticide, fatty acids and fatty acid esters, with more volatiles metabolites differentiating LT-ting from the raw LT-sorghum. A reduction of 4-chlorobenzonitrile (a pesticide) was however observed in HT-ting derived from fermentation with L. fermentum 3165, indicating the degradative ability of this strain, corresponding to results of mycotoxin reduction. Among the variations observed, only ethanone, 1-(2-hydroxyl-5-methylphenyl) (a ketone) was identified as the only significant (p ≤ 0.05) metabolite common to all the samples, an indication that this is an important volatile compound present in these samples.

iii

Addressing food security will not only be achieved by providing adequate nutrition, but also ensuring these food products contain functional components that can improve health and mitigate certain diseases among consumers. Fermentation, especially with L. fermentum strains was observed to yield a “modified” ting that is deemed safe and rich in beneficial food components. Subsequent adoption and provision of such WG-foods would also contribute to the utilization of sorghum as an indigenous food substrate.

Keywords: Sorghum, ting, optimization, metabolomics, chromatography, mycotoxins, and multivariate data analysis.

DECLARATION I, Oluwafemi Ayodeji Adebo, hereby declare that this study entitled “Metabolomics, physicochemical properties and mycotoxin reduction of whole grain ting (a Southern African fermented food) via natural and lactic acid bacteria (LAB) fermentation” is my work, conducted under the supervision of Dr. Eugenie Kayitesi and Prof. Patrick Berka Njobeh. This study represents my original work and has not been submitted for any degree or examination in any other university. All other sources used, have been duly cited in text and acknowledged by complete references.

______OLUWAFEMI AYODEJI ADEBO

v DEDICATION I dedicate this work to Almighty God for his abundant grace and to my late mum, Mrs. E. O. Adebo, for the sacrifices she made.

ACKNOWLEDGEMENT From his fullness we have all received grace upon grace (John 1:16). To God Almighty I give the glory, honour, praise and adoration for his grace, the privilege, opportunity and strength he gave me throughout the course of this study.

My profound gratitude goes to my family, my late Mum Mrs. E. O. Adebo and my Dad Elder. A. O. Adebo, my loving siblings Opeoluwa Adebo and Mrs. Omolara Adebo-Oloyede. My sincere appreciation also goes to my wife, Janet Adeyinka Adebiyi-Adebo, for her prayers, love, care, support and patience throughout my studies. Even the very little things you did, didn’t go unnoticed. I so much appreciate your prayers, support and encouragement, as without them and God’s grace, I wouldn’t even be here!

My sincere appreciation also goes to my supervisors, Dr. Eugenie Kayitesi and Prof. Patrick Njobeh who generously contributed to the success of this work. I thank you for your time, constructive criticisms, valuable comments, guidance and direction on my research work. My utmost appreciation also goes to Dr. Edwin Madala who “sold” the metabolomics idea to me as well as Dr. Fidele Tugizimana and Prof. Michael Gänzle for their kind advice and direction all along. To Dr. (Mrs) Olabisi Akinlua, I so much appreciate your tutelage and guidance. May God’s overflowing blessings never cease in your lives.

Special thanks also goes to Mr. Lekan Adeyemi, Rtd. Commodore Babatunde and family, the Ores’, the Adeoguns’, Mr. and Mrs. Joseph, Ms. Olajumoke Olutayo, Mrs. Grace Omonua, the Adebiyis’, Mr. Friday and Mama Afrika Msimang for their kind support and motivation. I wish to also appreciate Mr. Sefater Gbashi, Mr. Erick van Zyl, Prof. Green Ezekiel, staff of the Department of Biotechnology and Food Technology, Dr. Opeolu Ogundele, Prof. Derek Ndinteh, Ms. Denise Metcalfe, Mr. Mark Pieterse, the Southern Africa Network of Biosciences (SANBio), the Institute of Food Technologists (IFT), Mr. Alexander Whaley of LECO Pty, Dr Riaan Meyer and Mr Darryl Harris from Shimadzu South Africa for their technical assistance on LC-MS/MS, Dr. Paul Benton of the Scripps Center for Metabolomics, my friends, colleagues for their generous assistance, support and criticism throughout the course of this study. Appreciation also goes to the examiners of this thesis as well as reviewers of the journal articles, book chapters and conference presentations emanating from this study, for their time and constructive criticisms.

vii

For the financial support of this study, I wish to thank the University of Johannesburg (UJ), through the Global Excellence and Stature (GES) Doctoral Fellowship and the National Research Foundation (NRF), South Africa. The study was also supported in part under the NRF Centre of Excellence (CoE) in Food Security co-hosted by the University of Pretoria (UP) and University of the Western Cape (UWC), South Africa, NRF Research and Technology Funding (RTF) and the NRF National Equipment Programme (NEP).

viii

ARTICLES, CHAPTERS, CONFERENCE PRESENTATIONS EMANATING FROM THIS STUDY AND THOSE SUBMITTED Journal articles Adebo, O.A., Njobeh, P.B., Mulaba-Bafubiandi, A.F., Adebiyi, J.A., Desobgo, Z.S.C., Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, e13381, 1-10. Adebo, O.A., Njobeh, P.B., Adebiyi, J.A., Kayitesi, E. (2018). Co-influence of fermentation time and temperature on physicochemical properties, bioactive components and microstructure of ting (a Southern African food) from whole grain sorghum. Food Bioscience, 25,118-127. Adebo, O.A., Njobeh, P.B., Kayitesi, E. (2018). Fermentation by Lactobacillus fermentum strains (singly and in combination) enhances the properties of ting processed from two whole grain sorghum types. Journal of Cereal Science, 82, 49-56. Adebo, O.A., Kayitesi, E., Njobeh, P.B. Reduction of mycotoxins during the fermentation of whole grain (WG) sorghum to ting. To be Submitted. Adebo, O.A., Kayitesi, E., Tugizimana, F., Njobeh, P.B. Differential metabolic signatures in naturally and lactic acid bacteria (LAB) fermented ting, as revealed by gas chromatography mass spectrometry (GC-MS) based metabolomics. To be Submitted.

Book chapters Adebo, O.A., Njobeh, P.B., Adebiyi, J.A., Gbashi, S., Phoku, J.Z., Kayitesi, E. (2017). Food metabolomics (Foodomics), a new frontier in food analysis and its potential in understanding fermented foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M.C. (Ed.). InTech Publishers, Croatia. pp. 211-234. Adebo, O.A., Njobeh, P.B., Adeboye, A.S., Adebiyi, J.A., Sobowale, S.S., Ogundele, O.M., Kayitesi, E. (2018). Advances in fermentation technology for novel food products. In: Innovations in Technologies for Food and Beverage Industries, Panda, S., Shetty, P (Eds.). Springer Publishers, USA. pp. 71-87. Gbashi, S., Madala, N.E., De Seager, S., De Boevre, M., Adekoya, I.O., Adebo, O.A., Njobeh, P.B. The socio-economic impact of mycotoxin contamination in Africa. In: Fungi and Mycotoxins - Their Occurrence, Impact on Health and the Economy as well as Pre- and

Postharvest Management Strategies, Njobeh, P.B., Stepman, F. (Eds.). InTech, Croatia. Accepted, In Press.

Conference presentations Adebo, O.A., Njobeh, P.B., Adebiyi, J.A., Kayitesi, E. (2017). Optimizing fermentation conditions (time and temperature) of sorghum into ting: effects on health promoting and physicochemical properties. An oral presentation at the 2017 Institute of Food Technologists Annual Meeting and Food Expo (IFT17), Las Vegas, USA. 25th – 28th June 2017. Adebo, O.A., Njobeh, P.B., Adebiyi, J.A., Kayitesi, E. (2017). Optimization of fermentation conditions of sorghum for the production of ting: effects on physicochemical properties and microstructure. A poster presentation at the 22nd Biennial International Congress and Exhibition of South African Association of Food Science and Technology (SAAFoST), Cape Town, South Africa. 3rd – 6th September 2017. Adebo, O.A., Kayitesi, E., Adebiyi, J.A., Njobeh, P.B. (2018). Gas chromatography-high resolution time-of-flight mass spectrometry (GC-HRTOFMS) profiling of the volatile constituents in whole grain (WG) ting (a sorghum fermented product). A poster presentation at the 2018 Institute of Food Technologists Annual Meeting and Food Expo (IFT18), Chicago, USA. 15th – 18th July 2018. Adebo, O.A., Adebiyi, J.A., Njobeh, P.B., Kayitesi, E. (2018). Gas chromatography-high resolution time-of-flight mass spectrometry (GC-HRTOFMS) untargeted metabolomics of ting (a sorghum fermented food product). A poster presentation at the 19th World Congress of Food Science and Technology (IUFoST 2018), Mumbai, India. 23rd – 27th October 2018. Adebo, O.A., Adebiyi, J.A., Kayitesi, E., Njobeh, P.B. (2018). The reduction of mycotoxins during fermentation of whole grain (WG) sorghum into WG-ting. A poster presentation at the 19th World Congress of Food Science and Technology (IUFoST 2018), Mumbai, India. 23rd – 27th October 2018.

TABLE OF CONTENT EXECUTIVE SUMMARY ...... i DECLARATION ...... v DEDICATION ...... vi ACKNOWLEDGEMENT ...... vii ARTICLES, CHAPTERS, CONFERENCE PRESENTATIONS EMANATING FROM THIS STUDY AND THOSE SUBMITTED ...... ix TABLE OF CONTENT ...... xi LIST OF ABBREVIATIONS ...... xvii LIST OF UNITS/SYMBOLS ...... xxiii LIST OF FIGURES ...... xxiv LIST OF SUPLEMENTARY FIGURES ...... xxvii LIST OF TABLES ...... xxviii THESIS OUTLINE ...... xxix CHAPTER ONE ...... 1 1.0 GENERAL INTRODUCTION ...... 1 1.1 BACKGROUND ...... 1 1.2 PROBLEM STATEMENT AND JUSTIFICATION OF THE STUDY ...... 3 1.3 HYPOTHESIS ...... 4 1.4 AIMS AND OBJECTIVES ...... 5 1.4.1 Aims ...... 5 1.4.2 Objectives ...... 5 CHAPTER TWO ...... 6 2.0 LITERATURE REVIEW ...... 6 2.1 INTRODUCTION ...... 6 2.2 SORGHUM: AN OVERVIEW ...... 6 2.3 GRAIN STRUCTURE OF SORGHUM ...... 7 2.4 NUTRITIONAL PROPERTIES OF SORGHUM ...... 9 2.5 HEALTH PROMOTING PROPERTIES OF SORGHUM ...... 9 2.5.1 Polyphenols ...... 10 2.5.2 Lipids ...... 12

2.5.3 Starch ...... 12 2.5.4 Protein ...... 12 2.6 WHOLE GRAIN FOODS ...... 13 2.6.1 Health promoting properties of whole grain foods ...... 13 2.6.2 Whole grain sorghum food products ...... 14 2.7 FERMENTATION ...... 15 2.7.1 Natural (spontaneous) fermentation ...... 16 2.7.2 Controlled fermentation ...... 17 2.7.3 LAB fermentation of foods ...... 17 2.8 TING PRODUCTION AND ASSOCIATED CHALLENGES ...... 19 2.8.1 Ting production ...... 19 2.8.2 Challenges associated with ting production ...... 21 2.9 BENEFICIAL HEALTH PROPERTIES OF FERMENTED FOODS ...... 21 2.10 MODIFICATIONS AND CHANGES OCCURING DURING THE FERMENTATION OF SORGHUM ...... 22 2.11 FOOD METABOLOMICS (FOODOMICS) ...... 25 2.11.1 Fundamentals of food metabolomics ...... 26 2.11.2 GC-MS based food metabolomics ...... 36 2.12 MYCOTOXINS ...... 37 2.12.1 Factors influencing mycotoxin production ...... 38 2.12.2 Toxicity of mycotoxins ...... 38 2.12.3 Mycotoxins in sorghum and sorghum based foods...... 39 2.12.4 Mycotoxin control ...... 40 2.12.5 Regulation of mycotoxins in sorghum and sorghum based foods ...... 41 2.13 CONCLUDING REMARKS ...... 42 References ...... 43 CHAPTER THREE ...... 72 OPTIMIZATION OF FERMENTATION CONDITIONS AND ITS EFFECTS ON PHYSICOCHEMICAL PROPERTIES, BIOACTIVE COMPONENTS AND MICROSTRUCTURE OF “MODIFIED” TING FROM WHOLE GRAIN (WG) SORGHUM .72 Abstract ...... 72

xii

3.0 Introduction ...... 73 3.1 Materials and methods ...... 74 3.1.1 Raw material and sample preparation ...... 74 3.1.2 Spontaneous fermentation of whole grain sorghum flour into whole grain ting ...... 75 3.1.3 Optimization of WG-ting production process...... 75 3.1.4 Model validation ...... 76 3.1.5 pH and titratable acidity (TTA) ...... 76 3.1.6 Estimation of viable microbial counts ...... 76 3.1.7 Tannin content, total phenolic content, total flavonoid content, and antioxidant activity assay of the WG-ting samples...... 77 3.1.7.1 Tannin content (TNC) ...... 77 3.1.7.2 Total phenolic content (TPC)...... 77 3.1.7.3 Total flavonoid content (TFC) ...... 78 3.1.7.4 Antioxidant activity (AA) ...... 78 3.1.8 Scanning electron microscopy (SEM) analysis of the WG-ting samples ...... 78 3.1.9 Statistical analysis ...... 78 3.2 Results and discussion ...... 79 3.2.1 pH and TTA of the WG-ting samples ...... 79 3.2.2 Microbial load ...... 88 3.2.3 Tannin content, total phenolic content, total flavonoid content and antioxidant activity ..89 3.2.4 Multi response numerical optimization (MRNO) ...... 91 3.2.5 Scanning electron microscopy (SEM) of the WG-ting samples ...... 91 3.3 Conclusion ...... 93 References ...... 94 CHAPTER FOUR ...... 99 FERMENTATION BY LACTOBACILLUS FERMENTUM STRAINS (SINGLY AND IN COMBINATION) ENHANCES PHYSICOCHEMICAL PROPERTIES, BIOACTIVE COMPONENTS AND ANTIOXIDANT ACTIVITY OF TING PROCESSED FROM TWO WHOLE GRAIN SORGHUM TYPES ...... 99 Abstract ...... 99 4.0 Introduction ...... 100

xiii

4.1 Materials and methods ...... 101 4.1.1 Raw material and sample preparation ...... 101 4.1.2 Lactobacillus strains ...... 102 4.1.3 Fermentation of sorghum into WG-ting ...... 102 4.1.4 pH and titratable acidity (TTA) determination ...... 102 4.1.5 Estimation of viable microbial counts ...... 102 4.1.6 Tannin content (TNC) ...... 103 4.1.7 Total phenolic content (TPC)...... 103 4.1.8 Total flavonoid content (TFC) ...... 103 4.1.9 Antioxidant activity (AA) ...... 104 4.1.10 LC-MS/MS quantification of selected bioactive compounds ...... 104 4.1.11 Scanning electron microscopy (SEM) of the ting samples ...... 104 4.1.12 Statistical analysis ...... 104 4.2 Results and discussion ...... 105 4.2.1 pH and TTA of WG-ting...... 105 4.2.2 Microbial analysis of WG-ting ...... 108 4.2.3 Tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC), antioxidant activity (AA) and bioactive compounds of WG-ting ...... 109 4.2.4 Scanning electron microscopy (SEM) of the WG-ting samples ...... 112 4.3 Conclusion ...... 113 References ...... 114 CHAPTER FIVE ...... 118 REDUCTION OF MYCOTOXINS DURING THE FERMENTATION OF WHOLE GRAIN SORGHUM TO WHOLE GRAIN TING ...... 118 Abstract ...... 118 5.0 Introduction ...... 119 5.1 Materials and methods ...... 120 5.1.1 Raw material and sample preparation ...... 120 5.1.2 Lactobacillus strains ...... 120 5.1.3 Fermentation of sorghum into WG-ting ...... 120 5.1.4 Mycotoxin standards ...... 121

xiv

5.1.5 Mycotoxin extraction ...... 121 5.1.6 LC-MS/MS quantification of mycotoxins ...... 121 5.1.7 Statistical analysis ...... 122 5.2 Results and discussion ...... 122 5.2.1 Presence of mycotoxins ...... 123 5.2.2 Mycotoxin reduction ...... 125 5.3 Conclusion ...... 128 References ...... 128 CHAPTER SIX ...... 134 DIFFERENTIAL METABOLIC SIGNATURES IN NATURALLY AND LACTIC ACID BACTERIA (LAB) FERMENTED TING, AS REVEALED BY GAS CHROMATOGRAPGHY MASS SPECTROMETRY (GC-MS) BASED METABOLOMICS ...... 134 Abstract ...... 134 6.0 Introduction ...... 135 6.1 Materials and methods ...... 136 6.1.1 Raw material and sample preparation ...... 136 6.1.2 Lactobacillus strains ...... 137 6.1.3 Fermentation of sorghum into WG-ting ...... 137 6.1.4 Sample preparation for metabolite profiling ...... 137 6.1.5 GC-HRTOF-MS analysis...... 138 6.1.6 Data processing and statistical analysis ...... 139 6.2 Results and discussion ...... 139 6.2.1 Principal component analysis (PCA) of the GC-HRTOF-MS data set...... 139 6.2.2 Comparison of metabolites in the raw HT- and LT-sorghum samples ...... 141 6.2.3 Comparison of metabolites in raw HT-sorghum and subsequently obtained HT-ting samples ...... 144 6.2.4 Comparison of metabolites in raw LT-sorghum and subsequently obtained LT-ting samples ...... 148 6.2.5 Similarities between the raw-sorghum samples and the subsequently obtained WG-ting samples ...... 152 6.3 Conclusion ...... 153

References ...... 153 CHAPTER SEVEN ...... 165 7.0 GENERAL DISCUSSION, CRITIQUE OF METHODS, CONCLUSION AND RECOMMENDATION ...... 165 7.1 GENERAL DISCUSSION ...... 165 7.2 CRITIQUE OF METHODS ...... 169 7.2.1 Experimental approach ...... 169 7.2.2 Analytical methods ...... 170 7.3 CONCLUSION AND RECOMMENDATION ...... 173 References ...... 174

xvi

LIST OF ABBREVIATIONS 1HNMR Proton nuclear magnetic resonance α-ZEA α-Zearalenol β-ZEA β-Zearalenol AA Antioxidant activity AACC American Association of Cereal Chemists AAD Absolute average deviation ACN Acetonitrile AFs Aflatoxins

AFB1 Aflatoxin B1

AFB2 Aflatoxin B2

AFG1 Aflatoxin G1

AFG2 Aflatoxin G2

AFT Sum of AF (B1, B2, G1 and G2) AME Apparent metabolisable energy AMEn Nitrogen corrected apparent metabolisable energy ARR Apparent recovery rate CAST Council for Agricultural Science and Technology CCA Canonical correspondence analysis CCD Central composite design CE-MS Capillary electrophoresis mass spectrometry CE-TOF-MS Capillary electrophoresis time of flight mass spectrometry CE Collision energy cfu Colony forming units CoE Centre of Excellence CHD Coronary heart disease CIT Citrinin CVD Cardiovascular disease

xvii

DAFF Department of Agriculture, Forestry and Fisheries, South Africa DON Deoxynivalenol EC European Commission EI Electron ionization FAE Fatty acid ester FAEE Fatty acid ethyl ester FAME Fatty acid methyl ester FAPE Fatty acid pentyl ester FAO Food and Agriculture Organization FAOSTAT Food and Agriculture Organization Statistics FBs Fumonisins

FB1 Fumonisin B1

FB2 Fumonisin B2

FB3 Fumonisin B3 FC Fold change FTIR Fourier Transform Infrared Spectroscopy GC-FID Gas chromatography flame ionization detector GC-HRTOF-MS Gas chromatography high resolution time of flight mass spectrometry GC-MS Gas chromatography mass spectrometry GC-TOF-MS Gas chromatography time of flight mass spectrometry GRAS Generally recognized as safe HPLC-DAD High performance liquid chromatography-diode array detector HRT High resolution time of flight HT High tannin HT-2 HT-2 toxin HT0000 Raw high tannin-sorghum HT2872 Naturally fermented ting from HT-sorghum

xviii

HT2STRAINS HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321 HT3165 HT-sorghum fermented with L. fermentum FUA 3165 HT3321 HT-sorghum fermented with L. fermentum FUA 3321 IARC International Agency for Research on Cancer IBS Irritable bowel syndrome IFT Institute of Food Technologists KA Kojic acid LAB Lactic acid bacteria LC-MS Liquid chromatography-mass spectrometry LC-MS/MS Liquid chromatography tandem-mass spectrometry LOD Limit of detection LOQ Limit of quantification LT Low tannin LT0000 Raw low tannin-sorghum LT3165 LT-sorghum fermented with L. fermentum FUA 3165 LT3321 LT-sorghum fermented with L. fermentum FUA 3321 LT3424 Naturally fermented ting from LT-sorghum LT2STRAINS LT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321 MC Metabolite class MeOH Methanol MF Molecular formula MRNO Multi response numerical optimization MVDA Multivariate data analysis NFSS National Food Safety Standard NIV Nivalenol PAT Patulin PCA Plate count agar

xix

PCA Principal component analysis PDA Potato Dextrose Agar PUFA Polyunsaturated fatty acids ME:GE Metabolisable to gross energy ratio MRS de Man Rogosa and Sharpe agar NA Not available ND Not detected NEP National Equipment Programme NMN N’-methylnicotinamide NRF National Research Foundation NSPs Non starch polysaccharides OPLS-DA Orthogonal partial least square discriminant analysis OTA Ochratoxin A OTB Ochratoxin B PBS Phosphate buffer saline PC Principal components PCA Principal component analysis PFTBA Perfluorotributylamine PLS-DA Partial least square discriminant analysis QPS Qualified presumption of safety ROC Receiver operating characteristic RSM Response surface methodology RTF Research and Technology Funding Rt Retention time S/N Signal to noise ratio SAAFoST South African Association of Food Science and Technology SANBio Southern Africa Network of Biosciences SEM Scanning electron microscopy SEM Standard error of mean

xx sp. Specie SPME-GC-MS Solid phase microextraction- gas chromatography mass spectrometry spp. Species SSA sub-Saharan Africa T-2 T-2 toxin TBC Total bacteria count TEM Transmission electron microscopy TFC Total flavonoid content TFYC Total fungal and yeast count TH Trichothecenes TLABC Total lactic acid bacteria count TNC Tannin content TOF-MS Time of flight mass spectrometry TPC Total phenolic content TTA Titratable acidity USAID United States Agency for International Development UJ University of Johannesburg UP University of Pretoria UHPLC-LTQ-IT-MS/MS Ultra high pressure liquid chromatography linear ion trap-high resolution Orbitrap mass spectrometry UPLC-ESI-QTOF-MS Ultra high performance liquid chromatography with electrospray ionization quadrupole time of flight mass spectrometry UPLC-Q-TOF MS Ultra performance liquid chromatography quadrupole time of flight mass spectrometry USAID United States Agency for International Development USFDA United States Food and Drug Administration UV Ultraviolet UWC University of Western Cape VIP Variable influence projection

xxi

WG Whole grain WHO World Health Organization ZEA Zearalenone

xxii

LIST OF UNITS/SYMBOLS

퐴푓 Accuracy factor

퐵푓 Bias factor oC Degree celsius $ Dollar eV Electronvolt g Gram > Greater than ≥ Greater than or equal to h Hour kHz Kilohertz < Less than L Liter m/z Mass-to-charge ratio µl Microlitre µg/kg Micrograms/kilograms µg/mL Micrograms/milliliters µM Micromolar mg/mL Milligram/milliliters min Minute M Molar nm Nanometre % Percentage R2 Coefficient of determination R2X Cumulative variation explained by the models s Second Q2 Variation predicted by the model

xxiii

LIST OF FIGURES Figure 2.1 (A) Sorghum plant, (B) Sorghum seeds ...... 6 Figure 2.2 Map of Africa showing major sorghum producing countries in 2000 (left) and 2014 (right) ...... 7 Figure 2.3 Mature panicles of different types of sorghum in the field. Black (left), white (middle), and red (right) sorghum ...... 8 Figure 2.4 Longitudinal 10-µm bisection of a sorghum grain ...... 8 Figure 2.5 Ting fermentation process ...... 20 Figure 2.6 Various health benefits of fermented foods ...... 22 Figure 2.7 Potential mechanisms by which fermentation influence the composition of food 23 Figure 2.8 Different aspects of food metabolomics ...... 26 Figure 2.9 Diagrammatic workflow of the food metabolomics process ...... 28 Figure 2.10 Factors influencing mycotoxin production ...... 38 Figure 2.11 Strategies to prevent and control mycotoxin contamination...... 40 Figure 3.1 The sorghum grains used in this study (above) and their corresponding whole grains (below). HT- (left) and LT-sorghum (right) sorghum ...... 74 Figure 3.2 Response surface plots showing the effects of fermentation time and temperature on: A ‒ pH, B ‒ TTA (titratable acidity), C ‒ TBC (total bacteria count), D ‒ TLABC (total lactic acid bacteria count), E ‒ TFYC (total fungal and yeast count), F ‒ TNC (tannin content), G ‒ TPC (total phenolic content), H ‒ TFC (total flavonoid content), I ‒ AA (antioxidant activity) of the LT-ting samples ...... 84 Figure 3.3 Response surface plots showing the effects of fermentation time and temperature on: A ‒ pH, B ‒ TTA (titratable acidity), C ‒ TBC (total bacteria count), D ‒ TLABC (total lactic acid bacteria count), E ‒ TFYC (total fungal and yeast count), F ‒ TNC (tannin content), G ‒ TPC (total phenolic content), H ‒ TFC (total flavonoid content), I ‒ AA (antioxidant activity) of the HT-ting samples ...... 85 Figure 3.4 Scanning electron microscopy images of (A) raw HT-sorghum, (B) raw HT- sorghum...... 92 Figure 3.5 Scanning electron microscopy images of LT-ting samples: A ‒ 20 oC 36 h, B ‒ 20 oC 60 h, C ‒ 27 oC 24 h, D ‒ 27 oC 48 h, E ‒ 27 oC 72 h, F ‒ 34 oC 36 h, G ‒ 34 oC 60 h, H – 34 oC 24 h ...... 93

xxiv

Figure 3.6 Scanning electron microscopy images of HT-ting samples: A ‒ 20 oC 36 h, B ‒ 20 oC 60 h, C ‒ 27 oC 24 h, D ‒ 27 oC 48 h, E ‒ 27 oC 72 h, F ‒ 34 oC 36 h, G ‒ 34 oC 60 h, H – 34 oC 24 h ...... 93 Figure 4.1 Effect of fermentation by L. fermentum strains on phenolic compounds of ting from whole grain sorghum. (A) Ting samples obtained from the LT sorghum type; (B) Ting samples obtained from the LT sorghum type. 2872 – naturally fermented ting from HT-sorghum; 3424 – naturally fermented ting from LT-sorghum type; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321 ...... 112 Figure 4.2 Scanning electron microscopy images of ting samples: A ‒ 34 oC 24 h, B ‒ 34 oC 24 h with L. fermentum FUA 3165, C ‒ 34 oC 24 h with L. fermentum FUA 3321, D ‒ 34 oC 24 h with L. fermentum FUA 3165 and FUA 3321, E ‒ 28 oC 72 h, F ‒ 28 oC 72 h with L. fermentum FUA 3165, G ‒ 28 oC 72 h with L. fermentum FUA 3321, H – 28 oC 72 h with L. fermentum FUA 3165 and FUA 3321 ...... 113 Figure 5.1 Reduction of mycotoxin levels in ting from whole grain sorghum. (A) Ting samples obtained from the LT sorghum type; (B) Ting samples obtained from the LT sorghum type. 2872 – naturally fermented ting from HT-sorghum; 3424 – naturally fermented ting from LT-sorghum type; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321 ...... 126 Figure 6.1 Exploratory data analysis with unsupervised chemometric method (PCA) (A) raw sorghum and ting samples. HT0000 – raw HT-sorghum; HT2872 – naturally fermented ting from HT-sorghum; HT3165 – HT-sorghum fermented with L. fermentum FUA 3165; HT3321 – HT-sorghum fermented with L. fermentum FUA 3321; HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; LT0000 – raw LT-sorghum; LT3424 – naturally fermented ting from LT-sorghum; LT3165 – LT-sorghum fermented with L. fermentum FUA 3165; LT3321 – HT-sorghum fermented with L. fermentum FUA 3321; LT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum, (B)

xxv

HT-samples and LT-samples. HTs – HT-sorghum samples; LTs – LT-sorghum samples ...... 140 Figure 6.2 OPLS-DA modelling and variable selection: (A) OPLS-DA score plot separating raw HT- and raw LT-sorghum (R2X = 0.718 and Q2 = 0.977), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; LT0000 – raw low tannin- sorghum...... 141 Figure 6.3 Venn diagram showing comparison between the metabolites of (A) raw and fermented LT-sorghum, (B) raw and fermented HT-sorghum. HT0000 – raw high tannin-sorghum; HT2872 – naturally fermented ting from HT-sorghum; HT3165 – HT-sorghum fermented with L. fermentum FUA 3165; HT3321 – HT-sorghum fermented with L. fermentum FUA 3321; HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; LT0000 – raw low tannin-sorghum; LT3424 – naturally fermented ting from LT-sorghum; LT3165 – LT-sorghum fermented with L. fermentum FUA 3165; LT3321 – LT-sorghum fermented with L. fermentum FUA 3321; LT2STRAINS – LT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum...... 153

xxvi

LIST OF SUPPLEMENTARY FIGURES Figure 6.1 (A) OPLS-DA score plot separating raw HT0000 and HT3165 samples (R2X = 0.693 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model...... 161 Figure 6.2 (A) OPLS-DA score plot separating raw HT0000 and HT3321 samples (R2X = 0.787 and Q2 = 0.999), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 161 Figure 6.3 (A) OPLS-DA score plot separating raw HT0000 and HT2STRAINS samples (R2X = 0.797 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 162 Figure 6.4 A) OPLS-DA score plot separating raw HT0000 and raw HT2872 (R2X = 0.797 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS- DA loadings S-plot, (D) VIP plot for the same model ...... 162 Figure 6.5 (A) OPLS-DA score plot separating raw LT0000 and LT3165 (R2X = 0.581 and Q2 = 0.954), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 163 Figure 6.6 (A) OPLS-DA score plot separating raw LT0000 and LT3321 (R2X = 0.585 and Q2 = 0.966), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 163 Figure 6.7 (A) OPLS-DA score plot separating raw LT0000 and LT2STRAINS (R2X = 0.721 and Q2 = 0.984), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 164 Figure 6.8 (A) OPLS-DA score plot separating raw LT0000 and LT3424 (R2X = 0.446 and Q2 = 0.947), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model ...... 164

xxvii

LIST OF TABLES Table 2.1 Proximate compositions of sorghum reported in literature ...... 9 Table 2.2 Food products from whole grain sorghum ...... 14 Table 2.3 In vivo studies demonstrating the beneficial effects of WG-sorghum diets ...... 15 Table 2.4 Some African fermented food products from sorghum and associated lactic acid bacteria identified...... 18 Table 2.5 Summary of food metabolomic studies on fermented foods ...... 29 Table 3.1 Coded and real values for the Doehlert design ...... 75 Table 3.2 Experimental and predicted values obtained for the parameters investigated in LT- and HT-ting ...... 80 Table 3.3 Coefficient of regression and validation parameters for the different mathematical models obtained for LT- and HT-ting samples ...... 83 Table 3.4 Pearson correlation between investigated parameters in the LT- and HT-ting samples ...... 87 Table 3.5 Predicted and experimental values of the optimal HT- and LT-ting samples ...... 91 Table 4.1 Effect of fermentation by L. fermentum strains on biochemical, microbial and health promoting (bioactive) components of ting from whole grain sorghum ...... 106 Table 4.2 Pearson’s correlation between the investigated parameters of the WG-ting ...... 107 Table 4.3 Identity, properties and optimized parameters of the phenolic compounds investigated in the WG-ting ...... 110 Table 5.1 Identity and characteristics of the mycotoxins investigated on LC-MS/MS ...... 122 Table 5.2 Quantification of mycotoxins in sorghum and reduction after fermentation to whole grain-ting ...... 124 Table 6.1 Significant metabolites contributing to the differences in the raw HT- and LT- sorghum types ...... 142 Table 6.2 Significant metabolites contributing to the differences in the raw HT-sorghum and subsequently obtained HT-ting samples ...... 145 Table 6.3 Significant metabolites contributing to the differences in the raw LT-sorghum and subsequently obtained LT-ting samples...... 149

xxviii

THESIS OUTLINE A brief overview of the chapters presented in this thesis is given as follow:

Chapter One: General Introduction This chapter introduces the subject to be addressed, giving relevant background information and highlighting a statement of the problem to be addressed. The chapter ends with the hypothesis, aims and specific objectives of this study.

Chapter Two: Literature Review This chapter presents an in-depth literature review on key areas of this study, focusing on sorghum, fermentation, metabolomics and mycotoxins. Three book chapters published were produced from this literature review chapter.

Chapter Three: Optimization of fermentation conditions and its effects on physicochemical properties, bioactive components and microstructure of “modified” ting from whole grain (WG) sorghum Chapter Three describes the optimization of fermentation conditions (time and temperature) and their effect on the composition of ting from two whole grain sorghum types. A part of this study has been published in the Journal of Food Processing and Preservation and Food Bioscience.

Chapter Four: Fermentation by Lactobacillus fermentum strains (singly and in combination) enhances the physicochemical properties, bioactive components and antioxidant activity of ting processed from two whole grain sorghum types. The chapter describes the production of whole grain ting with two L. fermentum strains and their effect on the properties of ting from HT- and LT-sorghum types. Findings from this study has been published in Journal of Cereal Science.

Chapter Five: Reduction of mycotoxins during the fermentation of whole grain (WG) sorghum to ting The quantification of mycotoxins in WG-sorghum and subsequent reduction after fermentation into ting are reported in this chapter. This phase of the study will be submitted for publication.

xxix

Chapter Six: Differential metabolic signatures in naturally and lactic acid bacteria (LAB) fermented ting, as revealed by gas chromatography mass spectrometry (GC-MS) based metabolomics The chapter describes the metabolomic profiling of raw sorghum grains and the obtained ting products using GC-MS coupled with multivariate data analysis. An article emanating from this part of the study will be submitted for publication.

Chapter Seven: General discussion and conclusion This is the final chapter in the thesis, which juxtaposes the findings generated in each experimental chapter of the study, provides a brief critique of the analytical and experimental approach used, and a general conclusion drawn from the study. It was finalized with recommendations for future work that should be done on the subject.

xxx

CHAPTER ONE 1.0 GENERAL INTRODUCTION 1.1 BACKGROUND Sorghum is the fifth most important cereal crop in the world after rice, wheat, maize and barley (FAOSTAT, 2017) and the most grown cereal in sub-Saharan Africa (SSA), after maize (Mabhaudhi et al. 2016). Africa is the largest contributor to world sorghum production (Rosentrater and Evers, 2018a), with a production quantity of approximately 29 million tonnes in 2017 (FAOSTAT, 2017). Sorghum therefore remains one of the most versatile crops in SSA where it serves as a major staple and provides the daily diet for over 500 million people in more than 30 countries (DAFF, 2010; FAOSTAT, 2017; Adebo et al., 2018).

Sorghum whole grain (WG) consists of the three main anatomical parts – the pericarp (bran), germ and starchy endosperm, which are present in the same proportions as those in the complete grain (Schaffer-Lequart et al., 2017). Assuming that there is no loss of material, sorghum WG products would be expected to conatin these three anatomical parts in same proportions as in the unprocessed WG. Worldwide consumption of WG-products has tremendously grown due to increasing cognizance of their health-promoting benefits, importance in diets and their protective action against chronic diseases (Jones et al., 2002; McIntosh et al., 2003; Marquart et al., 2007; Mintel, 2011; Seal and Brownlee, 2015; McRae, 2017; Schaffer-Lequart et al., 2017; Lachman et al., 2018; Pang et al., 2018). Compared to refined grains (which has been decorticated and seed coat/pericarp removed), WGs are better sources of fiber, phenolic compounds, phytosterols, vitamins and other important bioactive compounds (Salawu et al., 2013; García-Mantrana et al., 2016; Laureati et al., 2016), thus contributing to increasing consumer interest in subsequently obtained WG-based foods. Such WG-based foods are commonly processed using conventional food processing techniques such as malting, boiling and fermentation.

Among all these food processing techniques, fermentation is one of the oldest, known to improve nutritional qualities, palatability and consumer appeal (Adebo et al., 2017a; Adebo et al., 2017b; Rosales et al., 2018). Subsequently derived fermented food products continue to constitute an important part of daily diet and are estimated to provide about a third of world food supplies (Galati et al., 2014; Adebiyi et al., 2018). Fermented sorghum-based foods have a long history and strong cultural ties with the African people in particular.

Due to the health promoting characteristics of these fermented foods, they are also depicted as being “functional” as they are reported to reduce malnutrition, diarrhea, encourage child growth and development, confer health promoting effects and may possess probiotic properties due to their unique microflora (Taylor et al., 2014; Taylor and Duodu, 2015; Adebo et al., 2017b; Marco et al., 2017). The available sorghum-based fermented foods exist in various forms as beverages, meals, porridges and gruels. According to Rosentrater and Evers (2018a), porridges and gruels are among the most frequently consumed forms of these sorghum-based foods in Southern Africa. One of such popular indigenous porridges is ting, also called leting, mabele, bogobe and motogo among some populated groups. It is a fermented cereal porridge or gruel made from sorghum, popularly known for its sour taste, unique flavor (Taylor and Taylor, 2002; Madoroba et al., 2009; Sekwati-Monang, 2011; Sekwati-Monang and Gänzle, 2011). It is commonly used as a weaning food for infants as well as consumed by adults as meals and during ceremonies (Boling and Eisener, 1982; Madoroba et al., 2009; Madoroba et al., 2011; Adebo et al., 2018).

The production of ting like most other fermented cereal products is usually through natural (spontaneous) fermentation. However, less acidification, variation in quality and inconsistencies in the final composition of the end-product are some challenges associated with spontaneous fermentation (Galati et al., 2014; Adebo et al., 2017b), which to an extent, limits the development of fermented foods. Further to this could be the growth of undesirable species of bacteria, yeasts and fungi that may contribute to the occurrence of pathogens leading to food spoilage. This has thus led to the need and desire for a more controlled fermentation process using starter cultures. Although many African cereal fermentations still largely rely on these spontaneous fermentation processes, the use of starter cultures is desirable to ensure consistency, maintain hygiene, improve quality and guarantee constant sensory appeal (Adebiyi et al., 2018). The use of starter cultures especially lactic acid bacteria (LAB) for cereal fermentation has been a convention over time, since they are dominant in the fermentation microbiota and reported to cause desirable changes in overall composition and sensory qualities (Mugula et al., 2003; Kockova et al., 2011; Galati et al., 2014; Gänzle, 2015; Liptakova et al., 2017; Marco et al., 2017). Subsequent use of these LAB starter cultures has been reported to enhance bioactive composition, contribute to aroma improvement, cause desirable structural changes, formation and metabolism of different compounds (Svensson et al., 2010; Soro-Yao et al., 2014; Adebo et al., 2017b).

The biochemical changes occuring during fermentation result in modification and/or formation of different metabolites. The various available techniques adopted for specific screening of these different analytes in fermented foods are limited by the diverse range of compounds to detect and determine. A comprehensive assessment of fermented foods therefore requires a broad approach of investigating and determining not only already-known constituents, but also considering the possibility of resolving any unexpected occurrence and degradation during fermentation. Concerted efforts at addressing this have led to a new discipline called “Food metabolomics”, “Foodomics” or “omics of food”, which entails the comprehensive profiling and identification of all metabolites in a food sample at a specific point in time (Cifuentes, 2009; Mozzi et al., 2013; Adebo et al., 2017a). Food metabolomics is a promising and valuable technique needed for adequate understanding of fermented foods as it offers vast possibilty of obtaining insight and comprehensive data that can be correlated to the composition of foods.

Despite the enormous potentials embedded in sorghum as a drough resistant crop that can address food insecurity, there have also been concerns about the safety of subsequent food products derived from it. Cereal crops including sorghum are generally susceptible to fungal proliferation along the food chain (Njobeh et al., 2010; Taye et al., 2016; Lahouar et al., 2017). These pervasive toxigenic fungal species produce toxic secondary metabolites some of which are known as mycotoxins (Njobeh et al., 2010; Makun et al., 2012; Bhatnagar-Mathur et al., 2015). By their nature, these toxins extend to subsequent derived food products. Considering the associated economic and health effects of these mycotoxins in food commodities, there is need to explore ways to reduce and/or eliminate them.

1.2 PROBLEM STATEMENT AND JUSTIFICATION OF THE STUDY Product formation during fermentation depends on extrinsic factors such as fermentation time and temperature, making the final quality of whole grain ting significantly dependent on the combination of these conditions. Although different fermentation conditions for preparing ting have been reported in the literature (Taylor and Taylor, 2002; Madoroba et al., 2009; Sekwati- Monang and Gänzle, 2011), there are variations in these fermentation conditions, with no available standardized or optimized condition for ting production from either refined or whole grain sorghum. This therefore necessitates the optimization of the fermentation process to obtain a product of better quality. A widely accepted optimization procedure is response surface

3 methodology (RSM), a statistical and mathematical method for obtaining optimal conditions of factors, for desired parameters (Yolmeh and Jafari, 2017). The Doehlert design of RSM is easily applied to optimize variables more effectively as it can explore the whole of an experimental domain with fewer experiments (Ferreira et al., 2004; Tabaraki and Heidarizadi, 2017).

Ting is usually prepared via a natural (spontaneous) fermentation of the grain substrate or through back-slopping, deeply embedded in tradition and practiced as techniques in Southern African households. The need to upscale ting production and obtain a product of better quality necessitates the use of starter cultures. Lactic acid bacteria are particularly important in this regard due to their important biochemical and physiological features including rapid acidification and their significant role in the development of functional foods (Wu et al., 2015; Liptakova et al., 2017). Although the application of LABs for production of ting from refined sorghum grains has been reported (Sekwati-Monang and Gänzle, 2011), there is currently no study on the application of LABs for ting production from whole grain sorghum and subsequent investigation of their effects on biochemical composition and morphological changes. Further to this is the dearth of vital metabolomic information on the composition and overview of significant metabolites formed after the fermentation (natural and LAB) of whole grain sorghum into whole grain ting.

The consumption of mycotoxin contaminated sorghum-based foods can lead to severe health and economic implications. Strategies for the reduction and/or elimination of these toxins in the food chain is therefore vital to safeguard intending consumers’ health. Several approaches of pre- and postharvest measures have not necessarily met the desired efficacy, safety levels and nutrient retention. Biological detoxification during fermentation especially with starter cultures has been proposed as an effective and acceptable technique able to improve beneficial composition and ensure the safety of food (Karlovsky et al., 2016; Adebo et al., 2017c; Okeke et al., 2018).

1.3 HYPOTHESIS It can be hypothesized in this study that when whole grain sorghum is fermented to form ting, certain biochemical and physicochemical properties of sorghum may be altered, metabolites may be modified, while mycotoxins present in the sorghum grains used to process this product can be reduced or degraded.

1.4 AIMS AND OBJECTIVES 1.4.1 Aims The aims of this study were to i) investigate the biochemical and physicochemical properties of ting obtained via natural and controlled fermentation; and ii) evaluate variations in its metabolite profile and safety thereof.

1.4.2 Objectives To achieve the aims as stated in Section 1.4.1, the objectives of the study set were to: • Optimize the fermentation conditions (temperature and time), for WG-ting production from WG-sorghum and investigate the composition thereof. • Determine the effects of natural and LAB fermentation on the biochemical and health promoting properties (total phenolic content, total tannins, total flavonoids antioxidant activity) of WG-ting • Investigate the effect of natural and LAB fermentation on mycotoxin reduction during the fermentation of WG-sorghum into WG-ting. • Investigate the volatile metabolite variations of the differently processed WG-ting samples using a gas chromatography mass spectrometry (GC-MS) metabolomics approach.

CHAPTER TWO 2.0 LITERATURE REVIEW 2.1 INTRODUCTION This review presents an overview of key variables and concepts adopted. This includes sorghum and a description of its functional and health beneficial components, fermentation, fermented foods and the use of LABs for food fermentation. The review also covers food metabolomics (“omics” of food), specifically highlighting the technique of gas chromatography mass spectrometry (GC- MS) based metabolomics used in this study in addition to mycotoxins as a food safety issue. This chapter is finalized with a concluding remark on the literature reviewed herein.

2.2 SORGHUM: AN OVERVIEW Sorghum (Sorghum bicolor L.) (Figure 2.1) is a drought resistant crop belonging to the Poaceae/Gramineae family (Kimber, 2000; Taylor and Duodu, 2017). It is popularly called amabele, amazinga in South Africa, mtama in Kenya, dura in Sudan, jowar in India and oka baba in Nigeria. Sorghum is an important source of calories for humans by direct intake as dietary foods in the semi-arid regions of Africa and Asia, supplying a variety of nutrients and beneficial food components (Schober and Bean, 2008; Taylor and Duodu, 2017). Although sorghum has numerous food purposes, a substantial part of it is equally used for animal feed, especially in United States of America (Wrigley 2017).

Figure 2.1: (A) Sorghum plant, (B) Sorghum seeds (Adapted from DAFF, 2010)

Sorghum is said to have originated from Africa and Eurasia and considered a tropical cereal crop (Taylor and Duodu, 2017). Though it is the third most produced cereal grain in Africa after maize and rice, it has not been fully utilized for industrial processing as compared to other major cereals. Rapid urbanization, cost of other imported cereal commodities, demand for high quality functional

6 foods and gluten-free meals have however driven the rise in the consumption of sorghum-based food products. Such upsurge is well reflected in the increase of sorghum production quantity in African countries from 18 million tonnes (in 2000) to 29 million tonnes (in 2014) (Figure 2.2). Nigeria remains the largest producer of sorghum in Africa (FAOSTAT, 2017), with Sudan, Ethiopia, Burkina Faso, Niger, Mali and Cameroon, being the other six countries with a production of over one million tonnes (Figure 2.2).

Figure 2.2: Map of Africa showing major sorghum producing countries in 2000 (left) and 2014 (right)

2.3 GRAIN STRUCTURE OF SORGHUM Sorghum is reported to have a remarkable genetic diversity with over 40, 000 reported accessions (Rooney and Awika, 2005). Sorghum grains are single-seeded cereals with their pericarp surrounding and tightly adhering to the seed coat (Eckhoff and Watson, 2009; Taylor and Duodu, 2017). Kernels are usually flattened spheres measuring about 4, 2.5 and 3.5 mm in length, thickness and width, respectively, with an average weight of about 25 mg (Eckhoff and Watson, 2009). Sorghum grains can typically be white, pale orange, tan, red, dark brown and brownish-red (Schober and Bean, 2008; Eckhoff and Watson, 2009; Awika, 2017), but the major commercially available ones are the black, white and red (Figure 2.3).

Figure 2.3: Mature panicles of different types of sorghum in the field. Black (left), white (middle), and red (right) sorghum (Adapted from Awika, 2017)

Sorghum types containing condensed tannins are pigmented and could be divided into types I, II and III relative to their tannin contents (Awika and Rooney, 2004; Awika, 2017; Taylor and Duodu, 2017). The principal structural parts in sorghum is the pericarp (outer layer/bran), germ (smallest inner most part) and endosperm (central starchy and largest part) (Figure 2.4), with the endosperm forming the largest structural component (Schober and Bean, 2008; van der Kamp et al., 2014). The different components entrenched in the sorghum grain confer various nutritional and health beneficial properties. These are discussed in the ensuing sections of this review.

Figure 2.4: Longitudinal 10-µm bisection of a sorghum grain (Adapted from Eckhoff and Watson, 2009)

2.4 NUTRITIONAL PROPERTIES OF SORGHUM In developing countries, especially in Africa, over 78% of the sorghum produced is used for food, with about 14% for animal feeding and 7% for other uses (Batey, 2017). Extensive studies on the composition of sorghum have indicated that the grain is a good source of energy, proteins, carbohydrates, polyunsaturated fatty acids (PUFAs), minerals, vitamins and essential amino acids (Rooney and Serna-Saldivar, 2000; Belton et al., 2006; Taylor and Emmambux, 2008; Afify et al., 2012; Awika, 2017). Proximate composition of sorghum from earlier studies has indicated that its protein content ranges from 7.8 to 12.8%, moisture (8.1–11.2%), carbohydrates (54.6–77.2%), crude fat (2.3–6.9%), ash (0.9–4.2%) and fibre (1.4–26.1%) as presented in Table 2.1.

Table 2.1: Proximate compositions of sorghum reported in literature

Proximate Adebiyi et Shawrang et Shargie Udachan et al. Ndimba et Singh et Vieira composition al. (2005) al. (2011) (2012) (2012) al. (2015) al. (2015a) Queiroz et (%) al. (2015) Moisture 10.66 ND ND 8.10-9.99 8.95-11.16 9.80 ND Protein 9.35 11.80 9.95 8.90-11.02 11.90-12.82 12.5 7.8-19 Carbohydrate 72.41 ND ND 70.65-76.20 ND ND 54.6-77.2 Fat 3.35 6.9 3.32 2.30-2.80 2.37-2.75 3.30 1.6-5.0 Fiber 2.25 19.5 1.83 1.40-2.70 ND 1.7 8.5-26.1 Ash 1.98 4.20 1.44 0.92-1.75 1.61-2.03 1.90 0.9-2.8 ND – not determined

Starch including dietary fiber derived from cellulosic cell wall carbohydrates is a major component of sorghum, constituting about 75% of the grain (Beta et al., 2000; Awika, 2017). Though deficient in lysine, sorghum proteins still contain albumins, prolamins, glutelins and high contents of glutamine and proline (Taylor and Schüssler, 1986; Hamaker et al., 1995; Bean et al., 2016), but do not contain gliadins responsible for gluten allergies (Awika, 2017), thus positioning it a suitable substrate for gluten free foods for individuals with celiac disease. Sorghum is also known to contain fatty acids, policosanols, fatty acid esters and fatty aldehydes (Gouni-Berthold and Berthold, 2002; Carr et al., 2005). With such properties, sorghum can be considered a health promoter. This is discussed subsequently herein.

2.5 HEALTH PROMOTING PROPERTIES OF SORGHUM Increasing evidence presented in the literature has suggested that sorghum grains and its subsequent products are excellent sources of health promoting constituents (Awika and Rooney, 2004; Stefoska-Needham et al., 2015; Taylor and Duodu, 2015; Singhal and Kaushik, 2016;

Awika, 2017). Such studies are appraised in this section focusing on those beneficial compounds in sorghum specifically being targeted for their health promoting properties.

2.5.1 Polyphenols According to Dykes and Rooney (2007), sorghum has one of the widest health beneficial components among other cereals. This is largely attributed to the high polyphenol contents and wide range of phenolic compounds present in it as compared to rye, millet, barley and wheat (Dykes and Rooney 2007; Srianta and Harijono, 2015). These phenolic compounds are mostly concentrated in the outer part of sorghum grain (bran) and majorly serve as a natural defense mechanism against pest and pathogens (Awika and Rooney, 2004; Awika, 2017). Predominant polyphenol compounds in sorghum include flavonoids, condensed tannins (unique to few cereal grains), deoxyanthocyanidins and phenolic acids (Taylor and Duodu, 2015; Awika, 2017).

The phenolic acids in sorghum are the most abundant and characterized group of polyphenols (Awika, 2017). These include the benzoic acids (gallic, p-Hydroxybenzoic, gentisic, salicyclic, vanillic, syringic), cinnamic acids and derivatives (ferulic, caffeic, p-Coumaric, cinnamic, sinapic), glycerol mono- and diesters and phenolic aldehydes (Awika and Rooney, 2004; Dykes and Rooney, 2007; Svensson et al., 2010) all occurring in free and bound forms. While the free ones are located in the outer layer of the pericarp, the bound ones are esterified to the cell walls (Taylor and Duodu, 2015). Likewise, are the flavonoids known to also contribute to pigmentation in sorghum grains and can be grouped into monomeric (flavonones, flavones, flavan-4-ols and 3- deoxyanthocyanins) and polymeric forms (majorly flavan-3-ols) (Mitaru et al., 1984; Awika and Rooney, 2004; Dykes and Rooney, 2007; Awika, 2017).

Sorghum dietary polyphenols are reported to show high-antioxidant capacity when compared to other grains such as rice, millet, maize and wheat (Awika et al., 2003; Dykes et al., 2005; Dykes and Rooney, 2006), particularly attributed to the redox chemistry of sorghum polyphenols (Awika, 2017). In vivo experiments have also suggested that sorghum phenolic extracts exert protective effect to neurodegenerative related diseases in rats (Oboh et al., 2010), confer anti-diabetic effects (Kim and Park, 2012), reduce swelling (oedema) (Burdette et al., 2010) and lower the incidence of oesophageal cancer (van Rensburg et al., 1981). This was also corroborated in other studies

10 reporting anticancer properties in sorghum phenolic extracts (Shih et al., 2007; Awika, 2011; Zhang and Hamaker, 2012; Taylor et al., 2014).

Equally important sorghum polyphenols are tannins, which are high molecular weight polymeric phenolic compounds known to contribute to the color of sorghum grains (Awika and Rooney, 2004; Sieniawska and Baj, 2017). These polyphenolic compounds have molecular weights of between 500-3000 g/mol, containing sufficient hydroxyls and other groups including carboxyl (Ghosh, 2015). Based on their properties and structures, tannins are classified as hydrolysable and condensed (Goel et al., 2005). The hydrolysable tannins are esters of ellagic acid (ellagitannins) or gallic acid (gallotannins) and are readily hydrolyzed by enzymes and acids, while the condensed tannins (called the polymeric proanthocyanidins) are composed of flavonoid units and are generally more abundant than the hydrolysable ones (Awika and Rooney, 2004; Goel et al., 2005; Kruger et al., 2012). The levels of tannin in sorghum has continued to be an issue of public debate among farmers/agriculturists and food scientists/processors/nutritionists. High tannin contents in sorghum reduces pest invasion and seed damage caused by birds and other rodents (Faquinello et al., 2004; Awika, 2017). While high tannin content is a pre-harvest measure to safeguard sorghum seeds, the subsequent seeds are less palatable and nutritive, since tannin decreases starch and protein digestibility (Bhat et al., 1998; Faquinello et al., 2004; Awika, 2017). They can also cause dysfunction of cellular membranes, mineral deprivation, severe gastroenteritis and abdominal pain as well as inhibit nutrient absorption in the colon (Bhat et al., 1998; Goel et al., 2005).

Contrary to its negative impact, tannins may be utilized medicinally as haemostatic, antidiarrheal and antihemorrhoidal compounds (Ashok and Upadhyaya, 2012). They can also heal burns and form a protective layer over an exposed tissue keeping the extent of wound infection to a minimal level (Goel et al., 2005; Ashok and Upadhyaya, 2012). Cork and Foley (1991) also recommended that some tannins may be required dietary components of mammals as it sometimes helps to precipitate, bind and detoxify alkaloids. Tannin-containing sorghum varieties are reported to have stronger in vitro antioxidant capacity, which is attributed to higher free radical scavenging power of tannins as compared to simple flavonoids and other phenolic compounds (Hagerman et al., 1998; Awika et al., 2005; Adebo et al., 2018). Increasing evidence has it that sorghum tannins can be used as ingredients to naturally improve the quality of starch for weight management and prevention of diabetes (Dunn et al., 2015; Amoako and Awika, 2016).

2.5.3 Lipids Sorghum grains have been described as having higher lipid levels compared to other cereal grains (Carr et al., 2005; Hwang et al., 2005). Several studies have shown that these lipids are composed of waxes and oils, with the wax fraction containing 4–8% fatty acids, 37–44% policosanols and 44–55% fatty aldehydes, while the oil fraction contains about 90% triacylglycerols, with oleic, palmitic and linoleic acids being the major ones (Neucere and Sumrell, 1980; Hwang et al., 2002a; Hwang et al., 2002b; Carr et al., 2005; Wang et al., 2007; Lee et al., 2014a; Althwab et al., 2015). Sorghum lipids have gained significant attention over the years due to their role as possible bioactive health agents. Quite high contents of phytosterols and policosanols have been reported in sorghum than for other grains and showed to reduce cholesterol levels (Singh et al., 2003; Leguizamón et al., 2009; Lee et al., 2014a). These sorghum policosanols are quite unique in that they contain both esterified and non-esterified forms (Adhikari et al., 2006).

2.5.4 Starch Starch is a major constituent of sorghum grains mainly comprising of amylopectin and amylose polysaccharides, with the proportion of both polysaccharides affecting rheological properties and starch digestibility. In vitro studies have shown the potential of sorghum-based foods in delivering slowly digested starch, with the possibility of controlling blood glucose levels (Yousif et al., 2012; Khan et al., 2013; Licata et al., 2014). The presence of non-starch polysaccharides (NSPs) in sorghum might also indicate their potential ability of lowering cholesterol levels, improving bowel function and reducing transit time in the small intestine (Topping, 1991; Warrand, 2006). Taylor and Emmambux (2010) however, indicated that sorghum polysaccharides do not confer this health promoting benefits in isolation, but their effects might rather be due to interactions between the polysaccharides and proteins, phenolics and other components in the grain.

2.5.5 Protein Proteins constitute about 8–12% of sorghum grains and are classified as globulins, albumins, glutelins and kafirins (de Mesa-Stonestreet et al., 2010; Taylor and Anyango, 2011). Sorghum grains are equally cited to contain a broad range of bioactive peptides, which are of significant interest due to their beneficial role in different human physiological processes (Lin et al., 2013; Stefoska-Needham et al., 2015). Bioactivities exerted by sorghum peptides include antihypertensive, antimicrobial, antioxidant, anticancer, antiviral, cholesterol-lowering and

12 immunomodulatory effects (Kamath et al., 2007; Stefoska-Needham et al., 2015; Cardoso et al., 2017). These activities are attributed to fractions isolated from the proteins including 2-kDa antiviral peptide, cationic peroxidase, protease, amylase and xylanase inhibitors (Dicko et al., 2006; Camargo-Filho et al., 2008; Sarmadi and Ismail, 2010; Lin et al., 2013; Cruz et al., 2015).

To varying extents, the presence of these aforementioned components can differ in refined grain and whole grain sorghum. A focus on WG-sorghum and its superior benefits over refined sorghum grain is highlighted in the subsequent sections of this Chapter.

2.6 WHOLE GRAIN FOODS According to Singh and Sharma (2017), whole grains are made up of the germ, bran and endosperm and contains all the important parts of the entire grain seed in their original proportions. For refined grains, the refining process removes the most potent protective components of the grains found in the bran and germ, leaving only the starchy-rich endosperm (McRae, 2017). As such, there is increasing evidence that consuming WG-products is much better as compared to their refined counterparts due to the beneficial components embedded in them.

2.6.1 Health promoting properties of whole grain foods The overall benefit derived from three anatomical components of WG (germ, bran and endosperm) all together is much better than any of the individual fractions (Harvard Health Letter, 1999; Schaffer-Lequart et al., 2017). A combination of these components makes WG to contain physiologically important components including vitamins, fatty acids, phytosterols, phytochemicals, fatty acids, dietary fiber, carotenoids, lignans and sphingolipids that can promote health either singly or in combination with each other, which may be synergistic (Jones et al., 2002; Okarter and Liu, 2010; Ye et al., 2012). Available epidemiological evidence and intervention studies has shown that the vital components in WG-diets have positive effects on markers of diseases such as blood pressure, diabetes and obesity (Hallfrisch et al., 2003; Juntunen et al., 2003; Marquart et al., 2003; Behall et al., 2004; Behall et al., 2006; USFDA, 2006; Marquart et al., 2007; Priebe et al., 2008; Maki et al., 2010; Tighe et al., 2010; Ross et al., 2011; Seal and Brownlee, 2015). A series of meta analyses combining the results of multiple scientific studies have equally reported an association between increasing intake of WG-foods and reduced risk of non-communicable diseases such as cardiovascular diseases (CVD), coronary heart diseases

(CHD), stroke (Anderson et al., 2000; Jones et al., 2002; Mellen et al., 2008; Ye et al., 2012), metabolic syndrome (Ye et al., 2010) and cancers (Jacobs et al., 1998; Mourouti et al., 2015).

2.6.2 Whole grain sorghum food products Although refined grains are mostly used in food processing, the use of WGs as staple foods equally has a long history of human consumption (Schaffer-Lequart et al., 2017). Findings from epidemiological studies and discoveries thereof have triggered renewed interest among governmental bodies of different nations that WG should form part of cereal servings (Kantor et al., 2001; Jones et al., 2002; Kyrø et al., 2012).

The incorporation of WG into diet is however, largely influenced by cultural beliefs, disadvantages of longer cooking time, the presence of phytates, tannins and limited variety of products made from them (Llopart et al., 2014). Further to this is that some of their components may adversely affect the functional characteristics, taste, texture and sensory appeal of subsequent formulations. Viable options for addressing this and incorporating WG-sorghum into diet would be through appropriate transformation into various other beneficial food forms which would ensure possibility of obtaining various value-added products. Available technologies for achieving this include extrusion, germination, steeping, baking and fermentation. To date, relatively few studies have investigated the use of WG-sorghum for food (Table 2.2) in contrast to refined grains with relatively more food products. This subsequently necessitates further intensified research on the development of food products from WG-sorghum.

Table 2.2: Food products from whole grain sorghum

Food Type Process Reference Biscuit Snack Baking Dovi (2013) Burukutu Beverage Sprouting & fermentation Ikediobi et al. (1988) Cookies Snack Baking Taylor and Taylor (2011) Cooked WG sorghum Meal Cooking N’Dri et al. (2012) Couscous Meal Steaming Taylor and Taylor (2011) Extrudate Snack Extrusion Dlamini et al. (2007); Llopart et al. (2014) Fermented balls Snack Fermentation & baking Ragaee and Abdel-Aal (2006) Fermented porridge Meal Fermentation Dlamini et al. (2007) Flat bread Meal Baking Yousif et al. (2012) Frybread Meal Baking Rose et al. (2013) Infant formula Weaned food Fermentation Nout (1991) Injera Meal Fermentation & baking Taylor and Taylor (2011) Muffin Snack Baking Poquette et al. (2014) Obushara Beverage Boiling Mukuru et al. (1992) Ogi Porridge Fermentation Akingbala et al. (1981)

Food Type Process Reference Omuramba Beverage Fermentation Mukuru et al. (1992) Pasta Meal Extrusion Khan et al. (2013) Pita bread Meal Baking Ragaee and Abdel-Aal (2006) Ting Porridge Fermentation Kruger et al. (2012); Adebo et al. (2018) Tô Porridge Cooking Bello et al. (1990); Taylor and Taylor (2011) Uji Porridge Fermentation & cooking Taylor and Taylor (2011) Ugali Porridge Boiling Kruger et al. (2012); Muhihi et al. (2012)

WG-sorghum diets are also reported to confer some beneficial effects being anti-diabetic, possessing higher antioxidant properties and better absorption in the digestive system (Wang and Kies, 1991; Kim et al., 2011a; Moss et al., 2017). Documented in vivo studies on these health beneficial effects are summarized in Table 2.3.

Table 2.3: In vivo studies demonstrating the beneficial effects of WG-sorghum diets

Meal Findings Reference Pasta containing WG-sorghum flour Higher plasma polyphenols and antioxidant Khan et al. (2013) capacity Plant protein diet consisting of Significant positive effects on minerals and Obizoba et al. (1979) different sorghum varieties vitamins as compared to the test diet. Ready-to-eat whole ground Decreased fecal transit time, lowered urinary Wang and Kies (1991) sorghum flour NMN excretions and better vitamin absorption. WG-sorghum feed Heavier gizzard and small intestine, with no Fernandes et al. (2013) detrimental effect on broilers Greater responses in AME, ME:GE ratios and AMEn to WGF Moss et al. (2017) WG-sorghum meal Lower glycemic responses Lakshmi et al. (1996) WG-sorghum muffin Reduced glucose and insulin responses in men, Poquette et al. (2014) reduced plasma glucose and insulin iAUC AME – apparent metabolisable energy; ME:GE – metabolisable to gross energy ratio; AMEn – nitrogen corrected apparent metabolisable energy; NMN – N’-methylnicotinamide and WG – whole grain.

2.7 FERMENTATION As earlier indicated in Section 2.6.2, available technologies for processing sorghum into food forms for subsequent consumption include steaming, steeping, germination, milling, boiling and fermentation. Of these processing techniques, fermentation is still regarded as the oldest form of processing sorghum and remains largely significant because of its beneficial functionalities to food (Blandino et al., 2003). It can simply be defined as a desirable biochemical process of intentionally transforming and modifying food substrates into different products through microbial action. Fermentation is usually carried out to purposely enhance properties including aroma, taste, texture, nutritional value and extend shelf life. These effects are induced by microbial actions brought

15 about by biochemical changes, leading to the production of alcohols, organic acids, amino acids, polysaccharides and other metabolites (Singh et al., 2015b; Adebiyi et al., 2018).

The categories of food fermentation include alcoholic, acetic acid, alkali and lactic acid fermentation (Blandino et al., 2003). Alcohol fermentation usually result in the production of ethanol, with yeasts being the predominant microorganisms responsible, while in acetic acid fermentation, alcohols are converted to acetic acids in the presence of excess oxygen (Oliveira et al., 2014). Acetobacter species are the main bacterial producers of acetic acid and thus major microorganisms present during acetic fermentation. Alkali fermentation is not common in cereals, but in fish, seeds and legumes, where Bacillus species usually ferment substrates leading to an increasing alkaline pH (Blandino et al., 2003; Adebo et al., 2017b). Lactic acid fermentation is the most common form of cereal fermentation type and mainly carried out by LABs (Gänzle, 2015; Liptakova et al., 2017). Although LAB fermentation is predominant in cereal fermented foods, the process could either be via spontaneous or controlled fermentation. This is subsequently discussed in the succeeding sections of this chapter.

2.7.1 Natural (spontaneous) fermentation For ages, fermentation has been done naturally otherwise referred to as spontaneous fermentation. This is usually carried out through the sequential action of different endogenous microorganisms present in the substrate. Likewise, is another technique called “back-slopping”, a process of adding a previously fermented product to shorten the fermentation process of a new lot and also to limit the risk of fermentation failure (Galati et al., 2014). During such spontaneous fermentation processes, a competitive action occurs between the plethora of microorganisms, with the best adapted one(s) eventually dominating the microbiota. This however, results in slow fermentation, less acidification, failure in the fermentation process, development of undesirable pathogenic microorganisms, production of undesirable product and metabolites as well as variations in sensory qualities of the final product (Galati et al., 2014; Adebiyi et al., 2018). Although most indigenous fermentation processes still largely rely on natural fermentation, a more controlled fermentation process is desirable to ensure consistency, maintain hygiene, improve quality and guarantee constant sensory quality and composition of fermented foods.

2.7.2 Controlled fermentation While the importance and cultural values of natural fermentation cannot be disregarded, the use of specific starter strains for fermentation is vital, especially when precise sensory, nutritional, technological properties, safety and other specific standards are desired. This has thus necessitated the selection, identification of specific strains (starter cultures) and use of such organisms as starter cultures in controlled fermentation process. As such, various studies over the years have investigated the dominant strains in sorghum fermented foods and subsequently isolated, purified, characterized and preserved these microorganisms with an objective of using them to obtain final fermented products with desired characteristics (Hamad et al., 1997; Madaroba et al., 2009; Correia et al., 2010; Yousif et al., 2010; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011; Marsh et al., 2014; Rao et al., 2015; Ogunremi et al., 2017). Reports from such studies have shown the prevalence of LABs including Lactobacillus amylovorous, L. pentosus, L. plantarum, L. brevis, L. paracasei, L. rhamnosus, L. curvatis and L. fermentum during sorghum fermentation. This is discussed in the ensuing section of this review based on the fact that it is of interest in this study.

2.7.3 LAB fermentation of foods Lactic acid bacteria are generally classified as gram-positive microorganisms and are known to be non-motile, non-spore forming, acid tolerant and rod- or coccus-shaped (Oliveira et al., 2014). They are microaerophilic organisms and their growth is dependent on available sugars, with other metabolic requirements including purines, pyrimidines, vitamins and amino acids (Collins et al., 2010; Valdez et al., 2010; Reis et al., 2012). This group of bacteria ferment carbohydrates to produce various end-products including lactic acid, CO2 and ethanol (Oliveira et al., 2014). Most LABs dominate the microbiota of most fermented foods through exploitative competition and inhibition of other microorganisms through rapid utilization of carbohydrates and accumulation of organic acids (Gänzle, 2015). In the presence of abundant fermentable carbohydrates, lactate is the major product of metabolism. The proteolytic activity of LABs also allows them to degrade peptides and proteins, generating metabolites that contribute to antimicrobial activity, flavor, texture and structure of different foods (Gänzle et al., 2008; Rathore et al., 2012).

According to Soro-Yao et al. (2014), LABs have been utilized in the production of fermented foods for decades and known to positively contribute to the final composition of foods. They are

17 listed under the generally recognized as safe (GRAS) and qualified presumption of safety (QPS) status, with regards to their use as starter and/or protective culture as well as food supplements (Collins et al., 2010; Bourdichon et al., 2012). They are technologically interesting as they produce a wide variety of metabolites including ethanol, acetic acid, aromatic compounds, several enzymes and bacteriocins (Todorov and Holzapfel, 2015; Adebo et al., 2017b). LABs also ensure food safety by decreasing pH through the production of acetic, propionic and lactic acid as end products which lead to a subsequent inhibitory effect on the growth of other pathogenic microorganisms (Leroy and De Vuyst, 2004; Sekwati-Monang and Gänzle, 2011; Russo et al., 2017). As stated by LeBlanc et al. (2011; 2013), LABs are vitamin producing microorganisms and thus provide an economical and natural alternative to fortification with synthesized vitamins. Acetate production during LAB fermentation also impacts food quality, increases redox potential, antioxidant capacity, antimicrobial properties and improves the flavor of the fermented product (Gänzle, 2015).

The prevalence of LABs in the various documented African sorghum fermented foods (Table 2.4) can be attributed to their versatile carbohydrate metabolism (Galati et al., 2014; Soro-Yao et al., 2014) and superior adaption of these strains to sorghum. Subsequent use of LABs during sorghum fermentation has been reported to increase acidification, accelerate the fermentation process, improve functionality, nutritional quality and health promoting components (Svensson, et al., 2010; Sekwati-Monang and Gänzle, 2011; Ray and Joshi, 2015). Using these LAB starter cultures can therefore positively impact on the quality of fermented sorghum products such as ting as these specific strains can provide the desired technological, nutritional and health qualities. This is further explicated subsequently in this chapter.

Table 2.4: Some African fermented food products from sorghum and associated lactic acid bacteria identified

Product name Country Product use LAB identified Reference Bushera Uganda Beverage L. brevis, L. delbrueckii, L. Marsh et al. (2014); paracasei, L. plantarum Mwale (2014) Chibuku Zimbabwe Alcoholic beverage Lactobacillus spp. Gadaga et al. (1999); Togo et al. (2002); Chingwaru and Vidmar (2017) Gowe Benin Porridge L. fermentum, L. mucosae Vieira-Dalodé et al. (2007); Adinsi et al. (2014)

Product name Country Product use LAB identified Reference Hulu mur Sudan Gruel Lactobacillus spp. Adams (1998) Hussuwa Sudan Porridge L. fermentum, L. Yousif et al. (2010); saccharolyticum Mwale (2014) Ikigage Rwanda Alcoholic beverage L. fermentum Lyumugabe et al. (2010) Kisra Sudan Pancake, flat bread L. amylovorus, L. brevis, L. Mohammed et al. confusus, L. fermentum (1991); Hamad et al. (1997); Ali and Mustafa (2009) Khamir Sudan Bread L. brevis, L. cellobiosus Gassem (1991) Mahewu South Africa Porridge gruel L. brevis, L. bulgaricus, L. Hesseltine (1979); delbruckii Kayitesi et al. (2017) Nasha Sudan Infant food Lactobacillus spp. Graham et al. (1986) Ogi West Africa Gruel L. acidophilus, L. agilis, L. Omemu and cellobiosus, L. confusus, L. Bankole (2015) murinus, L. plantarum Pito Nigeria Alcoholic beverage L. delbrueckii, L. fermentum Sawadogo-Lingani et al. (2008); Ajiboye et al. (2014) Tchapalo Ivory Coast Alcoholic beverage L. brevis, L. cellobiosus, L. Djè et al. (2008); coprophilus, L. fermentum, L. N’guessan et al. hilgardii, L. plantarum (2011) Ting Southern Africa Porridge L. casei, L. coryniformis, L. Madoroba et al. curvatus, L. fermentum, L. (2009); Madoroba et harbinensis, L. parabuchneri, al. (2011); Sekwati- L. plantarum, L. reuteri and L. Monang and Gänzle, rhamnosus (2011); Adebo et al. (2018) Umqombothi Southern Africa Beverage L. plantarum, Lactobacillus Katongole (2008) spp. LAB – lactic acid bacteria

2.8 TING PRODUCTION AND ASSOCIATED CHALLENGES Porridges made from cereals such as sorghum are among the most important dishes consumed by the people living in SSA. Both thick and thin porridges are made basically differing in the flour/water ratio required (Rosentrater and Evers, 2018b). As earlier indicated in Section 1.1, ting is a traditional, indigenous sour fermented food product, common in different Southern Africa settlements. It is usually consumed by all age groups as breakfast, lunch and supper and has a distinctive sour taste, reminiscent of yoghurt (Franz, 1971; Mavhungu, 2006).

2.8.1 Ting production Ting is conventionally prepared using traditional and natural (spontaneous) fermentation, brought about by the action of the endogenous microflora of the sorghum and those associated with the preparation equipment and local environments (Madoroba et al., 2011). As a result of this,

19 preparations vary greatly with respect to taste, acceptability and overall product quality (Mavhungu, 2006). Conventionally, sorghum slurry is prepared by thoroughly mixing sorghum flours with luke-warm water (1:1, w/v) in a container (calabash, plastic bucket or clay pot) (Figure 2.5). The container is then covered and left to incubate for 1-3 days to form ting (Franz, 1971; Mavhungu, 2006; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). Alternatively, using the back-slopping technique, sorghum slurries are inoculated with remnants from a previous fermentation or in previously used containers (Sekwati-Monang and Gänzle, 2011). Porridges are subsequently prepared from the ting by cooking the soured slurry in boiling water (kindly refer to Figure 2.5).

Figure 2.5: Ting fermentation process (Adapted from Sekwati-Monang and Gänzle, 2011).

Most studies presented in the literature on the fermentation of ting from sorghum have been based on the characterization of the microbial flora responsible for its production (Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). These studies have isolated, characterized and identified LABs during spontaneous fermentation of sorghum into ting and shown that the most dominant LAB species are L. fermentum, L. rhamnosus and L. plantarum (Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). Other predominant microorganisms reported in these studies include Enterobacteriaceae, Enterococcus mundtii, E. faecalis, Lactococcus lactis, L. curvatus, L. reuteri, L. harbinensis, L. parabuchneri, L. casei, L. coryniformis and Weissella cibaria.

2.8.2 Challenges associated with ting production The production of ting from spontaneously fermented sorghum has continually resulted in immense variations in its characteristics making its commercialization a daunting task (Madoroba et al., 2011). The traditional process is laden with challenges especially the variations in fermentation conditions, which influence the microbial interactions, composition and safety of the final ting. In such traditional settings, fermentation temperature and time are inconsistent, and different fermentation containers are sometime used with hygiene of these containers and other utensils rarely checked.

The fermentation process under such conditions is particularly unpredictable and unrepeatable. There is thus no standardized scientific method for producing ting. A combination of fermentation conditions (time and temperature) and ratio of sorghum grain to water are based on human judgement, causing variation in fermentation processes. This results in obtaining products of varying texture, taste, sourness and aroma (Mavhungu, 2006). This might further extend to the final nutritional composition and health promoting properties of ting as well as the safety and health risks posed by its consumption, necessitating an effective optimal technique of obtaining ting.

2.9 BENEFICIAL HEALTH PROPERTIES OF FERMENTED FOODS Although the benefits of WGs have been highlighted in the earlier part of this chapter (Section 2.6), this section reviews available studies in the literature with regards to the valuable properties imbedded in fermented foods. Studies presented on some indigenous fermented foods have shown that they have immense therapeutic and functional properties, possessing antimicrobial, antioxidant, probiotic and cholesterol-lowering attributes, and some other important bioactive compounds (Tamang, 2007; Farhad et al., 2010; Marco et al., 2017). These foods are in fact considered as sources of medical therapy for humans. They are also known to contain beneficial cultures, which translate to improved physiological benefits and better nutrition (Farhad et al., 2010; Sanlier et al., 2017). According to Marsh et al. (2014), fermented foods also provide crucial beneficial effects through direct microbial action and production of metabolites and other complex compounds. Human trials have also demonstrated that fermented foods aid in the treatment of irritable bowel syndrome (IBS), alleviate constipation (Tabbers et al., 2011), have modulatory effects on the brain and anti-cancer potentials (Kumar et al., 2012; Tillisch et al., 2013).

Fermented foods are also reported to contain high vitamin levels (Hugenholtz, 2013), being excellent sources of bioactive peptides, beneficial organic acids and functional components such as phenolic compounds and flavonoids, which can affect oxidative stress, hyperglycemia and inflammation (Svensson et al., 2010; Wang et al., 2013; Marsh et al., 2014; Taylor and Duodu, 2015). Studies have equally shown that sorghum fermented foods have probiotic and prebiotic properties (Willumsen et al., 1997; Taylor and Duodu, 2015). Owing to bacteriocin action, fermented sorghum reportedly showed growth inhibition against Bacillus subtilis, Escherichia coli, Campylobacter jejuni, Listeria monocytogenes, Shigella flexneri, Salmonella typhimurium and Staphylococcus aureus (Svanberg et al., 1992; Kingamkono et al., 1995; Svensson et al., 2010). Further health benefits of these fermented foods are summarized in Figure 2.6.

Figure 2.6: Various health benefits of fermented foods (Adapted from Farhad et al., 2010)

2.10 MODIFICATIONS AND CHANGES OCCURING DURING THE FERMENTATION OF SORGHUM As earlier indicated in Section 2.7, fermentation results in changes and subsequent improvement in nutritional qualities, taste, shelf life, aroma and modification in structure (Kohajdova and Karovicova, 2007). This process usually causes an increase and/or decrease in the level of

22 compounds and modification of some constituents (Figure 2.7). Detoxification of toxigenic compounds (including mycotoxins) and a decrease in trypsin, amylase inhibitory activities, tannins and phytates through enzymatic degradation have also been reported (Blandino et al., 2003; Gee and Harold, 2004; Badau et al., 2005; Kohajdova and Karovicova, 2007; Osman, 2011; Taylor and Duodu, 2015; Karlovsky et al., 2016; Awika, 2017; Okeke et al., 2018).

Figure 2.7: Potential mechanisms by which fermentation influence composition of food (Adapted from Poutanen et al., 2009)

According to Kohajdova and Karovicova (2007), fermentation can have multiple effects on primary substrates including carbohydrates and proteins, leading to modifications in the levels and formation of subsequent monomers or polymers (Correia et al., 2005; Kohajdova and Karovicova, 2007; Schons et al., 2011; Schons et al., 2012; Chaves-L’opez et al., 2014; Galati et al., 2014). Such modifications have also been reported to occur through hydrolysis by bacterial proteases, proteolysis and/or metabolic synthesis of inherent constituents (Nout and Ngoddy, 1997; Elkhalifa et al., 2006; Kohajdova and Karovicova 2007; Galati et al., 2014). Occurring during this process is the production of organic acids, leading to a decrease in pH and a corresponding rise in titratable acidity (TTA) (Correia et al., 2005; Correia et al., 2010; Adebo et al., 2017b). Such organic acids include succinic, formic, citric, pyruvic, lactic, acetic, uric and pyroglutamic acids (Correia et al., 2005; Correia et al., 2010). Changes in metabolites such as organic acids, amino acids, sugars, salts, isoflavones, saponins, tocopherol and polyphenols contribute to the sensorial properties,

23 nutritional composition, overall quality and health promoting properties of fermented food products (Namgung et al., 2010; Wardhani et al., 2010; Kang et al., 2011).

Depending on the type of fermentation process, metabolites differ and may consequently result in varying physicochemical properties of the respective products (Namgung et al., 2010; Kang et al., 2011; Galati et al., 2014). A study on the effect of natural and LAB fermentation on sorghum flour revealed that the number of total bacteria, LAB and proteolytic bacteria increased at the beginning of fermentation, but reduced after 32 h (Pranoto et al., 2013). Microbial growth of some microorganisms could be possibly inhibited due to an accumulation of metabolites and antibacterial substances that affect growth (Hutkins, 2006; Pranoto et al., 2013). The microbiota of fermented foods could also be influenced by production and accumulation of proteinases, aminopeptidases, presence of hydrogen peroxide, nitrogen oxides and peptides, which can alter LAB growth (Pranoto et al., 2013).

Changes in the functional properties during sorghum fermentation were studied by Elkhalifa et al. (2005) and an increase in emulsifying capacity and oil-binding capacity with a corresponding decrease in the water-binding capacity were noted. These changes were attributed to structural changes in the protein of the fermented samples and subsequent inactivation of anti-nutritional factors (Elkhalifa et al., 2005). As demonstrated subsequently by Elkhalifa et al. (2006), proteolytic actions and events occurring during the fermentation of sorghum cause a breakdown of protein and a disorientation of previously intact starch granules. Hydrolysis occurring during the fermentation process equally leads to a release of starch granules and breakdown of proteins (Elkhalifa et al., 2006; Adebo et al., 2018).

Sorghum tannins reduce during fermentation, subsequently improving protein digestibility (Hassan and Tinay, 1995; Osman, 2004; Dlamini et al., 2007). Schons et al. (2012) equally reported tannin degradation in sorghum by L. plantarum through a ring cleavage, gallic acid and glucose formation subsequent to decarboxylation of the gallic acid to pyrogallol. It has also been suggested that degaradation of tannins could be attributed to the utilization of tannic acid as a sole carbon source by microorganisms (Bhat et al., 1998).

Fermentation also affects the amount and composition of phenolic compounds in sorghum grains. An excellent and exhaustive review on the effect of processing methods including fermentation on

24 sorghum phenolic compounds has been provided by Taylor and Duodu (2015). A reduction in total phenolics, catechols, galloyls, p-hydroxybenzaldehyde, resorcinols and antioxidant activity by over 50% were documented by different authors (Towo et al., 2006; Dlamini et al., 2007), while an increase (over 100%) in total reactive phenolic hydroxyl groups in sorghum was reported by Kayodé et al. (2007). Strains of L. plantarum and L. fermentum have been reported to metabolize phenolic acids and their esters by phenolic acid decarboxylase, phenolic acid reductase and tannase activities (Curiel et al., 2009; Svensson et al., 2010). Decrease in phenolic compounds could be due to the rearrangement of phenolic structures and abstraction of hydride ions due to the acidic environment of fermentation (Porter et al., 1986; Taylor and Duodu, 2015). Other reasons could also be linked to the activity of polyphenol oxidase from the microflora of sorghum grain, polymerization and degradation or interaction with other molecules during the fermentation process, which may reduce extractability of phenolic compounds (Beta et al., 2000; Taylor and Duodu, 2015).

As reviewed, fermentation has numerous advantages. It is an affordable, low cost and efficient processing technique to increase safety, shelf life and improvement of nutrient, sensory qualities and health beneficial components. As with Africa and other developing nations, fermented foods play significant roles in the daily maintenance of nutrition and health with economic benefits as a source of income and contribute to food sovereignty and security. The fermentation process is also known to cause modification in different metabolites. Accordingly, a large proportion of this may not be detected and adequately differentiated using conventional targeted approaches. Metabolomics, which is discussed in the next section is a valuable technique that can effectively address this.

2.11 FOOD METABOLOMICS (FOODOMICS) Metabolomics is an emerging field within the “omics” (latin word meaning everything or total) sciences that began less than two decades ago (Mozzi et al., 2013). It encompasses the simultaneous detection, determination and quantitative/qualitative analysis of metabolites at a specific time and condition (Wishart, 2008; Capozzi and Bordoni, 2013). Metabolomic analysis has been utilized to address the limitations of other analytical methods, due to its robust capability for obtaining detailed information of metabolites present in different food metabolome (Adebo et al., 2017a). Metabolomic techniques can equally be effective for monitoring changes in the

25 metabolites of a substrate through statistical analysis of high-throughput data and have been widely adopted in numerous fields including pharmacology, toxicology, human nutrition and food science (Cifuentes, 2009; Wishart, 2008; Mozzi et al., 2013; Adebo et al., 2017a).

Within the context of food science, a new discipline called food metabolomics/foodomics was coined in 2009 and dedicated to the study of food and nutrition through the application of omics technologies (Cifuentes, 2009). It involves a combination of advanced analytical techniques to acquire comprehensive data on food composition and the use of data processing techniques/bioinformatics for the simultaneous characterization of compounds and metabolites (Wishart, 2008; Capozzi and Bordoni, 2013; Adebo et al., 2017a). Food metabolomics is thus applicable in different aspects of food science and technology, including food quality, food processing and nutrition, food safety, food microbiology and functional foods (Figure 2.8). The holistic approach of this technique positions it as a desirable tool for providing insight and understanding the multifunctionality and complexities of fermented foods (Mozzi et al., 2013; Adebo et al., 2017a).

Figure 2.8: Different aspects of food metabolomics

2.11.1 Fundamentals of food metabolomics Similar to other metabolomics studies, food metabolomics can be generally classified into either targeted or untargeted studies. Untargeted metabolomics is usually broad, and more focused on detection of a wide range and different metabolites to possibly obtain patterns without essentially quantifying specific metabolites (Cevallos-Cevallos et al., 2009; Tugizimana et al., 2013). It is thus more applied in hypothesis-generating studies. In contrast to untargeted analysis, targeted metabolomics analysis is more focused on specific group of metabolites and as such requires

26 subsequent quantification (Wishart, 2008; Mozzi et al., 2013). Targeted analysis is usually applied in validation studies and for the translation of findings and novel discoveries after an initial hypothesis-generating study. In both targeted and untargeted metabolomics approaches, different analytical platforms including CE-MS (capillary electrophoresis-mass spectrometry), liquid chromatography-mass spectrometry (LC-MS), gas chromatography-mass spectrometry (GC-MS), nuclear magnetic resonance (NMR) coupled with different multivariate data analysis (MVDA) tools, have been used to study variations in fermented foods.

A conventional metabolomics analysis follows a sequence of steps depicted in Figure 2.9. These include sample preparation, metabolite extraction (either targeted based or untargeted), data acquisition, subsequent data treatment/pre-processing/analysis and interpretation of results obtained (Figure 2.9). The sample preparation step is essentially needed to quench metabolism/reaction and prepare the sample in a ready state form prior to extraction (Tugizimana et al., 2013; Adebo et al., 2017a). Extraction is essentially needed to release metabolites from a sample and may require optimization, especially during an untargeted analysis as compared to a targeted one with a probable reported extraction procedure available in the literature. Depending on sample type/extraction technique adopted, further purification may still be needed in some cases and/or derivatization for specific compounds especially during GC-MS analysis.

Figure 2.9: Diagrammatic workflow of the food metabolomics process

Different analytical platforms for food metabolomics studies of fermented foods (Table 2.5) may be adopted, followed by data treatment/pre-processing/ analysis using appropriate bioinformatic software and MVDA tools. The process is subsequently finalized with interpretation of the results obtained, to confirm hypothesis earlier generated, discuss variation/similarities and quantify relevant metabolites (as with targeted analysis). Selection of the steps shown in Figure 2.9 is however, influenced by factors such as sample form, type of metabolomics approach to be adopted, available analytical platform, bioinformatic software and MVDA tool (Herrero et al., 2010; Tugizimana et al., 2013). The use of GC-MS based metabolomics will be further discussed as it is of interest and was adopted in this study.

Table 2.5: Summary of food metabolomic studies on fermented foods

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique Cheonggukjang Soybean Ala↑↓ 1,2,3,4-tetrakis[(trimethylsilyl)oxy]-butane, 2,3- 1HNMR*, PCA, Choi et al. bis(trimethylsilyl)-butane, δ-tocopherol, ɣ-tocopherol, D- CE-TOF- PLS-DA (2007); Baek ribitol, tyramine, glycerol, hydroxylamine, phytol MS*, GC- et al. (2010); Amb 1,3-diamino-propane, phenethylamine, putrescine, tryptamine, FIDδ, Park et al. serotonin, spermidine GC-TOF- (2010); Kim et AAc α -aminobutyric, β- alanine, ɣ-aminobutyric (GABA), g- MS*, al. (2011b); aminobutyric, 2,6-diaminopimelate, alanine, aminoadipate, LC- Kim et al. arginine, asparagine, aspartic, betanine, choline+, citrulline, MS/MS* (2012) DL-2-aminobutyric, DL-asparagine, DL-cysteine, DL- glutamine, DL-homoserine, DL-leucine, DL-methionine, DL- N-acetyl-serine, DL-ornithine, DL-phenylalanine, DL- threonine, DL-tryptophan, DL-valine, glutamic, glutamate, glycine, histidine, homotyrosine, homovaline, hydroxyproline, isoleucine, leucine, lysine, L-arginine, L-aspartic, L-cysteine, L-histidine, L-isoleucine, L-lysine, L-serine, L-tyrosine, methionine, N-a-acetylornithine, N-acetyl-glutamic acid, ornithine, phenylalanine, proline, pyroglutamate, pyroglutamic, serine, threonine, tryptophan, tyrosine, valine SUG, δ-trehalose, arabinose, arabitol, D-fructose, D-galactosamine, SUGDsd D-glucosamine, D-lactose, D-maltose, D-pintol, D-ribose, D- xylobiose, D-xylose, fructose, fructose-6-phosphate, galactose, galactinol, glucose, glucose-6-phosphate, inositol, isomaltose, lactate, lactitol maltose, mannose, mannotriose, melibiose, myo- ribitol, N-acetyl- raffinose, ribose, sorbitol, sucrose, xylose FAe Arachidic, behenic, linoleic, linolenic, myristic, oleic, palmitic, palmitoleic, stearic IFVNf 6”-O-acetyldaidzin, 6”-O-acetylgenistin, 6”-O- malonylglycitin, daidzin, glycitin, genistin, quercetin-tri-O- β- NTsg glucopyranoside, Adenine, adenosine, cytidine, cytosine, dihydrouracil, guanine, OAh guanosine, hypoxanthine, thymine, uracil, xanthine 2-hrdoxyisobutyric, 2-hydroxy-glutaric, 3-methyl-2- [(trimethylsilyl)oxy]-pentanoic acid, acetic, benzenepropanoic, calcium pantothenate, cis-aconitate, citric, citrilamic, D-

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique galacturonic, DL-isocitric, DL-lactic, DL-malic, formic, fumaric, galactaric, gluconic, glutamic, glutaric, glycerate, glycolic, itaconic, lactic, malic, malinic, malonic, n- octadecanoic, oxalic, palmitic, phenylpyruvate, quinate, saccharic, shikimic, succinic, succinate, tartaric, trans-aconitic SSAPNi acid, trans-caffeic, trans-sinapic, trimethylsilyl,3,5- Vj bis(trimethylsilyl)-3-methylvalerate Ok A3, Bg, I, II, IV, V Choline, nicotinic acid 3-amino-2-one-piperidin, allantonate, glycero-3-phosphate, mevalonolactone, phosphoric acid, R-(−)-1-amino-2-propanol, trigonelline, urea Crab paste Crab AAc Alanine, arginine, glutamate, glycine, histidine, isoleucine, 1HNMR* PCA, Chen et al. leucine, methionine, phenylalanine, tryptophan, tyrosine, OPLS-DA (2016) valine OAh Acetate, formate, fuarate, lactate, succinate, taurine OBl Betaine, trimethylamine (TMA), trimethylamine-N-oxide PUR, PYRm 2-pyridinemethanol, adenosine diphosphate (ADP), SUGd hypoxanthine, inosine, trigonelline Sucrose Daqu Barley and Ala Ethanol, glycerol, isopropanol 1HNMRδ PCA Yan et al. peas AAc 2-Aminobutyrate, cysteine, glutamate, glycine, glycylproline, (2015) homoserine, isoleucine, proline, serine, threonine SUG, Arabinitol, fructose, galactitol, galactose, glucose, gluticol, SUGDsd lactose, maltose, mannitol, myo-inositol, ribose, sucrose 2-hydroxyisobutyrate, 2-phosphoglycerate, acetate, glycerate, OAh glycolate, isobutyrate, lactate, pyruvate, succinate, taurine Betaine, cis-aconitate Acetone, allantoin, ascorbate, choline, ethylene glycol, OBl galactonate, maltate, malonate, N-nitrosodimethylamine, O- Ok phosphocholine, O-phosphoserine, oxypurinol, propionate, propylene glycol, S-sulfocysteine, urea Doenjang Soybean AAc ɣ-aminobutyric, alanine, aminoaldiphic, aminobutyric, 1HNMRδ PCA, PLS- Yang et al. asparagine, aspartic, glutamine, glutamic, glycine, histidine, GC-TOF- DA (2009a); Lee et isoleucine, leucine, lysine, methionine, ornithine, MS*, al. (2014b) phenylalanine, proline, pyroglutamic, sarcosine, serine, UPLC–Q- thioproline, threonine, tryptophan, tyrosine, valine TOF-MS*

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique SUG, α-glucose, β-glucose, arabinose, arabitol, erythrose, fructose, SUGDsd galactonic, galactose, glucitol, glucose, glycerol, glucosamine, inositol, mannitol, mannose, maltose, melibiose, myo-inositol, raffinose, ribitol, ribonic acid, sucrose, xylitol, Arachidic, behenic, caproc, eicosanic, eicosadienoic, lauric, FAe linoleic, linolenic, magaric, myristic, oleic, palmitic, palmitoleic, pentadecyclic, stearic, tricosanoic Acetyldaidzin, acetylgenistin, acetylglycitin, daidzin, daidzein, IFVNf genistin, glycitin, glycitein, malonyldaidzin, malonyglycitin, malonygenistin OAh 2-ketoglutaric, acetic, carbonic, citric, formic, fumaric, glucaric, glycolic, lactate, lactic, maleic, malic, malomic, malonic, manelic, oxalic, pipecolic, propionic, pyroglutamic, succinic, vanilic SSAPNi ɣg, ɣa, Bd, Be, I, II, III, IV, V Ok Choline, phosphocholine Fermented Wheat Ala 1-decanol, 1-dodecanol, 1-octanol, 1,2-dodecanediol, 7- SPME-GC- HCA Ferri et al. cereal methyl-4-octanol, dimethyl-1-octanol, ethylalcohol, hexanol, MS* (2016) isoamylalcohol, methyl-2-buten-1-ol, methyl-3-heptanol, HPLC- octadien-2-ol, octen-3-ol, pentanol, phenethylalcohol DADδ Cn 1,1,3-trimethyl-3-cyclohexene-5-one, 6-methyl-5-hepten-2- one, acetoin, decadienal, dodecanal heptanal, hexanal, methy- pentanal, nonadienal, nonanone, octanone, octenal, pentanal Ho 1,2-dimethyl-benzene, 1,3-hexadiene, 2-ethyl-furan, 2- penthyl-furan, 2-methyldecane, 3-methyl-dodecane, 4-methyldodecane, 5-methyldodecane, 10-methylnonadecane, 10- methyl-eicosane, furanone OAh 2-methylbutanoic, 3-methylbutanoic, acetic, dodecanoic, pentanoic, hexanoic, heptanoic, Ok Ester Fermented Cocoa beans CTH, CTHdp Epicatechin, O-hexoside-proanthocyanidin A5', O-pentoside- UPLC-ESI- PCA, PLS- Mayorga- cocoa beans proanthocyanidin A5, procyanidin QTOF-MS* DA, Gross et al. Ok Tripeptide, sucrose (2016) Fermented tea Green tea, AAc Glutamine, glutamic acid, glucoside, histamine, leucine, 1HNMR*, PCA Lee et al. black tea phenylalanine, proline, theanine, theanine-glucoside, tyrosine, UHPLC- (2011); Tan et tryptophan, valine QTOF-MS* al. (2016) Akq Caffeine, choline, glycerophosphocholine, theobromine

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique CTH, CTHdp 3-galloylprocyanidin B1, Cathechin, epiafzelechin, epicatechin-3-gallate, epicatechin, epicatechingallate, epigallocatechin, epigallocatechin gallate, epigallocatechin methylgallate, theaflavin-3-gallate, theaflavin 3,3’-digallate, theaflavin-3’-gallate, theasinensin A, theasinensin F, pigallocatechin-3-gallate, procyanidin B1, procyanin B2 FVNG, Apigenin-6,8-C-diglucoside, apigenin 6-C-glucoside 8-C- VOGr arabinoside, apigenin-6-C-arabinoside-8-C-glucoside, isoquercitrin, isovitexin, kaempferol 3-O-galactosylrutinoside, kaempferol 3-O-glucosylrutinoside, kaempferol-3-O- galactoside, kaempferol-3-O-glucoside, kaempferol-3-O- rutinoside myricetin 3-galactoside, quercetin-3-O-galactoside, quercetin 3-O-glucosylrutinoside, rutin Ls LysoPC, MG NTsg (S)-5’-deoxy-5’(methylthio)adenosine, 5’-deoxy- 5’(methylthio)adenosine, adenine, guanosine, inosine OAh 3-O-p-coumaroylquinic, 4-O-p-coumaroylquinic, p-coumaric, caffeoylshikimic, theogallin SUGd α-glucose, β-glucose, sucrose Ok Caffeine, gallic acid, N-(1-deoxy-1-fructosyl)leucine, N-(1- deoxy-1-fructosyl)tyrosine, N-vinyl-2-pyrrolidone, O- demethylfonsecin, theanine, unknown compounds Fermented milk Milk AAc↑↓ 3-aminobutyric, alanine, arginine, asparagine, aspartic, GABA, CE-TOF- NR Hagi et al. glutamine, glycine, isoleucine, methionine, threonine MS* (2016) Amb↓ Cyclohexylamine OAh↓↑ 2-oxoglutaric, citric, isocitric PUR↑↓ Adenine, guanine, hypoxanthine Pt↑ Ala-Pro, Leu-Pro, Pro-Pro, Val-Leu, Val-Pro, Val-Pro-Pro SUGd↓ Fructose 1,6-diphosphate Vj↓ Pyridoxamine Fermented Soymilk AAc↓ Phenylalanine, tyrosine 1HNMR* PCA Yang et al. soymilk OAh↑↓ Acetic, citric, fumaric, lactate, lactic, malic, oxalacetic, (2009b) succinic SUGd↓ Raffinose, stachyose, sucrose Ok Choline

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique Fermented Soybean AAc↑↓ Aspartic, GABA, glutamic, glycine, pyroglutamic, serine, GC-TOF- PCA, PLS- Lee et al. soybean threonine MSδ, LC- DA (2014c) SUG, Arabitol, fructose, galactose, maltose, mannitol, myo-inositol, ESI-MSδ SUGDsd↑↓ ribose, sorbitol, tagatose FAe↑ Palmitic, pentadecanoic, stearic IFVNf↑↓ 8-hydroxydaidzein, acetyldaidzin, acetylglycitin, acetylgenistin, daidzein, aidzin, genistin, glycitein, glycitin, hydroxygenistein, hydroxyglycitein NTg↑ Uracil OAh↓ Cinnamic, citric, malomic SSAPNi I Gochujang Wheat/rice AAc Alanine, GABA, glycine, glutamic, isoleucine, leucine, UPLC-Q- PCA, PLS- Lee et al. phenylalanine, proline, pyroglutamic, serine, threonine, TOF-MS*, DA (2016a) tyrosine, valine GC-TOF- Akq, DPHu Alnustone, dihydrocapsaicin, capsaicin MS* IFVNf, Apigenin-diglucoside, daidzein, glycitein, genistein, FLVDv hydroxydaidzen, kaempferol, luteolin-diglucoside Lw Lyso (PC16:0, PC18:1, PC18:2) SUG, Adonitol, arabinose, erythritol, fructose, fumaric, gentibiose, SUGDsd glucitol, gluconic, glucose, glycerol, glyceryl-glucoside, lactose, inositol, myo-inositol, xylose, xylitol OAh Citric, malic, malonic, phosphoric, propanoic, succinic SSAPNi I, III, V Ok glyceryl-glucoside, unknown compounds Kimchi Vegetables AAc↑↓ δ-aminobutyric, alanine, asparagine, aspartic, glycine, GC-MS* PCA, PLS- Park et al. glutamic, glutamine, leucine, ornithine, proline, threonine, DA (2016) valine SUG, D-fructose, galactose, glucose, glycerol, mannitol, myo- SUGDsd↑↓ inositol, sucrose, xylose OAh↑↓ 1-Propene-1,2,3-tricarboxylic acid, 2-keto-L-gluconic acid, 2,3,4-trihydroxybutyric acid, citric, fumaric, gluconic, isocitric, lactic, malic, octadecanoic, palmitic, pentanedioic, propanoic, pyrotartaric, ribonic, succinic Ok↑↓ Adenine, urea Koji Rice AAc↑↓ Alanine, aspartic, GABA, glutamic, glycine, isoleucine, GC-TOF- PCA, PLS- Lee et al. leucine, lysine, methionine, ornithine, phenylalanine, proline, MS*, DA (2016b) pyroglutamic, serine, threonine, tryptophan, tyrosine, valine UHPLC-

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique FAe↑↓ Hydroxy-oxo-octadecenoic, linoleic, linolenic, oleic, palmitic, LTQ-IT- pinellic, stearic MS/MS* FLVNv↑↓ Apigenin-C-glucosyl-C-arabinoside, chrysoeriol-hexoside, chrysoeriol-rutinoside, isovitexin-O-glucoside, tricin, tricin-7- O-rutinoside, tricin-O-glucoside LPLx↑↓ Lyso (PE14:0, PC14:0, PC18:3, PC16:1, PE18:2, PC18:2, PE16:0, PC16:0, PC18:1) OAh↑↓ Citric, fumaric, gluconic, glyceric, kojic, lactic, malic, malonic, shikimic, succinic, oxalic, PAy↑↓ 4-hydroxybenzoic acid, ferulic acid SUG, Erythritol, fructose, glucose, glycerol, maltose, myo-inositol, SUGDsd↑↓ pentitol, sorbitol, xylose, xylitol, Vj↑↓ Nicotinic acid Ok↑↓ Bacillibactin, unknowns Makgeolli Rice AAc Alanine, asparagine, glutamic, glutamine, glycine, leucine, GC-MS* OPLS-DA Seo et al. lysine, ornithine, proline, pyroglutamic, tryptophan, tyrosine (2016) Ala↑ 4-hydroxyphenylethanol OAh↑ 2-hydroxyglutaric, citric, lactic, malic, succinic SUG, Erythritol, fructose, glucose, glycerol, myo-inositol, ribose SUGDsd↑↓ Ok↑↓ 1,2-propanediol, phosphoric Meju Soybean AAc↑↓ ɣ-aminobutyric, acetylornithine, alanine, arginine, citrulline, UPLC-Q- PLS-DA Kang et al. glutamic, glutamine, histidine, isoleucine, leucine, lysine, TOF MS* OPLS-DA (2011) methionine, ornithine, phenylalanine, proline, pyroglutamic, threonine, tryptophane, tyrosine valine NTs↑↓ Adenine, hypoxanthine, uracil, xanthine OAh↑↓ Citric, pipecolic Pt↑↓ Glu-Gln, Glu-Tyr, Leu-Gln, Leu-Glu, Glu-Phe, Leu-Pro, Ser- Pro, Val-Glu, Val-Thr, Val-Leu, Leu-Val-Pro-Pro Miso Soybean AAc↑↓ Arginine, aspartate, glutamate, glutamine, lysine, LC-MSδ PCA Yoshida et al. phenylalanine, pyroglutamic, (2009) OAh↑ Citric Ok Fructosyl-leucine, fructosyl-phenylalanine Myeolchi- Fish AAc↑↓ Alanine, arginine, aspartate, glutamate, glutamic, glutamine, 1HNMR* PCA Jung et al. aekjeot glycine, isoleucine, leucine, serine, threonine (2016) Amb, NTsg↑↓ Betanine, choline, creatine, inosine, methyl amines OAh↑↓ Acetate, lactate

Produce Raw Metabolite Metabolite forms Data Data Reference material group acquisition processing method technique SUG, Glucose, glycerol SUGDsd↑↓ Saeu-jeot Shrimp AAc↑ Alanine, arginine, asparagine, aspartate, glutamate, glutamine, 1HNMR* CCA Lee et al. glycine, isoleucine, leucine, lysine, methionine, phenylalanine, (2014d) proline, pyroglutamate, serine, threonine, tryptophan, tyrosine, valine Amb↑↓ Dimethylamine, trimethylamine OAh↑ Acetate, butyrate, lactate SUG, Glucose, glycerol SUGDsd↑↓ a – alcohols; b – amines; c – amino acids; d – carbohydrates; sugars and sugar derivatives; e – fatty acids; f – isoflavonoids; g – nucleotides; h – organic acids; i – soyasaponins; j – vitamins; k – others (not classified); l – organic bases; m – purines and pyrimidines; n – carbonils; o – hydrocarbons; p – catechin and catechin derivatives; q – alkaloids; r – flavonol glycosides and flavone glycosides; s – lipids; t – peptides; u – diphenylheptanoid; v – flavonoids; w – lipids; x – lysophospholipids; y – phenolic acids; ↑ – increase in metabolites; ↓ ‒ decrease in metabolites; ↑↓ ‒ both increase and decrease in metabolites; * ‒ untargeted/profiling metabolomics; δ ‒ targeted metabolomics; CE-TOF-MS – capillary electrophoresis time of flight mass spectrometry; CCA – canonical correspondence analysis; FTIR – Fourier transform infrared spectroscopy; GC-FID – gas chromatography flame ionization detector; GC-MS – gas chromatography mass spectrometry; GC-TOF-MS – gas chromatography time of flight mass spectrometry; 1HNMR – proton nuclear magnetic resonance; HPLC-DAD – high performance liquid chromatography-diode array detector; LC-MS/MS – liquid chromatography tandem-mass spectrometry; NR – not reported; OPLS-DA – orthogonal partial least square discriminant analysis; PCA – principal component analysis; PLS-DA – partial least square discriminant analysis; SPME-GC- MS – solid phase microextraction- gas chromatography mass spectrometry; UPLC-ESI-QTOF-MS – ultra high performance liquid chromatography with electrospray ionization quadrupole time of flight mass spectrometry; UHPLC-LTQ-IT-MS/MS –ultra high pressure liquid chromatography linear ion trap-high resolution Orbitrap mass spectrometry and UPLC-Q-TOF MS – ultra performance liquid chromatography quadrupole time of flight mass spectrometry.

2.11.2 GC-MS based food metabolomics As highlighted in Table 2.5, advances and improvements in the sensitivity of analytical instruments, bioinformatics and MVDA tools have further increased the adoption of metabolomics in the past decade. Compared with other analytical platforms, GC-MS is one of the most efficient, robust, reliable and sensitive tools for metabolomic studies (Qiu and Reed, 2014). In a typical GC-MS analytical run, analytes are attached to the surface of a column and individual constituents are intermittently eluted based on their volatility using a temperature gradient. The signal subsequently gives a response proportional to the concentration of the analyte (compound), which is useful for quantification. In addition to ion peaks obtained, masses and fragmentation patterns are also acquired, which are vital for compound identification (Fancy and Rumpel, 2008; Swyngedouw and Lessard, 2018).

GC-MS is a suitable technique for the comprehensive analysis as it has a high separation efficiency, produces reproducible molecular fragmentation patterns, with sensitive and selective mass detection (Koek et al., 2011; Qiu and Reed, 2014). Challenges associated with ion suppression encountered with LC-MS are virtually non-existent in GC-MS (Koek et al., 2011). In comparison to other metabolomic platforms, the assignment of peaks through available database is relatively direct, due to reproducible fragmentation patterns all acquired with a constant electron voltage (70eV) (Garcia and Barbas, 2011). This largely makes it an integral tool for metabolite identification. Although the application of GC-MS is limited to volatiles (non-polar analytes), a large proportion of small molecular metabolites in fermented foods are within the GC-range of separation and identification (Table 2.5). Due to poor volatility of certain compounds such as amino acids, sugars and some organic acids in fermented foods, derivatization of GC-samples prior to analysis may be required to improve volatility and make them GC-amenable. It is nevertheless clear that not all metabolites in a biological sample can be acquired and detected using a single analytical platform. Hence, a compromise in the range of metabolites acquired will always have to be reached in GC-MS metabolomics analysis, either with derivatized or underivatized analytes.

As highlighted in Table 2.3, different studies have indicated the robustness of GC-MS for metabolomic studies. Using a GC-TOF-MS, Baek et al. (2010) investigated the metabolite profile of cheonggukjang, a fermented Korean cuisine inoculated with different Bacillus species. The authors reported increase and/or decrease of amino acids (20), sugars (10), organic acids (7), sugar alcohols (5) in the product after fermentation with different Bacillus sp.

Subsequent MVDA using PCA separated the unfermented samples from the fermented ones and amino acids were particularly observed to have contributed to the distinct cluster of later stages of fermentation. Same approach was also adopted in understanding L. plantarum fermentation of vegetables into kimchi (Park et al., 2010). Likewise, decreasing and/or increasing levels of organic acids, amino acids, sugars and polyols was reported. Obtained PCA plots were observed to have differentiated the LAB fermented kimchi samples from each other, suggesting that the LAB strains produced varying metabolites, which influence the characteristics of the kimchi samples (Park et al., 2010). Other available studies on GC-MS based metabolomics studies of fermented foods including cheonggukjang (Kim et al., 2012), doenjang (Lee et al., 2014b), fermented milk (Hagi et al., 2016), gochujangs (Lee et al., 2016a), koji (Lee et al., 2016b) and makgeolli (Seo et al., 2016) (Table 2.5) are published. There is however, a dearth of metabolomics studies on sorghum fermented foods, WG- fermented foods and no available metabolomics study exist on ting, which informed and necessitated this study.

2.12 MYCOTOXINS The global dimension and complexities of the food chain have their marked effect on food safety. Mycotoxin contamination remains one of most daunting challenges hampering the safety of food. These fungal toxins cause deleterious health and economic effects. The mycotoxins are naturally occurring compounds produced by certain filamentous fungal strains as secondary metabolites, with no significant link to fungal development (Pascari et al., 2018). The word ‘mycotoxin’ comes from two Greek words “mykes” and “toxicum” meaning ‘fungus’ and ‘poison’, respectively (Devreese et al., 2013).

Although over 300 mycotoxins and some of their metabolites have been characterized to date, those of significant economic and health impact are the aflatoxins (AFs), fumonisins (FBs), ochratoxin A (OTA), patulin (PAT), the trichothecenes [deoxynivalenol (DON), HT-2 toxin, nivalenol (NIV), T-2 toxin (T-2)] and zearalenone (ZEA) (Njobeh et al., 2010; Adebo, 2016; Pascari et al., 2018). Other notable ones are alternariol (AOH), citrinin (CIT), alpha-zearalenol (α-ZEA) and beta-zearalenol (β-ZEA) (Bryden, 2012). These mycotoxins are majorly produced by toxigenic fungal strains of Aspergillus (AFs, OTA, PAT), Fusarium (DON, NIV, HT-2, T- 2, ZEA) and Penicillium (OTA, PAT) genera (Pitt, 2008; Pascari et al., 2018). Although mycotoxins have been exhaustively reviewed in available literature, an overview is presented herein.

2.12.1 Factors influencing mycotoxin production The survival of mycotoxigenic fungi in substrates and attendant mycotoxin production can principally be attributed to chemical, environmental and biological factors (Klich, 2007; Milani, 2013) summarized in Figure 2.10. Chemical factors influencing mycotoxin production include carbon to nitrogen ratio, minerals, available nutrients and the presence of other compounds, while environmental factors could be pH, temperature, available moisture, humidity, climatic condition and the level of atmospheric gases (Milani, 2013). Equally important are biological factors including the mycotoxigenic fungal strain, possible competition with other microorganisms, genotype/composition of substrate, presence of insecticides/fungicides and agricultural practices. Furthermore, some other factors influencing mycotoxin production are prevailing conditions during storage, distribution and processing, which in addition to the other factors singly or those acting synergistically in combination, impact on the production of mycotoxins in agricultural commodities.

Figure 2.10: Factors influencing mycotoxin production

2.12.2 Toxicity of mycotoxins Numerous available studies in the literature have indicated that sorghum and its subsequent products can contain high levels of mycotoxins (Ayalew et al., 2006; Makun et al., 2009; Oueslati et al., 2012; Warth et al., 2012; Taye et al., 2018). It therefore, follows that consumption of such products by humans daily, may cause mycotoxicosis (disease derived from mycotoxin exposure), which can result in a chronic (long term, developing) or an acute

(short term and severe) poisoning. The toxic effects of these mycotoxins however, vary according to the amount (quantity) and type of toxin, as well as the duration of exposure (Anater et al., 2016). The consumption of mycotoxin infested commodities results in different human and animal health challenges such as intestinal dysfunction, growth faltering, cancer, immune suppression and hemorrhage (Bhat and Vasanthi, 2003; Gong et al., 2012; IARC, 2015).

Particularly notable is AFB1, which is known to be the most potent natural carcinogen that affects mainly the liver, causing hepatocellular carcinoma (Makun et al., 2012). On the other hand, OTA and CIT usually affect the kidney, causing impaired function and cell damage (Anater et al., 2016). High incidence of esophageal cancer has been linked to increased intake of FB contaminated maize in the Transkei region of South Africa (Marasas, 1996; Shephard et al., 2000), some parts of China and Iran (Sun et al., 2007). Meanwhile, the trichothecenes (THs) can trigger immune-suppression (Anater et al., 2016). Also, DON and PAT are known to affect the gastrointestinal and respiratory system and ZEA, the reproductive system causing hyperestrogen and vulvovaginitis (Pitt, 2008; Anukul et al., 2013). Ingestion of mycotoxins can equally cause reduced food intake, blood clotting, oedema, jaundice, abdominal pain, growth faltering, DNA mutation and in extreme cases, could lead to death (Wagacha and Muthomi, 2008; Dhanasekaran et al., 2011; USAID, 2012; Wu et al., 2014). Unfortunately, several of these mycotoxicosis cases occur mainly among humans living in developing countries, especially SSA, but regrettably, they go unnoticed and/or not reported.

2.12.3 Mycotoxins in sorghum and sorghum based foods Mycotoxins are produced throughout the food chain, during crop production in the field, to harvest and storage. Through synthesis by mycotoxigenic fungi that might have earlier proliferated in these crops, the presence of mycotoxins has been reported in cereals, legumes, vegetables, oilseeds and nuts (Njobeh et al., 2010; Makun et al., 2012; Pascari et al., 2018). Even though the case in SSA can be alarming, there is absolutely no region of the world free from the challenge of mycotoxin contamination. Stringent quality control programmes and implemented regulations have rather reduced the burden considerably in developed countries. Favorable conditions of high temperatures and humidity coupled with poor harvesting practices and improper storage common in SSA further aggravate the risk of mycotoxin production. This is further exacerbated by the poor or non-existent food/feed control programmes and total lack of regulations in some nations.

Relative to other food commodities, cereals and their subsequent products mostly suffer mycotoxin contamination. Excellent and exhaustive reviews presented in the literature have indicated that sorghum grains and subsequent products can be proliferated with mycotoxigenic fungi and could thus contain mycotoxins (Njobeh et al., 2010; Makun et al., 2012; Neme and Mohammed, 2017). As such, FB levels of >2000 μg/kg have been reported in sorghum grains from Ethiopia (Ayalew et al., 2006; Taye et al., 2016). Likewise, ZEA levels reaching 1454 ug/kg (Makun et al., 2009) and high levels of AFs, DON, KA, OTA, OTB and other significant mycotoxins, have equally been reported in this crop from SSA (Dada and Muller, 1983; Ayalew et al., 2006; Makun et al., 2009; Oueslati et al., 2012; Warth et al., 2012; Ezekiel et al., 2015; Taye et al., 2016; Taye et al., 2018). Owing to the relative stability of mycotoxins, studies in recent years have equally documented the severity of mycotoxin contamination in sorghum and their subsequent products from SSA (Ezekiel et al., 2015; Sirma et al., 2016; Adedeji et al., 2017; Adinsi et al., 2017; Chilaka et al., 2017; Misihairabgwi et al., 2017; Pleadin et al., 2017; Udomkun et al., 2017). The fact that sorghum and its food products are staple food sources to millions of African dwellers, negatively impacts food security in the continent and increases the risk of mycotoxin exposure among humans. Subsequent exposure to these mycotoxins via consumption of contaminated food poses some significant health risks, therefore necessitating their control. This is reviewed in the proceeding section of this chapter.

2.12.4 Mycotoxin control Sequel to the detrimental effects and challenges of mycotoxin contamination described in this review, different strategies have been proposed for the prevention and reduction of mycotoxin along the food value chain. These strategies can broadly be categorized as pre-harvest, harvest and post-harvest strategies (Figure 2.11).

Figure 2.11: Strategies to prevent and control mycotoxin contamination

According to Halasz et al. (2009), the most preferred strategy of preventing the occurrence of mycotoxins is to breed crops resistant to fungal invasion and to enhance genetic composition for the suppression of mycotoxin production. Pre-harvest measures could also include the adoption of good agricultural practices such as crop rotation, suitable irrigation, use of appropriate soil type, pest management and generally preventing plant stress. Equally important during harvest is the use of effective harvesting methods that would prevent mechanical damage that might expose plant tissue and ensuring that only mature crops are harvested.

Although results from these aforementioned strategies can be promising, adopting them under real world situations may not totally eliminate mycotoxin contamination, therefore requiring further post-harvest measures to reduce mycotoxin exposure. Available physical means aimed at reducing mycotoxins include sorting, decortication (to remove seed coat) and segregating visibly contaminated portion from a batch. Chemical techniques including the use of ammonia and nixtamalization have also been adopted. These approaches have however, not necessarily met the desired safety and efficacy (Wu et al., 2009; Adebo et al., 2017c), leading to the search for more effective strategies that could be put in place.

Biological detoxification/degradation of mycotoxins offers a more desirable and promising alternative of controlling mycotoxins. It encompasses reduction/removal of mycotoxins through microbial pathways to less/non-toxic intermediates and end products (Kolosova and Stroka, 2011; Adebo et al., 2017c). Further to the numerous advantages of fermentation, studies have also shown that this food processing operation can equally reduce/eliminate mycotoxins (Ezekiel et al., 2015; Karlovsky et al., 2016; Adebo et al., 2017c; Okeke et al., 2018). This has been attributed to the ability of fermenting organisms (LABs and yeasts) to detoxify mycotoxins through a binding mechanism and possible adsorption of these toxins to their cell wall (El-Khoury et al., 2011; Kolosova and Stroka, 2011; Bovo et al., 2014; Karlovsky et al., 2016). Such additional benefits make fermentation a viable traditional food processing technique that does not only improve beneficial composition, but also contributes to safety.

2.12.5 Regulation of mycotoxins in sorghum and sorghum based foods Mycotoxins can not only endanger human and animal health but can equally result in severe economic losses. It is therefore imperative that regulations are established to safeguard against the harmful effects of these toxins. Although hundreds of these mycotoxins are known to exist, regulatory limits/maximum tolerated levels are only available for a few. For these few

41 regulated mycotoxins, only about 15 African countries have established regulations for mycotoxins in food (FAO, 2004; Darwish et al., 2014).

Despite the huge consumption patterns and significant utilization of sorghum as staple foods in African, only Zimbabwe has an established national regulation for AFB1 (5 µg/kg) in sorghum (Viljoen, 2003). Although research on FBs was pioneered in South Africa, there is currently no existing regulation for FBs in sorghum in the country and other neighboring Southern African nations. The lack of mycotoxin legislation on FBs and other mycotoxins has been attributed to lack of data on the prevalence of some mycotoxins and insufficient capacity and resources to obtain vital exposure and toxicological data (FAO, 2004; Wild and Gong, 2010).

As for regulations on existing ones, the biggest challenge from an African context has been the lack of enforcement, partly attributed to the presence of numerous subsistence farmers and informal market systems in these countries. Under such circumstances, agricultural produce and products are directly sold to consumers without being inspected for mycotoxins. High costs of sampling and analysis, associated research, training initiatives and extension programmes are some other challenges limiting the implementation of mycotoxin regulation (Misihairabgwi et al., 2017). In addition to these are the inadequate control mechanisms in developing countries, which cannot be applied at house hold level. The mycotoxin issue in SSA still needs to be viewed with respect to indigenous food safety, health and associated agricultural practices. Accordingly, an efficient surveillance and food quality control program that will ensure that agricultural commodities destined for human consumption are free of mycotoxins at detrimental levels, must be in place to limit mycotoxin exposure.

2.13 CONCLUDING REMARKS From the literature review herein, it can be discerned that whole grain sorghum has better beneficial components embedded in it as compared to its refined counterpart. Mycotoxins do also contaminate sorghum and is equally noted to pose a significant threats to health and the economy, necessitating an efficient biological strategy of controlling their occurrence in foods including those derived from sorghum. Accordingly, fermentation is a beneficial food processing technique that does not only improve health beneficial components of foods but can also enhance mycotoxin reduction, thus ensuring the safety of products. The application of food metabolomics offers enormous opportunities to obtain detailed and exhaustive information on fermented foods such as WG-ting, a product that has not been studied in-depth.

Due to a dearth of information on optimization of WG-ting production, use of LAB strains to produce WG-ting, mycotoxin reduction during the fermentation process and the use of metabolomics to understand the variations in WG-ting samples, there is need to perform studies that will provide data that may contribute to indigenous knowledge.

REFERENCES Adams, M. R. (1998). Fermented weaning foods. In: Microbiology of fermented foods, Wood, B. J. B. (Ed.). Thomson Science, London. pp. 790-811. Adebiyi, A. O., Adebiyi, A. P., & Olaniyi, E. O. (2005). Nutritional composition of sorghum bicolor starch hydrolyzed with amylase from Rhizopus sp. African Journal of Biotechnology, 4, 1089-1094. Adebiyi, J. A., Obadina, A. O., Mulaba-Bafubiandi, A. F., Adebo, O. A., & Kayitesi, E. (2016). Effect of fermentation and malting on the physicochemical properties of millet flour and biscuit. Journal of Cereal Science, 70, 132-139. Adebiyi, J. A., Obadina, A.O., Adebo, O. A., & Kayitesi, E. (2018). Fermented and malted millet products in Africa: Expedition from traditional/ethnic foods to industrial value added products. Critical Reviews in Food Science and Nutrition, 58, 463-474.

Adebo (2016). Degradation of aflatoxin B1 (AFB1) by lysates of bacteria obtained from a gold mine aquifer. An MTech Dissertation submitted to the Faculty of Science, University of Johannesburg, South Africa. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., & Kayitesi, E. (2017a). Food metabolomics (Foodomics), a new frontier in food analysis and its potential in understanding fermented foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M.C. (Ed.). InTech, Croatia. pp. 211-234. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., Phoku, J. Z., & Kayitesi, E. (2017b). Fermented pulse-based foods in developing nations as sources of functional foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M.C. (Ed.). InTech, Croatia. pp. 77-109. Adebo, O. A., Njobeh, P. B., Gbashi, S., Nwinyi, O. C., & Mavumengwana, V. (2017c). Review on microbial degradation of aflatoxin. Critical Reviews in Food Science and Nutrition, 57, 3208-3217. Adebo, O. A., Njobeh, P. B., Mulaba-Bafubiandi, A. F., Adebiyi, J. A., Desobgo, S. C. Z., & Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, 1-10.

Adedeji, B. S., Ezeokoli, O. T., Ezekiel, C. N., Obadina, A. O., Somorin, Y. M., Sulyok, M., Adeleke, R. A., & Warth, B. (2017). Bacterial species and mycotoxin contamination associated with locust bean, melon and their fermented products in south-western Nigeria. International Journal of Food Microbiology, 258, 73-80. Adinsi, L., Vieira-Dalode, G., Akissoe, N. H., Anihouvi, V., Mestres, C., Jacobs, A., Dlamini, N., Pallet, D., & Hounhouigan, J. D. (2014). Processing and quality attributes of gowe: a malted and fermented cereal-based beverage from Benin. Food Chain, 4, 171-183. Adinsi, L., Mestres, C., Akissoe, N., Vieria-Dalode, G., Anihouvi, V., Durand, N., & Hounhouigan, J. (2017). Comprehensive quality and potential hazards of gowe, a malted and fermented cereal beverage from West Africa. A diagnostic for a future re-engineering. Food Control, 82, 18-25. Adhikari, P., Hwang, K. T., Park, J. N., & Kim, C. K. (2006). Policosanol content and composition in Perilla seeds. Journal of Agricultural and Food Chemistry, 54, 5359-5362. Afify, A. M. R., El-Beltagi, H. S., Abdelsalam, S. M., & Omran, A. A. (2012). Effect of soaking, cooking, germination and fermentation processing on proximate analysis and mineral content of three white sorghum varieties (Sorghum bicolor L. Moench). Notulae Botanicae Horti Agrobotanici Cluj-Napoca, 40, 92-98. Ajiboye, T. O., Iliasu, G. A., Adeleye, A. O., Abdussalam, F. A., Akinpelu, S. A., Ogunbode, S. M., Jimoh, S. O., & Oloyede, O. B. (2014). Nutritional and antioxidant dispositions of sorghum/millet beverages indigenous to Nigeria. Food Science and Nutrition, 2, 597-604. Akingbala, J. O., Rooney, L. W., & Fauboin, J. M. (1981). Physical, chemical, and sensory evaluation of ogi from sorghum of differing kernel characteristics. Journal of Food Science, 76, 1532-1536. Ali, A. A., & Mustafa, M. M. (2009). Isolation, characterization and identification of lactic acid bacteria from fermented sorghum dough used in Sudanese Kisra preparation. Pakistan Journal of Nutrition, 8, 1814-1818. Althwab, S., Carr, T. P., Weller, C. L., Dweikat, I. M., & Schlegel. V. (2015). Advances in grain sorghum and its co-products as a human health promoting dietary system. Food Research International, 77, 349-359. Amoako, D. B., & Awika, J. M. (2016). Polymeric tannins significantly alter properties and in vitro digestibility of partially gelatinized intact starch granule. Food Chemistry, 208, 10-17. Anater, A., Manyes, L., Meca, G., Ferrer, E., Luciano, B., Pimpão, C. T., & Font, G. (2016). Mycotoxins and their consequences in aquaculture: A review. Aquaculture, 451, 1-10.

Anderson, J. W., Hanna, T. J., Peng, X., & Kryscio, R. J. (2000). Whole grain foods and heart disease risk. Journal of the American College of Nutrition, 19, 291S-299S. Anukul, N., Vangnai, K., & Mahakarnchanakul, W. (2013). Significance of regulation limits in mycotoxin contamination in Asia and risk management programs at the national level. Journal of Food and Drug Analysis, 21, 227-241. Ashok, P. K., & Upadhyaya, K. (2012). Tannins are astringent. Journal Pharmacognosy and Phytochemistry, 1, 45-50. Awika, J. M., Rooney, L.W., Wu, X., Prior, R. L., & Cisneros-Zevallos, L. (2003). Screening methods to measure antioxidant activity of sorghum (Sorghum bicolor) and sorghum products. Journal of Agricultural and Food Chemistry, 51, 6657-6662. Awika, J. M., & Rooney, L.W. (2004). Sorghum phytochemicals and their potential impact on human health. Phytochemistry, 65, 1199-1221. Awika, J. M., Mcdonough, C. M., & Rooney, L. W. (2005). Decorticating sorghum to concentrate healthy phytochemicals. Journal of Agricultural and Food Chemistry, 53, 6230- 6234. Awika, J. M. (2011). Sorghum flavonoids: Unusual compounds with promising implications for health. In: Advances in Cereal Science: Implications to Food Processing and Health Promotion. American Chemical Society, Washington DC. pp. 171-200. Awika, J. M. (2017). Sorghum: its unique nutritional and health-promoting attributes. In: Gluten-Free Ancient Grains, Taylor, J. R. N., & Awika, J. M. (Eds.). Woodhead Publishing, United Kingdom. pp. 21-54. Ayalew, A., Fehrmann, H., Lepschy, J., Beck, R., & Abate, D. (2006). Natural occurrence of mycotoxins in staple cereals from Ethiopia. Mycopathologia, 162, 57-63. Badau, M., Nkama, H. I., & Jideani, A. I. (2005). Phytic acid content and hydrochloric acid extractability of minerals in pearl millet as affected by germination time and cultivar. Food Chemistry, 92, 425-435. Baek, J. G., Shim, S. M., Kwon, D. Y., Choi, H. K., Lee, C. H., & Kim, Y. S. (2010). Metabolite profiling of cheonggukjang, a fermented soybean paste, inoculated with various Bacillus strains during fermentation. Bioscience Biotechnology and Biochemistry, 74, 1860-1868. Batey, I. (2017). The diversity of uses for cereal grains. In: Cereal Grains, Wrigley, C., Batey, I., & Miskelly, D. (Eds.). Elsevier, Amsterdam. pp. 41-53. Bean, S. R., Wilson, J. D., Moreau, R. A., Galant, A., Awika, J. M., Kaufman, R. C., Adrianos, S. L., & Ioerger, B. P. (2016). Structure and composition of the sorghum grain. In: Sorghum:

State of the Art and Future Perspectives, Agronomy Monographs (Vol. 58), Ciampitti, I., & Prasad, V. (Eds.). ASA and CSSA, Madison. pp. 1-26. Behall, K. M., Scholfield, D. J., & Hallfrisch, J. (2004). Diets containing barley significantly reduce lipids in mildly hypercholesterolemic men and women. American Journal of Clinical Nutrition, 80, 1185-1193. Behall, K. M., Scholfield, D. J. and Hallfrisch, J. (2006). Whole-grain diets reduce blood pressure in mildly hypercholesterolemic men and women. Journal of the American Dietician Association, 106, 1445-1449. Bello, A. B., Rooney, L. W., & Waniska, R. D. (1990). Factors affecting quality of sorghum Tô, a thick porridge. Cereal Chemistry, 67, 20-25. Belton, P. S., Delgadillo, I., Halford, N. G., & Shewry, P. R. (2006). Kafirin structure and functionality. Journal of Cereal Science, 44, 272-286. Beta, T., Corke, H., & Taylor, J. R. N. (2000). Starch properties of Barnard red, a South African red sorghum variety of significance in traditional African brewing. Starch, 52, 467-470. Bhat, T. K., Singh, B., & Sharma, O. P. (1998). Microbial degradation of tannins – a current perspective. Biodegradration, 9, 343-347. Bhat, R. V., & Vasanthi, S. (2003). Food safety in food security and food trade: Mycotoxin food safety risk in developing countries. Unnevehr, L. J. (Ed.). International Food Policy Research Institute, (Brief), Washington, DC. Bhatnagar-Mathur, P., Sunkara, S., Bhatnagar-Panwar, M., Waliyar, F., & Sharma, K. K. (2015). Biotechnological advances for combating Aspergillus flavus and aflatoxin contamination in crops. Plant Science, 234, 119-132. Blandino, A., Al-Aseeri, M. E., Pandiella, S. S., Cantero, D., & Webb, C. (2003). Cereal-based fermented foods and beverages. Food Research International, 36, 527-543. Boling, M. B., & Eisener, N. (1982). Bogobe: sorghum porridge of Botswana. In: Proceedings on the International Symposium on Sorghum Grain Quality, Mertin, J. V. (Ed.). International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru. pp. 32-35. Bourdichon, F., Casaregola, S., Farrokh, C., Frisvad, J. C., Gerds, M. L., Hammes, W. P., Harnett, J., Huys, G., Laulund, S., Ouwehand, A., Powell, I. B., Prajapati, J. B., Seto, Y., Schure, E. T., Boven, A. V., Vankerckhoven, V., Zgoda, A., Tuijtelaars, S., & Hansen, E. B. (2012). Food fermentations: Microorganisms with technological beneficial use. International Journal of Food Microbiology, 154, 87-97.

Bovo, F., Franco, L. T., Rosim, R. E., Trindade, C. S. F., & de Oliveira, C. A. F. (2014). The ability of Lactobacillus rhamnosus in solution, spray dried or lyophilized to bind

aflatoxin B1. Journal of Food Research, 3, 35-42. Bryden, W. I. (2012). Mycotoxin contamination of the feed supply chain: Implications for animal productivity and feed security. Animal Feed Science and Technology, 173, 134-158. Burdette, A., Garner, P. L., Mayer, E. P., Hargrove, J. L., Hartle, D. K., & Greenspan, P. (2010). Anti-inflammatory activity of select sorghum (Sorghum bicolor) brans. Journal of Medicinal Foods,13, 879-887. Camargo-Filho, I., Cortez, D. A. G., Ueda-Nakamura, T., Nakamura, C. V., & Dias-Filho, B. P. (2008). Antiviral activity and mode of action of a peptide isolated from Sorghum bicolor. Phytomedicine, 15, 202-208. Capozzi, F., & Bordoni, A. (2013). Foodomics: A new comprehensive approach to food and nutrition. Genes and Nutrition, 8, 1-4. Cardoso, L. D., Pinheiro, S. S., Martino, H. S. D., & Pinheiro-Sant’Ana, H. M. (2017). Sorghum (Sorghum bicolor L.): Nutrients, bioactive compounds, and potential impact on human health. Critical Reviews in Food Science and Nutrition, 2, 372-390. Carr, T. P., Weller, C. L., Schlegel, V. L., Cuppett, S. L., Guderian, D. M., & Johnson, K. R. (2005). Grain sorghum lipid extract reduces cholesterol absorption and plasma non-HDL cholesterol concentration in hamsters. The Journal of Nutrition, 135, 2236-2240. CAST (Council for Agricultural Science and Technology) (2003). Mycotoxins: risks in plant animal and human systems. Report No. 139. Ames, Iowa, USA. Cevallos-Cevallos, J. M., Reyes-De-Corcuera, J. I., Etxeberria, E., Danyluk, M. D., & Rodrick, G. E. (2009). Metabolomic analysis in food science: A review. Trends in Food Science and Technology, 20, 557-566. Chaves-L´opez, C., Serio, A., Grande-Tovar, C. D., Cuervo-Mulet, R., Delgado-Ospina, J., & Paparella, A. (2014). Traditional fermented foods and beverages from a microbiological and nutritional perspective: The Colombian heritage. Comprehensive Reviews in Food Science Safety, 13, 1031-1048. Chen, D., Ye, Y., Chen, J., & Yan, X. (2016). Evolution of metabolomics profile of crab paste during fermentation. Food Chemistry, 192, 886-892. Chilaka, C. A., De Boevre, M., Atanda, O. O., & De Saeger, S. (2017). The status of Fusarium mycotoxins in sub-Saharan Africa: A review of emerging trends and post-harvest mitigation strategies towards food control. Toxins, 9, 1-36.

Chingwaru, W., & Vidmar, J. (2017). Potential of Zimbabwean commercial probiotic products and strains of Lactobacillus plantarum as prophylaxis and therapy against diarrhoea caused by Escherichia coli in children. Asian Pacific Journal of Tropical Medicine, 10, 57-63. Choi, H. K., Yoon, J. H., Kim, Y. S., & Kwon, D. Y. (2007). Metabolomic profiling of cheonggukjang during fermentation by 1H NMR spectrometry and principal components analysis. Process Biochemistry, 42, 263-266. Cifuentes, A. (2009). Food analysis and foodomics. Journal of Chromatography A, 1216, 7109. Collins, B., Cotter, P. D., Hill, C., & Ross, R. P. (2010). Applications of lactic acid bacteria produced bacteriocins. In: Biotechnology of Lactic Acid Bacteria Novel Applications, Mozzi, F., Raya, R., & Vignolo, G. (Eds.). Wiley-Blackwell, Iowa. pp. 89-109. Cork, S. J., & Foley, W. J. (1991). Digestive and metabolic strategies of arboreal mammalian folivores in relation to chemical defenses in temperature and tropical forests. In: Plant Defenses against Mammalian Herbivory, Chivers, D. J., & Langer, P. (Eds.). CRC Press, Boca Raton. pp. 133-166. Correia, I., Nunes, A., Duarte, I. F., Barros, A., & Delgadillo, I. (2005). Sorghum fermentation followed by spectroscopic techniques. Food Chemistry, 90, 853-859. Correia, I., Nunes, A., Guedes, S., Barros, A. S., & Delgadillo, I. (2010). Screening of lactic acid bacteria potentially useful for sorghum fermentation. Journal of Cereal Science, 52, 9- 15. Cruz, R. A. O., López, J. L. C., Aguilar, G. A. G., García, H. A., Gorinstein, S., Romero, R. C., & Sánchez, M. R. (2015). Influence of sorghum karifin on serum lipid profile and antioxidant activity in hyperlipidemic rats (in vitro and in vivo studies). BioMed Research International, 1-8. Article ID 164725. Curiel, J. A., Rodrı´guez, H., Acebroń, I., Manchenõ, J. M., de las Rivas, B., & Munõz, R. (2009). Production and physicochemical properties of recombinant Lactobacillus plantarum tannase. Journal of Agricultural and Food Chemistry, 57, 6224-6230. Dada, L. O., & Muller, H. G. (1983). The fate of aflatoxin in the production of ogi, a Nigerian fermented sorghum porridge. Journal of Cereal Science, 1, 63-70. DAFF (Department of Agriculture, Forestry and Fisheries, South Africa) (2010). Sorghum production guideline. Department of Agriculture, Forestry and Fisheries, Pretoria, South Africa. Darwish, W. S., Ikenaka, Y., Nakayama, S. M. M., & Ishizuka, M. (2014). An overview of mycotoxin contamination of foods in Africa. Journal of Veterinary Medical Sciences, 76, 789-797.

48 de Mesa-Stonestreet, N. J., Alavi, S., & Bean, S.R. (2010). Sorghum proteins: The concentration, isolation, modification, and food applications of kafirins. Journal of Food Science, 75, R90-R104. Devreese, M., De backer, P., & Croubels, S. (2013). Overview of the most important mycotoxins for the pig and poultry husbandry. Vlaams Diergeneeskundig Tijdschrift, 82, 171-180. Dhanasekaran, D., Shanmugapriya, S., Thajuddin, N., & Panneerselvam, A. (2011). Aflatoxins and aflatoxicosis in human and animals. In: Aflatoxins - Biochemistry and Molecular Biology, Guevara-Gonzalez, R. G. (Ed.). InTech, Croatia. pp. 221-254. Dicko, M. H., Gruppen, H., Hilhorst, R., Voragen, A. G., & van Berkel, W. J. (2006). Biochemical characterization of the major sorghum grain peroxidase. FEBS Journal, 273, 2293-2307. Djè, K. M., Aka, S., Zinzendorf, N. Y., Yao, K. C., & Loukou, Y. G. (2009). Predominant lactic acid bacteria involved in the spontaneous fermentation step of tchapalo process, a traditional sorghum beer of Côte d’Ivoire. Research Journal of Biological Sciences, 4, 789- 795. Dlamini, N. R., Taylor, J. R. N., & Rooney, L. W. (2007). The effect of sorghum type and processing on the antioxidant properties of African sorghum-based foods. Food Chemistry, 105, 1412-1419. Dovi, K. A. P. (2013). Whole grain sorghum and whole grain cowpea biscuits as a complementary food for improved child nutrition. A MSc Dissertation submitted to the Faculty of Natural and Agricultural Sciences, University of Pretoria, South Africa. Dunn, K. L., Yang, L. Y., Girard, A., Bean, S., & Awika, J. M. (2015). Interaction of sorghum tannins with wheat proteins and effect on in vitro starch and protein digestibility in a baked product matrix. Journal of Agricultural and Food Chemistry, 63, 1234-1241. Dykes, L., Rooney, L. W., Waniska, R. D., & Rooney, W. L. (2005). Phenolic compounds and antioxidant activity of sorghum grains of varying genotypes. Journal of Agricultural and Food Chemistry, 53, 6813-6818. Dykes, L., & Rooney, L. W. (2006). Sorghum and millet phenols and antioxidants. Journal of Cereal Science, 44, 236-251. Dykes, L., & Rooney, L. W. (2007). Phenolic compounds in cereal grains and their health benefits. Cereal Foods World, 52, 105-111.

Eckhoff, S. R., & Watson, S. A. (2009). Corn and sorghum starch production. In: Starch: Chemistry and Technology (3rd ed), BeMiller, J., & Whistler, R. (Eds.). Elsevier, Netherlands. pp. 373-439. Elkhalifa, A. E. O., Schiffler, B., & Bernhardt, R. (2005). Effect of fermentation on the functional properties of sorghum flour. Food Chemistry, 92, 1-5. Elkhalifa, A. E. O., Bernhardt, R., Bonomi, F., Iametti, S., Pagani, M. A., & Zardi, M. (2006). Fermentation modifies protein/protein and protein/starch interactions in sorghum dough. European Food Research and Technology, 222, 559-564.

El-Khoury, A., Atoui, A., & Yaghi, J. (2011). Analysis of aflatoxin M1 in milk and yoghurt

and AFM1 reduction by lactic acid bacteria used in Lebanese industry. Food Control, 22, 1695-1699. Ezekiel, C. N., Abia, W. A., Ogara, I. M., Sulyok, M., Warth, B., & Krska, R. (2015). Fate of mycotoxins in two popular traditional cereal-based beverages (kunu-zaki and pito) from rural Nigeria. LWT-Food Science and Technology, 60, 137-141. FAO (Food and Agricultural Organization) (2004). Worldwide regulations for mycotoxins in food and feed in 2003. FAO Food and Nutrition Paper No. 81. Rome, Italy FAOSTAT (Food and Agriculture Organization Statistics) (2017). http://www.fao.org/faostat/en/#data/QC (Accessed 21/10/2017). Fancy, S. A., & Rumpel, K. (2008). GC-MS-based metabolomics. In: Biomarker Methods in Drug Discovery and Development, Wang, F. (Ed.). Humana Press, New Jersey. pp. 317-340. Faquinello, P., Murakami, A. E., Cella, P. S., Franco, J. R. G., Sakamoto, M. I., & Bruno, L. D. G. (2004). High tannin sorghum in diets of Japanese quails (Coturnix coturnix japonica). Revista Brasileira de Ciencia Avícola, 6, 81-86. Farhad, M., Kailasapathy, K., & Tamang, J. P. (2010). Health aspects of fermented food. In: Fermented Foods and Beverages of the World, Tamang, J. P., & Kailasapathy, K. (Eds.). CRC Press, London. pp. 391-414. Fernandes, E. A., Pereira, W. J. S., Hackenhaar, L., Rodrigues, R. M., & Terra, R. (2013). The use of whole grain sorghum in broiler feeds. Brazilian Journal of Poultry Science, 15, 169- 286. Ferreira, S. L. C., Dos Santos, W. N. L., Quintella, C. M., Neto, B. B., & Bosque-Sendra, J. M. (2004). Doehlert matrix: a chemometric tool for analytic chemistry–review. Talanta, 63, 1061-1067. Ferri, M., Serrazanetti, D. I., Tassoni, A., Baldissarri, M., & Gianotti, A. (2016). Improving the functional and sensorial profile of cereal-based fermented foods by selecting Lactobacillus

plantarum strains via a metabolomics approach. Food Research International, 89, 1095- 1105. Franz, H.C. (1971). The traditional diet of the Bantu in the Pietersburg district. South African Medical Journal, 45, 1232-1235. Gadaga, T. H., Mutakumira, A. N., Narvhus, J. A., & Ferusu, S. B. (1999). A review of traditional fermented foods and beverages in Zimbabwe. International Journal of Food Microbiology, 53, 1-11. Galati, A., Oguntoyinbo, F. A., Moschetti, G., Crescimanno, M., & Settanni, L. (2014). The cereal market and the role of fermentation in cereal-based food production in Africa. Food Reviews International, 30, 317-337. Gänzle, M. G., Loponen, J., & Gobbetti, M. (2008). Proteolysis in sourdough fermentations: mechanisms and potential for improved bread quality. Trends in Food Science and Technology, 19, 513-521. Gänzle, M. G. (2015). Lactic metabolism revisited: metabolism of lactic acid bacteria in food fermentations and food spoilage. Current Opinion in Food Science, 2, 106-117. Garcia, A., & Barbas, C. (2011). Gas chromatography-mass spectrometry (GC-MS)-based metabolomics. In: Metabolic Profiling, Methods in Molecular Biology, Metz, D. O. (Ed.). Springer Science+Business Media, New York. pp. 191-204. García-Mantrana, I., Yebra, M. J., Haros, M., & Monedero, V. (2016). Expression of bifidobacterial phytases in Lactobacillus casei and their application in a food model of whole-grain sourdough bread. International Journal of Food Microbiology, 216, 18-24. Gassem, M. A. A. (1999). Study of micro-organisms associated with the fermented bread (khamir) produced from sorghum in Gizan region, Saudi Arabia. Journal of Applied Microbiology, 86, 221-225. Gee, M. C., & Harold, G. W. (2004). An assessment of the tannin content of wild sorghum. Journal of Crop Science, 43, 1850-1870. Ghosh, D. (2015). Tannins from foods to combat diseases. International Journal of Pharma Research and Review,4, 40-44. Graham, G. G., MacLean, W. C., Morales, E., Hamaker, B. R., Kirleis, A. W., Mertz, E. T., & Axtell, J. D. (1986). Digestibility and utilization of protein and energy from Nasha, a traditional Sudanese fermented sorghum weaning food. The Journal of Nutrition, 116, 978- 984. Goel, G., Puniya, A. K., Aguliar, C. N., & Singh, K. (2005). Interaction of gut microflora with tannins in feeds. Naturwissenschaften, 92, 497-503.

Gong, Y. Y., Wilson, S., Mwatha, J. K., Routledge, M. N., Castelino, J. M., & Zhao, B., Kimani, G., Kariuki, H. C., Vennervald, B. J., Dunne, D. W., & Wild, C. P. (2012). Aflatoxin exposure may contribute to chronic hepatomegaly in Kenyan school children. Environmental Health Perspectives, 120, 893-896. Gouni-Berthold, I., & Berthold, H. K. (2002). Policosanol: clinical pharmacology and therapeutic significance of a new lipid-lowering agent. American Heart Journal, 143, 356- 365. Hagerman, A. E., Riedl, K. M., Jones, G. A., Sovik, K. N., Ritchard, N. T., Hartzfeld, P. W., & Riechel, T. L. (1998). High molecular weight plant polyphenolics (Tannins) as biological antioxidants. Journal of Agricultural and Food Chemistry, 46, 1887-1892. Hagi, T., Kobayashi, M., & Nomura, M. (2016). Metabolome analysis of milk fermented by - aminobutyric acid-producing Lactococcus lactis. Journal of Dairy Science, 99, 994-1001. Hallfrisch, J., Scholfield, D. J., & Behall, K. M. (2003). Blood pressure reduced by whole grain diet containing barley or whole wheat and brown rice in moderately hypercholesterolemic men. Nutrition Research, 23, 1631-1642. Halasz, A., Lasztity, R., Abonyi, T., & Bata, A. (2009). Decontamination of mycotoxin- containing food and feed by biodegradation. Food Reviews International, 25, 284-298. Hamad, S. H., Dieng, M. C., Ehrmann, M. A., & Vogel, R. F. (1997). Characterization of the bacterial flora of Sudanese sorghum, flour and sorghum sourdough. Journal of Applied Microbiology, 28, 764-770. Hamaker, B., Mohamed, A., Habben, J., Huang, C., & Larkins, B. (1995). Efficient procedure for extracting maize and sorghum kernel proteins reveals higher prolamin contents than the conventional method. Cereal Chemistry, 72, 583-588. Hassan, I. A. G., & Tinay, A. H. E. (1995). Effect of fermentation on tannin content and in- vitro protein and starch digestibilities of two sorghum cultivars. Food Chemistry, 53, 149- 151. Harvard Health Letter. (1999). The whole is greater than the sum of the parts. I. Whole grain. Harvard Health Letters, 25, 8. Herrero, M., Garcıa-Canas, V., Simo, C., & Cifuentes, A. (2010). Recent advances in the application of CE methods for food analysis and foodomics. Electrophoresis, 31, 205-228. Hesseltine, C. W. (1979). Some important fermented foods in mid-Asia, the Middle East and Africa. Journal of American Oil Chemists’ Society, 56, 367-374. Hugenholtz, J. (2013). Traditional biotechnology for new foods and beverages. Current Opinion in Biotechnology, 24, 155-159.

Hwang, K. T., Cuppett, S. L., Weller, C. L., & Hanna, M. A. (2002a). Properties, composition, and analysis of grain sorghum wax. Journal of the American Oil Chemists' Society, 79, 521- 527. Hwang, K. T., Cuppett, S. L., Weller, C. L., & Hanna, M. A. (2002b). HPLC of grain sorghum wax classes highlighting separation of aldehydes from wax esters and steryl esters. Journal of Separation Science, 25, 619-623. Hwang, K. T., Kim, J. E., & Weller, C. L. (2005). Policosanol contents and compositions in wax-like materials extracted from selected cereals of Korean origin. Cereal Chemistry, 82, 242-245. IARC (International Agency for Research on Cancer). (2015). Mycotoxin control in low- and middle income countries. IARC Working Group Reports, 9, 54. Lyon, France. Ikediobi, C. O., Olugboji, O., & Okoh, P. N. (1988). Cyanide profile of component parts of sorghum (Sorghum bicolor L. Moench) sprouts. Food Chemistry, 27, 167-175. Jacobs, D., Marquart, L., Slavin, J., & Kushi, L. H. (1998). Whole grain intake and cancer: an expanded review and meta-analysis. Nutrition and Cancer, 30, 85-96. Jones, J. M., Reicks, M., Adams, J., Fulcher, G., Weaver, G., Kanter, M., & Marquart, L. (2002). The importance of promoting a whole grain foods message. Journal of the American College of Nutrition, 21, 293-297. Jung, J. Y., Lee, H. J., Chun, B. H., & Jeon, C. O. (2016). Effects of temperature on bacterial communities and metabolites during fermentation of myeolchi-aekjeot, a traditional Korean fermented anchovy sauce. PLoSOne, 11, 1-20. Juntunen, K. S., Laaksonen, D. E., Poutanen, K. S., Niskanen, L. K., & Mykkanen, H. M. (2003). High-fiber rye bread and insulin secretion and sensitivity in healthy postmenopausal women. American Journal of Clinical Nutrition, 77, 385-391. Kamath, V., Niketh, S., Chandrashekar, A., & Rajini, P. S. (2007). Chymotryptic hydrolysates of α-kafirin, the storage protein of sorghum (Sorghum bicolor) exhibited angiotensin converting enzyme inhibitory activity. Food Chemistry, 100, 306-311. Kang, H. J., Yang, H. J., Kim, M. J., Han, E. S., Kim, H. J., & Kwon, D. Y. (2011). Metabolomic analysis of meju during fermentation by ultra-performance liquid chromatography-quadrupole-time of flight mass spectrometry (UPLC-Q-TOF-MS). Food Chemistry, 127, 1056-1064. Kantor, L. S., Variyam, J. N., Allshouse, J. E., Putnam, J. J., & Lin, B. H. (2001). Choose a variety of grains daily, especially whole grains: a challenge for consumers. Journal of Nutrition, 131, 473S-486S.

Karlovsky, P., Suman, M., Berthiller, F., De Meester, J., Eisenbrand, G., Perrin, I., Oswald, I. P., Speijers, G., Chiodini, A., Recker, T., & Dussort, P. (2016). Impact of food processing and detoxification treatments on mycotoxin contamination. Mycotoxin Research, 32, 179- 205. Katongole, J. N. (2008). The microbial succession in indigenous fermented maize products. A MSc Dissertation submitted to the Faculty of Natural and Agricultural Sciences, University of the Free State, South Africa. Kayitesi, E., Behera, S. K., Panda, S. K., Dlamini, B., & Mulaba-Bafubiandi, A. F. (2017). Amasi and Mageu: Expedition from ethnic Southern African foods to cosmopolitan markets. In: Fermented Food-Part II: Technological Interventions, Ray, R. C., & Montet, D. (Eds.). CRC Press, Boca Raton. pp. 384-399. Kayodé, A. P. P., Hounhouigan, J. D., & Nout, M. J. R. (2007). Impact of brewing process on phytate, phenolic compounds and in vitro solubility of iron and zinc in opaque sorghum beer. Lebensmittel-Wissenschaft and Technologie, 40, 834-841. Khan, I., Yousif, A., Johnson, S. K., & Gamlath, S. (2013). Effect of sorghum flour addition on resistant starch content, phenolic profile and antioxidant capacity of durum wheat pasta. Food Research International, 54, 578-586. Kim, J., Hyun, T. K., & Kim, M. (2011a). The inhibitory effects of ethanol extracts from sorghum, foxtail millet and proso millet on α-glucosidase and α-amylase activities. Food Chemistry, 124, 1647-1651. Kim, J., Choi, J. N., John, K. M. M., Kang, D., Son, G. H., Kim, Y. S., Choi, H. K., Kwon, D. Y., & Lee, C. H. (2011b). Correlation between antioxidative activities and metabolite changes during cheonggukjang fermentation. Bioscience, Biotechnology, and Biochemistry, 75, 732-739. Kim, J., Choi, J. N., John, K. M. M., Kusano, M., Oikawa, A., Saito, K., & Lee, C. H. (2012). GC-TOF-MS- and CE-TOF-MS-based metabolic profiling of cheonggukjang (fast- fermented bean paste) during fermentation and its correlation with metabolic pathways. Journal of Agricultural and Food Chemistry, 60, 9746-9753. Kim, J., & Park, Y. (2012). Anti-diabetic effect of sorghum extract on hepatic gluconeogenesis of streptozotocin-induced diabetic rats. Nutrition and Metabolism, 9, 1-7. Kimber, C. T. (2000). Origins of domesticated sorghum and its early diffusion to China and India. In: Sorghum: Origin, History, Technology, and Production, Smith, C. W., & Frederiksen, R. A. (Eds.). Wiley, New York. pp. 3-97.

Kingamkono, R., Sjögren, E., Svanberg, U., & Kaijser, B. (1995). Inhibition of different strains of enteropathogens in a lactic-fermenting gruel. World Journal of Microbiology and Biotechnology, 11, 664-669. Klich, M. A. (2007). Environmental and developmental factors influencing aflatoxin production by Aspergillus flavus and Aspergillus parasiticus. Mycoscience, 48, 71-80. Kockova, M., Gerekova, P., Petrulakova, Z., Hybenova, E., Sturdik, E., & Valik, L. (2011). Evaluation of fermentation properties of lactic acid bacteria isolated from sourdough. Acta Chimica Slovaca, 4, 78-87. Koek, M. M., Jelleme, R. H. van der Greef, J., Tas, A. C., & Hankemeier, T. (2011). Quantitative metabolomics based on gas chromatography mass spectrometry: status and perspectives. Metabolomics, 7, 307-328. Kohajdova, Z., & Karovicova, J. (2007). Fermentation of cereals for specific purpose. Journal of Food Nutrition Research, 46, 51-57. Kolosova, A., & Stroka, J. (2011). Substances for reduction of the contamination of feed by mycotoxins: a review. World Mycotoxin Journal, 4, 225-256. Kruger, J., Taylor, J. R. N., & Oelofse, A. (2012). Effects of reducing phytate content in sorghum through genetic modification and fermentation on in vitro iron availability in whole grain porridges. Food Chemistry, 131, 220-224. Kumar, M., Verma, V., Nagpal, R., Kumar, A., Behare, P. V., & Singh, B. (2012).

Anticarcinogenic effect of probiotic fermented milk and chlorophyllin on aflatoxin-B1- induced liver carcinogenesis in rats. British Journal of Nutrition, 107, 1006-1016. Kyrø, C., Skeie, G, Dragsted, L. O., Christensen, J., Overvad, K., Hallmans, G., Johansson, I., Lund, E., Slimani, N., Johnsen, J. F., Halkjaer, J., Tjonneland, A., & Olsen, A. (2012). Intake of whole grain in Scandinavia: Intake, sources and compliance with new national recommendations. Scandinavian Journal of Public Health, 40, 76-84. Kwon, D. Y., Daily, J. W., Kim, H. J., & Park, S. (2010). The antidiabetic effects of fermented soybean products in type 2 diabetes. Nutrition Research, 30, 1-13. Lachman, J., Hejtmánková, A., Orsás, M., Popov, M., & Martinek, P. (2018). Tocotrienols and tocopherols in colored-grain wheat, tritordeum and barley. Food Chemistry, 240, 725-735. Lakshmi, K. B., & Vimala, V. (1996). Hypoglycemic effect of selected sorghum recipes. Nutrition Research, 16, 1651-1658. Lahouar, A., Marin, S., Crespo-Sempere, A., Saïd, S., & Sanchis, V. (2017). Influence of temperature, water activity and incubation time on fungal growth and production of ochratoxin A and zearalenone by toxigenic Aspergillus tubingensis and Fusarium

incarnatum isolates in sorghum seeds. International Journal of Food Microbiology, 242, 53- 60. Laureati, M., Conte, A., Padalino, L., Del Nobile, M. A., & Pagliarini, E. (2016). Effect of fiber information on consumer’s expectation and liking of wheat bran enriched pasta. Journal of Sensory Studies, 31, 348-359. LeBlanc, J. G., Laino, J. E., del Valle, M. J., Vannini, V., van Sinderen, D., Taranto, M. P., de Valdez, G. F., de Giori, G. S., & Sesma, F. (2011). B-group vitamin production by lactic acid bacteria--current knowledge and potential applications. Journal of Applied Microbiology, 111, 1297-1309. LeBlanc, J. G., Milani, C., De Giori, G. S., Sesma, F., van Sinderen, D., & Ventura, M. (2013). Bacteria as vitamin suppliers to their host: A gut microbiota perspective. Current Opinion in Biotechnology, 24, 160-168. Lee, J. E., Lee, B. J., Chung, J. O., Shin, H. J., Lee, S. J., Lee, C. H., & Hong, Y. S. (2011). 1H NMR-based metabolomic characterization during green tea (Camellia sinensis) fermentation. Food Research International, 44, 597-604. Lee, B. H., Carr, T. P., Weller, C. L., Cuppett, S., Dweikat, I. M., & Schlegel, V. (2014a). GS whole kernel oil lowers plasma and liver cholesterol in male hamsters with minimal wax involvement. Journal of Functional Foods, 7, 709-718. Lee, S. Y., Lee, S., Lee, S., Oh, J. Y., Jeon, E. J., Ryu, H. S., & Lee, C. H. (2014b). Primary and secondary metabolite profiling of doenjang, a fermented soybean paste during industrial processing. Food Chemistry, 165, 157-166. Lee, S., Seo, M. H., Oh, D. K., & Lee, C. H. (2014c). Targeted metabolomics for Aspergillus oryzae mediated biotransformation of soybean isoflavones, showing variations in primary metabolites. Bioscience Biotechnology and Biochemistry, 78, 167-174. Lee, S. H., Jung, J. Y., & Jeon, C. O. (2014d). Effects of temperature on microbial succession and metabolite change during saeu-jeot fermentation. Food Microbiology, 38, 16-25. Lee, D. E., Shin, G. R., Lee, S., Jang, E. S., Shin, H. W., Moon, B. S., & Lee, C. H. (2016a). Metabolomics reveal that amino acids are the main contributors to antioxidant activity in wheat and rice gochujangs (Korean fermented red pepper paste). Food Research International, 87, 10-17. Lee, D. E., Lee, S., Jang, E. S., Shin, H. W., Moon, B. S., & Lee, C. H. (2016b). Metabolomic profiles of Aspergillus oryzae and Bacillus amyloliquefaciens during rice koji fermentation. Molecules, 21, 1-15.

Leguizamón, C., Weller, C. L., Schlegel, V. L., & Carr, T. P. (2009). Plant sterol and policosanol characterization of hexane extracts from grain sorghum, corn and their DDGS. Journal of the American Oil Chemists' Society, 86, 707-716. Leroy, F., & De Vuyst, L. (2004). Lactic acid bacteria as functional starter cultures for the food fermentation industry. Trends in Food Science and Technology, 15, 67-78. Licata, R., Chu, J., Wang, S., Coorey, R., James, A., Zhao, Y., & Johnson, S. (2014). Determination of formulation and processing factors affecting slowly digestible starch, protein digestibility and antioxidant capacity of extruded sorghum-maize composite flour. International Journal Food Science Technology, 49, 1408-1419. Lin, P., Wong, J. H., Ng, T. B., Ho, V. S., & Xia, L. (2013). A sorghum xylanase inhibitor-like protein with highly potent antifungal, antitumor and HIV-1 reverse transcriptase inhibitory activities. Food Chemistry, 141, 2916-2922. Liptakova, D., Matejcekova, Z., & Valik, L. (2017). Lactic acid bacteria and fermentation of cereals and pseudocereals. In: Fermentation Processes, Jozala, A. F. (Ed.). InTech, Croatia. pp. 223-254. Llopart, E. E., Drago, S. R., De Greef, D. M., Torres, R. L., & Gonzalez, A. R. (2014). Effects of extrusion conditions on physical and nutritional properties of extruded whole grain red sorghum (sorghum spp). International Journal of Food Science and Nutrition, 65, 34-41. Lorri, W., & Svanberg, U. (1993). Lactic-fermented gruels with improved in vitro protein digestibility. International Journal of Food Science Nutrition, 44, 29-36. Lyumugabe, F., Kamaliza, G., Bajyana, E., & Thonart, P. H. (2010). Microbiological and physico-chemical characteristic of Rwandese traditional beer “ikigage”. African Journal of Biotechnology, 9, 4241-4246. Mabhaudhi, T., O’Reilly, P., Walker, S., & Mwale, S. (2016): Opportunities for underutilised crops in Southern Africa’s post-2015 development agenda. Sustainability, 302, 1-16. Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Madoroba, E., Steenkamp, E. T., Theron, J., Scheirlinck, I., Cloete, T. E., & Huys, G. (2011). Diversity and dynamics of bacterial populations during spontaneous sorghum fermentations used to produce ting, a South African food. Systematic and Applied Microbiology, 34, 227- 234.

Maki, K. C., Beiseigel, J. M., Jonnalagadda, S. S., Gugger, C. K., Reeves, M. S., Farmer, M. V., Kaden, V. N., & Rains, T. M. (2010). Whole-grain ready-to-eat oat cereal, as part of a dietary program for weight loss, reduces low-density lipoprotein cholesterol in adults with overweight and obesity more than a dietary program including low-fiber control foods. Journal of the American Dietetic Association, 110, 205-214. Makun, H. A., Gbodi, T. A., Akanya, H. O., Salako, E. A., & Ogbadu, G. H. (2009). Fungi and some mycotoxins found in mouldy sorghum in Niger State, Nigeria. World Journal of Agricultural Sciences, 5, 5-17. Makun, H. A., Dutton, M. F., Njobeh, P. B., Gbodi, T. M., & Ogbadu, G. H. (2012). Aflatoxin contamination in foods and feeds: A special focus on Africa. In: Trends in Vital Food and Control Engineering, Eissa, A. A. (Ed.). InTech, Croatia. pp. 187-234. Marasas, W. F., Jaskiewicz, K., Venter, F. S., & Van Schalkwyk, D. J. (1988). Fusarium moniliforme contamination of maize in oesophageal cancer areas in Transkei. South African Medical Journal, 74, 110-114. Marco, M. L., Heeney, D., Binda, S., Cifelli, C. J., Cotter, P. D., Foligne, B., Gänzle, M. G., Kort, R., Pasin, G., Pihlanto, A., Smid, E. J., & Hutkins, R. (2017). Health benefits of fermented foods: microbiota and beyond. Current Opinion in Biotechnology, 44, 94-102. Marsh, A. J., Hill, C., Ross, R. P., & Cotter, P. D. (2014). Fermented beverages with health- promoting potential: Past and future perspectives. Trends in Food Science and Technology, 38, 113-124. Marquart, L., Wiemer, K. L., Jones, J. M., & Jacob, B. (2003). Wholegrains health claims in the USA and other efforts to increase wholegrain consumption. The Proceedings of the Nutrition Society, 62, 151-160. Marquart, L., Jones, J. M., Chohen, E. A., & Poutanen, K. (2007). The future of whole grains. In: Whole Grains and Health, Marquart, L., Jacobs, D. R., McIntosh, G. H., Poutanen, K., & Reicks, M. (Eds.). Black Well Publishing, Iowa. pp. 3-16. Mavhungu, J. (2006). Isolation and characterization of lactic acid bacteria from “ting” in the Northern Province of South Africa. A MSc Dissertation submitted to the Faculty of Natural and Agricultural Science, University of Pretoria, South Africa. Mayorga-Gross, A. L., Quiros-Guerrero, L. M., Fourny, G., & Vaillant, F. (2016). An untargeted metabolomic assessment of cocoa beans during fermentation. Food Research International, 89, 901-909.

McIntosh, G. H., Noakes, M., Royle, P. J., & Foster, P. R. (2003). Whole-grain rye and wheat foods and markers of bowel health in overweight middle-aged men. Journal of Clinical Nutrition, 77, 967-974. McRae, M. P. (2017). Health benefits of dietary whole grains: an umbrella review of meta- analyses. Journal of Chiropractic Medicine, 16, 10-18. Mellen, P. B., Walsh, T. F., & Herrington, D. M. (2008) Whole grain intake and cardiovascular disease: a meta-analysis. Nutrition, Metabolism and Cardiovascular Diseases, 18, 283-290. Milani, J. M. (2013). Ecological conditions affecting mycotoxin production in cereals: a review. Veterinarni Medicina, 58, 405-411. Mintel (2011). Mintel global new products database. Whole grain statistics, Whole Grains Council. https://wholegrainscouncil.org/newsroom/whole-grain-statistics (Accessed 15/03/2017). Misihairabgwi, J. M., Ezekiel, C. N., Sulyok, M., Shepard, G. S., & Krska, R. (2017). Mycotoxin contamination of foods in Southern Africa: A 10-year review (2007–2016). Critical Reviews in Food Science and Nutrition. In Press. https://doi.org/10.1080/10408398.2017.1 357003. Mitaru, B. N., Reichert, R. D., & Blair, R. (1984). The binding of dietary protein by sorghum tannins in the digestive tract of pigs. The Journal of Nutrition, 114, 1787-1796. Mohammed, S. I., Steenson, L. R., & Kirleis, A. W. (1991). Isolation and characterization of microorganisms associated with the traditional sorghum fermentation for production of Sudanese kisra. Applied and Environmental Microbiology, 57, 2529-2533. Moss, M. O. (2002). Mycotoxin review - 2. Fusarium. Mycologist, 16, 158-161. Moss, A. F., Sydenham, C. J., Truong, H. H., Liu, S. Y., & Selle, P. H. (2017). The interactions of exogenous phytase with whole grain feeding and effects of barley as the whole grain component in broiler diets based on wheat, sorghum and wheat-sorghum blends. Animal Feed Science and Technology, 227, 1-12. Mourouti, N., Papavagelis, C., Plytzanopoulou, P., Kontogianni, M., Vassilakou, T., Malamos, N., Linos, A., & Panagiotakos, D. (2015). Dietary patterns and breast cancer: a case-control study in women. European Journal of Nutrition, 54, 609-617. Mozzi, F., Ortiz, M. E., Bleckwedel, J., Vuyst, L. D., & Pescuma, M. (2013). Metabolomics as a tool for the comprehensive understanding of fermented and functional foods with lactic acid bacteria. Food Research International, 54, 1152-1161.

Mugula, J. K., Narvhus, J. A., & Sorhaug, T. (2003). Use of starter cultures of LAB and yeasts in the preparation of togwa, a Tanzanian fermented food. International Journal of Food Microbiology, 83, 307-318. Muhihi, A. J., Shemaghembe, E., Njelekela, M. A., Gimbi, D., Mwambene, K., Malik, V. S., Wedick, N. M., Spiegelman, D., Hu, F. B., & Willet, W. C. (2012). Perceptions, facilitators, and barriers to consumption of whole grain staple foods among overweight and obese Tanzanian adults: a focus group study. ISRN Public Health, 1-7. Article ID 790602. Mukuru, S. Z., Butler, L. G., Rogler, J. C., Kirleis, A. W., Ejeta, G., Axtell, J. D., & Mertz, E. T. (1992). Traditional processing of high-tannin sorghum grain in Uganda and its effect on tannin, protein digestibility, and rat growth. Journal of Agricultural and Food Chemistry, 40, 1172-1175. Mwale, M. M. (2014). Microbiological quality and safety of the Zambian fermented cereal beverage: Chibwantu. A PhD Thesis submitted to the Faculty of Natural and Agricultural Sciences, University of Free State, South Africa. N’Dri, D., Mazzeo, T., Zaupa, M., Ferracane, R., Fogliano, V., & Pellegrini, N. (2012). Effect of cooking on the total antioxidant capacity and phenolic profile of some whole-meal African cereals. Journal of the Science of Food Agriculture, 93, 29-36. N’guessan, K. F., Brou, K., Jacques, N., Casaregola, S., & Djè, K. M. (2011). Identification of yeasts during alcoholic fermentation of tchapalo, a traditional sorghum beer from Côte d’Ivoire. Antonie Van Leeuwenhoek, 99, 855-864. Namgung, H. J., Park, H. J., Cho, I. H., Choi, H. K., Kwon, D. Y., & Shim, S. M. (2010). Metabolite profiling of doenjang, fermented soybean paste, during fermentation. Journal of the Science of Food Agriculture, 90, 1926-1935. Ndimba, R., Grootboom, A. W., Mehlo, L., Mkhonza, N. L., Kossmann, J., Barnabas, A. D., Mtshali, C., & Pineda-Vargas, C. (2015). Detecting changes in the nutritional value and elemental composition of transgenic sorghum grain. Nuclear Instruments and Methods in Physics Research B, 363, 183-187. Neucere, N. J., & Sumrell, G. J. (1980). Chemical composition of different varieties of grain sorghum. Journal of Agricultural and Food Chemistry, 28, 19-21. Njobeh, P. B., Dutton, M. F., & Makun, H. A. (2010). Mycotoxins and human health: significance, prevention and control. In: Smart Biomolecules in Medicine, Ajay, K. M., Ashutosh, T., & Shivanti, B. M. (Eds.). VBRI Press, India. pp. 132-177. Nout, M. J. R. (1991). Ecology of accelerated natural lactic fermentation of sorghum-based infant formulas. International Journal of Food Microbiology, 12, 217-224.

Nout, M. J. R., & Ngoddy, P. O. (1997). Technological aspects of preparing affordable fermented complementary foods. Food Control, 8, 279-287. Obizoba, I. C., Ezekwe, M. O., & Akaigwe, B. N. (1979). Utilization of sorghum, wheat, and navy beans by human adults: Protein metabolism. Nutrition Reports International, 20, 291- 301. Oboh, G., Akomolafe, T. L., & Adetuyi, A. O. (2010). Inhibition of cyclophosphamide-induced oxidative stress in brain by dietary inclusion of red dye extracts from sorghum (Sorghum bicolor) stem. Journal of Medicinal Food, 13, 1075-1080. Ogunremi, O. R., Banwo, K., & Sanni, A. I. (2017). Starter-culture to improve the quality of cereal-based fermented foods: trends in selection and application. Current Opinion in Food Science, 13, 38-43. Okarter, N., & Liu, R. H. (2011). Health benefits of whole grain phytochemicals. Critical Reviews in Food Science and Nutrition, 50, 193-208. Okeke, C. A., Ezekiel, C. N., Sulyok, M., Ogunremi, O. R., Ezeamagu, C. O., Sarkanj, B., Warth, B., & Krska, R. (2018). Traditional processing impacts mycotoxin levels and nutritional value of ogi – A maize-based complementary food. Food Control, 86, 224-233. Oliveira, P. M., Zannini, E., & Arendt, E. K. (2014). Cereal fungal infection, mycotoxins, and lactic acid bacteria mediated bioprotection: From crop farming to cereal products. Food Microbiology, 37, 78-95. Omemu, A. M., & Bankole, M. O. (2015). Consumer’s knowledge, attitude, usage and storage pattern of ogi – a fermented cereal gruel in South West, Nigeria. Food Public Health, 5, 77- 83. Osman, M. A. (2004). Changes in sorghum enzyme inhibitors, phytic acid, tannins and in vitro protein digestibility occurring during Khamir (local bread) fermentation. Food Chemistry, 88, 129-134. Osman, M. A. (2011). Effect of traditional fermentation process on the nutrient and antinutrient contents of pearl millet during preparation of Lohoh. Journal of Saudi Society of Agriculture and Science, 10, 1-6. Oueslati, S., Romero-González, R., Lasram, S., Frenich, A. G., & Martínez Vidal, J. L. (2012). Multi-mycotoxin determination in cereals and derived products marketed in Tunisia using ultra-high performance liquid chromatography coupled to triple quadrupole mass spectrometry. Food and Chemical Toxicology, 50, 2376-2381.

Pang, Y., Ahmed, S., Xu, Y., Beta, T., Zhu, Z., Shao, Y., & Bao, J. (2018). Bound phenolic compounds and antioxidant properties of whole grain and bran of white, red and black rice. Food Chemistry, 240, 212-221. Park, M. K., Cho, I. H., Lee, S., Choi, H. K., Kwon, D. Y., & Kim, Y. S. (2010). Metabolite profiling of cheonggukjang, a fermented soybean paste, during fermentation by gas chromatography mass spectrometry and principal component analysis. Food Chemistry, 122, 1313-1319. Park, S. E., Yoo, S. A., Seo, S. H., Lee, K. I., Na, C. S., & Son, H. S. (2016). GC-MS based metabolomics approach of Kimchi for the understanding of Lactobacillus plantarum fermentation characteristics. LWT-Food Science and Technology, 68, 313-321. Pascari, X., Ramos, A. J., Marín, S., & Sanchís, V. (2018). Mycotoxins and beer. Impact of beer production process on mycotoxin contamination. A review. Food Research International, 103, 121-129. Pitt, J. (2008). Understanding plant–fungus associations as a key to mycotoxin control. In: Food Contaminants, Siantar, D., Trucksess, M., Scott, P., & Herman, E. (Eds.). ACS, Washington. pp. 96-108. Pleadin, J., Zadravec, M., Brnic, D., Perkovic, I., Skrivanko, M., & Kovacevic, D. (2017). Moulds and mycotoxins detected in the regional speciality fermented sausage ‘slavonski kulen’ during a 1-year production period. Food Additives and Contaminants: Part A, 34, 282-290. Porter, L. J., Hrstich, L. N., & Chan, B. G. (1986). The conversion of procyanidins and prodelphinidins to cyanidin and delphinidin. Phytochemistry, 25, 223-230. Pranoto, Y., Anggrahini, S., & Efendi, Z. (2013). Effect of natural and Lactobacillus plantarum fermentation on in-vitro protein and starch digestibilities of sorghum flour. Food Bioscience, 2, 46-52. Priebe, M. G., van Binsbergen, J. J., de Vos, R., & Vonk, R. J. (2008). Whole grain foods for the prevention of type 2 diabetes mellitus. The Cochrane Database of Systematic Reviews, 23, CD006061. Poquette, N. M., Gu, X., & Lee, S. O. (2014). Grain sorghum muffin reduces glucose and insulin responses in men. Food and Function, 5, 894-899. Qiu, Y., & Reed, D. (2014). Gas chromatography in metabolomics study. In: Advances in Gas Chromatography, Guo, X. (Ed.). InTech, Croatia. pp. 83-101. Ragaee, S., & Abdel-Aal, E. M. (2006). Pasting properties of starch and protein in selected cereals and quality of their food products. Food Chemistry, 95, 9-18.

Rathore, S., Salmeron, I., & Pandiella, S. (2012). Production of potentially probiotic beverages using single and mixed cereal substrates fermented with lactic acid bacteria cultures. Food Microbiology, 30, 239-244. Rao, K. P., Chennappa, G., Suraj, U., Nagaraja, H., Raj, A. P., & Sreenivasa, M. Y. (2015). Probiotic potential of Lactobacillus strains isolated from sorghum-based traditional fermented food. Probiotics and Antimicrobial Proteins, 7, 146-156. Ray, R. C., & Joshi, V. K. (2015). Fermented foods: past, present and future. In: Microorganisms and fermentation of traditional foods, Ray, R. C., & Montet, D. (Eds.). CRC Press, Boca Raton. pp. 1-36. Reis, J. A., Paula, A. T., Casarotti, S. N., & Penna, A. L. B. (2012). Lactic acid bacteria antimicrobial compounds: characteristics and applications. Food Engineering Reviews, 4, 124-140. Rooney, L. W., & Serna-Saldivar, S. O. (2000). Sorghum. In: Handbook of Cereal Science and Technology (2nd ed), Kulp, K., & Ponte, J. G. (Eds.). Marcel Dekker, New York. pp. 149- 175. Rooney, L. W., & Awika, J. M. (2005). Specialty sorghums for healthful foods. In: Specialty Grains for Food and Feed, Abdel-Aal, E. S. M., & Wood, P. (Eds.). American Association of Cereal Chemists, St Paul, Minnesota. pp. 283-312. Rosales, E., Pasoz, M., & Sanroman, M. A. (2018). Solid-state fermentation for food applications. In: Current Developments in Biotechnology and Bioengineering, Pandey, A., Larroche, C., & Soccol, C. (Eds.). Elsevier, Amsterdam. pp. 319-355. Rosentrater, K. A., & Evers, A. D. (2018a). Introduction to cereals and pseudocereals and their production. In: Kent’s Technology of Cereals (5th ed), Elsevier, Amsterdam. pp. 1-76. Rosentrater, K. A., & Evers, A. D. (2018b). Other food products. In: Kent’s Technology of Cereals (5th ed), Elsevier, Amsterdam. pp. 699-727. Rose, D. J., Williams, E., Mkandawire, N. L., Weller, C. L., & Jackson, D. S. (2013). Use of whole grain and refined flour from tannin and non-tannin sorghum (Sorghum bicolor (L.) Moench) varieties in frybread. Food Science and Technology International, 20, 333-339. Ross, A. B., Bruce, S. J., Blondel-Lubrano, A., Oguey-Araymon, S., Beaumont,M., Bourgeois, A., Nielsen-Moennoz, C., Vigo, M., Fay, L. B., Kochhar, S., Bibiloni, R., Pittet, A. C., Emady-Azar, S., Grathwohl, D., & Rezzi, S. (2011). A whole-grain cereal-rich diet increases plasma betaine, and tends to decrease total and LDL-cholesterol compared with a refined- grain diet in healthy subjects. British Journal of Nutrition, 105, 1492-1502.

Russo, P., Spano, G., & Capozzi, V. (2017). Safety evaluation of starter cultures. In: Starter cultures in food production, Speranza, B., Bevilacqua, A., Corbo, M. R., & Sinigaglia, M. (Eds.). Wiley-Blackwell, UK. pp 101-128. Salawu, S. O., Bester, M. J., & Duodu, K. G. (2013). Phenolic composition and bioactive properties of cell wall preparations and whole grains of selected cereals and legumes. Journal of Food Biochemistry, 38, 62-72. Sanlier, N., Gókcen, B. B., & Sezgin, A. C. (2017). Health benefits of fermented foods. Critical Reviews in Food Science and Nutrition. In Press. https://doi.org/10.1080/10408398.2017.1 383355 Sarmadi, B. H., & Ismail, A. (2010). Antioxidative peptides from food proteins: a review. Peptides, 31, 1949-1956. Sawadogo-Lingani, H., Diawara, B., Traore, A. S., & Jakobsen, M. (2008). Technological properties of Lactobacillus fermentum in the processing of dolo and pito, West African sorghum beers, for selection of starter cultures. Journal of Applied Microbiology, 104, 873- 882. Schaffer-Lequart, C., Lehmann, U., Ross, A. B., Roger, O., Eldridge, A. L., Ananta, E., Bietry, M. F., King, L. R., Moroni, A. V., Srichuwong, S., Wavreille, A. S., Redgwell, R., Labat, E., & Robin, F. (2017). Whole grain in manufactured foods: Current use, challenges and the way forward. Critical Reviews in Food Science and Nutrition, 57, 1562-1568. Schober, T. J., & Bean, S. R. (2008). Sorghum and maize. In: Gluten-free cereal products and beverages, Arendt, E. K., & Bello, F. D. (Eds.). Elsevier, Netherlands. pp. 101-118. Schons, P. F., Ries, E. F., Battestin, V., & Macedo, G. A. (2011). Effect of enzymatic treatment on tannins and phytate in sorghum (Sorghum bicolor) and its nutritional study in rats. International Journal of Food Science and Technology, 46, 1253-1258. Schons, P. F., Battestin, V., & Macedo, G. A. (2012). Fermentation and enzyme treatments for sorghum. Brazilian Journal of Microbiology, 43, 89-97. Seal, C. J., & Brownlee, I. A. (2015). Whole-grain foods and chronic disease: evidence from epidemiological and intervention studies. The Proceedings of the Nutrition Society, 74, 313- 319. Sekwati-Monang, B. (2011). Microbiological and chemical characterization of ting, a sorghum based gluten-free fermented cereal product from Botswana. A PhD Thesis submitted to the Faculty of Graduate Studies and Research, University of Alberta, Canada.

Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Sekwati-Monang, B., Valcheva, R., & Ganzle, M. G. (2012). Microbial ecology of sorghum sourdoughs: Effect of substrate supply and phenolic compounds on composition of fermentation microbiota. International Journal of Food Microbiology, 159, 240-246. Seo, S. H., Park, S. E., Yoo, S. A., Lee, K. I., Na, C. S., & Son, H. S. (2016). Metabolite profiling of Makgeolli for the understanding of yeast fermentation characteristics during fermentation and aging. Process Biochemistry, 51, 1363-1373. Shargie, N. (2012). Physico-chemical characteristics and nutritional value of sorghum grain. http://www.grainsa.co.za/physio-chemical-characteristics-and-nutritional-value-of-sorghu m-grain (Accessed 21/10/2017). Shawrang, P., Sadeghi, A. A., Behgar, M., Zareshahi, H., & Shahhoseini, G. (2011). Study of chemical compositions, anti-nutritional contents and digestibility of electron beam irradiated sorghum grains. Food Chemistry, 125, 376-379. Shephard, G. S., Marasas, W. F., Leggott, N. L., Yazdanpanah, H., Rahimian, H., & Safavi, N. (2000). Natural occurrence of fumonisins in corn from Iran. Journal of Agricultural and Food Chemistry, 48, 1860-1864. Shih, C. H., Siu, S. O., Ng, R., Wong, E., Chiu, L. C. M., Chu, I. K., & Lo, C. (2007). Quantitative analysis of anticancer 3-deoxyanthocyanidins in infected sorghum seedlings. Journal of Agricultural and Food Chemistry, 55, 254-259. Sieniawska, E., & Baj, T. (2017). Tannins. In: Pharmacognosy: Fundamentals, Applications and Strategies, McCreath, S. B., & Delgoda, R. (Eds.). Elsevier, Amsterdam. pp. 199-232. Singh, V., Moreau, R. A., & Hicks, K. B. (2003). Yield and phytosterol composition of oil extracted from grain sorghum and its wet-milled fractions. Cereal Chemistry, 80, 126-129. Singh, E., Jain, P. K., & Sharma, S. (2015a). Effect of different household processing on nutritional and anti-nutritional factors in Vigna aconitifolia and Sorghum bicolour (L.) Moench seeds and their product development. Journal of Medical Nutrition and Nutraceuticals, 4, 95-100. Singh, A. K., Rehal, J., Kaur, A., & Jyot, A. (2015b). Enhancement of attributes of cereals by germination and fermentation: A review. Critical Reviews in Food Science and Nutrition, 55, 1575-1589. Singhal, P., & Kaushik, G. (2016). Therapeutic effect of cereal grains: A review. Critical Reviews in Food Science and Nutrition, 56, 748-759.

Singh, A., & Sharma, S. (2017). Bioactive components and functional properties of biologically activated cereal grains: A bibliographic review. Critical Reviews in Food Science and Nutrition, 14, 3051-3071. Sirma, A. J., Senerwa, D. M., Grace, D., Makita, K., Mtimet, N., Kang’ethe, E. K., & Lindahl,

J. F. (2016). Aflatoxin B1 occurrence in millet, sorghum and maize from four agro-ecological zones in Kenya. African Journal of Food, Agriculture, Nutrition and Development, 16, 10991-11003. Soro-Yao, A. A., Brou, K., Amani, G., Thonart, P., & Dje. K. M. (2014). The use of lactic acid bacteria starter cultures during the processing of fermented cereal-based foods in West Africa: A review. Tropical Life Sciences Research, 25, 81-100. Srianta, I., & Harojono, A. (2015). Monascus-fermented sorghum: pigments and monacolin K produced by Monascus purpureus on whole grain, dehulled grain and bran substrates. International Food Research Journal, 22, 377-382. Stefoska-Needham, A., Beck, E. J., Johnson, S. K., & Tapsell, L. C. (2015). Sorghum: An underutilized cereal whole grain with potential to assist in the prevention of chronic disease. Food Reviews International, 31, 401-437. Sun, G., Wang, S., Hu, X., Su, J., Huang, T., Yu, J., Tang, L., Gao, W., & Wang, J. S. (2007).

Fumonisin B1 contamination of home-grown corn in high-risk areas for esophageal and liver cancer in China. Food Additives and Contaminants, 24, 181-185. Swyngedouw, C., & Lessard, R. (2018). Gas chromatographic methods of chemical analysis of organics and their quality control. In: Environmental Geochemistry (2nd ed), De Vivo, B., Belkin, H. E., & Lima, A. (Eds.). Elsevier, Amsterdam. pp. 103-113. Svanberg, U., Sjögren, E., Lorri, W., Svennerholm, A-M., & Kaijser, B. (1992). Inhibited growth of common enteropathogenic bacteria in lactic-fermented gruels. World Journal of Microbiology and Biotechnology, 8, 601-606. Svensson, L., Sekwati-Monang, B., Lutz, D. L., Schieber A., & Gänzle, M. G. (2010). Phenolic acids and flavonoids in nonfermented and fermented red sorghum (Sorghum bicolor (L.) Moench). Journal of Agricultural and Food Chemistry, 58, 9214-9220. Tabaraki, R., & Heidarizadi, E. (2017). Simultaneous multidye biosorption by chemically modified Sargassum glaucescens: Doehlert optimization and kinetic, equilibrium, and thermodynamic study in ternary system. Separation Science and Technology, 52, 583-595. Tabbers, M. M., Chmielewska, A., Roseboom, M. G., Crastes, N., Perrin, C., & Reitsma, J. B. (2011). Fermented milk containing Bifidobacterium lactis DN-173 010 in childhood constipation: a randomized, double-blind, controlled trial. Pediatrics, 127, 1392-1399.

Tamang, J. P. (2007). Fermented foods for human life. In: Microbes for Human Life, Chauhan, A. K., Verma, A., & Kharakwal, H. (Eds.). I.K International Publishing House Pvt., New Delhi, India. pp. 73-87. Tan, J., Dai, W., Lu, M., Lv, H., Guo, L., Zhang, Y., Zhu, Y., Peng, Q., & Lin, Z. (2016). Study of the dynamic changes in the non-volatile chemical constituents of black tea during fermentation processing by a non-targeted metabolomics approach. Food Research International, 79, 106-113.

Taye, W., Ayalew, A., Chala, A., & Denjene, M. (2016). Aflatoxin B1 and total fumonisin contamination and their producing fungi in fresh and stored sorghum grain in East Hararghe, Ethiopia. Food Additives and Contaminants: Part B, 9, 237-245. Taye, W., Ayalew, A., Denjene, M., & Chala, A. (2018). Fungal invasion and mycotoxin contamination of stored sorghum grain as influenced by threshing methods. International Journal of Pest Management, 64, 66-76. Taylor, J., & Schüssler, L. (1986). The protein compositions of the different anatomical parts of sorghum grain. Journal of Cereal Science, 4, 361-369. Taylor, J., & Taylor, J. R. N. (2002). Alleviation of the adverse effect of cooking on sorghum protein digestibility through fermentation in traditional African porridges. International Journal of Food Science and Technology, 37, 129-137. Taylor, J. R. N., Schober, T. J., & Bean, S. R. (2006). Novel food and non-food uses for sorghum and millets. Journal of Cereal Science, 44, 252-271. Taylor, J. R. N., & Emmambux, M. N. (2008). Gluten-free cereal products and beverages. In: Gluten-Free Foods and Beverages from Millets, Arendt, E. K., & Bello, F. D. (Eds.). Elsevier, Amsterdam. pp. 119-148. Taylor, J. R. N., & Emmambux, M. N. (2010). Review: Developments in our understanding of sorghum polysaccharides and their health benefits. Cereal Chemistry, 87, 263-271. Taylor, J., & Taylor, J. R. N. (2011). Protein biofortified sorghum: Effect of processing into traditional African foods on their protein quality. Journal of Agricultural and Food Chemistry, 59, 2386-2392. Taylor, J. R. N., & Anyango, J. O. (2011). Sorghum flour and flour products: Production, nutritional quality, and fortification. In: Flour and Breads and their Fortification in Health and Disease Prevention, Preedy, V., Watson, R., & Patel, V. (Eds.). Academic Press, Oxford. pp. 127-139. Taylor, J. R. N., Belton, P. S., Beta, T., & Duodu, K. G. (2014). Increasing the utilization of sorghum, millets and pseudocereals: Developments in the science of their phenolic

phytochemicals, biofortification and protein functionality. Journal of Cereal Science, 59, 257-275. Taylor, J. R. N., & Duodu, K. G. (2015). Effects of processing sorghum and millets on their phenolic phytochemicals and the implications of this to the health-enhancing properties of sorghum and millet food and beverage products. Journal of the Science of Food and Agriculture, 95, 225-237. Taylor, J. R. N., & Duodu, K. G. (2017). Sorghum and millets: grain quality characteristics and management of quality requirements. In: Cereal Grains, Wrigley, C., Batey, I., & Miskelly, D. (Eds.). Elsevier, Amsterdam. pp. 317-351. Tighe, P., Duthie G., Vaughan, N., Brittenden, J., Simpson, W. G., Duthie, S., Mutch, W., Wahle, K., Horgan, G., & Thies, F. (2010). Effect of increased consumption of whole-grain foods on blood pressure and other cardiovascular risk markers in healthy middle-aged persons: A randomized controlled trial. American Journal of Clinical Nutrition, 92, 733-740. Tillisch, K., Labus, J., Kilpatrick, L., Jiang, Z., Stains, J., & Ebrat, B. (2013). Consumption of fermented milk product with probiotic modulates brain activity. Gastroenterology, 144, 1394-1401. Todorov, S. D., & Holzapfel, W. H. (2015). Traditional cereal fermented foods as sources of functional microorganisms. In: Advances in Fermented Foods and Beverages: Improving Quality, Technologies and Health Benefits, Holzapfel, W. H. (Ed.). Elsevier, UK. pp. 123- 153. Togo, A. H., Ferusa, B. S., & Mutukumira, N. A. (2002). Identification of lactic acid bacteria isolated from opaque beer (chibuku) for potential use as a starter culture. The Journal of Food Technology in Africa, 7, 93-97. Topping, D. L. (1991). Soluble fiber polysaccharides: Effects on plasma cholesterol and colonic fermentation. Nutrition Reviews, 49, 195-203. Towo, E., Matuschek, E., & Svanberg, U. (2006). Fermentation and enzyme treatment of tannin sorghum gruels: effects on phenolic compounds, phytate and in vitro accessible iron. Food Chemistry, 94, 369-376. Tugizimana, F., Paiter, L., & Dubery, I. (2013). Plant metabolomics: a new frontier in phytochemical analysis. South African Journal of Science, 109, 1-11. Udachan, I. S., Sahoo, A. K., & Hend, G. M. (2012). Extraction and characterization of sorghum (Sorghum bicolor L. Moench) starch. International Food Research Journal, 19, 315-319.

Udomkun, P., Wiredu, A. N., Nagle, M., Bandyopadhyay, R., Muller, J., & Vanlauwe, B. (2017). Mycotoxins in Sub-Saharan Africa: Present situation, socio-economic impact, awareness, and outlook. Food Control, 72, 110-122. USAID (United States Agency for International Development) (2012). Aflatoxin: A synthesis of the research in health, agriculture and trade. In: Feed the Future, The Office of Regional Economic Integration USAID East Africa Regional Mission Nairobi, Kenya. http://www.danya.com/portfolio/aflatoxin_report.pdf (Accessed, 11/12/2017). USFDA (US Food and Drug Administration). (2006). Draft guidance for industry and FDA staff: Whole grains label statements; availability. https://www.federalregister.gov/documents/20 06/02/17/06-1509/draft-guidance-for- industry-and-fda-staff-whole-grains-label-statement s- availability (Accessed 31/10/2017). Valdez, G. F., Gerez, C. L., Torino, M. I., & Rollán, G. (2010). New trends in cereal-based products using lactic acid bacteria. In: Biotechnology of Lactic Acid Bacteria Novel Applications, Mozzi, F., Raya, R. R., & Vignolo, G. M. (Eds.). Wiley-Blackwell, Iowa. pp. 273-288. van der Kamp, J. W., Poutanen, K., Seal, C. J., & Richardson, D. P. (2014). The health grain definition of ‘whole grain’. Food and Nutrition Research, 58, 1-8. van Rensburg, S. J. (1981). Epidemiological and dietary evidence for a specific nutritional disposition to esophageal cancer. Journal of the National Cancer Institute, 67, 243-241. Vieira-Dalodé, G., Jespersen, L., Hounhouigan, J., Moller, P. L., Nago, C. M., & Jakobsen, M. (2007). Lactic acid bacteria and yeasts associated with gowé production from sorghum in Bénin. Journal of Applied Microbiology, 103, 342-349. Vieira Queiroz, V. A., da Silva, C. S., de Menezes, C. B., Schaffert, E., Guimaraes, F. F. M., Guimaraes, L. J. M., de Oliveira Guimaeraes, P. E., & Tardin, F. D. (2015). Nutritional composition of sorghum [Sorghum bicolor (L.) Moench] genotypes cultivated without and with water stress. Journal of Cereal Science, 65, 103-111. Viljoen, J. H. (2003). Mycotoxins in grain and grain products in South Africa and proposals for their regulation. A PhD Thesis submitted to the Faculty of Natural and Agricultural Sciences, University of Pretoria, South Africa. Wagacha, J. M., & Muthomi, J. W. (2008). Mycotoxin problem in Africa: current status, implications to food safety and health and possible management strategies. International Journal of Food Microbiology, 124, 1-12. Wardhani, D. H., Vázquez, J. A., & Pandiella, S. S. (2010). Optimisation of antioxidants extraction from soybeans fermented by Aspergillus oryzae. Food Chemistry, 118, 731-739.

Wang, R. S., & Kies, C. (1991). Niacin status of humans as affected by eating decorticated and whole ground sorghum (Sorghum gramineae) grain, ready-to-eat breakfast cereals. Plant Foods for Human Nutrition, 41, 355-369. Wang, L., Weller, C. L., Schlegel, V. L., Carr, T. P., & Cuppett, S. L. (2007). Comparison of

supercritical CO2 and hexane extraction of lipids from sorghum distillers grains. European Journal of Lipid Science and Technology, 109, 567-574. Wang, C. Y., Wu, S. J., & Shyu, Y. T. (2013). Antioxidant properties of certain cereals as affected by food-grade bacteria fermentation. Journal of Bioscience and Bioengineering, 117, 449-456. Warrand, J. (2006). Healthy polysaccharides the next chapter in food products. Food Technology and Biotechnology, 44, 355-370. Warth, B., Parich, A., Atehnkeng, J., Bandyopadhyay, R., Schuhmacher, R., Sulyok, M., & Krska, R. (2012). Quantitation of mycotoxins in food and feed from Burkina Faso and Mozambique using a modern LC-MS/MS multitoxin method. Journal of Agricultural and Food Chemistry, 60, 9352-936. Wild, C. P., & Gong, Y. Y. (2010). Mycotoxins and human disease: a largely ignored global health issue. Carcinogenesis, 31, 71-82. Willumsen, J. F., Darling, J. C., Kitundu. J. A., Kingamkono, R. R., Msengi, A. E., & Mduma, B. (1997). Dietary management of acute diarrhoea in children: effect of fermented and amylase-digested weaning foods on intestinal permeability. Journal of Pediatric Gastroenterology Nutrition, 24, 235-242. Wishart, D. S. (2008). Metabolomics: Applications to food science and nutrition research. Trends in Food Science and Technology, 19, 482-493. Wrigley, C. (2017). The cereal grains: providing our food, feed and fuel needs. In: Cereal Grains, Wrigley, C., Batey, I., & Miskelly, D. (Eds.). Elsevier, Netherlands. pp. 27-40. Wu, Q., Jezkova, A., Yuan, Z., Pavlikova, L., Dohnal, V., & Kuca, K. (2009). Biological degradation of aflatoxins. Drug Metabolism Reviews, 41, 1-7. Wu, H., Rui, X., Li, W., Chen, X., Jiang, M., & Dong, M. (2015). Mung bean (Vigna radiata) as probiotic food through fermentation with Lactobacillus plantarum B1-6. LWT-Food Science and Technology, 63, 445-451. Wu, F., Groopman, J. D., & Pestka, J. J. (2014). Public health impact of foodborne mycotoxins. Annual Reviews in Food Science and Technology, 5, 351-372.

Yan, Z., Zheng, X. W., Han, B. Z., Yan, Y. Z., Zhang, X., & Chen, J. Y. (2015). 1H NMR- based metabolomics approach for understanding the fermentation behaviour of Bacillus licheniformis. Journal of the Institute of Brewing, 121, 425-431. Yang, S. O., Kim, M. S., Liu, K. H., Auh, J. H., Kim, Y. S., Kwon, D. Y., & Choi, H. K. (2009a). Classification of fermented soybean paste during fermentation by 1H nuclear magnetic resonance spectroscopy and principal component analysis. Bioscience Biotechnology and Biochemistry, 73, 502-507. Yang, S. O., Kim, S. H., Cho, S., Lee, J. H., Kim, Y. S., Yun, S. S., & Choi, H. K. (2009b). Classification of fermented soymilk during fermentation by 1H NMR coupled with principal component analysis and elucidation of free-radical scavenging activities. Bioscience Biotechnology and Biochemistry, 73, 1184-1188. Ye, E. Q., Chacko, S. A., Chou, E. L., Kugizaki, M., &Liu, S. (2012). Greater whole-grain intake is associated with lower risk of type 2 diabetes, cardiovascular disease, and weight gain. Journal of Nutrition, 142, 1304-1313. Yolmeh, M., & Jafari, S. M. (2017). Applications of response surface methodology in the food industry processes. Food Bioprocess Technology, 10, 413-433. Yoshida, H., Yamazaki, J., Ozawa, S., Mizukoshi, T., & Miyano, H. (2009). Advantage of LC- MS metabolomics methodology targeting hydrophilic compounds in the studies of fermented food samples. Journal of Agricultural and Food Chemistry, 57, 1119-1126. Yousif, N. M. K., Huch, M., Schuster, T., Cho, G. S., Dirar, H. A., Holzapfel, W. H., & Franz, C. M. A. P. (2010). Diversity of lactic acid bacteria from Hussuwa, a traditional African fermented sorghum food. Food Microbiology, 6, 757-768. Yousif, A., Nhepera, D., & Johnson, S. (2012). Influence of sorghum flour addition on flat bread in vitro starch digestibility, antioxidant capacity and consumer acceptability. Food Chemistry, 134, 880-887. Zhang, G., & Hamaker, B. (2012). Nutraceutical and health properties of sorghum and millet. In: Cereals and Pulses: Nutraceutical Properties and Health Benefits, Liangli, Y., Tsao, R., & Shahidi, F. (Eds.). Wiley, Ames. pp. 165-186.

CHAPTER THREE1 Optimization of fermentation conditions and its effects on physicochemical properties, bioactive components and microstructure of “modified” ting from whole grain sorghum

Abstract Whole grain (WG) foods contain high levels of bioactive and health functional components that confer beneficial effects. In line with this, the effect of fermentation conditions (time and temperature) of two sorghum varieties [low tannin (LT) and high tannin (HT)] on the composition of WG-ting subsequently obtained from them, was investigated in this study. Using the Doehlert design of response surface methodology (RSM), fermentation conditions were optimized and pH, titratable acidity (TTA), total viable bacteria count (TBC), total lactic acid bacteria count (TLABC), total fungal and yeast count (TFYC), tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC) and antioxidant activities (AA) were determined. Experimental and predicted values obtained were similar, with statistical indices indicating the validity of the models generated (R2 between 89.47–99.93%, AAD values close to 0, Bf and Af values close to 1). Numerical multi-response optimization of parameters indicated the optimal fermentation conditions for LT-ting to be 34 oC for 24 h and HT-ting to be 28oC for 72 h. At these conditions, reduced pH, high TTA, desired microbial growth accompanied with relatively high values of TNC, TPC, TFC and AA were obtained for the LT- and HT-ting samples. Scanning electron microscopy (SEM) showed slight changes in the morphology of LT-ting samples and no considerable change in that of HT-ting. Findings from this study provide information that would facilitate production of quality WG-ting, with better health beneficial composition, potential for commercialization and improved intake of health promoting components. Keywords: Sorghum, ting, fermentation, optimization, physicochemical properties, and whole grain.

1Part of this study has been published in the Journal of Food Processing and Preservation and Food Bioscience.

3.0 Introduction The significant impact of diet on health has led to an increased demand for functional foods such as whole grain (WG) meals (García-Mantrana et al., 2016). Worldwide consumption of products containing WGs has tremendously grown due to increasing awareness of their health- promoting benefits and importance in diets (Marquart et al., 2007; Mintel, 2011; Schaffer- Lequart et al., 2017). Different epidemiological studies have equally indicated the significant and important role of WG diets, including their protective action against cancers, diabetes, obesity and cardiovascular diseases (Jones et al., 2002; Seal and Brownlee, 2015; McRae, 2017).

Sorghum is an important cereal crop and major source of food for millions of people. Phenolic compounds in sorghum and resultant sorghum-based food products have been identified as antioxidants conferring functional and nutraceutical effects (Awika and Rooney, 2004; Taylor and Duodu, 2015; Awika, 2017). These health beneficial components are mostly located in the seed coat of sorghum (Awika, 2017) and their presence in WG sorghum-based foods is thus more beneficial as compared to their refined counterpart. Due to these health effects, eating habits are gradually changing and the potential of functional and nutraceutical foods in mitigating health problems, are recently being encouraged.

Sorghum like other cereals is transformed into edible forms using fermentation, a food processing technique known to enhance nutritional qualities, shelf life, bioavailability of nutrients, palatability, beneficial health promoting components and consumer appeal (Taylor and Duodu, 2015; Adebiyi et al., 2018). Ting is a sorghum fermented product commonly consumed as a gruel (motogo) or porridge (bogobe) in Botswana, South Africa and other neighboring countries. It is known for its sour taste, unique flavor and used as a weaning food for infants as well as consumed during ceremonies (Sekwati-Monang and Gänzle, 2011).

Different fermentation conditions for the preparation of ting have been reported in the literature (Taylor and Taylor, 2002; Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011), with variations in their fermentation process and no available standardized or optimized conditions. Product formation during fermentation depends on extrinsic factors such as fermentation time and temperature, making the final quality of the product significantly dependent on the regime and combination of these conditions. Differences in these conditions affect ting quality, therefore necessitating an optimization of the fermentation process to obtain a product of better quality. A widely accepted optimization procedure is response surface methodology (RSM),

73 which is a collection of statistical and mathematical methods for obtaining the optimum conditions of factors for desirable responses. The Doehlert design of RSM are easily applied to optimize variables more effectively as they can explore the whole of an experimental domain with fewer experiments (Ferreira et al., 2004; Tabaraki and Heidarizadi, 2017).

Few studies have been presented in the literature on characterizing the microbiota and selected chemical properties of ting (Taylor and Taylor, 2002; Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). None of these studies have however, optimized fermentation variables and subsequently investigated their effects on the composition and physicochemical properties of WG-ting. Hence, this study investigated the effect of fermentation conditions (time and temperature) on the biochemical properties, bioactive components and microstructure of WG-ting.

3.1 Materials and Methods 3.1.1 Raw material and sample preparation Sorghum (Sorghum bicolor L.) grains Titan (low tannin) and Avenger (high tannin) types were purchased from Agricol (Pty) Ltd., Potchefstroom, South Africa. The sorghum grains were milled through a 2mm screen using a Perten Laboratory Mill 3600 (Perten Instruments, Sweden) to obtain whole grain (WG) sorghum flour (Figure 3.1). Tannin content (TNC) of sorghum types were determined, with values recorded as 17.97 and 31.68 mg CE/g, and as such, subsequently classified as low tannin (LT) and high tannin (HT) sorghum, respectively.

Figure 3.1: The sorghum grains used in this study (above) and their corresponding whole grain flours (below). HT- (left) and LT-sorghum (right) sorghum

3.1.2 Spontaneous fermentation of whole grain sorghum flour into whole grain ting Using the method of Madoroba et al. (2009), WG-sorghum flour was processed into WG-ting by mixing 50 mL sterile distilled water (40 oC) and the 50 g sorghum flour (1:1, w/v). The mixture was subsequently allowed to ferment by endogenous microflora using some fermentation conditions. For each experimental run, the fermentation process was done in triplicates.

3.1.3 Optimization of WG-ting production process A response surface methodology (RSM) approach using the Doehlert design was used to model and optimize the effect of fermentation conditions on the parameters investigated. The independent variables were fermentation time (푋1) and fermentation temperature (푋2), with intervals of 24–72 h and 20–34 oC respectively. The selection of the parameter levels was based on other studies in the literature on the production of ting from refined grains (Taylor and Taylor, 2002; Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011). Mathematical Eq. (1) and (2) were used to convert the coded independent variables into real values.

푋푖 = 푋0푖 + 푥푖 × ∆푋푖 (1) 2 푁 = 푘 + 푘 + 푘0 (2)

Where Xi is a real variable, X0i is the center of the variable; 푥푖 is a coded variable given by the

Doehlert design, ∆Xi is the increment, k is the number of variables, k0 is the number of center points and N is the number of experiments.

The two-factor Doehlert design gave a total of eight experimental runs (Table 3.1) and the parameters determined in the WG-ting samples were pH, total titratable acidity (TTA), total bacteria count (TBC), total lactic acid bacteria count (TLABC), total fungal and yeast count (TFYC), tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC) and antioxidant activity (AA), with corresponding annotations given as 푌1, 푌2, 푌3, 푌4, 푌5, 푌6,

푌7, 푌8 and 푌9 (for LT-ting) and 푌10, 푌11, 푌12, 푌13, 푌14, 푌15, 푌16, 푌17 and 푌18 (for HT-ting).

Table 3.1: Coded and real values for the Doehlert design Experimental runs Coded values Real values o o 푋1 (h) 푋2 ( C) X1 (h) X2 ( C) 1 0 0 48 27 2 0 0 48 27 3 1 0 72 27 4 0.5 0.866 60 34 5 -1 0 24 27

Experimental runs Coded values Real values 6 -0.5 -0.866 36 20 7 0.5 -0.866 60 20 8 -0.5 0.866 36 34

푋1 – fermentation time, 푋2 – fermentation temperature

Mathematical models describing the relationship between the process variables in terms of their linear, quadratic and interactive effects were described by a second-order polynomial equation presented in Eq. (3). 푘 푘 2 푘 푌 = 훽0 + ∑푖=1 훽푖푥푖 + ∑푖=1 훽푖푖푥푖 + ∑ ∑푖<푗 훽푖푖 푥푖푥푗 (3) where Y is the response, xi and xj are factors, β0 the constant, βi, βii and βij the coefficients of linear, quadratic and interaction terms respectively. The response surfaces were subsequently represented with model equations and respective coefficients obtained using Minitab 16 statistical software (Minitab Lt. Coventry, UK).

3.1.4 Model validation The different statistical parameters utilized in validating the adequacy of the models generated, were average absolute deviation (AAD), bias factor (Bf) and accuracy factor (Af) [Eqn. (4) – (6)], respectively, as well as the coefficient of determination (R2).

|푌푖,푒푥푝− 푌푖,푐푎푙| [∑푁 ( )] 푖=1 푌 퐴퐴퐷 = 푖,푒푥푝 (4) 푁 1 푁 푌푖,푐푎푙 퐵푓 = 10푁 ∑푖=1 log ( ) (5) 푌푖,푒푥푝

1 푁 푌푖,푐푎푙 퐴푓 = 10푁 ∑푖=1 |log ( )| (6) 푌푖,푒푥푝 where 푌푖,푒푥푝 and 푌푖,푐푎푙 are the experimental and calculated (predicted) responses respectively and 푁 is the total number of experiments done.

3.1.5 pH and titratable acidity (TTA) At the end of each fermentation process, pH of the WG-ting was measured using a pH meter (pH 510, Eutech Pte Ltd, Singapore). Titratable acidity was determined by titrating a mixture of 2 g of ting sample in 20 mL distilled water against 0.1 N NaOH, using phenolphthalein as an indicator, the TTA was expressed in g of tartaric acid/kg sample.

3.1.6 Estimation of viable microbial counts For viable microbial counts, 1 g of the respective WG-ting sample was added to 9 mL sterile distilled water and vortexed. Counts were determined by surface plating tenfold serial dilutions

76 of ting sample on plate count agar (Oxoid, South Africa), potato dextrose agar (Merck, South Africa) and MRS agar (Sigma Aldrich, Germany) in Petri dishes for bacterial (TBC), fungal and yeast (TFYC) and lactic acid bacteria (LAB) counts, respectively (Njobeh et al., 2009; Madoroba et al., 2011; Nyambane et al., 2014). Plates were incubated (IncoShake, Labotec, South Africa) at 30 oC for 72 h for TBC, 25 oC for 120 h for TFYC and anaerobically at 35 oC for 72 h for TLABC. Viable microbial counts were determined after incubation.

3.1.7 Tannin content, total phenolic content, total flavonoid content and antioxidant activity assay of the WG-ting samples Sample extraction was done by adding 10 mL of acidified methanol (1% HCl in methanol) to 0.3 g of freeze-dried (-55 oC for 24 h) milled WG-ting sample in a centrifuge tube (Kayitesi et al., 2012). The content was sealed with an aluminum foil, stirred for 2 h and centrifuged at 3500 rpm for 10 min (Eppendorf 5702R, Merck South Africa). The supernatant was decanted and kept while the residue was re-extracted using 10 mL acidified methanol as earlier described. The extraction process was repeated until a total of 30 mL acidified methanol (1% HCl in methanol) was used. After extraction, the supernatants were pooled together and stored at –4 oC prior to analysis.

3.1.7.1 Tannin content (TNC) Using the methods of Price et al. (1978), 1 mL of extract was added to a test tube containing 5 mL of an equal volume of 8% HCl in methanol and 1% vanillin (Merck, South Africa). The resulting mixture was vortexed and incubated in a water bath (30 oC) for 20 min. A blank experiment was performed repeating the earlier step but this time using 5 mL of 4% HCl. The absorbance of the mixture was read on a spectrophotometer (Biomate, Thermo Spectronic, Rochester, USA), set at a wavelength of 500 nm. Catechin (Sigma Aldrich, Germany) was used as a standard and results obtained expressed in mg catechin equivalents (CE)/g.

3.1.7.2 Total phenolic content (TPC) The TPC of the WG-ting sample was determined according to Folin–Ciocalteu method as described by Ainsworth and Gillespie (2007). To a 500 µL of distilled water, 10 µL of the extract was added and reacted with 50 µL of the Folin–Ciocalteu phenol reagent (Sigma Aldrich, Germany). This was allowed to stand in the dark for 3 min followed by the addition of 200 µL of 20% Na2CO3 (g/v) and finally 245 µL of distilled water and mixed. The mixture (300 µL) was accurately pipetted into a 96-well microplate, wrapped in aluminum foil and further incubated in the dark for 30 min and absorbance read on a microplate reader (iMark,

Biorad, South Africa), at a wavelength of 750 nm. Gallic acid (Sigma Aldrich, Germany) was used as a standard and results obtained expressed in mg gallic acid equivalents (GAE)/g.

3.1.7.3 Total flavonoid content (TFC) Using the method of Al-Farsi and Lee (2008), TFC was determined by mixing 30 µL of the extract with 20 µL of 36 mM NaNO2 followed by incubation in the dark for 5 min. Thereafter,

20 µL of 94 mM AlCl3 was added and after incubation for another 5 min (in the dark), 100 µL of NaOH was added. The absorbance of the mixture was read on a microplate reader (iMark, Biorad, South Africa) at 450 nm. Catechin (Sigma Aldrich, Germany) was used as a standard and data obtained were expressed as mg CE/g.

3.1.7.4 Antioxidant activity (AA) The free radical scavenging potential of WG-ting sample was determined using the trolox equivalent antioxidant capacity (TEAC) assay also known as ABTS modified methods of Awika et al. (2003) and Kayitesi et al. (2012). To a 20 µL of the extract, 180 µL of ABTS free radical cation solution (equal volumes of 7 mM ABTS and 2.45 mM K2S2O8 previously incubated for 12 h) was added and incubated for 5 min in the dark. Absorbance of the solution was measured on a microplate reader (iMark, Biorad, South Africa) set at 750 nm. Trolox (Sigma Aldrich, Germany) was used as a standard solution and results obtained expressed as µM trolox equivalents (TE)/g sample).

3.1.8 Scanning electron microscopy (SEM) analysis of the WG-ting samples The freeze-dried WG-ting samples were mounted on an aluminum stub and sprayed-coated in a carbon coater (Quorum Q150TE, Quorum Technologies, UK). The samples were then transferred to a SEM specimen chamber, subjected to electron beam and viewed under a scanning electron microscope under vacuum (Vega 3 XMU, TESCAN, Czech Republic).

3.1.9 Statistical analysis All other analyses were done in triplicates. To determine the significance of the generated models, an analysis of variance (ANOVA) was conducted on Minitab 16 (Minitab Lt. Coventry, UK) and differences were considered statistically significant if p < 0.05. Response surface plots were obtained using Sigmaplot 12.5 (Systat Software Inc., California, USA).

3.2 Results and discussion 3.2.1 pH and TTA of the WG-ting samples pH and TTA are important biochemical parameters peculiar to fermented foods and values obtained in this study are presented in Table 3.2. As anticipated, with increasing time and temperature, pH decreased (increased acidity) throughout the entire fermentation conditions. This were from an initial value of 6.93 and 6.94, for the LT-sorghum and HT-sorghum, respectively (Table 3.2). It could be observed that the pH values of the LT-ting samples are lower as compared to their HT counterpart, which can be attributed to the varying composition of the sorghum varieties.

Significant (p  0.05) increased acidity of the WG-ting samples with increasing fermentation time and temperature suggests an increased microbial action during fermentation and accumulation of acids, especially because most fermenting microorganisms prefer acidic conditions (Jay et al., 2005). Although the pH values ranged between 3.5 – 4.0 for ting from refined grains in other studies (Taylor and Taylor, 2002; Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011), the values observed for modified WG-ting in this study could be attributed to difference in sample sources, but most importantly reduced metabolic activities of the fermenting microbiota, influenced by the presence of seed coat (pericarp) in the whole sorghum grain substrate. Similar occurrence of such high pH values after fermentation of WGs have been reported and attributed to the presence of bioactive compounds in the non-decorticated grains (Nuobariene et al., 2015; García-Mantrana et al., 2016).

Table 3.2: Experimental and predicted values obtained for the parameters investigated in LT- and HT-ting Variables pH TTA TBC TLABC TFYC TNC TPC TFC AA (g/kg) (×106 cfu/g) (×105 cfu/g) (×105 cfu/g) (mg CE/g) (mg GAE/g) (mg CE/g) (µM TE/g)

X1 X2 Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre LT-sorghum samples Control LT 6.94 - 0.04 - 0.03 - - - 0.35 - 17.97 - 32.83 - 13.43 - 1.88 - (0.02) (0.07) (1.29) (2.16) (0.13) (0.23) (0.14) (0.13) 48 27 6.18d 6.18 0.56c 0.57 2.25b 2.24 14.3d 14.2 7.07d 7.06 9.96d 9.97 16.07c 16.06 8.71d 8.70 1.15cd 1.15 (0.02) (0.06) (1.86) (2.28) (2.99) (0.22) (0.09) (0.25) (0.03) 48 27 6.18d 6.18 0.57c 0.57 2.23b 2.24 14.0d 14.2 7.09d 7.06 9.97d 9.97 16.05c 16.06 8.68d 8.70 1.14cd 1.15 (0.01) (0.04) (1.59) (2.31) (2.64) (0.19) (0.13) (0.28) (0.20) 72 27 5.59b 5.66 1.29ef 1.18 4.87d 4.50 27.1e 28.8 11.7e 11.0 7.95c 7.43 12.42b 12.74 8.43c 8.74 1.09ab 1.09 (0.03) (0.04) (2.02) (3.82) (2.61) (0.13) (0.21) (0.14) (0.11) 60 34 5.20a 5.14 1.31g 1.42 7.80f 8.17 57.3g 55.6 14.4f 15.1 0.97a 1.49 9.02a 8.70 7.96a 7.64 1.08a 1.08 (0.01) (0.09) (1.99) (3.16) (2.88) (0.08) (0.18) (0.13) (0.03) 24 27 6.62f 6.56 0.52ab 0.63 1.83a 2.20 10.2c 8.48 4.95c 5.64 10.89e 11.41 21.94e 21.62 11.46f 11.14 1.16cd 1.16 (0.04) (0.05) (3.01) (2.81) (3.01) (0.11) (0.20) (0.09) (0.07) 36 20 6.69g 6.76 0.50a 0.39 2.40c 2.03 7.61a 9.33 1.31a 0.62 13.66g 13.14 25.14f 25.46 11.98g 12.30 1.21f 1.21 (0.02) (0.08) (3.88) (4.02) (2.97) (0.21) (0.05) (0.16) (0.10) 60 20 6.58e 6.52 0.60cd 0.71 2.43c 2.80 7.73b 6.01 1.41b 2.10 11.52f 12.04 20.98d 20.66 10.93e 10.61 1.18e 1.17 (0.01) (0.07) (2.81) (3.04) (2.59) (0.09) (0.26) (0.18) (0.05) 36 34 5.73c 5.79 1.28e 1.17 7.01e 6.64 30.2f 31.9 11.9e 11.2 4.90b 4.38 12.46b 12.78 8.03b 13.18 1.12c 1.11 (0.01) (0.04) (2.07) (3.19) (2.86) (0.14) (0.16) (0.07) (0.06) HT-sorghum samples Control HT 6.93 - 0.06 - 0.01 - - - 0.09 - 31.68 - 64.87 - 51.08 - 4.03 - (0.04) (0.07) (1.13) (1.36) (0.17) (0.28) (0.13) (0.10) 48 27 6.43d 6.43 0.55bc 0.56 1.84e 1.82 1.21d 1.22 8.57e 8.56 18.43e 18.39 55.46e 55.38 34.19d 34.31 3.88cd 3.87 (0.01) (0.01) (3.00) (4.64) (1.80) (0.25) (0.17) (0.07) (0.05) 48 27 6.42d 6.43 0.57c 0.56 1.81d 1.82 1.22e 1.22 8.56d 8.56 18.34e 18.39 55.30e 55.38 34.42e 34.31 3.86cd 3.87 (0.01) (0.05) (2.65) (3.22) (2.55) (0.28) (0.06) (0.04) (0.02) 72 27 6.01b 6.08 1.12de 1.05 2.95f 2.88 1.63f 1.62 10.1f 10.2 16.73c 15.95 46.99b 47.03 38.37f 40.81 3.76b 3.81 (0.01) (0.10) (3.21) (1.78) (0.53) (0.11) (0.19) (0.11) (0.03) 60 34 5.91a 5.78 1.13e 1.40 3.99h 4.45 2.26h 2.47 11.8h 12.1 3.45a 1.28 42.83a 40.79 31.04b 26.44 3.46a 3.30

Variables pH TTA TBC TLABC TFYC TNC TPC TFC AA (g/kg) (×106 cfu/g) (×105 cfu/g) (×105 cfu/g) (mg CE/g) (mg GAE/g) (mg CE/g) (µM TE/g)

X1 X2 Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre Exp Pre (0.01) (0.14) (3.61) (2.21) (1.28) (0.06) (0.07) (0.11) (0.04) 24 27 6.80f 6.73 0.50ab 0.59 1.59b 1.66 1.14c 1.16 7.83c 7.77 22.21g 22.99 57.65f 57.61 50.21g 47.77 3.88d 3.83 (0.01) (0.03) (1.53) (2.97) (1.58) (0.17) (0.06) (0.15) (0.03) 36 20 6.83g 6.95 0.46a 0.37 1.57a 1.60 1.11a 1.15 5.21a 4.85 19.64f 17.62 60.00g 59.90 31.60c 32.20 3.78bc 3.78 (0.03) (0.01) (2.08) (2.00) (1.83) (0.18) (0.08) (0.09) (0.03) 60 20 6.75e 6.75 0.49ab 0.58 1.62c 1.71 1.13b 1.17 6.97b 6.55 17.65d 17.67 54.79d 54.61 31.04b 26.44 3.87cd 3.77 (0.01) (0.03) (1.05) (1.32) (2.33) (0.16) (0.05) (0.11) (0.03) 36 34 6.22c 6.24 1.11d 1.13 3.07g 3.33 1.88g 2.04 10.9g 11.4 11.62b 8.37 48.03c 46.08 27.66a 27.64 3.38a 3.32 (0.00) (0.06) (1.53) (1.16) (2.44) (0.09) (0.08) (0.09) (0.04) AA – antioxidant activity; Control HT – raw high tannin sorghum; Control LT – raw low tannin sorghum; Exp – Experimental value; Pre – Predicted value; TBC – total bacteria count; TFC – total flavonoid content; TFYC – total fungal and yeast count; TLABC – total lactic acid bacteria; TNC – tannin content; TPC – total phenolic content; TTA – titratable acidity; X1 – fermentation o time (h); X2 – fermentation temperature ( C); Values in parentheses represent the standard deviation of triplicate measurements. Means with no common letters within a column significantly differ (p  0.05).

The regression model representing the effect of fermentation time and temperature on the pH of the LT and HT-ting samples is provided in Eq. (7) and (8), respectively.

푌1 = 6.18 − 0.45푥1 − 0.67552푥2 − 0.07500푥12 − 0.14834푥22 − 0.24249푥1푥2 (7)

푌2 = 6.425 − 0.32833푥1 − 0.48336푥2 − 0.02푥12 + 0.01333푥22 − 0.15334푥1푥2 (8)

The model equations presented in Eq. (7) and (8) had R2 values of 97 and 99%, respectively, indicating that the model could explain over 90% variability of the data obtained. The AAD values (0.01) and the closeness of both 퐵푓 and 퐴푓 to 1 (Desobgo et al., 2015; Sobowale et al., 2017), indicate the adequacy of the model in describing the pH values. As observed from Table

3.3, the linear negative effects of fermentation time (푋1) and temperature (푋2) on the LT- and HT-ting samples were both significant (p  0.05) on the pH values. This is equally reflected on the response surface plots (Figures 3.2A and 3.3A).

The TTA values of the WG-ting ranged from 0.5 - 1.31 (LT-ting) and 0.46 - 1.19 g/kg (HT- ting), with the significant (p  0.05) highest production of organic acids recorded for WG-ting samples fermented at higher temperature and longer time (34 oC for 60 h). Corresponding least values were observed in WG-ting obtained at lower temperature and shorter time (20 oC for 36 h) (Table 3.2). The high TTA values were observed in the LT-ting samples as compared to low values in the HT-ting samples (Table 3.2), suggest that the metabolic activities of the fermenting organisms in the HT-sorghum samples were relatively low.

Table 3.3: Coefficient of regression and validation parameters for the different mathematical models obtained for LT- and HT-ting samples Coefficient pH TTA TBC TLABC TFYC TNC TPC TFC AA LT-ting samples

β0 6.18 0.565 2240000 1415000 706000 9.965 16.06 8.695 1.145 * * * * * β1 ‒0.45 0.27833 1150000 1017000 268333 ‒1.9917 ‒4.44 ‒1.1967 ‒0.03500 * * * * * * * * β2 ‒0.677552 0.43014 2881062 2083141 680716 ‒5.5745 ‒7.1132 ‒1.9977 ‒0.05485 β11 ‒0.075 0.34 1110000 450000 126500 ‒0.5450 1.12 1.2500 ‒0.02 * * β22 ‒0.14834 0.36335 3190187 1391415 ‒16168 ‒2.7552 0.7467 0.9567 0.01 * β12 ‒0.24249 ‒0.04042 438799 1557737 138568 ‒1.0335 0.4157 0.5658 ‒0.00577 R2 (%) 98.77 93.45 97.95 99.11 98.28 98.58 99.71 96.80 99.63 AAD 0.01 0.12 0.01 0.11 0.19 0.12 0.02 0.12 0.01 푩풇 1.00 1.00 1.01 0.98 0.96 1.04 1.00 1.07 1.00 푨풇 1.01 1.13 1.10 1.12 1.23 1.11 1.02 1.10 1.00 HT-ting samples

β0 6.425 0.56 182244 121500 85650 18.387 55.3772 34.306 3.87450 * * * * * * β1 ‒0.32833 0.23167 61394 23000 12066.7 ‒3.52 ‒5.2885 ‒3.477 ‒0.01083 * * * * * * * * β2 ‒0.48336 0.45003 129057 63337 35068.7 ‒7.404 ‒7.9766 ‒1.316 ‒0.26835 * * β11 ‒0.02 0.26 44861 17000 4100 1.079 ‒3.0546 9.984 ‒0.05375 * * * * β22 0.01333 0.33337 111459 60007 933.4 ‒9.899 ‒5.6918 ‒11.499 ‒0.42283 * β12 ‒0.15334 0.03334 58349 24001 ‒5733 ‒4.122 ‒0 2.632 ‒0.00667 R2 (%) 97.00 93.83 99.55 99.89 99.93 98.50 99.99 89.47 95.48 AAD 0.01 0.12 0.02 0.01 0.03 0.16 0.01 0.06 0.02 푩풇 1.00 1.00 1.04 1.03 0.99 0.81 0.99 0.96 0.99 푨풇 1.01 1.12 1.05 1.04 1.03 1.24 1.01 1.07 1.02

β represents the coefficients of equations of the different models with β0 representing the constant term; β1 and β2 the linear effects of fermentation time and temperature respectively; β11 and β22 their quadratic effects and β12 their interactions. AA – antioxidant activity; TBC – total bacteria count; TFC – total flavonoid content; TFYC – total fungal and yeast count; TLABC – total lactic acid bacteria; TNC – tannin content; TPC – total phenolic content; TTA – titratable acidity. * Significant at p  0.05.

Figure 3.2: Response surface plots showing the effects of fermentation time and temperature on: A ‒ pH, B ‒ TTA (titratable acidity), C ‒ TBC (total bacteria count), D ‒ TLABC (total lactic acid bacteria count), E ‒ TFYC (total fungal and yeast count), F ‒ TNC (tannin content), G ‒ TPC (total phenolic content), H ‒ TFC (total flavonoid content), I ‒ AA (antioxidant activity) of the LT-ting samples

Figure 3.3: Response surface plots showing the effects of fermentation time and temperature on: A ‒ pH, B ‒ TTA (titratable acidity), C ‒ TBC (total bacteria count), D ‒ TLABC (total lactic acid bacteria count), E ‒ TFYC (total fungal and yeast count), F ‒ TNC (tannin content), G ‒ TPC (total phenolic content), H ‒ TFC (total flavonoid content), I ‒ AA (antioxidant activity) of the HT-ting samples

An increase in TTA values with increase in fermentation time and temperature indicates an accumulation of organic acids with increasing microbial activity and metabolism of the fermenting organisms. An inverse relationship between the pH and TTA observed in this study was in agreement with the significant (p  0.01) negative correlation coefficient obtained between pH and TTA (Tables 3.4). This indicates that a decrease in pH (increased acidity) would result in increased organic acids and vice versa. This could also mean that the production of organic acids was suppressed at higher pH (lower acidity) when compared to lower pH conditions. Furthermore, most ting fermenting organisms generally tolerate acidic conditions (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011; Madoroba et al., 2011) and could thus explain the better production of organic acids at these pH conditions.

The regression model describing the effect of fermentation time and temperature on TTA of LT- and HT-ting is respectively given in Eq. (9) and (10), with their corresponding regression coefficient of determination values provided in Tables 3.3.

푌3 = 0.565 + 0.27833푥1 + 0.43014푥2 + 0.34푥12 + 0.36335푥22 − 0.04042푥1푥2 (9)

푌4 = 0.56 + 0.23167푥1 + 0.45003푥2 + 0.26푥12 + 0.33337푥22 + 0.03334푥1푥2 (10)

The values of AAD (0.01) and other validation indices depicts an agreement between experimental and predicted values further showing that the models adequately described the TTA values (Tables 3.3). As further observed in Table 3.3, only the positive linear effect of fermentation temperature (푋2) had a significant (p  0.05) effect on the TTA of the HT-ting samples (Figure 3.3B), suggesting that increase in 푋2 would translate to a significantly (p  0.05) higher TTA. The surface plots of the TTA in LT-ting samples equally showed an increase with increasing fermentation time and temperature (Figure 3.2B).

Table 3.4: Pearson correlation between investigated parameters in the LT- and HT-ting samples pH TTA TBC TLABC TFYC TNC TPC TFC AA LT-ting pH ‒0.914** ‒0.889** ‒0.934** ‒0.966** 0.930** 0.968** 0.880** 0.950** TTA ‒0.914** 0.940** 0.849** 0.886** ‒0.857* ‒0.846** ‒0.723* ‒0.860** TBC ‒0.889** 0.940** 0.921** 0.852** ‒0.928** ‒0.812* ‒0.682 ‒0.770* TLABC ‒0.934** 0.849** 0.921** 0.888** ‒0.967** ‒0.859** ‒0.708* ‒0.850** TFYC ‒0.966** 0.886** 0.852** 0.888** ‒0.921** ‒0.958** ‒0.887** ‒0.962** TNC 0.930* ‒0.857* ‒0.928** ‒0.967** ‒0.921** 0.909** 0.788 0.877** TPC 0.968** ‒0.846** ‒0.812* ‒0.859** ‒0.958** 0.909** 0.958** 0.955** TFC 0.880** ‒0.723* ‒0.628 ‒0.708* ‒0.887** 0.788* 0.958** 0.863** AA 0.950** ‒0.860** ‒0.770* ‒0.850** ‒0.962** 0.877** 0.955** 0.863** HT-ting pH ‒0.921** ‒0.914** ‒0.873** ‒0.919** 0.775* 0.935** 0.298 0.636 TTA ‒0.921** 0.959** 0.935** 0.893** ‒0.791* ‒0.957** ‒0.285 ‒0.809* TBC ‒0.914** 0.959** 0.992** 0.897** ‒0.919** ‒0.968** ‒0.340 ‒0.843** TLABC ‒0.873** 0.935** 0.992** 0.892** ‒0.942** ‒0.946** ‒0.368 ‒0.892** TFYC ‒0.919** 0.893** 0.897** 0.892** ‒0.779** ‒0.925** ‒0.171 ‒0.712** TNC 0.775* ‒0.791* ‒0.919** ‒0.942** ‒0.779** 0.870** 0.544 0.845** TPC 0.935** ‒0.957** ‒0.968** ‒0.946** ‒0.925** 0.870** 0.318 0.753* TFC 0.298 ‒0.285 ‒0.340 ‒0.368 ‒0.171 0.544 0.318 0.518 AA 0.636 ‒0.809* ‒0.843** ‒0.892** ‒0.712* 0.845** 0.753* 0.518 ** ‒ Correlation is significant at 1%; * ‒ Correlation is significant at 5%

3.2.2 Microbial load Counts obtained on the different selective media used in this study (Table 3.2), suggested a diverse microbial flora on WG-ting samples. LABs usually dominate the fermentation microbiota of ting (Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011), while the relatively low counts of yeasts and fungi suggest that spontaneous cereal fermentation involves the competitive action of different endogenous microorganisms with LABs dominating (Meroth et al., 2003; Adebiyi et al., 2018). While this study did not focus on characterizing these microorganisms, previous studies on the microbiota of ting from refined (decorticated) sorghum reported different species of LABs, Enterobacteriaceae, Enterococcus, Weissella and yeast (Madoroba et al., 2009; Madoroba et al., 2011; Sekwati- Monang and Gänzle, 2011). Although the presence of fungi was not investigated in these earlier studies, their presence in microflora of the WG-ting samples could be attributed to ability of these fungi to utilize available nutrients in the raw sorghum for their metabolism.

Similar to pH and TTA values, microbial counts were also observed to be significantly (p  0.05) different (Table 3.2), increasing with increase in fermentation time and temperature (Figures 3.1 and 3.2). While the TBC, TLABC and TFYC are significantly (p  0.05) lower at 20 oC and 36 h, same parameters were notably higher at elevated temperatures and longer times as observed with samples obtained at 34oC and 60 h (Table 3.2). The microbial counts (TBC, TLABC and TFYC) were observed to negatively and significantly (p  0.01) correlate with the pH (Table 3.4) thereby, suggesting that a decrease in pH would translate to an increase in microbial counts and vice versa. This also validates the earlier proposition that viability of the microorganisms was dependent on pH and the availability of acids for metabolism. Higher TLABC over TFYC indicates the dominance of LABs during the fermentation process, equally reported in other studies on ting (Madoroba et al., 2009; Madoroba et al., 2011; Sekwati- Monang and Gänzle, 2011).

Regression models representing the effect of fermentation time (푋1) and fermentation temperature (푋2) on the TBC, TLABC and TFYC is presented in Eq. (11) – (16).

푌5 = 2240000 + 1150000푥1 + 2881062푥2 + 1110000푥12 + 3190187푥22 + 438799푥1푥2 (11)

푌6 = 182244 + 61394푥1 + 129057푥2 + 44861푥12 + 111459푥22 + 58349푥1푥2 (12)

푌7 = 1415000 + 1017000푥1 + 2083141푥2 + 450000푥12 + 1391415푥22 + 1557737푥1푥2 (13)

푌8 = 121500 + 23000푥1 + 63337푥2 + 17000푥12 + 60007푥22 + 24001푥1푥2 (14)

푌9 = 706000 − 268333푥1 + 680716푥2 + 126500푥12 − 16168푥22 + 138568푥1푥2 (15)

푌10 = 85650 + 12066.7푥1 + 35068.7푥2 + 4100푥12 + 933.4푥22 − 5733.7푥1푥2 (16)

Where 푌5, 푌7 and 푌9 represents the model equation of TBC, TLABC and TFYC of the LT-ting samples, respectively and 푌6, 푌8 and 푌10, the TBC, TLABC and TFYC of the HT-ting samples.

As provided in Table 3.3, the linear effect of fermentation temperature (푋2) was particularly significant (p  0.05) on the microbial cell counts. The various AAD, 퐵푓 and 퐴푓 values (Table 3.3) obtained for the regression models [Eq. (11) – (16)] indicate the effectiveness of the proposed mathematical models for the prediction of the microbial cell count in this study. The observed R2 values (>90%) show that the proposed mathematical models of the microbial counts can explain more than 90% experimental observations as a function of the fermentation time and temperature. R2 values should be at least 80% to have a good fit of the model and the closer it is to 100%, the better the empirical model fits the actual data (Filli et al., 2011).

3.2.3 Tannin content, total phenolic content, total flavonoid content and antioxidant activity The observed tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC) and antioxidant activity (AA) for the LT- and HT-ting samples is presented in Tables 3.2. For the LT-sorghum samples, initial levels of TNC, TPC, TFC and AA were 17.97 mg CE/g, 32.83 mg GAE/g, 13.44 mg CE/g and 1.88 µM TE/g, respectively, while for HT- sorghum samples, these were 31.68 mg CE/g (TNC), 64.87 mg GAE/g (TPC), 51.08 mg CE/g (TFC) and 4.03 µM TE/g (AA). The varying values in the TNC, TPC, TFC and AA values of the sorghum types can be attributed to difference in physicochemical composition of the grains (Beta et al., 1999; Dykes et al., 2005; Wu et al., 2012).

A general decrease of TNC, TPC, TFC and AA was observed with increase in fermentation time and temperature (Tables 3.2). This could be attributed to reduced extractability of the phenolic compounds due to de-polymerization and/or interaction of these compounds with other macromolecules (Beta et al., 2000; Taylor and Duodu, 2015). It could also mean that the phenolic compounds and tannin polymers have been transformed to health beneficial monomers (such as flavanols and anthocyanidins), since sorghum fermented products are known to be rich in health promoting components (Awika and Rooney 2004; Apea-Bah et al., 2014; Taylor and Duodu 2015; Awika, 2017). Such reduction and degradation of phenolic compounds in fermented sorghum have also been reported by Svensson et al. (2010) probably due to the actions of decarboxylases, reductases, esterases and the ability of fermenting LABs to metabolize phenolic compounds.

A strong positive correlation between the AA and TPC (Table 3.4) strongly suggests that the phenolic contents of the LT-ting samples, largely contributed to the antioxidant activities. This was not the case with the HT-ting samples with an equally strong positive correlation between the AA and TNC, suggesting that the tannin-related components contributed more to the antioxidant activity values. While phenolic compounds are generally known to exhibit antioxidant activities, the strong correlation of the AA and TNC is in agreement with the reports indicating that sorghum tannins exhibit high antioxidant levels as compared to other cereals and fruits (Awika and Rooney, 2004; Gu et al., 2004; Taylor and Duodu, 2015; Awika, 2017). Tannin-sorghum have also been used and preferred for centuries in the preparation of certain food products because foods made from them “stay longer in the stomach” in addition to the belief that they are therapeutic against diseases (Awika et al., 2003; Awika and Rooney, 2004; Rose et al., 2013; Awika, 2017). Studies have equally shown that tannins from sorghum grains are beneficial to human due to their anticarcinogenic activity against human cells, anti- inflammatory effects and potential use as anticaloric agents for obesity (Gómez-Cordovés et al., 2001; Ross and Kasum, 2002; Burdette et al., 2010; Awika, 2017).

Regression models describing the effect of fermentation time and temperature on the TNC, TPC, TFC and AA of LT-ting samples are presented in Eqn. (17) – (20) and that of HT-ting 2 samples in Eqn. (21) – (24). High R values and adequate values of AAD, 퐵푓 and 퐴푓 indicates the validity of the models in Eqn. (17) – (24).

푌11 = 9.9650 − 1.9917푥1 − 5.5745푥2 − 0.5450푥12 − 2.7552푥22 − 1.0335푥1푥2 (17)

푌12 = 16.06 − 4.44푥1 − 7.1132푥2 + 1.12푥12 + 0.7467푥22 + 0.4157푥1푥2 (18)

푌13 = 8.695 − 1.1967푥1 − 1.9977푥2 + 1.2500푥12 + 0.9567푥22 + 0.5658푥1푥2 (19)

푌14 = 1.145 − 0.035푥1 − 0.05485푥2 − 0.02푥12 + 0.01푥22 − 0.00577푥1푥2 (20)

푌15 = 18.387 − 3.52푥1 − 7.404푥2 + 1.079푥12 − 9.899푥22 − 4.122푥1푥2 (21)

푌16 = 55.3772 − 5.2885푥1 − 7.9766푥2 − 3.0546푥12 − 5.6918푥22 − 0푥1푥2 (22)

푌17 = 34.306 − 3.477푥1 − 1.316푥2 + 9.984푥12 − 11.499푥22 + 2.632푥1푥2 (23)

푌18 = 3.8745 − 0.01083푥1 − 0.26835푥2 − 0.05375푥12 − 0.42283푥22 − 0.00667푥1푥2 (24)

Figures 3.1 and 3.2 represent the response surface plot showing effect of fermentation time and temperature on these parameters. As observed in Table 3.3, the linear effect of fermentation temperature (X2) was significantly (p  0.05) negative on all health promoting properties of LT-ting analyzed. A similar observation was noted for the HT-ting samples, except for TFC.

This thus, indicates that at increased temperature and sufficiently longer time, the amount of these parameters will decrease (Figures 3.2 and 3.3).

3.2.4 Multi response numerical optimization (MRNO) The surface plots (Figures 3.2 and 3.3) show the effect of different process variables (fermentation time and temperature) on the investigated parameters on LT- and HT-ting. A MRNO technique was however adopted to determine the best experimental conditions for obtaining WG-ting, yielding maximal values of TNC, TPC, TFC and AA, good microbial growth, reduced pH and high titratable acidity. This was done on Minitab 16 statistical software

(Minitab Lt. Coventry, UK). The optimum derived conditions were fermentation time (푋1) and o o temperature (푋2) of 24 h and 34 C respectively (for LT-ting) and 72 h and 28 C (for HT-ting). The corresponding predicted parameters at these conditions are presented in Table 3.5. Confirmatory experiments (in triplicates) done to verify the predicted values gave experimental results closely related to those predicted (Table 3.5), suggesting the adequacy of the models generated.

Table 3.5: Predicted and experimental values of the optimal HT- and LT-ting samples Investigated LT-ting HT-ting o o parameters (푿ퟏ = 24 h) & (푿ퟐ = 34 C) (푿ퟏ = 72 h) & (푿ퟐ = 28 C) Exp Pre Exp Pre pH 6.00±0.02 6.09 5.87±0.04 5.99 TTA (g/kg) 1.34±0.08 1.31 1.13±0.03 1.11 TBC (× 105 cfu/g) 61.9±1.51 67.1 3.84 ±1.39 3.15 TLABC (× 105 cfu/g) 20.1±1.90 23.5 1.68 ±1.84 1.74 TFYC (× 105 cfu/g) 9.91±1.67 10.2 1.17 ±2.01 1.24 TNC (mg CE/g) 5.36±0.31 5.41 14.11±0.16 14.30 TPC (mg GAE/g) 15.48±0.16 15.66 46.09±0.22 45.91 TFC (mg CE/g) 9.67±0.11 9.64 40.86±0.18 40.79 AA (mg TE/g) 1.18±0.05 1.13 3.71±0.08 3.77 HT – high tannin; LT – low tannin; Exp – Experimental value; Pre – Predicted value; TBC – total bacteria count; TFC – total flavonoid content; TFYC – total fungal and yeast count; TLABC – total lactic acid bacteria; TNC – tannin content; TPC – o total phenolic content; TTA – titratable acidity; X1 – fermentation time (h); X2 – fermentation temperature ( C); Values represent mean ± standard deviation of triplicate measurements.

3.2.5 Scanning electron microscopy (SEM) of the WG-ting samples The SEM images of samples of ting were compared to investigate possible morphological changes after fermentation. As observed on the obtained micrographs (Figures 3.4), the granular structure of the WG-ting samples was predominantly spherical, similar to that obtained for fermented cereals (Amadou et al., 2014; Adebiyi et al., 2016).

Figure 3.4: Scanning electron microscopy images of (A) raw HT-sorghum, (B) raw HT- sorghum

Increase in fermentation time and temperature led to gradual change from a more compact, fused and consistent structure to a more loosened, disoriented one, forming pits (Figures 3.5 and 3.6). This was more pronounced in LT-ting samples (Figure 3.5) fermented for longer time and higher temperatures, suggesting increased degradation and hydrolysis of inherent components such as starch and protein. This could also be caused by molecular disintegration, probable modifications in particle sizes and intermolecular interactions (Odunmbaku et al., 2018).

A slightly different trend was however observed for the HT-ting samples, as the effect of fermentation time and temperature did not appear to have induced considerable visible modification to microstructure (Figure 3.6). This could also be tandem with the earlier observations of the pH, TTA and microbial counts (Table 3.2), depicting the less microbial activity during the fermentation process. It has also been suggested that the presence and amount of tannins (as observed in this study) can inhibit the growth of microorganisms and consequently affect proteolysis and degradation of other compounds (Tabacco et al. 2006). Due to the large size of these tannins and available hydroxyl groups, they interact and form complexes with proteins, exerting steric effects and preventing enzyme access and subsequent proteolysis (Duodu et al., 2003).

Figure 3.5: Scanning electron microscopy images of LT-ting samples: A ‒ 20 oC 36 h, B ‒ 20 oC 60 h, C ‒ 27 oC 24 h, D ‒ 27 oC 48 h, E ‒ 27 oC 72 h, F ‒ 34 oC 36 h, G ‒ 34 oC 60 h, H – 34 oC 24 h

Figure 3.6: Scanning electron microscopy images of HT-ting samples: A ‒ 20 oC 36h, B ‒ 20oC 60 h, C ‒ 27 oC 24 h, D ‒ 27 oC 48 h, E ‒ 27 oC 72 h, F ‒ 28 oC 72 h, G ‒ 34 oC 36 h, H – 34 oC 60 h

3.3 Conclusion Using a Doehlert RSM approach, this study investigated the effects of fermentation variables on some selected parameters i.e., pH, titratable acidity, microbial count, tannin content, total phenolic content, total flavonoid content and antioxidant activity. Sorghum type was noted to

93 largely influence optimal fermentation time and temperature conditions with numerical optimization establishing that optimal conditions for WG-ting production from LT- and HT- sorghum as 34 oC for 24 h and 28 oC for 72 h. At these conditions, maximal phenolic, tannin, flavonoid contents and antioxidant activity were derived, complemented with good microbial growth, reduced pH and high production of organic acids. Although the seed coats (pericarp) in the whole grains reduced microbial activity and subsequent metabolic actions, appreciable amounts of bioactive and health promoting compounds in the WG-ting samples will confer beneficial effects to intending consumers. Adoption of the optimal conditions will be beneficial for the production of high quality ting.

References Adebiyi, J. A., Obadina, A. O., Mulaba-Bafubiandi, A. F., Adebo, O. A., & Kayitesi, E. (2016). Effect of fermentation and malting on the microstructure and selected physicochemical properties of pearl millet (Pennisetum glaucum) flour and biscuit. Journal of Cereal Science, 70, 132-139. Adebiyi, J. A., Obadina, A. O., Adebo, O. A., & Kayitesi, E. (2018). Fermented and malted millet products in Africa: Expedition from traditional/ethnic foods to industrial value added products. Critical Reviews in Food Science and Nutrition, 58, 463-474. Adebo, O. A., Njobeh, P. B., Mulaba-Bafubiandi, A. F., Adebiyi, J. A., Desobgo, Z. S. C., & Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, e13381 (1-10). Ainsworth, E. A., & Gillespie, K. M. (2007). Estimation of total phenolic content and other oxidation substrates in plant tissues using Folin-Ciocalteu reagent. Nature Protocols, 2, 875- 877. Amadou, I., Gounga, M. E., Shi, Y. H., & Le, G. W. (2014). Fermentation and heat moisture treatment induced changes on the physicochemical properties of foxtail millet (Setaria italica) flour. Food Bioproducts Processing, 92, 38-45. Apea-Bah, F. B., Minnaar, A., Bester, M. J., & Duodu, K. G. (2014). Does a sorghum-cowpea composite porridge hold promise for contributing to alleviating oxidative stress? Food Chemistry, 15, 157-166. Al-Farsi, M. A., & Lee, C.Y. (2008). Optimization of phenolics and dietary fibre extracts from date seeds. Food Chemistry, 108, 977-985.

Awika, J. M., Rooney, L. W., Wu, X., Prior, R. L., & Zevallos, L. C. (2003). Screening methods to measure antioxidant activity of sorghum (Sorghum bicolor) and sorghum products. Journal of Agricultural and Food Chemistry, 51, 6657-6662. Awika, J. M., & Rooney, L.W. (2004). Sorghum phytochemicals and their potential impact on human health. Phytochemistry. 65, 1199-1221. Awika, J. M. (2017). Sorghum: its unique nutritional and health-promoting attributes. In: Gluten-Free Ancient Grains, Taylor, J. R. N., & Awika, J. M. (Eds.). Woodhead Publishing, United Kingdom. pp. 21-54. Beta, T., Rooney, L. W., Marobatsanga, L. T., & Taylor, J. R. N. (1999). Phenolic compounds and kernel characteristics of Zimbabwean sorghums. Journal of the Science of Food and Agriculture, 79, 1003-1010. Beta, T., Rooney, L. W., Marovatsanga, L. T., & Taylor, J. R. N. (2000). Effect of chemical treatments on polyphenols and malt quality in sorghum. Journal of Cereal Science, 31, 295- 302. Burdette, A, Garner, P. L., Mayer, E. P., Hargrove, J. L., Hartle, D. K., & Greenspan, P. (2010). Anti-inflammatory activity of select sorghum (Sorghum bicolor) brans. Journal of Medicinal Food, 13, 879-887. Desobgo, Z. S. C., Stafford, R. A., & Metcalfe, D. J. A. (2015). Dimethyl sulfide stripping behavior during wort boiling using response surface methodology. Journal of the American Society of Brewing Chemists, 73, 84-89. Duodu, K. G., Taylor, J. R. N., Belton, P. S., & Hamaker, B. R. (2003). Factors affecting sorghum protein digestibility. Journal of Cereal Science, 38, 117-131. Dykes, L., Rooney, L. W., Waniska, R. D., & Rooney, W. L. (2005). Phenolic compounds and antioxidant activity of sorghum grains of varying genotypes. Journal of Agricultural and Food Chemistry, 53, 6813-6818. Ferreira, S. L. C., Dos Santos, W. N. L., Quintella, C. M., Neto, B. B., & Bosque-Sendra, J. M. (2004). Doehlert matrix: a chemometric tool for analytic chemistry–review. Talanta, 63, 1061-1067. Filli, K. B., Nkama, I., Jideani, V. A., & Abubakar, U. M. (2011). Application of response surface methodology for the study of composition of extruded millet-cowpea mixtures for the manufacture of fura: A Nigerian food. African Journal of Food Science, 5, 884-896. García-Mantrana, I., Yebra, M. J., Haros, M., & Monedero, V. (2016). Expression of bifidobacterial phytases in Lactobacillus casei and their application in a food model of whole-grain sourdough bread. International Journal of Food Microbiology, 216, 18-24.

Gómez-Cordovés, C., Bartolomé, B., Vieira, W., & Virador, V. M. (2001). Effects of wine phenolics and sorghum tannins on tyrosinase activity and growth of melanoma cells. Journal of Agricultural and Food Chemistry, 49, 1620-1624. Gu, L., Kelm, M. A., Hammerstone, J. F., Beecher, G., Holden, J., Haytowitz, D., Gebhardt, S., & Prior, R. L. (2004). Concentrations of proanthocyanidins in common foods and estimations of normal consumption. The Journal of Nutrition, 134, 613-617. Jay, J. M., Loessner, M. J., & Golden, D. A. (2005). Modern Food Microbiology (7th ed). Springer, USA. Jones, J. M., Reicks, M., Adams, J., Fulcher, G., Weaver, G., Kanter, M., & Marquart, L. (2002). The importance of promoting a whole grain foods message. Journal of the American College of Nutrition, 21, 293-297. Kayitesi, E., De Kock, H. L., Minnaar, A., & Duodu, K. G. (2012). Nutritional quality and antioxidant activity of marama–sorghum composite flours and porridges. Food Chemistry, 131, 837-842. Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Madoroba, E., Steenkamp, E. T., Theron, J., Scheirlinck, I., Cloete, T. E., & Huys, G. (2011). Diversity and dynamics of bacterial populations during spontaneous sorghum fermentations used to produce ting, a South African food. Systematic and Applied Microbiology, 34, 227- 234. Marquart, L., Jones, J. M., Chohen, E. A., & Poutanen, K. (2007). The future of whole grains. In: Whole Grains and Health, Marquart, L., Jacobs, D. R., McIntosh, G. H., Poutanen, K., & Reicks, M. (Eds.). Black Well Publishing, Iowa. pp. 3-16. McRae, M. P. (2017). Health benefits of dietary whole grains: an umbrella review of meta- analyses. Journal of Chiropractic Medicine, 16, 10-18. Meroth, C. B., Hammes, W. P., & Hertel, C. (2003). Identification and population dynamics of yeasts in sourdough fermentation processes by PCR-denaturing gradient gel electrophoresis. Applied and Environmental Microbiology, 69, 7453-7461. Mintel (2011). Mintel global new products database: Whole grain statistics. Whole Grains Council. https://wholegrainscouncil.org/newsroom/whole-grain-statistics (Accessed 15/03/2017).

Njobeh, P. B., Dutton, M. F., Kock, S. H., Chuturgoon, A., Stoev, S., & Seifert, K. (2009). Contamination with storage fungi of human food from Cameroon. International Journal of Food Microbiology, 135, 193-198. Nuobariene, L., Cizeikiene, D., Gradzeviciute, E., Hansen, A. S., Rasmussen, S. K., Juodeikiene, G., & Vogensen, F. K. (2015). Phytase-active lactic acid bacteria from sourdoughs: Isolation and identification. LWT-Food Science and Technology, 63, 766-722. Nyambane, B., Thari, W. M., Wangoh, J., & Njage, P. M. K. (2014). Lactic acid bacteria and yeasts involved in the fermentation of amabere amaruranu, a Kenyan fermented milk. Food Science and Nutrition, 2, 692-699. Odunmbaku, L. A., Sobowale, S. S., Adenekan, M. K., Oloyede, T., Adebiyi, J. A., & Adebo, O. A. (2018). Influence of steeping duration, drying temperature and duration on the chemical composition of sorghum-starch. Food Science and Nutrition, 6, 348-355. Price, M. L., Van Scoyoc, S., & Butler, L. G. (1978). A critical evaluation of the vanillin reaction as an assay for tannin in sorghum grain. Journal of Agricultural and Food Chemistry, 26, 1214-1218. Rose, D. J., Williams, E., Mkandawire, N. L., Weller, C. L., & Jackson, D. S. (2013). Use of whole grain and refined flour from tannin and non-tannin sorghum (Sorghum bicolor (L.) Moench) varieties in frybread. Food Science and Technology International, 20, 333-339. Ross, J. A., Kasum, C. M. (2002). Dietary flavonoids: Bioavailability, metabolic effects, and safety. Annual Reviews in Nutrition, 22, 19-34. Schaffer-Lequart, C., Lehmann, U., Ross, A. B., Roger, O., Eldridge, A. L., Ananta, E., Bietry, M. F., King, L. R., Moroni, A. V., Srichuwong, S., Wavreille, A. S., Redgwell, R., Labat, E., & Robin, F. (2017). Whole grain in manufactured foods: Current use, challenges and the way forward. Critical Reviews in Food Science and Nutrition, 57, 1562-1568. Seal, C. J., & Brownlee, I. A. (2015). Whole-grain foods and chronic disease: evidence from epidemiological and intervention studies. The Proceedings of the Nutrition Society, 74, 313- 319. Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Sobowale, S. S., Adebiyi, J. A., & Adebo, O. A. (2017). Effect of optimizing processing conditions (blanching and frying) on the qualities of deep-fat fried bonga fish (Ethmalosa fimbriata). Journal of Food Process Engineering, 40, e12551 (1-8).

Tabacco, E., Borreani, G., Crovetto, G.M., Galassi, G., Colombo, D., & Cavallarin, L. (2006). Effect of chestnut tannin on fermentation quality, proteolysis, and protein rumen degradability of alfalfa silage. Journal of Diary Science, 89, 4736-4746. Tabaraki, R., & Heidarizadi, E. (2017). Simultaneous multidye biosorption by chemically modified Sargassum glaucescens: Doehlert optimization and kinetic, equilibrium, and thermodynamic study in ternary system. Separation Science and Technology, 52, 583-595. Taylor, J., & Taylor, J. R. N. (2002). Alleviation of the adverse effect of cooking on sorghum protein digestibility through fermentation in traditional African porridges. International Journal of Food Science and Technology, 37, 129-137. Taylor, J. R. N., & Duodu, K. G. (2015). Effects of processing sorghum and millets on their phenolic phytochemicals and the implications of this to the health-enhancing properties of sorghum and millet food and beverage products. Journal of the Science of Food and Agriculture, 95, 225-237. Wu, Y., Li, X., Xiang, W., Zhu, C., Lin, Z., Wu, Y., Li, J., Pandravada, S., Ridder, D. D., Bai, G., Wang, M. L., Trick, H. N., Bean, S. R., Tuinstra, M. R., Tesso, T. T., & Yu, J. (2012). Presence of tannins in sorghum grains is conditioned by different natural alleles of Tannin1. Proceedings of the National Academy of Science USA, 109, 10281-10286.

CHAPTER FOUR2 Fermentation by Lactobacillus fermentum strains (singly and in combination) enhances the physicochemical properties, bioactive components and antioxidant activity of ting processed from two whole grain sorghum types

Abstract The effect of single and co-starter culture addition of L. fermentum (FUA 3165 and 3321) on ting properties obtained from whole grain sorghum types [high tannin and low tannin] were investigated. Ting samples were obtained from low tannin and high tannin sorghum types after fermentation at 34 oC for 24 h and 28 oC for 72 h, respectively. Starter culture addition (singly and in combination) yielded significantly lower pH (4.94 – 5.17), tannin content (0.41 – 3.01 mg CE/g), total phenolic content (8.11 – 34.89 mg GAE/g) and flavonoid content (7.53 – 29.61 mg CE/g), with higher titratable acidity (1.98 – 2.67 g/kg) and antioxidant activity (4.82 – 7.81 µM TE/g). Liquid chromatography-tandem mass spectrometry (LC-MS/MS) quantification of some phenolic compounds showed that whole grain ting samples from the high tannin sorghum, fermented with L. fermentum FUA 3321 had significantly higher bioactive compounds (catechin, gallic acid and quercetin). Scanning electron microscopy (SEM) also showed pronounced variations in the cellular morphology of the L. fermentum ting samples. Sorghum type and fermentation with L. fermentum significantly influenced fermentation and subsequently the whole grain ting composition. Fermenting HT sorghum type with L. fermentum FUA 3321 yielded whole grain ting desirable biochemical properties and better phenolic composition. Keywords: Sorghum fermentation, sorghum ting, lactic acid bacteria, whole grain sorghum, and Lactobacillus fermentum.

2This part of the study has been publsihed in Journal of Cereal Science

4.0 Introduction Sorghum (Sorghum bicolor) is a drought resistant grass majorly cultivated for its grain use and the 3rd most important cereal crop in Africa (FAOSTAT, 2017). It is an important food source for millions of people in the continent and known to consist of a wide range of beneficial bioactive compounds such as phenolic compounds and other compounds constituting the grain (Awika, 2017).

Fermentation is one of the oldest methods used to transform sorghum into subsequent food products, thereby improving nutritional qualities, increasing shelf life, palatability and functional constituents (Adebiyi et al., 2018). Porridge and gruels are one of the most common forms of fermented food products, with sorghum being the most popular cereal grain used in developing countries. Ting is one of such popular indigenous fermented sorghum porridges used for the preparation of bogobe (a hard and firm porridge, eaten for lunch or dinner) and motogo (a soft porridge, mostly eaten for breakfast). It is predominantly consumed in South Africa, Botswana and other neighbouring countries, serving an important meal for both adults and children (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011; Adebo et al., 2018).

According to Madoroba et al. (2011), challenges associated with variations in the sensory characteristics and quality of spontaneously fermented ting hampers its large scale production and subsequent commercialization. Variations in quality, inconsistencies in sensory properties and final composition of the end-product are some of these challenges. Although many fermentation processes in Africa still largely rely on spontaneous fermentation, the use of starter cultures is desirable to ensure consistency, maintain hygiene, improve quality and guarantee constant sensory appeal (Adebo et al., 2017a). The use of lactic acid bacteria (LAB) as starter cultures for cereal fermentation has been a convention, since they are dominant in the fermentation microbiota and have been reported to cause desirable effect on sensory and biochemical properties of food (Liptakova et al., 2017). Furthermore, standardized fermentation process, enhanced nutritional quality, increased overall acceptability, contribution to aroma and taste formation and increased levels of phenolic compounds favours the use of LABs (Svensson et al., 2010; Pistarino et al., 2013; Adebo et al., 2017a; Adebo et al., 2017b).

Worldwide consumption of whole grain (WG) products has tremendously grown due to increasing awareness of their health-promoting benefits and importance (Schaffer-Lequart et al., 2017). Different epidemiological studies have equally highlighted the significant and

100 important role of WG-diets, including protective action against such non-communicable diseases as cancers, diabetes, obesity and cardiovascular diseases (Schaffer-Lequart et al., 2017). As compared to refined grains, WGs are better sources of minerals, fiber, phenolic compounds, phytosterols, vitamins, ligands and other important bioactive compounds (García- Mantrana et al., 2016). Although most fermented cereal foods are made from refined grains, studies on fermented WG-cereal products have shown better composition, improved satiety and higher content and availability of bioactive compounds (Nuobariene et al., 2015; García- Mantrana et al., 2016).

Previous studies on the microbiota of ting have equally reported the dominance of LABs (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011; Madoroba et al., 2011). This study thus utilized LAB strains earlier isolated from ting (Sekwati-Monang and Gänzle, 2011) and investigated its potential use as starter culture for ting production from WG-sorghum. Lactobacillus fermentum was particularly selected due to its dominance in ting (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011; Madoroba et al., 2011), its reported acidification, competitive advantage, antimicrobial activity and its ability to produce bacteriocins (Svensson et al., 2010; Owusu-Kwarteng et al., 2015). In this study, two strains of L. fermentum starter cultures (singly and in combination) were used for ting fermentation from WG sorghum and its effects on physicochemical properties, some bioactive components and antioxidant activities were determined. Accordingly, this is the first study investigating the use of L. fermentum starter cultures (singly and in combination) to produce ting from WG- sorghum.

4.1 Materials and methods 4.1.1 Raw material and sample preparation Sorghum (Sorghum bicolor L.) grain types [high tannin (HT) and low tannin (LT)] were purchased from Agricol (Pty) Ltd. Potchefstroom, South Africa. The grains were milled through a 2 mm screen using a Perten Laboratory Mill 3600 (Perten Instruments, Sweden) to obtain WG-sorghum flour. Tannin content (TNC) of both raw LT and HT sorghum types were investigated, with values recorded as 17.97 and 31.68 mg CE/g, and as such, were subsequently classified as LT and HT sorghum types, respectively (Chapter Three).

4.1.2 Lactobacillus strains

101

Two L. fermentum strains (L. fermentum FUA 3165 and L. fermentum FUA 3321) were used singly and in combination for the controlled fermentation of WG-ting. These strains were earlier isolated from ting (Sekwati-Monang and Gänzle, 2011) and subsequently grown in MRS broth using a modified method of Sekwati-Monang and Gänzle (2011). The strains were grown in 10 mL MRS broth (HiMedia, India) for 24 h in an incubator (IncoShake, Labotec, South Africa) with temperature set at 34 oC. The liquid culture obtained was subsequently centrifuged (Eppendorf 5702R, Merck South Africa) at 3000 rpm and 10 oC for 5 min, to obtain cells. This was washed thrice with sterile phosphate buffer saline (PBS) and reconstituted in 10 mL of PBS (Sekwati-Monang and Gänzle, 2011).

4.1.3 Fermentation of sorghum into WG-ting Ting was processed by mixing milled WG-sorghum flour and sterile distilled water (40 oC) (1:1, w/v) (Adebo et al., 2018). The L. fermentum strains (both singly and in combination) (cell counts of approximately 105 cfu/mL) were then inoculated in the mixture and fermented for 72 h at 28 oC (HT-sorghum type) and for 24 h at 34 oC (LT-sorghum type). The selection of these fermentation time and temperature conditions was guided by optimal results earlier obtained in Chapter Three, where fermentation time and temperature of these sorghum types were optimized. Control WG-ting samples were also obtained by spontaneously fermenting sorghum under similar conditions (without the strains).

4.1.4 pH and titratable acidity (TTA) determination pH of the WG-ting samples was measured using a pH meter (pH 510, Eutech Pte Ltd, Singapore), while TTA was determined by titrating a mixture of 2 g of the ting samples in 20 mL distilled water against 0.1 N NaOH, using phenolphthalein as an indicator (Adebo et al., 2018).

4.1.5 Estimation of viable microbial counts Into a sterile test tube containing 9 mL sterile distilled water, 1 g of WG-ting sample was added and vortexed. A ten-fold serial dilution of the WG-ting samples were surface plated on plate count agar (Oxoid, South Africa), potato dextrose agar (Merck, South Africa) and MRS agar (Sigma Aldrich, Germany) for bacterial, fungal and yeast and lactic acid bacteria counts respectively. Plates were incubated (IncoShake, Labotec, South Africa) at 30 oC for 72 h for total bacteria count (TBC), 25 oC for 120 h for total fungal and yeast count (TFYC) and anaerobically at 35 oC for 72 h for total lactic acid bacteria count (TLABC). Viable microbial counts were determined after incubation (Adebo et al., 2018).

102

4.1.6 Tannin content (TNC) Extraction was done using acidified methanol (1% HCl in methanol) (Adebo et al., 2018) and extracts obtained were used for the determination of tannin content, total phenolic content, total flavonoid content and antioxidant activity assays. The TNC was determined using the vanillin method described by Price et al. (1978). To a 1 mL of extract in a test tube, 5 mL each of 8% HCl in methanol and 1% vanillin (Merck, South Africa) were added. The resulting mixture was vortexed and incubated in a water bath (30 oC) for 20 mins. A blank experiment was also performed following similar steps but this time using 5 mL of 4% HCl. The absorbance of the mixture was read on a spectrophotometer (Biomate, Thermo Spectronic, Rochester, USA) set at a wavelength of 500 nm. Catechin (Sigma Aldrich, Germany) was used as a standard and results obtained expressed in mg catechin equivalents (CE)/g.

4.1.7 Total phenolic content (TPC) The TPC was determined according to the Folin–Ciocalteu method of Ainsworth and Gillespie, (2007). Accordingly, 10 µL of the extract was put in a test tube containing 500 µL of distilled water and reacted with 50 µL of Folin–Ciocalteu phenol reagent (Sigma Aldrich, Germany).

This was left to stand in the dark for 3 min followed by the addition of 200 µL of 20% Na2CO3 (g/v) and finally 245 µL of distilled water. The mixture (300 µL) was pipetted into a 96-well microplate, wrapped with aluminum foil, incubated in the dark for 30 min and absorbance read on a microplate reader (iMark, Biorad, South Africa), set at a wavelength of 750 nm. Gallic acid (Sigma Aldrich, Germany) was used as a standard and results obtained were expressed in mg GAE/g sample.

4.1.8 Total flavonoid content (TFC) Total flavonoid content (TFC) was determined following the method described by Al-Farsi and

Lee (2008) by mixing 30 µL of the extract with 20 µL of 36 mM NaNO2 followed by incubation in the dark for 5 min. Afterwards, 20 µL of 94 mM AlCl3 was added and incubated for another 5 min in the dark. Thereafter, 100 µL of NaOH were added and absorbance of the resulting mixture read on a microplate reader (iMark, Biorad, South Africa) at 450 nm. Catechin (Sigma Aldrich, Germany) was used as a standard and results obtained were expressed in mg CE/g sample.

4.1.9 Antioxidant activity (AA)

103

The antioxidant activity (AA) was determined using the trolox equivalent antioxidant capacity (TEAC) assay (also known as ABTS method) of Awika et al. (2003). To 20 µL extract, 180 µL of the ABTS free radical cation solution (equal volumes of 7 mM ABTS and 2.45 mM

K2S2O8 previously incubated for 12 h) were added and the content incubated for 5 min in the dark. Trolox (Sigma Aldrich, Germany) was used as a standard. Absorbance of each solution was measured on a micro plate reader (iMark, Biorad, South Africa) set at 750 nm and results obtained were expressed as µM TE/g sample.

4.1.10 LC-MS/MS quantification of selected bioactive compounds Phenolic compounds were quantified in triplicate samples of WG-ting. This was done in a multiple reaction monitoring (MRM) mode by injecting 2 µL of each extract into a liquid chromatography-tandem mass spectrometry (LC-MS/MS) system. The UPLC instrument (Shimadzu Kyoto, Japan) used is equipped with an auto-sampler (SIL-30AC), communication bus module (CBM-20A), column oven (CTO-30A), degassing unit (DGU-20A5R) and a liquid chromatograph (LC-30AD) interfaced with a triple quadrupole mass spectrometer (LC-MS-

8030). A Raptor C18 column (2.7 µm particle size × 100 mm length × 2.1 mm ID, Restek, USA) was used and the analysis done at a constant flow rate of 0.2 mL/min. All standards (catechin, quercetin, gallic acid and vanillin) purchased from Sigma Aldrich, South Africa were dissolved in LC-grade methanol at concentrations of 0.5, 1, 2, 4, 5, 10 and 40 ppm. The choice of these standards was based on availability. The mobile phases, solvents A and B consisted of 0.1% formic acid in water and 0.1% formic acid in acetonitrile, respectively. The solvent gradients were 15% solvent B for 1 min, 95% solvent B for 9 min and 15% solvent B for 5 min. This was followed by 3% solvent B to re-equilibrate the column. The column temperature was maintained at 40 oC throughout the chromatographic runs.

4.1.11 Scanning electron microscopy (SEM) of the ting samples The ting samples were mounted on aluminum stubs and sprayed-coated with carbon in a coater (Quorum Q150TE, Quorum Technologies, UK). The samples were subsequently transferred to the SEM specimen chamber, subjected to electron beam and viewed on a SEM instrument (Vega 3 XMU, TESCAN, Czech Republic) under vacuum.

4.1.12 Statistical analysis All experiments were done in triplicates. A one-way analysis of variance (ANOVA) was conducted and differences at 5% (p  0.05) level were considered statistically significant.

104

4.2 Results and discussion Adaptability, behavior and the effect of other microorganisms during fermentation are difficult to control and maintain due to the plethora of other microorganisms present. This has consequently informed the use of selected starter cultures for fermentation of foods that ensures uniformity in production, standardization of the fermentation process and improvement in quality (Holzapfel, 2002; Adebo et al., 2017a). This study therefore, investigated the effect of two starter cultures on the composition of ting from whole grain (WG) sorghum.

4.2.1 pH and TTA of WG-ting The pH and titratable acidity (TTA) are both important indicators of the fermentation process, as their changes are consequent of metabolic activity of microorganisms. Irrespective of the sorghum type used, the pH values of WG-ting samples decreased with a corresponding increase in TTA values, after fermentation. Inoculation with both L. fermentum strains resulted in a significant (p  0.05) lowering of the pH (Table 4.1).

For both sorghum types, significantly (p  0.05) lower pH values of 4.98 and 4.94 were observed for WG-ting samples obtained with L. fermentum FUA 3321, while the highest pH values (5.87 and 6.00) were recorded for the control (spontaneously fermented) samples (Table 4.1). As observed with the pH, the highest TTA values were obtained from WG-ting samples fermented with the LAB strains, but values were significantly (p  0.05) low in spontaneously fermented samples with TTA values of 1.13 (HT-ting) and 1.34 g/kg (LT-ting). The relatively higher pH and corresponding low TTA values of both HT and LT-ting could be ascribed to suppressed metabolic activities of the L. fermentum strains, influenced by the presence of seed coats (pericarp). Similar trends have been reported for fermented whole grain (WG) products and reportedly attributed to the presence of bioactive compounds in the seed coat that limits microbial activity (Nuobariene et al., 2015; García-Mantrana et al., 2016; Adebo et al., 2018).

105

Table 4.1: Effect of fermentation by L. fermentum strains on biochemical, microbial and health promoting (bioactive) components of ting from whole grain sorghum

pH TTA TBC TLABC TFYC TNC TPC TFC AA (g/kg) (×105 cfu/g) (×105 cfu/g) (cfu/g) (mg CE/g) (mg GAE/g) (mg CE/g) (µM TE/g) HT sorghum type HT0000 6.93d 0.06a 0.10a - 0.09× 105a 31.68d 64.87e 51.08e 4.03b (0.04) (0.07) (1.13) (1.36) (0.17) (0.28) (0.13) (0.10) 2872 5.87c 1.13b 3.84b 1.68a 1.17 × 105e 14.11c 46.09d 40.86d 3.71a (0.02) (0.03) (1.39) (1.84) (2.01) (0.16) (0.22) (0.18) (0.08) 2872 (3165) 5.17b 1.98c 5.63c 4.69c 7.07× 102d 2.83ab 32.13b 26.38ab 7.07c (0.07) (0.03) (1.97) (1.39) (1.97) (0.21) (1.41) (1.02) (0.09) 2872 (3321) 4.98a 2.24d 5.68d 4.70c 7.13× 102c 2.61a 31.38a 25.44a 7.81e (0.02) (0.14) (2.02) (1.88) (1.55) (0.18) (1.88) (1.11) (0.08) 2872 (3165+3321) 5.04a 2.09c 5.69d 4.67b 6.99× 102b 3.01b 34.89c 29.61c 7.42cd (0.04) (0.04) (1.91) (1.73) (1.49) (0.09) (1.53) (1.37) (0.08) LT sorghum type LT0000 6.94c 0.04a 0.30a - 0.35× 105a 17.97e 32.83e 13.43e 1.88b (0.02) (0.07) (1.29) (2.16) (0.13) (0.23) (0.14) (0.13) 3424 6.00b 1.34b 71.90b 20.10a 9.91 × 106e 5.36d 15.48d 9.67d 1.18a (0.02) (0.06) (2.11) (2.39) (1.48) (1.09) (0.12) (0.72) (0.05) 3424 (3165) 5.01a 2.23c 87.90c 66.50b 3.15× 103d 0.52ab 8.72bc 7.80ab 4.87c (0.02) (0.02) (2.05) (1.63) (1.48) (0.26) (0.94) (1.09) (0.04) 3424 (3321) 4.94a 2.67e 88.00c 67.10c 3.03× 103c 0.41a 8.11a 7.53a 4.99d (0.08) (0.10) (1.63) (1.70) (1.64) (0.18) (0.94) (1.13) (0.09) 3424 (3165+3321) 4.95a 2.52d 88.30d 66.80b 2.98 × 103b 0.63bc 9.01c 8.69c 4.82e (0.01) (0.03) (1.96) (1.68) (1.73) (0.10) (0.18) (0.81) (0.06) HT – high tannin sorghum; LT – low tannin sorghum; HT0000 – raw HT-sorghum type; LT0000 – raw LT-sorghum type; 2872 – naturally fermented ting from HT-sorghum type at 28 oC for 72 h; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum type at 34 oC for 24 h. TTA – titratable acidity, TBC – total viable bacteria count; TLABC – total lactic acid bacteria count; TFYC – total fungal and yeast count; TNC – tannin content; TPC – total phenolic content; TFC – total flavonoid content; AA – antioxidant activity. *Values in parentheses represents the standard deviation of replicated determinations. Means with no common letters within a column significantly (p  0.05) differ.

106

Addition of starter cultures were observed to cause an accelerated acidification of the WG-ting samples obtained from both sorghum types. Fermentation with starter cultures is known to enhance acid production and ensure rapid reduction in pH levels (Sekwati-Monang and Gänzle, 2011) via the production of organic acids. Such rapid acidification by these starter cultures causing pH reduction is considered crucial from a food safety perspective (Leroy and de Vuyst, 2004; Adebo et al., 2018) and plays a significant role in reducing food pathogens thereby, extending the shelf life of the product. The higher pH and low TTA values observed with the naturally fermented samples may be ascribed to initial low numbers of LABs causing longer lag periods, consequently reducing metabolic and microbial activity. A Pearson’s correlation between the pH and TTA values showed a statistically significant (p  0.01) negative correlation for HT-ting (R2 = ‒0.998) and LT-ting (R2 = ‒0.968) samples (Table 4.2). This indicates a reduction of pH values with an associated increase in titratable acidity levels.

Table 4.2: Pearson’s correlation between the investigated parameters of the WG-ting pH TTA TBC TLABC TFYC TNC TPC TFC AA HT sorghum type pH ‒ ‒ ‒0.981* 0.981* 0.983* 0.957* 0.950 0.999** 0.998** 0.986* TTA ‒ 0.981* 0.978* 0.977* ‒ ‒ ‒ ‒ 0.998** 0.980* 0.966* 0.962* 0.998** TBC ‒ 0.981* 0.999** ‒ ‒ ‒ ‒ ‒ 0.986* 1.000** 0.999** 0.972* 0.964* 0.990** TLABC ‒ 0.978* 0.999** ‒ ‒ ‒ ‒ ‒ 0.981* 1.000** 1.000** 0.977* 0.970* 0.988* TFYC ‒ 0.977* ‒ ‒1.000** 1.000** 0.975* 0.968* ‒ 0.981* 1.000** 0.987* TNC 0.983* ‒ ‒ ‒1.000** 1.000** 0.981* 0.974* ‒ 0.980* 0.999** 0.989* TPC 0.957* ‒ ‒ ‒0.977* 0.975* 0.981* 1.000** 0.971* 0.966* 0.972* TFC 0.950 ‒ ‒ ‒0.970* 0.968* 0.974* 1.000* ‒ 0.962* 0.964* 0.965* AA ‒ 0.998** 0.990** 0.988* ‒ ‒ ‒ ‒ 0.999** 0.987* 0.989* 0.971* 0.965* LT sorghum type pH ‒ ‒ ‒0.999** 0.998** 0.998** 0.994** 0.852 ‒ 0.968* 0.999** 0.998** TTA ‒ 0.954* 0.955* ‒ ‒ ‒ ‒0.824 0.957* 0.968* 0.952* 0.955* 0.963* TBC ‒ 0.954* 1.000** ‒ ‒ ‒ ‒0.849 0.999** 0.999** 1.000** 0.999** 0.993** TLABC ‒ 0.955* 1.000** ‒ ‒ ‒ ‒0.860 0.999** 0.999** 1.000** 0.999** 0.995** TFYC 0.998** ‒ ‒ ‒1.000** 0.999** 0.994** 0.859 ‒ 0.952* 1.000** 0.999**

107

pH TTA TBC TLABC TFYC TNC TPC TFC AA TNC 0.998** ‒ ‒ ‒0.999** 0.999** 0.997** 0.877 ‒ 0.955* 0.999** 1.000** TPC 0.994** ‒ ‒ ‒0.995** 0.994** 0.997** 0.903 ‒ 0.963* 0.993** 0.998** TFC 0.852 ‒0.824 ‒0.849 ‒0.860 0.859 0.877 0.903 ‒0.857 AA ‒ 0.957* 0.999** 0.999** ‒ ‒ ‒ ‒0.875 0.998** 0.999** 1.000** 0.998** ** ‒ Correlation is significant at 1%; * ‒ Correlation is significant at 5%. HT – high tannin sorghum; LT – low tannin sorghum; TTA – titratable acidity, TBC – total viable bacteria count; TLABC – total lactic acid bacteria count; TFYC – total fungal and yeast count; TNC – tannin content; TPC – total phenolic content; TFC – total flavonoid content; AA – antioxidant activity.

4.2.2 Microbial analysis of WG-ting The microbial counts of both the control (spontaneously fermented) and starter culture fermented WG-ting samples are presented in Table 4.1. The mean total viable bacteria count (TBC) of samples fermented with starter cultures was significantly (p  0.05) higher than that of the spontaneously fermented ones. A similar trend in the total lactic acid bacteria content (TLABC) of WG-ting samples fermented with L. fermentum, suggests that the starter culture dominated and constituted major portion of the viable bacteria growing on the WG-ting samples. Despite being strains of the same species, the L. fermentum starter cultures exhibited different characteristics (Table 4.1). This thus suggests that the competitiveness and reactions of the fermenting microbiota was dependent on the type of L. fermentum strain and substrate composition. Such differences have also been reported to determine and affected microbial community dynamics and metabolite kinetics during fermentation (De Vuyst et al., 2014). This might have equally influenced the significantly (p  0.05) higher mean TBC and TLABC generally observed for the LT-ting samples analyzed (Table 4.1), indicating dominance and a better adaptability of the strains to the LT-sorghum substrate. As compared to the HT-ting samples, relatively lower TPC, TNC and TFC levels might have equally encouraged microbial growth in LT-sorghum. Such high viable counts are needed to influence desired acid production that invariably lead to pH reduction, which subsequently affect organoleptic properties, shelf-life and safety of ting.

Total fungal and yeast counts (TFYC) in both sorghum types were significantly (p  0.05) lower in starter culture fermented samples as compared to the control samples. This could be attributed to inhibitory and antagonistic effect of LABs on the growth and survival of other microorganisms, including fungi and yeasts. The observed significant (p  0.05) reduction in the TFYC would have consequently influenced the concomitant reduced pH values of LAB fermented WG-ting samples. Lactic acid bacterial strains are known to inhibit the growth and

108 proliferation of spoilage micro-organisms by various mechanisms including the reduction of pH, production of bacteriocins, antagonistic and competitive actions (Holzafpel, 2002; Adebiyi et al., 2018). A perfect negative correlation (R2 = ‒1.000) between the TLABC and TFYC of both the HT- and LT-ting samples (Table 4.2) further confirms this assertion that an increase in TLABC would correspond to a significant decrease in TFYC.

4.2.3 Tannin content (TNC), total phenolic content (TPC), total flavonoid content (TFC), antioxidant activity (AA) and bioactive compounds of WG-ting Irrespective of the sample type, fermentation using starter cultures generally caused a significant (p  0.05) decrease in TNC, TPC and TFC studied (Table 4.1). When compared to the raw sorghum samples (Chapter Three), a decrease in TNC of 56 and 70% were obtained for spontaneously fermented ting from the HT- and LT-sorghum, respectively. Decreases of over 90% in TNC were however, found in ting fermented with the LAB strains. A similar trend but lesser decrease in TPC and TFC in the spontaneously fermented samples when compared to the LAB-fermented ones were equally recorded (Table 4.1). The higher decrease observed for the LAB-fermented ting than spontaneously fermented samples could be attributed to improved metabolic activities due to the inoculated L. fermentum strains. The lesser decrease generally observed in HT-ting samples relates to reduced microbial action in the raw HT- sorghum substrate. The higher amounts of TNC, TPC and TFC in the HT-samples could have equally influenced the observed low microbial counts in this study (Table 4.1). Studies in this regard have shown that phenolic compounds can exhibit toxicity to fermenting microorganisms, influence microbial growth by causing an extension in lag phase, inhibit enzyme activity, decrease growth rate as these compounds can bind to bacterial cell and membranes, as well as interact with lipids and proteins, altering membrane permeability (Scalbert, 1991; Bvochora et al., 2005; Kemperman et al., 2010; Pacheco-Ordaz et al., 2017).

The general decrease in these constituents could be attributed to possible degradation and hydrolysis of these compounds as well as binding of phenolic structures with other compounds, thereby reducing their extractability. Through the action of fermenting microorganisms, subsequent enzymes produced could have also depolymerized these phenolic compounds (Othman et al., 2009). It is established that fermentation can cause a decrease in the polyphenolic content due to oxidation and condensation reactions of polyphenols forming different compounds (Shrestha et al., 2010; Suazo et al., 2014; Taylor and Duodu, 2015). This reduction in the content of phenolic compounds after fermentation has equally been

109 demonstrated in other studies, attributing such changes to rearrangement of phenolic structures and cleavage of phenolic compounds (Towo et al., 2006; Taylor and Duodu, 2015; Adebo et al., 2018). It is noteworthy to mention that the same L. fermentum FUA 3165 strain used in this study has been reported to exhibit glucosidase, phenolic acid reductase and phenolic acid decarboxylase activities, which contributed to the polyphenol metabolism during sorghum fermentation (Svensson et al., 2010).

Tannins are large polyphenol polymers that can inhibit digestive enzymes, bind and possibly reduce the digestibility of nutrients (Awika, 2017). They are thus generally referred to as “antinutritional factors”. Fermentation, especially with the L. fermentum strains was observed to significantly (p  0.05) reduce tannin content (TNC), suggesting that these tannin-related compounds have been broken down to low molecular weight compounds, that can be more bio- accessible. Such reduction in the content of tannins and consequent increase in bio-accessibility after fermentation has been demonstrated in sorghum gruels (Towo et al., 2006). Tannins have equally been reported to confer some antioxidant activities and health benefits in sorghum (Awika, 2017). Accordingly, a significant (p  0.05) increase in the antioxidant activity of the LAB-fermented WG-ting samples could mean that bioactive compounds (including non- phenolic components) contributing to the radical scavenging properties of the ting were possibly regenerated and released after fermentation with the L. fermentum strains.

Further to the availability of standards, some phenolic compounds (gallic acid, quercetin, catechin and vanillin) were quantified by LC-MS/MS, with vanillin occurring at traceable levels. Masses, fragments and retention times of these compounds and other optimized parameters are presented in Table 4.3.

Table 4.3: Identity, properties and optimized parameters of the phenolic compounds investigated in the WG-ting

Rt Standard MW Parent ion m/z MS/MS fragments (Daughter CE (min) [M-H]– ions) 1.27 Gallic acid 170 169 143, 125, 79 35, 16, 25 1.66 Catechin 290 289 245, 109 13, 24 3.84 Vanillin 152 151 136, 108 16, 25 6.14 Quercetin 302 301 179, 151 19, 21 Rt – retention time, MW – molecular weight, m/z – mass-to-charge ratio, CE – collision energy.

Quantification of the phenolic compounds using specific standards (Figure 4.1A and B) were observed to have relatively lower values when compared to the TPC and TFC levels (Table

110

4.1). The relatively higher quantity of the phenolic compounds as presented in Table 4.1 could mean that some other compounds, such as amino acids, small peptides, aldehydes, amines and reducing sugars all present in the WG-ting samples were initially quantified when colorimetric methods were used. Whole grain ting samples from the HT-sorghum, obtained with L. fermentum FUA 3321 had higher levels of catechin (14.94 µg/g) and quercetin (1.94 µg/g) (Figure 4.1A) than the spontaneously fermented ones and control samples. A similar trend was also observed for WG-ting samples obtained from the LT-sorghum (Figure 4.1B). A slightly general increase in the levels of flavonoids and phenolic acid suggests a better release of these bioactive compounds in LAB-fermented WG-ting samples, possibly via hydrolysis and activities of glycosidases (Ju et al., 2009; Svensson et al., 2010). During the fermentation process, proteolytic actions of the LAB strains might have equally contributed to this observation, causing release of bound phenolic compounds (Shrestha et al., 2010; Ademiluyi and Oboh, 2011), leading to the formation of biologically active compounds.

A 16 c 14 b b a a 12

8 µg/g Catechin 6 Gallic acid 4 Quercetin c c d b c bc 2 bc b a a 0

111

B 0,7 d c c 0,6

b 0,5

0,4

µg/g c Catechin 0,3 c a c c Gallic acid b b 0,2 b a Quecertin a 0,1 a

Figure 4.1: Effect of fermentation by L. fermentum strains on phenolic compounds of ting from whole grain sorghum. (A) Ting samples obtained from the LT sorghum type; (B) Ting samples obtained from the LT sorghum type. 2872 – naturally fermented ting from HT-sorghum; 3424 – naturally fermented ting from LT-sorghum type; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321

The observed increases in the amounts of quercetin, catechin and gallic acid might have contributed to the increased antioxidant activities observed in LAB fermented WG-ting samples (Table 4.1). This has been demonstrated in other studies (Oliveira et al., 2012; Pistarino et al., 2013; Dey and Kuhad, 2014), suggesting that an increase in the levels of phenolic compounds can result in increased antioxidant activity. Equally important are other bioactive compounds generated during the fermentation process that might have contributed to this observation. Significantly (p  0.05) higher flavonoid levels are desirable considering the recent upsurge and renewed interest in these compounds. This is particularly related to their ability to enhance intestinal barrier function and other epidemiological studies indicating a strong correlation between consumption of phenolic rich compounds and mitigation of diseases (Adebo et al., 2017a; Awika, 2017).

4.2.4 Scanning electron microscopy of the WG-ting samples Scanning electron microscopy (SEM) was used to investigate and study the morphological differences in naturally (spontaneously) fermented and LAB-fermented ting samples. The SEM images obtained are presented in Figure 4.2 and as observed from the micrographs, fermentation with the LAB strains caused better degradation of the ting granular structure. More loosened, disoriented granular arrangements with pits were seen in ting samples

112 fermented with the L. fermentum strains. Such pits suggest better hydrolysis and degradation of starch and protein molecules by amylases and proteases, respectively, potentially caused by increased metabolic activity of L. fermentum strains (Table 4.1). The structurally constricted morphology obtained in the naturally fermented ting samples could possibly be attributed to lesser degradation of organelles within the grain endosperm (especially starch granules, protein bodies, protein matrix and cell wall material), which were better disrupted after fermentation with L. fermentum strains.

Figure 4.2: Scanning electron microscopy images of ting samples: A ‒ 34 oC 24 h, B ‒ 34 oC 24 h with L. fermentum FUA 3165, C ‒ 34 oC 24 h with L. fermentum FUA 3321, D ‒ 34 oC 24 h with L. fermentum FUA 3165 and FUA 3321, E ‒ 28 oC 72 h, F ‒ 28 oC 72 h with L. fermentum FUA 3165, G ‒ 28 oC 72 h with L. fermentum FUA 3321, H – 28 oC 72 h with L. fermentum FUA 3165 and FUA 3321.

4.3 Conclusion Fermentation using L. fermentum strains as starter cultures was observed to significantly influence the composition of whole grain ting produced from the two sorghum types. While whole grain ting samples from the high tannin sorghum type has higher bioactive components including antioxidants, as compared to their low tannin counterparts, fermentation with LAB strains largely enhanced the release of catechin, quercetin and gallic acid. Rapid acidification, relatively higher organic acid production and reduction in fungal load with higher bacterial load and bioactive components further suggests the suitability of the LAB strains for ting fermentation from whole grain sorghum. Antagonism, probable competitive inhibition and

113 conflicting modes of similar metabolism and action by the L. fermentum strains when used in combination, might be attributed to the observations made in this study. Results obtained show that fermentation of high tannin sorghum type especially with L. fermentum FUA 3321 gave better composition of bioactive compounds than for low tannin sorghum type ting. This can further be exploited for the development of whole grain ting on a larger scale and delivery of improved health-promoting sorghum product that can increase its utilization. In future, comprehensive studies are vital to elucidate the evolution and metabolism of phenolic compounds by the LAB strains and presence of other bioactive compounds, involved in processing whole grain ting.

References Adebiyi, J. A., Obadina, A. O., Mulaba-Bafubiandi, A. F., Adebo, O. A., & Kayitesi, E. (2016). Effect of fermentation and malting on the microstructure and selected physicochemical properties of pearl millet (Pennisetum glaucum) flour and biscuit. Journal of Cereal Science, 70, 132-139. Adebiyi, J. A., Obadina, A. O., Adebo, O. A., & Kayitesi, E. (2018). Fermented and malted millet products in Africa: Expedition from traditional/ethnic foods to industrial value added products. Critical Reviews in Food Science and Nutrition, 58, 463-474. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., Phoku, J. Z., & Kayitesi, E. (2017a). Fermented pulse-based foods in developing nations as sources of functional foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 77-109. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., & Kayitesi, E. (2017b). Food metabolomics (Foodomics), a new frontier in food analysis and its potential in understanding fermented foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 211-234. Adebo, O. A., Njobeh, P. B., Mulaba-Bafubiandi, A. F., Adebiyi, J. A., Desobgo, S. C. Z., & Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, e13381, (1-10). Ademiluyi, A. O., & Oboh, G. (2011). Antioxidant properties of condiment produced from fermented bambara groundnut (Vigna subterranean L. Verdc). Journal of Food Biochemistry, 35, 1145-1160.

114

Ainsworth, E. A., & Gillespie, K. M. (2007). Estimation of total phenolic content and other oxidation substrates in plant tissues using Folin-Ciocalteu reagent. Nature Protocols, 2, 875- 877. Al-Farsi, M. A., & Lee, C. Y. (2008). Optimization of phenolics and dietary fibre extracts from date seeds. Food Chemistry, 108, 977-985. Awika, J. M., Rooney, L. W., Wu, X., Prior, R. L., & Zevallos, L. C. (2003). Screening methods to measure antioxidant activity of sorghum (Sorghum bicolor) and sorghum products. Journal of Agricultural and Food Chemistry, 51, 6657-6662. Awika, J. M. (2017). Sorghum: its unique nutritional and health-promoting attributes. In Gluten-Free Ancient Grains, Taylor, J. R. N., & Awika, J. M. (Eds.). Woodhead Publishing, United Kingdom. pp. 21-54. Bvochora, J. M., Daner, H., Miyafuji, H., Braun, R., & Zvauya, R. (2005). Variation of sorghum phenolic compounds during the preparation of opaque beer. Process Biochemistry, 40, 1207-1213. De Vuyst, L., Van Kerrebroeck, S., Harth, H., Huys, G., Daniel, H. M., & Weckx, S. (2014). Microbial ecology of sourdough fermentations: Diverse or uniform? Food Microbiology, 37, 11─29. Dey, T. B., & Kuhad, R. C. (2014). Enhanced production and extraction of phenolic compounds from wheat by solid-state fermentation with Rhizopus oryzae RCK2012. Biotechnology Reports, 4, 120-127. FAOSTAT (Food and Agriculture Organization of the United Nations Statistics) (2017). Crops statistics: Sorghum. http://www.fao.org/faostat/en/#data/QC (Accessed 16/10/2017). García-Mantrana, I., Yebra, M. J., Haros, M., & Monedero, V. (2016). Expression of bifidobacterial phytases in Lactobacillus casei and their application in a food model of whole-grain sourdough bread. International Journal of Food Microbiology, 216, 18─24. Holzafpel, W. H. (2002). Appropriate starter culture technologies for small-scale fermentation in developing countries. International Journal of Food Microbiology, 75, 197-212. Ju, H. K., Cho, E. J., Jang, M. H., Lee, Y. Y., Hong, S. S., Park, J. H., & Kwon, S. W. (2009). Characterization of increased phenolic compounds from fermented Bokbunja (Rubus coreanus Miq.) and related antioxidant activity. Journal of Pharmaceutical and Biomedical Analysis, 49, 820-827. Kemperman, R. A., Bolca, S., Roger, L. C. and Vaughan, E. E. (2010) Novel approaches for analysing gut microbes and dietary polyphenols: challenges and opportunities. Microbiology, 156, 3224-3231.

115

Leroy, F., & De Vuyst, L. (2004). Lactic acid bacteria as functional starter cultures for the food fermentation industry. Trends in Food Science and Technology, 15, 67-78. Liptakova, D., Matejcekova, Z., & Valik, L. (2017). Lactic acid bacteria and fermentation of cereals and pseudocereals. In Fermentation Processes, Jozala, A. F. (Ed.). InTech, Croatia. pp. 223-254. Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Madoroba, E., Steenkamp, E. T., Theron, J., Scheirlinck, I., Cloete, T. E., & Huys, G. (2011). Diversity and dynamics of bacterial populations during spontaneous sorghum fermentations used to produce ting, a South African food. Systematic and Applied Microbiology, 34, 227- 234. Nuobariene, L., Cizeikiene, D., Gradzeviciute, E., Hansen, A. S., Rasmussen, S. K., Juodeikiene, G., & Vogensen, F. K. (2015). Phytase-active lactic acid bacteria from sourdoughs: Isolation and identification. LWT-Food Science and Technology, 63, 766-722. Oliveira, M. S., Cipolatti, E. P., Furlong, E. B., & Soares, L. (2012). Phenolic compounds and antioxidant activity in fermented rice (Oryza sativa) bran. Ciência e Tecnologia de Alimentos Campinas, 32, 531-537. Othman, N. B., Roblain, D., Chammen, N., Thonart, P., & Hamdi, M. (2009). Antioxidant phenolic compounds loss during the fermentation of Chetoui olives. Food Chemistry, 116, 662-669. Owusu-Kwarteng, J., Tano-Debrah, K., Akabanda, F., & Jespersen, L. (2015). Technological properties and probiotic potential of Lactobacillus fermentum strains isolated from West African fermented millet dough. BMC Microbiology, 15, 1-10. Pacheco-Ordaz, R., Wall-Medrano, A., Goni, M. G., Ramos-Clamont-Montfort, G., Ayala- Zavala, J. F., & González-Aguilar, G. A. (2017). Effect of phenolic compounds on the growth of selected probiotic and pathogenic bacteria. Letters in Applied Microbiology, 66, 25-31. Pistarino, E., Aliakbarian, B., Casazza, A. A., Paini, M., Cosulich, M. E., & Perego, P. (2013). Combined effect of starter culture and temperature on phenolic compounds during fermentation of Taggiasca black olives. Food Chemistry, 138, 2043–2049.

116

Price, M. L., Van Scoyoc, S., & Butler, L. G. (1978). A critical evaluation of the vanillin reaction as an assay for tannin in sorghum grain. Journal of Agriculture and Food Chemistry, 26, 1214-1218. Scalbert, A. (1991). Antimicrobial properties of tannins. Phytochemistry, 30, 3875-3883. Schaffer-Lequart, C., Lehmann, U., Ross, A. B., Roger, O., Eldridge, A. L., Ananta, E., Bietry, M. F., King, L. R., Moroni, A. V., Srichuwong, S., Wavreille, A. S., Redgwell, R., Labat, E., & Robin, F. (2017). Whole grain in manufactured foods: Current use, challenges and the way forward. Critical Reviews in Food Science and Nutrition, 57, 1562-1568. Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Shrestha, A. K., Dahal, N. R., & Ndungustse, V. (2010). Bacillus fermentation of soybean: A review. Journal of Food Science and Technology Nepal, 6, 1-9. Suazo, Y., Davidov-Pardo, G., & Arozarena, I. (2014). Effect of fermentation and roasting on the phenolic concentration and antioxidant activity of cocoa from Nicaragua. Journal of Food Quality, 37, 50-56. Svensson, L., Sekwati-Monang, B., Lutz, D. L., Schieber A., & Gänzle, M. G. (2010). Phenolic acids and flavonoids in nonfermented and fermented red sorghum (Sorghum bicolor (L.) Moench). Journal of Agricultural and Food Chemistry, 58, 9214-9220. Taylor, J. R. N., & Duodu, K. G. (2015). Effects of processing sorghum and millets on their phenolic phytochemicals and the implications of this to the health-enhancing properties of sorghum and millet food and beverage products. Journal of the Science of Food and Agriculture, 95, 225-237. Towo, E., Matuschek, E., & Svanberg, U. (2006). Fermentation and enzyme treatment of tannin sorghum gruels: effects on phenolic compounds, phytate and in vitro accessible iron. Food Chemistry, 94, 369-376. Wouters, D., Grosu-Tudor, S., Zamfir, M., & De Vuyst, L. (2013). Applicability of Lactobacillus plantarum IMDO 788 as a starter culture to control vegetable fermentations. Journal of the Science of Food and Agriculture, 93, 3352-3361. Wouters, D., Grosu-Tudor, S., Zamfir, M., & De Vuyst, L. (2013). Applicability of Lactobacillus plantarum IMDO 788 as a starter culture to control vegetable fermentations. Journal of the Science of Food and Agriculture, 93, 3352-3361.

117

CHAPTER FIVE3 Reduction of mycotoxins during the fermentation of whole grain sorghum to whole grain ting

Abstract Mycotoxins are fungal secondary metabolites that pose health risks to exposed individuals, requiring necessary measures to reduce them. Using liquid chromatography-tandem mass spectrometry (LC-MS/MS), mycotoxins were quantified in whole grain (WG) sorghum and ting subsequently obtained from two sorghum varieties (high and low tannin). The WG-ting samples were obtained by fermenting sorghum with Lactobacillus fermentum strains (FUA 3165 and FUA 3321) 34 oC for 24 h and 28 oC for 72 h. Naturally (spontaneously) fermented WG-ting under these conditions were equally analyzed. The mycotoxins investigated included aflatoxin B1 (AFB1), B2 (AFB2), G1 (AFG1), G2 (AFG2), deoxynivalenol (DON), fumonisin B1

(FB1), B2 (FB2), B3 (FB3), ochratoxin A (OTA), ochratoxin B (OTB), T-2 toxin (T-2), zearalenone (ZEA), α-ZEA and β-ZEA. Results obtained showed that sorghum grains were contaminated with different mycotoxins, but concentrations significantly (p  0.05) reduced after fermentation, especially when using L. fermentum strains. Accordingly, over an 80% reduction was observed for FB2, T-2, α-ZEA, meanwhile and a 98% reduction of FB1 levels was noted. Other mycotoxins were reduced to varying degrees. Fermentation, particularly with L. fermentum strains showed potential to effectively sequester and limit mycotoxin contamination in WG-ting and can thus be recommended as potential starter cultures. Keywords: Sorghum, ting, Lactobacillus fermentum, LC-MS/MS, mycotoxins, and food safety.

3This part of the study will be submitted for publication.

118

5.0 Introduction Mycotoxins are deleterious and of global public health concern, with numerous reported adverse health and economic effects (Egal et al., 2005; Lewis et al., 2005; Darwish et al., 2014; Udomkun et al., 2017). These naturally occurring toxic compounds are frequent contaminants of agricultural commodities, significantly contributing to food losses, with most of the world’s population (FAO, 2011; Atherstone et al., 2014).

Sorghum is an important cereal crop for millions of people in Africa and other developing countries of the world (Odunmbaku et al., 2018). However, sorghum like other cereal crops, is susceptible to fungal proliferation during cultivation, harvest, storage and processing. Such colonization of sorghum by toxigenic fungi could be accompanied by the production of secondary metabolites including mycotoxins (Njobeh et al., 2010; Oueslati et al., 2012; Chala et al., 2014; Taye et al., 2016), further aggravated by favorable tropical climatic conditions that prevail in the continent. When ingested, these toxins elicit harmful health effects including cancer and in extreme cases, may lead to death (Lewis et al., 2005; Njobeh et al., 2010; Makun et al., 2012). Cereals have been identified as a major route of dietary exposure to mycotoxins (Karlovsky et al., 2016) and of global concern especially when these mycotoxins are carried over to subsequent products derived from them (Bankole et al., 2006; FAO, 2010).

Fermentation is a traditional, age-old technique of transforming sorghum grains like any other grain into diverse food forms that constitute the daily diets of most African populations. This processing technique has been well documented to improve shelf life, nutrient bioavailability and health beneficial composition (Adebo et al., 2017a; Adebiyi et al., 2018). Porridges are the most common forms of dishes consumed by inhabitants of Southern Africa (Rosentrater and Evers, 2018). A form of this is ting, an indigenous fermented sorghum porridge, commonly consumed as bogobe or motogo in South Africa, Botswana and other neighboring countries (Sekwati-Monang and Gänzle, 2011; Adebo et al., 2018). The product is frequently fed to infants and regularly consumed by adults (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011).

Considering the deleterious effects of mycotoxins, there is need to explore viable, safe and practicable strategies that can reduce and or at best eliminate their presence in foods. Available reports in the literature have suggested fermentation as an effective and promising technique for reducing the presence of mycotoxins, while improving nutritional composition and conferring preservative effects (Ezekiel et al., 2015; Okeke et al., 2015; Karlovsky et al., 2016;

119

Nyamete et al., 2016; Adebo et al., 2017b; Dawlal et al., 2017; Okeke et al., 2018). This study thus, investigated the reduction of mycotoxin levels by natural and lactic acid bacteria (LAB) fermentation of whole grain (WG) sorghum to ting.

5.1 Materials and Methods 5.1.1 Raw material and sample preparation Sorghum (Sorghum bicolor L.) grain cultivars, i.e., high tannin (HT) and low tannin (LT) were purchased from Agricol (Pty) Ltd. Potchefstroom, South Africa. The sorghum grains were milled through a 2 mm screen using a Perten Laboratory Mill 3600 (Perten Instruments, Sweden) to obtain whole grain (WG) sorghum flour. Tannin content (TNC) of both raw LT and HT sorghum types were investigated, with values recorded as 17.97 and 31.68 mg CE/g, and as such, were subsequently classified as LT and HT sorghum types, respectively (Chapter Three).

5.1.2 Lactobacillus strains Two L. fermentum strains (L. fermentum FUA 3165 and L. fermentum FUA 3321) were used singly and in combination for controlled fermentation of ting. These strains were earlier isolated from ting (Sekwati-Monang and Gänzle, 2011) and donated by Prof. Michael Gänzle of the University of Alberta, Canada. The strains were grown in MRS broth (HiMedia, India) using a modified method of Sekwati-Monang and Gänzle (2011). The strains were grown in 10 mL MRS broth (HiMedia, India) for 24 h at 34 oC (IncoShake, Labotec, South Africa). The liquid culture obtained was subsequently centrifuged (3000 rpm at 10 oC for 5 min, Eppendorf 5702R, Merck South Africa), cells washed thrice with sterile phosphate buffer saline (PBS) and reconstituted with 10 mL of PBS.

5.1.3 Fermentation of sorghum into WG-ting Ting was processed by mixing 50 g of WG-sorghum flour and sterile distilled water (40 oC) (1:1, w/v) (Chapter Three). The mixture was then inoculated (cell counts of approximately 105 cfu/mL) with the L. fermentum strains (both singly and in combination) and fermented for 72 h at 28 oC (HT-sorghum type) and for 24 h at 34 oC (LT-sorghum type). The selection of these fermentation time and temperature conditions was guided by optimal results obtained in our earlier study (Chapter Three) for WG-ting production. Control WG-ting samples were also obtained by spontaneously fermenting sorghum under similar conditions (without the strains). All samples were freeze-dried at -55 oC for 24 h, prior to analysis.

120

5.1.4 Mycotoxin standards

Mycotoxin standards including AFB1, AFB2, AFG1, AFG2 and DON were purchased from

Sigma Aldrich, Germany, while FB1, FB2, FB3, OTA, OTB, T-2, ZEA, α-ZEA and β-ZEA were obtained from the Council for Scientific and Industrial Research (CSIR), South Africa. They were all dissolved in LC-grade methanol and prepared at five different concentration levels ranging between 0.0002 – 0.2 μg/L (OTA, OTB), 0.001 – 1 μg/L (AFB1, AFB2, AFG1,

AFG2), 0.0025 – 2.5 μg/L (FB1, FB2, FB3, ZEA, α-ZEA, β-ZEA) and 0.05 – 5 μg/L (DON, T- 2). This was subsequently used to obtain calibration curves for mycotoxin quantification.

5.1.5 Mycotoxin extraction A modified Quick Easy Cheap Effective Rugged and Safe (QuEChERS) method earlier adapted by Oueslati et al. (2012) and Motloung et al. (2018) was followed for mycotoxin extraction. Freeze-dried raw sorghum and ting samples were finely milled, homogenized and 1 g weighed into an extraction tube containing 5 mL of distilled water, vortexed and left for 30 min. Thereafter, 5 mL [acetonitrile (ACN)/1% formic acid, v/v] extraction solvent was added and sonicated for 20 min. NaCl (0.5 g) and 2 g of MgSO4 anhydrous salt were added, the tube capped and briefly shaken, to avoid agglomeration. The tubes were vortexed, centrifuged for 15 min at 4,000 × g and the supernatant layer filtered (0.22 µm, Millipore, Bedford, MA, USA) into vials for LC-MS/MS analysis.

5.1.6 LC-MS/MS quantification of mycotoxins For the quantification of the mycotoxins, triplicates of each extract were analyzed in a multiple reaction monitoring (MRM) mode by injecting 10 µL of each into the LC-MS/MS system that consisted of an UPLC instrument (Shimadzu Kyoto, Japan) equipped with an auto-sampler (SIL-30 AC, Nexera), communication bus module (CBM-20A), column oven (CTO-30A), degassing unit (DGU-20A5R) and a liquid chromatograph (LC-30AD) interfaced with a triple quadrupole mass spectrometer (LC-MS-8030). A Raptor C18 column (2.7 µm × 100 × 2.1 mm, Restek, USA) was used and the analysis performed at a constant flow rate of 0.2 mL/min. The mobile phases, solvents A and B consisted of 0.1% formic acid in water and 0.1% formic acid in ACN:MeOH (50:50, v/v), respectively. The solvent gradients were 10% B for 0.1 min, ramped to 95% B through 8.4 min, held at 95% B for 3 min and then 10% B for 1 min, after which the column was re-equilibrated at this condition for 4.5 min. The column temperature was maintained at 40 oC throughout the chromatographic runs and MS data acquired in a positive mode. The interface nebulizing gas flow rate was 3 L/min, DL temperature was 250

121

ᵒC, heat block temperature was 400 ᵒC, and drying gas flow rate was 15 L/min. Other optimal LC-MS/MS parameters are equally presented in Table 5.1. Apparent recovery rates (ARR) were accessed by spiking blank samples (1g) with known concentrations of mycotoxins in triplicates and kept at room temperature overnight for solvent evaporation (Njobeh et al., 2012). The mycotoxin was re-extracted and quantified as earlier described in this section. Subsequent data presented in this study were adjusted based on the ARR. Percentage mycotoxin reduction after fermentation was calculated as [(A – B)/A] × 100, where A and B are initial and final mycotoxin concentrations, respectively (Adebo et al., 2017c).

Table 5.1: Identity and characteristics of the mycotoxins investigated on LC-MS/MS

2 Rt Mycotoxin MW Parent ion MS/MS CE R ARR (min) standard m/z Fragments (%) (precursor) 2.78 DON 296 297.10 231, 249 12 0.9899 89 6.87 AFG2 330 331.00 245, 313 32 0.9994 82 7.12 AFG1 328 329.00 243, 311 28 0.9997 85 7.30 AFB2 314 315.00 259, 287 31 0.9995 91 7.48 AFB1 312 313.00 241, 285 24 0.9990 90 7.58 FB1 721 722.40 352, 334 42 0.9972 98 8.03 β-ZEA 322 323.00 277, 305 11 0.9980 90 8.20 FB2 705 706.10 336, 318 38 0.9995 99 8.25 FB3 705 706.30 336, 354 35 0.9995 94 8.28 OTB 369 370.10 205, 324 14 0.9985 94 8.38 α-ZEA 322 323.10 277, 305 9 0.9995 87 8.53 T-2 466 467.20 245, 305 11 0.9998 96 8.74 ZEA 318 319.10 185, 187 21 0.9999 92 8.78 OTA 403 404.00 239, 221 38 0.9997 86 ARR – overall recovery rate; CE – collision energy; Rt – retention time; MW – molecular weight; m/z – mass-to-charge ratio; 2 MS/MS – tandem mass spectrometry; R – coefficient of determination; AFB1 – aflatoxin B1; AFB2 – aflatoxin B2; AFG1 – aflatoxin G1; AFG2 – aflatoxin G2; DON – deoxynivalenol; FB1 – fumonisin B1; FB2 – fumonisin B2; FB3 – fumonisin B3; OTA – ochratoxin A; OTB – ochratoxin B; T-2 – T-2 toxin; ZEA – zearalenone; α-ZEA – alpha-ZEA and β-ZEA – beta-ZEA.

5.1.7 Statistical analysis All analyses were done in triplicates and results presented represent the average of triplicate determinations. An analysis of variance (ANOVA) was performed using SPSS 22 (IBM, USA) (Adebo et al., 2017c) and mean values among treatments for each sample type were considered to differ significantly if p  0.05.

5.2 Results and discussion Since mycotoxins are naturally occurring toxic compounds, quite stable and frequently contaminate agricultural commodities, they potentiate hazardous risk to humans and animals. They equally affect international trade and have negative impact on the economy. The

122 consumption of mycotoxin contaminated foods can seriously compromise health and may cause mycotoxicoses (mycotoxin-related diseases), as some are immunosuppressive, cause intestinal dysfunction, cancer, cell damage, reduced food intake and in extreme cases, lead to death (IARC, 1993; Njobeh et al., 2010; Makun et al., 2012; Anater et al., 2016). With increasing levels of mycotoxins in food products derived from these commodities used as raw materials, a viable strategy of reducing mycotoxins without compromising quality is ideal to boost food safety.

5.2.1 Presence of mycotoxins Of the 14 mycotoxins investigated in this study (Table 5.1), seven including the fumonisins

(FB1, FB2 and FB3), T-2 and the zearalenones (ZEA, α-ZEA and β-ZEA) were detected in raw WG-sorghum grain and subsequent WG-ting samples (Table 5.2). It was observed that the raw

LT-sorghum samples generally had higher levels of mycotoxins, notably FB1, FB3 and β-ZEA than the HT-sorghum samples (Table 5.2).

While LT-sorghum samples had FB1, FB2 and FB3 levels of 163, 12 and 400 μg/kg, respectively, the HT-sorghum samples contained neither FB1 nor FB2 but the same sample had

FB3 at a level of 148 μg/kg. Higher mycotoxin levels in LT-sorghum samples when compared to their HT counterpart was equally recorded for T-2, ZEA and α-ZEA (Table 5.2). This observation could be attributed to relatively higher concentration and types of bioactive compounds (such as polyphenols, flavonoids and tannins) as established in an earlier study (Chapters Three and Four). This might have contributed to the inhibition of mycotoxigenic fungi, microbial action and attendant mycotoxin production (Mahoney and Moleneux, 2004; Atanasova-Penichon et al., 2016; Telles et al., 2017; Adebo et al., 2018).

The levels of mycotoxins in the WG-sorghum samples recorded in this study are below the regulatory recommended mycotoxin limits in Southern Africa and European Union (EU) (Viljoen, 2003; EC, 2006, 2013), indicating that the LT- and HT-sorghum samples are quite “safe” for intending consumers. The relatively low mycotoxin levels observed in this study as compared to other available studies in the literature on sorghum (Ayalew et al., 2006; Makun et al., 2009; Oueslati et al., 2012; Warth et al., 2012; Ezekiel et al., 2015; Taye et al., 2016; Taye et al., 2018) could also reflect on effective agricultural practices that might have limited initial fungal contamination in the sorghum grains.

123

Table 5.2: Quantification of mycotoxins (µg/kg) in sorghum and reduction after fermentation to whole grain-ting

FB1 FB2 FB3 T-2 ZEA α-ZEA β-ZEA LT-sorghum Control 162.67d±3.90 12.00b±0.99 400.00d±2.98 7.39b±1.20 6.67c±0.50 28.00d±0.45 37.33d±0.53 3424 34.68c±5.58 2.67a±0.45 170.67c±2.35 2.32a±0.74 5.67b±0.28 11.00c±0.99 24.67c±0.49 3424+3165 9.33b±1.40 2.31a±0.98 155.33b±1.45 1.68a±0.78 4.00a±0.82 7.00b±0.97 13.33b±0.75 3424+3321 4.00a±2.63 1.33a±1.06 133.33a±2.19 1.17a±0.63 4.00a±0.05 5.00a±0.92 11.83a±0.63 3424 (3165+3321) 6.67ab±2.10 2.00a±0.92 156.67b±2.09 1.28a±0.75 5.00b±0.06 9.67c±0.81 11.94a±0.07 HT-sorghum Control – – 148.00d±1.93 6.67b±1.00 6.04b±0.13 20.89c±0.82 25.33c±0.44 2872 – – 105.33c±1.80 4.06a±0.06 5.33ab±0.43 10.33b±0.44 19.31b±0.44 2872+3165 – – 91.07b±1.74 3.94a±0.85 4.82a±0.1 8.69a±0.21 12.00a±0.87 2872+3321 – – 84.06a±2.29 3.17a±0.17 4.67a±0.3 8.67a±0.51 10.67a±0.46 2872 (3165+3321) – – 93.38b±1.82 3.85a±0.33 5.00a±0.71 9.00a±0.23 18.33b±1.21

HT – high tannin; LT – low tannin; 2872 – naturally fermented ting from HT-sorghum type; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum type. Each value is a mean ± standard deviation of triplicates. Means with no common letters within a column under each sample type significantly (p  0.05) differ. FB1 – fumonisin B1, FB2 – fumonisin B2; FB3 – fumonisin B3, T-2 – T-2 toxin; ZEA – zearalenone; α-ZEA – α-zearalenone and β-ZEA – β-zearalenone.

124

5.2.2 Mycotoxin reduction The presence of all the mycotoxins was significantly (p  0.05) reduced after fermentation of WG- sorghum to WG-ting (Table 5.2). Residual mycotoxin levels in LT-ting samples were FB1 (4 –

34.68 μg/kg), FB2 (1.33 – 2.67 μg/kg), FB3 (133.33 – 170.67 μg/kg), T-2 (1.17 – 2.32 μg/kg), ZEA (4.00 – 5.67 μg/kg), α-ZEA (5 – 11 μg/kg) and β-ZEA (11.83 – 24.67 μg/kg). For HT-ting samples, residual mycotoxin levels ranged between 84.06 – 105.33, 3.17 – 4.06, 4.67 – 5.33, 8.67 – 10.33 and 10.67 – 19.31 μg/kg for FB3, T-2, ZEA, α-ZEA and β-ZEA, respectively (Table 5.2). Although the sorghum varieties were fermented at different conditions, i.e., 24 h at 34 oC (LT-sorghum type) and 72 h at 28 oC (HT-sorghum type), fermentation with L. fermentum strains both singly and in combination was observed to be more effective in reducing mycotoxins in subsequent WG-ting samples, as compared to spontaneously fermented samples (Figures 5.1A and 1B).

100

3424 60 3424(3165) 3424(3321)

40 3424(3165+3321) Percentage Percentage reduction (%)

0 FB1 FB2 FB3 T-2 toxin ZEA α-ZEA β-ZEA

125

B 70

40 2872 2872 (3165)

30 2872(3321) 2872(3165+3321)

Percentage Percentage reduction (%) 20

0 FB3 T-2 toxin ZEA α-ZEA β-ZEA

Figure 5.1: Reduction of mycotoxin levels in ting from whole grain sorghum. (A) Ting samples obtained from the LT sorghum type; (B) Ting samples obtained from the HT sorghum type. 2872 – naturally fermented ting from HT-sorghum; 3424 – naturally fermented ting from LT-sorghum type; (3165) – fermentation with L. fermentum FUA 3165; (3321) – fermentation with L. fermentum FUA 3321; (3165+3321) – fermentation with L. fermentum FUA 3165 and L. fermentum FUA 3321

Percentage reduction of the mycotoxins after fermentation with L. fermentum strains were more pronounced and above 60% in the LT-ting samples (Figure 5.1A) and above 25% for the HT-ting samples. These reductions were relatively lower in naturally (spontaneously) fermented WG-ting samples obtained from both LT- and HT-sorghum samples (Figure 5.1A and B). Although fermentation time for HT-ting samples were longer, percentage mycotoxin reduction in LT-ting samples were higher, implying that initial substrate composition could have influenced the extent of mycotoxin reduction. This is equally reflected in higher microbial population observed in the LT-ting samples, as compared to their HT-counterpart (Chapter Four).

Initial microbiota largely influences the products of microbial metabolism during fermentation. Relatively lower pH, high contents of alcohols, lactic acid and more production of other metabolites relevant to mycotoxin reduction, during the fermentation with the LABs might have equally instigated better mycotoxin reduction during LAB-fermentation. Consequent reduction could be attributed to a possible breakdown and/or degradation of mycotoxins to less toxic

126 products by the fermenting microbiota. Fermentation has been identified as an effective process to reduce mycotoxins due to breakdown by endogenous enzymes and compounds secreted and released into the food matrix by these fermenting organisms (Motarjemi and Nout, 1996; Milani and Maleki, 2014; Adebo et al., 2017b; Neme and Mohammed, 2017; Okeke et al., 2018). The production of bacteriocins, antagonistic and proteinaceous compounds might have also contributed to this observation.

Significantly (p  0.05) lower residual mycotoxin levels and corresponding higher percentage mycotoxin reduction in WG-ting samples obtained using L. fermentum FUA 3321 (singly) demonstrate better mycotoxin reduction caused by this strain. This can be attributed to an enhanced interaction between mycotoxins and cell wall of the L. fermentum FUA 3321 strain (Zhao et al., 2016). Relatively lower mycotoxin reduction in HT-ting samples (Figure 5.1B) as compared to LT-ting samples (Figure 5.1A) and the generally higher reduction with LAB strains (Figures 5.1A and B) as compared to natural fermentation, could be attributed to accelerated fermentation and increased microbial action. It has equally been established that LABs can better detoxify mycotoxins to less toxic forms during cereal fermentation (Mokoena et al., 2005; Dalié et al., 2010; Nyamete et al., 2016; Adebo et al., 2017b). It could be speculated that these toxins have been completely detoxified, hydrolysed and degraded to less toxic forms.

Although effective mycotoxin reduction in this study was highest for L. fermentum FUA 3321 alone, followed by L. fermentum FUA 3165 alone, combination of L. fermentum FUA 3321 and 3165 and natural fermentation (Figures 5.1A and B), this trend was particularly different for T-2 (Table 5.2, Figures 5.1A and B). While L. fermentum FUA 3321 was still the most effective strain, a combination of L. fermentum FUA 3321 and 3165 proved more effective than L. fermentum FUA 3165 when used in isolation. It can thus be suggested that the combined effect of the mixed LABs on T-2 reduction was more effective in binary combination (L. fermentum FUA 3165 and L. fermentum FUA 3321) as compared to pure culture of L. fermentum FUA 3165. It could also be speculated that single strain of L. fermentum FUA 3321 and the binary combination of the strains had relatively higher affinity for the 12, 13-epoxy ring, responsible for the toxicity of T-2 (Sudakin, 2003; Adhikari et al., 2017).

127

5.3 Conclusion Fermentation significantly reduced the levels of mycotoxins in ting samples obtained from whole grain sorghum samples, especially with the use of lactic acid bacteria strains. Although both L. fermentum strains exhibited good mycotoxin reduction ability, the use of L. fermentum FUA 3321 enhanced a better mycotoxin reduction. It can thus be deduced that L. fermentum strains are promising and suitable starter cultures for dietary detoxification of mycotoxins in fermented food commodities. Further research is still needed to investigate the activity of these strains on other fermented foods and to clarify whether the mycotoxins apparently lost during WG-ting preparation are indeed destroyed, hydrolysed, or bound to the food matrix to become non-recoverable.

References Adebiyi, J. A., Obadina, A. O., Adebo, O. A., & Kayitesi, E. (2018). Fermented and malted millet products in Africa: Expedition from traditional/ethnic foods to industrial value added products. Critical Reviews in Food Science and Nutrition, 58, 463-474. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., Phoku, J. Z., & Kayitesi, E. (2017a). Fermented pulse-based food products in developing nations as functional foods and ingredients. In: Functional food – improve health through adequate food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 77-109 Adebo, O. A., Njobeh, P. B., Gbashi, S., Nwinyi, O. C., & Mavumengwana, V. (2017b). Review on microbial degradation of aflatoxins. Critical Reviews in Food Science and Nutrition, 57, 3208-3217. Adebo, O. A., Njobeh, P. B., Sidu, S., Adebiyi, J. A., & Mavumengwana, V. (2017c). Aflatoxin

B1 degradation by culture and lysate of a Pontibacter specie. Food Control, 80, 99-103. Adebo, O. A., Njobeh, P. B., Mulaba-Bafubiandi, A. F., Adebiyi, J. A., Desobgo, Z. S. C., & Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, e13381, 1-10. Adhikari, M., Negi, B., Kaushik, N., Adhikari, A., Al-Khedhairy, A. A., Kaushik, N. K., & Choi, E. H. (2017). T-2 mycotoxin: toxicological effects and decontamination strategies. Oncotarget, 8, 33933-33952. Anater, A., Manyes, L., Meca, G., Ferrer, E., Luciano, B., Pimpão, C. T., & Font, G. (2016). Mycotoxins and their consequences in aquaculture: A review. Aquaculture, 451, 1-10.

128

Atanasova-Penichon, V., Barreau, C., & Richard-Forget, F. (2016). Antioxidant secondary metabolites in cereals: potential involvement in resistance to Fusarium and mycotoxin accumulation. Frontiers in Microbiology, 7, 1-16. Atherstone, C., Grace, D., Waliyar, F., Lindahl, J., & Osiru, M. (2014). Aflatoxin literature synthesis and risk mapping: Special emphasis on Sub-Saharan Africa. International Livestock Research Institute, Kenya. pp. 1-104. Ayalew, A., Fehrmann, H., Lepschy, J., Beck, R., & Abate, D. (2006). Natural occurrence of mycotoxins in staple cereals from Ethiopia. Mycopathologia, 162, 57-63. Bankole, S., Schollenberger, M., & Drochner, W. (2006). Mycotoxins in food systems in sub Saharan Africa: A review. Mycotoxin Research, 22, 163-169. Chala, A., Taye, W., Ayalew, A., Krska, R., Sulyok, M., & Logrieco, A. (2014). Multimycotoxin analysis of sorghum (Sorghum bicolor L. Moench) and finger millet (Eleusine coracana L. Garten) from Ethiopia. Food Control, 45, 29-35. Dalié, D. K. D., Deschamps, A. M., & Richard-Forget, F. (2010). Lactic acid bacteria – Potential for control of mould growth and mycotoxins: A review. Food Control, 21, 370- 380. Darwish, W. S., Ikenaka, Y., Nakayama, S. M. M., & Ishizuka, M. (2014). An overview on mycotoxin contamination of foods in Africa. Journal of Veterinary Medicine and Science, 76, 789-797. Dawlal, P., Brabet, C., Thantsha, M. S., & Buys, E. M. (2017). Potential of lactic acid bacteria for the reduction of fumonisin exposure in African fermented maize based foods. World Mycotoxin Journal, 10, 309-318. EC (European Commission) (2006). Commission regulation No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs. http://eur- lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONSLEG:2006R1881:20100701:EN:PDF (Accessed 23/12/2017). EC (European Commission) (2013). Commission recommendation of 27 March 2013 on the presence of T-2 and HT-2 toxin in cereals and cereal products. http://eur- lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2013:091:0012:0015:EN:PDF (Accessed 23/12/2017).

129

Egal, S., Hounsa, A., Gong, Y. Y., Turner, P. C., Wild, C. P., Hall, A. J., Hell, K., & Cardwell, K. F. (2005). Dietary exposure to aflatoxin from maize and groundnut in young children from Benin and Togo, West Africa. International Journal of Food Microbiology, 104, 215-224. Ezekiel, C. N., Abia, W. A., Ogara, I. M., Sulyok, M., Warth, B., & Krska, R. (2015). Fate of mycotoxins in two popular traditional cereal-based beverages (kunu-zaki and pito) from rural Nigeria. LWT-Food Science and Technology, 60, 137-141. FAO (Food and Agricultural Organization) (2010). Current status and options for biotechnologies in food processing and in food safety in developing countries. FAO International Technical Conference, Guadalajara, Mexico, 1-4 March 2010. FAO (Food and Agricultural Organization) (2011). Global food losses and food waste: Extent causes and prevention. http://www.fao.org/docrep/014/mb060e/mb060e00.pdf (Accessed 05/01/2018). Hidy, P. H., Baldwin, R. S., Greasham, R. L., Keith, C. L., & McMullen, J. R. (1977) Zearalenone and some derivatives: production and biological activity. Advances in Applied Microbiology, 22, 59-82. IARC (International Agency for Research on Cancer) (1993). Toxins derived from Fusarium

moniliforme: Fumonisin B1 and B2 and fusarin C. Lyon; France. Kazanas, N., Ely, R. W., Fields, M. L., & Erdman, J. W. (1984). Toxic effects of fermented and unfermented sorghum meal diets naturally contaminated with mycotoxins. Applied and Environmental Microbiology, 47, 118-1125. Karlovsky, P., Suman, M., Berthiller, F., De Meester, J., Eisenbrand, G., Perrin, I., Oswald, I. P., Speijers, G., Chiodini, A., Recker, T., & Dussort, P. (2016). Impact of food processing and detoxification treatments on mycotoxin contamination. Mycotoxin Research, 32, 179-205. Keller, L., Abrunhosa, L., Keller, K., Rosa, C. A., Cavaglieri, L., & Venâncio, A. (2015). Zearalenone and its derivatives α-Zearalenol and β-Zearalenol decontamination by Saccharomyces cerevisiae strains isolated from bovine forage. Toxins, 7, 3297-3308. Lewis, L., Onsongo, M., Njapau, H., Schurz-Rogers, H., Luber, G., Kieszak, S., Nyamongo, J., Backer, L., Dahiye, A. M., Misore, A., DeCock, K., & Rubin, C. (2005). Aflatoxin contamination of commercial maize products during an outbreak of acute aflatoxicosis in eastern and central Kenya. Environmental Health Perspectives, 113, 1763-1767.

130

Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Mahoney, N., & Molyneux, R. J. (2004). Phytochemical inhibition of aflatoxigenicity in Aspergillus flavus by constituents of walnut (Juglans regia). Journal of Agricultural and Food Chemistry, 52, 1882-1889. Makun, H. A., Gbodi, T. A., Akanya, H. O., Salako, E. A., & Ogbadu, G. H. (2009). Fungi and some mycotoxins found in mouldy sorghum in Niger State, Nigeria. World Journal of Agricultural Sciences, 5, 5-17. Makun, H. A., Dutton, M. F., Njobeh, P. B., Gbodi, T. M., & Ogbadu, G. H. (2012). Aflatoxin contamination in foods and feeds: A special focus on Africa. In: Trends in Vital Food and Control Engineering, Eissa, A. A. Eissa. (Ed.). InTech, Croatia. pp. 187-234 Milani, J., & Maleki, G. (2014). Effects of processing on mycotoxin stability in cereals. Journal of the Science of Food and Agriculture, 94, 2372-2375.

Mokoena, M. P., Chelule, P. K., & Ggaleni, N. (2005). Reduction of fumonisin B1 and zearalenone by lactic acid bacteria in fermented maize meal. Journal of Food Protection, 68, 2095-2099. Motarjemi, Y., & Nout, M. J. R. (1996). Food fermentation: a safety and nutritional assessment. Bulletin of the World Health Organization, 74, 553-559. Motloung, L., De Saeger, S., De Boevre, M., Detavernier, C., Audenaert, K., Adebo, O. A., & Njobeh, P. B. (2018). Study on mycotoxin contamination in South African food spices. World Mycotoxin Journal. Accepted, In Press. Neme, K., & Mohammed, A. (2017). Mycotoxin occurrence in grains and the role of postharvest management as a mitigation strategies. A review. Food Control, 78, 412-425. Njobeh, P. B., Dutton, M. F., & Makun, H. A. (2010). Mycotoxins and human health: significance, prevention and control. In: Smart Biomolecules in Medicine, Ajay, K. M., Ashutosh, T., & Shivanti B. M. (Eds.). VBRI Press, India. pp. 132-177. Njobeh, P. B., Dutton, M. F., Aberg, A. T., & Haggblom, P. (2012). Estimation of multi-mycotoxin contamination in South African compound feeds. Toxins, 4, 836-848.

131

Nyamete, F. A., Bennink, M., & Mugula, J. K. (2016). Potential of lactic acid fermentation in

reducing aflatoxin B1 in Tanzania maize-based gruel. African Journal of Food Agriculture, Nutrition and Development, 16, 11139-11151. Odunmbaku, L. A., Sobowale, S. S., Adenekan, M. K., Oloyede, T., Adebiyi, J. A., Adebo, O. A. (2018). Influence of steeping duration, drying temperature and duration on the chemical composition of sorghum-starch. Food Science and Nutrition, 6, 348-355. Okeke, C. A., Ezekiel, C. N., Nwangburuka, C. C., Sulyok, M., Ezeamagu, C. O., Adeleke, R. A., Dike, S. K., & Krska, R. (2015). Bacterial diversity and mycotoxin reduction during maize fermentation (steeping) for ogi production. Frontiers in Microbiology, 6, 1-12. Okeke, C. A., Ezekiel, C. N., Sulyok, M., Ogunremi, O. R., Ezeamagu, C. O., Sarkanj, B., Warth, B., & Krska, R. (2018). Traditional processing impacts mycotoxin levels and nutritional value of ogi – A maize-based complementary food. Food Control, 86, 224-233. Oueslati, S., Romero-González, R., Lasram, S., Frenich, A. G., & Martínez Vidal, J. L. (2012). Multi-mycotoxin determination in cereals and derived products marketed in Tunisia using ultra- high performance liquid chromatography coupled to triple quadrupole mass spectrometry. Food and Chemical Toxicology, 50, 2376-2381. Rosentrater, K. A., & Evers, A. D. (2018). Other food products. In: Kent’s Technology of Cereals (5th ed). Amsterdam: Elsevier. pp. 699-727. Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Sudakin, D. L. (2003). Trichothecenes in the environment: relevance to human health. Toxicology Letters, 143, 97-107.

132

Udomkun, P., Wiredu, A. N., Nagle, M., Bandyopadhyay, R., Mu¨ller, J., & Vanlauwe, B. (2017). Mycotoxins in sub-Saharan Africa: present situation, socio-economic impact, awareness, and outlook. Food Control, 72, 110-122. Viljoen, J. H. (2003). Mycotoxins in grain and grain products in South Africa and proposals for their regulation. A PhD Thesis submitted to the Faculty of Natural and Agricultural Sciences, University of Pretoria, South Africa. Warth, B., Parich, A., Atehnkeng, J., Bandyopadhyay, R., Schuhmacher, R., Sulyok, M., & Krska, R. (2012). Quantitation of mycotoxins in food and feed from Burkina Faso and Mozambique using a modern LC-MS/MS multitoxin method. Journal of Agricultural and Food Chemistry, 60, 9352-936. Zhao, H., Wang, X., Zhang, J., Zhang, J., & Zhang, B. (2016). The mechanism of Lactobacillus

strains for their ability to remove fumonisins B1 and B2. Food and Chemical Toxicology, 97, 40-46.

133

CHAPTER SIX4 Differential metabolic signatures in naturally and lactic acid bacteria (LAB) fermented ting, as revealed by gas chromatography mass spectrometry (GC-MS) based metabolomics

Abstract Fermented whole grain (WG) sorghum food products including WG-ting can be obtained from different sample sources and fermentation conditions, leading to variations in subsequent metabolites. There is however, a lack of understanding regarding the underlying metabolites defining differences in derived products. The current study is a nontargeted gas chromatography- mass spectrometry (GC-MS)-based metabolomics approach to elucidate and profile differences in volatile metabolite composition of two WG-sorghum types [high tannin (HT) and low tannin (LT)] and their derived WG-ting samples obtained via fermentation. Metabolites were extracted using 80% aqueous methanol and profiled on a gas chromatography high resolution time of flight mass spectrometry (GC-HRTOF-MS) system. Different chemometric techniques such as principal component analysis (PCA) and orthogonal partial least square-discriminant analysis (OPLS-DA) were applied to mine the generated data. The results showed that the metabolite signatures that differentiated raw HT- and LT-sorghum included groups of acids, cyclic compounds, pesticides, 2,4-di-tert-butylphenol, an acid, ester, a ketone, fatty acid esters, and sugar derivatives. Furthermore, following the fermentation of the HT- and LT-sorghum into WG-ting, an increase in the levels of 9-hexadecenoic acid, 9,12-octadecadienoic acid (Z,Z), trans-13-octadecenoic acid, 2(5H)-furanone, ethyl 2-isocyanatopropionate, 2,4-di-tert-butylphenol and fatty acid esters were noted with a reduction of some phenols, cyclic compounds, a pesticide and ketone. As such, this GC-MS-based metabolomics provides some key metabolic markers that could differentiate the two sorghum types. Furthermore, the study gives some insights into the dynamic metabolite profiles generated during the sorghum fermentation process. These include increases in certain compounds especially acids, fatty acids and their esters, which are vital from a dietary and health context. The notable decrease in 4-chlorobenzonitrile shows the degradative ability of L. fermentum FUA 3165 on this pesticide. Results obtained suggests that inherent composition of raw sorghum and fermentation conditions affected volatile metabolites during WG-ting production.

4To be submitted for publication.

134

This study also highlights the applicability of GC-MS metabolomics in understanding WG-ting fermentation. Keywords: Metabolomics, ting, fermentation, Lactobacillus fermentum, and gas chromatography mass spectrometry (GC-MS).

6.0 Introduction Fermented foods are embedded into culture and societies around the globe. They are well-known for their nutritional and health benefits, with studies suggesting their significant role in mitigating certain chronic diseases (Adebo et al., 2017a; 2017b; Ayash et al., 2018). A significant and important form of such fermented food is ting, a commonly consumed sorghum fermented food in Southern Africa. It is served as an important indigenous and cultural meal for both infants and adults in this part of the continent (Madoroba et al., 2009; Sekwati-Monang and Gänzle, 2011; Adebo et al., 2018). Although the numerous benefits of fermented foods have been described in the literature (Jay et al., 2005; Adebo et al., 2017a; Adebiyi et al., 2018), the use of WGs as substrates for fermented food products are recently being encouraged due to their role in the prevention of certain diseases (Giusti et al., 2017; Schaffer-Lequart et al., 2017).

Most analytical techniques so far applied for the study of ting have focused on measuring specific group of analytes, which usually do not provide a complete overview of the constituents present. An untargeted approach for studying the composition of ting is desirable to detect, identify and explain variations among sample forms, types and/or fermentation conditions. Metabolomics, particularly metabolic profiling (untargeted metabolomics) is considered useful and important in this regard in that, it investigates all the metabolites in sample in a single run, at a specific point in time (Wishart et al., 2008; Mozzi et al., 2013). Irrespective of the analytical platform used, targeted metabolomics is focused on a specific group of metabolites and as such, requires subsequent quantification, unlike the case of untargeted metabolomics that is usually broad and focused on detecting a wide range of metabolites (Wishart, 2008; Tugizimana et al., 2013; Mozzi et al., 2013). This thus, provides the possibility of picking up both expected and unexpected compounds that would provide more insight into a food metabolome (Cevallos-Cevallos et al., 2009; Adebo et al., 2017b).

135

Although different untargeted metabolomic techniques are available including capillary electrophoresis-mass spectrometry, liquid mass spectrometry and nuclear magnetic resonance (Mozzi et al., 2013; Tugizimana et al., 2013; Adebo et al., 2017b), gas chromatography mass spectrometry (GC-MS) is the technique of choice for volatile, semi-volatile and compounds of low polarity (Hernández et al., 2011). Several GC-MS metabolite profiling studies have thus, been reported, suggesting its frequent use, which is attributed to effective separation, constant electron ionisation energy (EI) to obtain reproducible fragmentation pattern, as well as satisfactory sensitivity and fewer matrix effects (Park et al., 2010; Ibánez et al., 2014; Fernando et al., 2014; Lee et al., 2016). Recent advances in instrumentation design and improved sensitivity have led to the birth of high-resolution time of flight mass spectrometry (HRTOF-MS) instruments, with high mass resolutions and data acquisition rates (Ibánez et al., 2014; Brits et al., 2018). As demonstrated in other studies, GC-HRTOF-MS is an effective analytical platform that can satisfactorily screen different compounds at a high sensitivity in a single run, with better deconvolution at a fast scan rate (Hernández et al., 2011; Ibánez et al., 2014; Fernando et al., 2014; Ubukata et al., 2015; Brits et al., 2018), giving useful data that can possibly answer desired biological questions.

The data generated from metabolomics are high-dimensional, complex and difficult to explore. To mine such an information-rich data, chemometric methodologies are applied. Such methods include multivariate unsupervised approaches such as PCA and supervised tools such as OPLS- DA (Tugizimana et al., 2013). Both approaches have been utilized in numerous food metabolomic studies (Wishart et al., 2008; Cevallos-Cevallos et al., 2009; Mozzi et al., 2013; Lee et al., 2014a; Adebo et al., 2017b), providing an overview of the dataset, revealing patterns with respect to treatments and identifying metabolites that significantly contribute to such variations. Thus, as earlier mentioned, this study is a GC-MS-based metabolomics approach to unravel differentiating markers between two sorghum types and to understand changes in the metabolites as a function of fermentation conditions and sample type.

6.1 Materials and Methods 6.1.1 Raw material and sample preparation Sorghum (Sorghum bicolor L.) grain types [high tannin (HT) and low tannin (LT)] were purchased from Agricol (Pty) Ltd. Potchefstroom, South Africa. The grains were milled (Perten Laboratory Mill 3600, Perten Instruments, Sweden) and passed through a 2000 µm aperture size sieve

136

(Analysette 3 Spartan, Fritsch, Germany) to obtain whole grain (WG) sorghum flour. Tannin content (TNC) in both raw LT and HT sorghum types were investigated, with values recorded as 17.97 and 31.68 mg CE/g, respectively, and as such, were subsequently classified as LT and HT sorghum types, respectively (Chapter Three).

6.1.2. Lactobacillus strains Two L. fermentum strains (L. fermentum FUA 3165 and L. fermentum FUA 3321) were used singly and in combination for controlled fermentation of WG-ting. These strains were earlier isolated from ting (Sekwati-Monang and Gänzle, 2011) and donated by Prof. Michael Gänzle of the University of Alberta, Canada. The strains were grown in 10 mL MRS broth (HiMedia, India) for 24 h at 34 oC in an incubator (IncoShake, Labotec, South Africa). The liquid culture obtained was subsequently centrifuged (Eppendorf 5702R, Merck South Africa) at 3000 rpm and 10 oC for 5 min to obtain cells. Cells were washed thrice with sterile phosphate buffer saline (PBS) and reconstituted in 10 mL of PBS (Sekwati-Monang and Gänzle, 2011).

6.1.3 Fermentation of sorghum into WG-ting Ting was processed by mixing sorghum flour and sterile distilled water (40 oC) (1:1, w/v) (Chapter Three). The mixture was then inoculated (cell counts of approximately 105 cfu/mL) with the L. fermentum strains (both singly and in combination) and fermented for 72 h at 28 oC (HT sorghum) and 24 h at 34 oC (LT sorghum). The selection of these conditions was based on the optimized results earlier obtained in Chapter Three. Control WG-ting samples were also obtained by spontaneously fermenting sorghum types (HT and LT) under similar conditions without any strains. All samples thereafter obtained were freeze-dried (at −55 oC for 24 h) and ground prior to analysis.

6.1.4 Sample preparation for metabolite profiling Different extracting solvents [100% acetonitrile (ACN), 100% methanol (MeOH), 100% water

(H2O), 80% ACN in H2O, 80% MeOH in H2O, 50% ACN:MeOH (v/v), 1% HCl in MeOH,

ACN:MeOH:H2O (4,4,2, v/v/v) and isopropanol:ACN:H2O (4,4,2, v/v/v)] were initially investigated for a range of possible obtainable metabolites. An informed compromise was reached and extraction using 80% MeOH in H2O was eventually adopted based on a wider range of relevant metabolites detected on the GC-MS system. Accordingly, 1 g of each ground freeze-dried sample

137 was weighed into centrifuge tubes and 10 mL of selected extraction solvent added. This was thoroughly shaken and sonicated in an ultrasonic bath (Scientech 704, Labotech, South Africa) for 1 h at 4 oC. This was followed by centrifugation at 3500 rpm at 4 oC for 5 min (Eppendorf 5702R, Merck South Africa). The supernatant was subsequently transferred to a round bottom flask and concentrated using a rotavapor under vacuum at 40 oC. The dried extract was reconstituted with 1 mL of chromatographic grade MeOH and filtered into dark amber vials for analysis. The extraction was done in triplicates for each biological sample.

6.1.5 GC-HRTOF-MS analysis To ensure that accurate mass data were collected on the LECO Pegasus GC-HRTOF-MS system (LECO Corporation, St Joseph, MI, USA), mass calibration of the instrument was initially performed and passed before analysis. Perfluorotributylamine (PFTBA) was used as the mass calibration compound and eleven masses were used for pre-analysis calibration: CF3 (m/z

68.9952), C2F4 (m/z 99.9936), C2F4N (113.9967), C2F5 (m/z 130.9920) C3F6 (m/z 149.9904), C4F9

(m/z 218.9856), C5F10N (m/z 263.9871), C8F16N (m/z 413.9775), C9F18N (m/z 463.9743) and

C9F20N (m/z 501.9711). The observed intensity and resolution were 41392 and 40200, respectively, with a mass accuracy root mean square (RMS) of less than 1 ppm.

Samples were subsequently analyzed in a randomized order on the GC-HRTOF-MS system equipped with an Agilent 7890A gas chromatograph (Agilent Technologies, Inc., Wilmington, DE, USA) operating in high-resolution, equipped with a Gerstel MPS multipurpose autosampler (Gerstel Inc. Germany) and a Rxi®-5ms column (30 m × 0.25 mm ID × 0.25 µm) (Restek, Bellefonte, USA). One microliter of each sample was injected in a spitless mode using helium as a carrier gas pumped at a constant flow rate of 1 mL/min. The inlet and transfer line temperature were 250 and 225 oC, respectively, the initial oven temperature was set at 70 oC, held for 0.5 min, ramped at 10 °C/min to 150 °C, held for 2 min, ramped at 10 °C/min to 330 °C and held for 3 min for the column to bake out. The MS data acquisition rate was a recommended rate of 13 spectra/s, m/z range of 30–1000, electron ionization at 70 eV, ion source temperature at 250 °C and a system recommended extraction frequency of 1.25 kHz. Sample extracts of three biological replicates were analyzed twice, yielding a total of six analytical injections for each sample.

138

6.1.6 Data processing and statistical analysis The collected GC-HRTOF-MS dataset was converted to mzML format using the LECO ChromaTOF-HRT software and then processed (peak picking and alignment) on the XCMS open- source tool (https://xcmsonline.scripps.edu/). The resulting peak list showed 6270 variables with corrected peak retention times (min), mass-to-charge ratios (m/z) and integrated peak areas. The processed data were then imported to SIMCA 14.1 software (Umetrics, Umea, Sweden) for downstream multivariate statistical analyses. The X-dataset were Pareto scaled prior to principal component analysis (PCA) (data dimensionality reduction, and explorative analyses) and orthogonal partial least square discriminant analysis (OPLS-DA) (explicative analyses). The computed models were validated (e.g. 7-fold cross validation) and the quality of the models was evaluated using diagnostic tools such as cumulative R2X (variation explained by the models) and cumulative Q2 (variation predicted by the model, according to the cross-validation). Discriminant variables were selected using OPLS-DA loadings S-plots and validated using different methods such as variable influence projection (VIP) scores and descriptive statistics (Lee et al., 2014a; Seo et al., 2016). Statistically significant metabolites were identified based on their mass spectra and retention time using the NIST, Mainlib and Feihn metabolomics libraries.

6.2 Results and Discussion 6.2.1 Principal component analysis (PCA) of the GC-HRTOF-MS data set To understand the differential metabolic signatures in raw whole grain sorghum and subsequently derived ting, an GC-MS untargeted metabolomic profiling of the volatile components was investigated in this study. To first reduce dimensionality of the GC-HRTOF-MS dataset obtained, volatile metabolites in the differently obtained WG-ting and raw sorghum samples were analyzed using PCA. The PCA adopted provided an exploration of the dataset revealing sample grouping and clusters (Figures 6.1A and B). The PCA score plot showed differences among the samples, accounting for 62.4% of the variation with principal components 1 (PC1) and 2 (PC2) contributing 51.2 and 11.2% of the variation, respectively (Figure 6.1A). Such variations based on this profiling is described subsequently.

139

Figure 6.1: Exploratory data analysis with unsupervised chemometric method (PCA) (A) raw sorghum and ting samples, (B) HT-samples and LT-samples. HT0000 – raw high tannin-sorghum; HT2872 – naturally fermented ting from HT-sorghum; HT3165 – HT-sorghum fermented with L. fermentum FUA 3165; HT3321 – HT-sorghum fermented with L. fermentum FUA 3321; HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; LT0000 – raw low tannin-sorghum; LT3424 – naturally fermented ting from LT-sorghum; LT3165 – LT-sorghum fermented with L. fermentum FUA 3165; LT3321 – LT-sorghum fermented with L. fermentum FUA 3321; LT2STRAINS – LT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum. HTs – high tannin-sorghum samples; LTs – high tannin-sorghum samples.

PC1 separated WG-ting samples obtained from HT-sorghum to the left and corresponding WG- ting samples from LT-sorghum to the right. As further observed in Figure 6.1A, the raw HT- and LT-sorghum samples were distinctly separated from the subsequently fermented samples (WG- ting). The observed separation and clusters are suggestive of levels of metabolic changes during

140 fermentation as HT-samples were more clustered when compared to LT-clusters, which were more dispersed on the surface plot (Figure 6.1B).

6.2.2 Comparison of metabolites in raw HT- and LT-sorghum samples To answer an important biological question in this study related to the raw sorghum types, the dataset was further explored and investigated using OPLS-DA. As observed in Figure 6.2A, the score plot of the constructed OPLS-DA model showed differences between the raw sorghum types, with HT-sorghum clusters on the left and LT-ones on the right.

Figure 6.2: OPLS-DA modelling and variable selection: (A) OPLS-DA score plot separating raw HT- and raw LT- sorghum (R2X = 0.718 and Q2 = 0.977), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; LT0000 – raw low tannin-sorghum.

Further investigation into the metabolites contributing to this discrimination using the generated OPLS-DA loadings S-plot (Figure 6.2C) indicated that 16 compounds significantly (p ≤ 0.05) contributed and explained the difference in these samples (Table 6.1). The selected discriminating features were assessed based on the VIP-plots and receiver operating characteristic (ROC) curve (Figures 6.2B and D, respectively). These metabolites were an acid, ester, cyclic compounds, a sugar derivative, pesticides, a benzene, phenol, ketone and fatty acid esters (FAEs) (Table 6.1). Their respective MS fragments and retention times are presented in Table 6.1.

141

Table 6.1: Significant metabolites contributing to the differences in the raw HT- and LT- sorghum types Rt Observed Fragments MF Metabolite MC p value FC m/z

3.00 98.0364 53, 81 C5H6O2 Methylene cyclopropanecarboxylic acid Acid 0.005 12.52 5.22 112.9998 61, 218 C9H5Cl3F2O2 3,4-difluorobenzoic acid, 2,2,2-trichloroethyl ester Ester 0.002 3.91 6.29 144.042 95, 130 C7H4ClN 4-chlorobenzonitrile Pesticide 0.018 0.10 7.34 120.057 65, 91 C8H8O Dihydrobenzofuran Benzene 0.011 2.96 8.45 73.0467 147, 326 C12H36O6Si6 Dodecamethylcyclohexasiloxane Cyclic 0.030 2.10 12.06 191.143 41, 68 C14H22O 2,4-di-tert-butylphenol Phenol 0.008 2.68 -5 17.36 276.201 55, 74 C17H24O3 7,9-Di-tert-butyl-1-oxaspiro(4,5)deca-6-9-diene- Ketone 1.35×10 4.25 2,8-dione -4 17.41 206.117 77, 160 C15H21NO4 Metalaxyl Pesticide 1.27×10 3.18 18.17 88.0519 70, 101 C10H20O6 Glucopyranoside, methyl 2,3,4-tri-O-methyl Sugar derivative 0.010 3.09 -4 19.22 82.0778 67, 95 C19H34O2 Methyl 9-cis, 11-trans-octadecadienoate FAE 6.95×10 5.26 19.26 70.0755 55, 110 C19H36O2 Trans-13-octadecenoic acid, methyl ester FAME 0.014 20.10 -5 20.02 111.081 55, 69 C20H38O2 (E)-9-octadecenoic acid ethyl ester FAEE 6.95×10 4.12 -4 20.12 67.0938 81, 109 C19H34O2 11,14-octadecadienoic acid, methyl ester FAME 9.99×10 14.64 -7 21.23 74.0358 87, 143 C21H42O2 Eicosanoic acid, methyl ester FAME 1.86×10 3.85 -5 22.42 71.0730 58, 72 C12H25NO2 Octanoic acid, 2-dimethylaminoethylester Fatty acid derivative 4.25×10 10.68 23.02 74.0362 43, 143 C27H54O2 Hexacosanoic acid, methyl ester FAME 0.031 6.43 FAEE – fatty acid ethyl ester, FAME – fatty acid methyl ester, FC – fold change (value of the average peak area of LT/HT), HT – raw high tannin sorghum, LT – raw low tannin sorghum, m/z – mass-to-charge ratio, MC – metabolite class, MF – molecular formula, Rt – retention time (min)

142

While derivatives of cyclopropanecarboxylic acids are potential pharmacologically important compounds (Bonnaud et al., 1987; Bender et al., 2008; Dharni et al., 2014), both 2,4-di-tert- butylphenol and 7,9-Di-tert-butyl-1-oxaspiro(4,5)deca-6-9-diene-2,8-dione are known to possess radical scavenging activity, metal chelating, antioxidant, anti-androgen, estrogen and glucocorticoid effects (Yoon et al., 2006; Grover and Patni, 2013; Dharni et al., 2014; Song et al., 2018).

Presence of FAEs as observed in this study is in agreement with other studies reporting WG- sorghum as rich sources of fat-related compounds (Carr et al., 2005; Mehmood et al., 2008; Hassan et al., 2017). These lipid fractions have been reported to exhibit antibacterial activities, modulate immune function, mitigate against cardiovascular diseases, scavenge for free radicals, exhibit antioxidant properties and render inhibitory effects on low density lipoprotein oxidation (Tapiero et al., 2002; Villa et al., 2002; Hadbaoui et al., 2010; Alabdulkarim et al., 2012; Hassan et al., 2017; Sun et al., 2018). Particularly in WG-sorghum, these fat-related compounds have been reported to lower plasma and liver cholesterol levels through inhibition of cholesterol absorption (Carr et al., 2005; Lee et al., 2014b). These essential dietary lipid deriavtives are also known to perform regulatory activities in nutrient metabolism and cell functions by controlling gene function (Nagao and Yanagita, 2008; Hassan et al., 2017). The presence of these FAEs is of significant nutritional value in foods (Hassan et al., 2017), indicative of their potential use as food ingredients or dietary supplements in food.

Other significant compounds including metalaxyl, 4-chlorobenzonitrile, 2,4-di-tert-butylphenol and some others (not deemed significant by the OPLS-DA model) might have contributed to the reduced fungal proliferation and low attendant mycotoxin in raw WG-sorghum flour earlier found in this study (Chapter Five). Of significant interest is 4-chlorobenzonitrile, a known benzonitrile pesticide (Holtze et al., 2008) identified in this study (Table 6.1). Although pesticides are globally needed for pest control, their residues in food crops, however potentiate some adverse and long- term health effects (Ruediger et al., 2005; Kaushik et al., 2009). This could in fact be more worrisome when these pesticides occasionally find themselves subsequently in cereal-based foods at unacceptably high levels (González-Curbelo et al., 2012; Yang et al., 2014; Al-Zahraa et al., 2016).

143

The presence and the relatively high amounts of this compound in HT-sorghum (based on the fold change) could also explain the reduced microbial activity observed during the fermentation of this sample into HT-ting (Chapters Three, Four and Five). Such observation has been reported in the literature, attributed to inhibition of membrane and cell wall activities of the fermenting microbiota, growth suppression and toxic effects of pesticides on microorganisms resulting in longer lag phases (Kaushi et al., 2009; Pale and Taonda, 2010; Muturi et al., 2017). Subject to the higher fold changes for other compounds in LT-sorghum, it could be exemplified that these metabolites might have contributed to increased microbial actions and subsequent metabolic reactions as recorded in the earlier parts of this study.

6.2.3 Comparison of metabolites in raw HT-sorghum and subsequently obtained HT-ting samples In addition to the PCA, further multivariate comparison of metabolites in raw HT-sorghum and fermented samples (HT-ting) using OPLS-DA showed a separation in sample clusters, suggesting a difference in the sample metabolites as found on the OPLS-DA score plot (Supplementary Figures 6.1 - 6.4). A probe of the significant metabolites contributing to these differences based on their respective S-plots are summarized in Table 6.2. Although all the fermentation substrates were WG-sorghum from the HT-type, earlier studies (Chapter Four and Five) have shown that subsequent derived WG-ting samples differ in physicochemical and phenolic compound composition. As depicted in Table 6.2, significant metabolites contributing to the variation between raw HT-sorghum and HT-ting samples were cyclic compounds, phenols, a ketone, pesticide, fatty acids and fatty acid esters.

144

Table 6.2: Significant metabolites contributing to the differences in the raw HT- and subsequently obtained HT-ting samples Rt Observed Fragments MF Metabolite MC p value FC m/z HT0000 vs. HT3165

6.29 144.042 95, 130 C7H4ClN 4-chlorobenzonitrile Pesticide 0.027 0.17 8.45 73.0467 147, 326 C12H36O6Si6 Dodecamethylcyclohexasiloxane Cyclic 0.031 0.56 -4 8.47 150.0676 107, 135 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 1.54×10 0.11 11.44 147.0658 221, 355 C14H42O7Si7 Tetradecamethylcycloheptasiloxane Cyclic 0.075 0.62 14.25 355.0714 147, 221 C16H48O8Si8 Hexadecamethylcyclooctasiloxane Cyclic 0.007 0.56 14.31 73.0284 45 C10H20O2 1,3-dioxolane, 2-heptyl Cyclic 0.001 0.68 16.26 429.0908 207, 355 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 0.002 0.59 20.14 69.0700 55, 69, 81 C16H30O2 9-Hexadecenoic acid Fatty acid 0.033 4.44 20.19 95.0857 55, 67 C18H32O2 9,12-octadecadienoic acid (Z,Z) Fatty acid 0.024 2.42 22.21 161.096 149, 155 C21H38O2 Phenol, 2,2’-methylenebis[6-(1,1-dimethyl)-4- Phenol methyl 9.25×10-7 0.45 HT0000 vs. HT3321

8.45 73.0467 147, 326 C12H36O6Si6 Dodecamethylcyclohexasiloxane Cyclic 0.003 0.45 -4 8.47 150.0676 107, 135 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 3.0×10 0.15 11.44 147.0658 221, 355 C14H42O7Si7 Tetradecamethylcycloheptasiloxane Cyclic 0.008 0.51 -5 14.25 355.071 147, 221 C16H48O8Si8 Hexadecamethylcyclooctasiloxane Cyclic 4.93×10 0.44 -4 14.31 73.0284 45 C10H20O2 1,3-dioxolane, 2-heptyl Cyclic 5.68×10 0.72 -4 16.26 429.0908 207, 355 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 9.82×10 0.50 -4 20.19 95.0857 55, 67 C18H32O2 9,12-octadecadienoic acid (Z,Z) Fatty acid 1.90×10 2.31 20.26 97.1015 43, 55 C18H34O2 Trans-13-octadecenoic acid Fatty acid 0.014 4.8 -7 22.20 161.096 149, 155 C23H32O2 Phenol, 2,2’-methylenebis[6-(1,1-dimethyl)-4- Phenol 4.81×10 0.40 methyl HT0000 vs. HT2STRAINS

8.45 73.0467 147, 326 C12H36O6Si6 Dodecamethylcyclohexasiloxane Cyclic 0.003 0.43 8.47 150.0676 107, 135 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 0.001 0.21 11.44 147.0658 221, 355 C14H42O7Si7 Tetradecamethylcycloheptasiloxane Cyclic 0.056 0.54 14.25 355.0714 147, 221 C16H48O8Si8 Hexadecamethylcyclooctasiloxane Cyclic 0.002 0.46 -4 16.26 429.0908 207, 355 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 3.41×10 0.49 19.46 67.1135 79, 118 C18H14O4 3,3’-(p-phenylenedioxy)diphenol Phenol 0.0483 1439.16 20.26 97.1015 43, 55 C18H34O2 Trans-13-octadecenoic acid Fatty acid 0.0011 5.79 -5 22.20 161.096 149, 155 C23H32O2 Phenol, 2,2’-methylenebis[6-(1,1-dimethyl)-4- Phenol 1.4×10 0.39 methyl HT0000 vs. HT2872 -4 8.47 150.0676 107, 135 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 6.51×10 0.22 14.25 355.0714 147, 221 C16H48O8Si8 Hexadecamethylcyclooctasiloxane Cyclic 0.025 0.59

145

Rt Observed Fragments MF Metabolite MC p value FC m/z 16.26 429.0908 207, 355 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 0.007 0.60 20.14 69.0700 55, 69 C16H30O2 9-Hexadecenoic acid Fatty acid 0.004 4.27 20.27 97.1015 43, 55 C18H34O2 Trans-13-octadecenoic acid Fatty acid 0.011 5.58 21.41 95.0857 67, 96 C21H38O6 N-propyl linoleate FAE 0.075 2.74 21.43 84.0572 43, 83 C23H44O2 9-octadecenoic acid (Z)-, pentyl ester FAPE 2.772 9.14 FAE – fatty acid ester, FAEE – fatty acid ethyl ester, FAME – fatty acid methyl ester, FAPE – fatty acid pentyl ester, FC – fold change (value of the average peak area of derived ting sample/HT0000), HT0000 – raw high tannin sorghum, HT3165 – HT-sorghum fermented with L. fermentum FUA 3165, HT3321 – HT-sorghum fermented with L. fermentum FUA 3321, HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321, HT2872 – naturally fermented ting from HT-sorghum, m/z – mass- to-charge ratio, MC – metabolite class, MF – molecular formula, Rt – retention time (min).

146

It would seem interesting to note that different fermentation behaviors were observed for each starter culture as the OPLS-DA model indicated some different metabolites in HT-ting samples (Table 6.2). Although common metabolites among the LAB-obtained ting were all noted to follow the same decreasing trend, the rate of decrease however varied. As observed in Table 6.2, 4- chlorobenzonitrile and 9-hexadecenoic acid were identified as significant metabolites only in HT3165 but not in other LAB-obtained HT-ting samples. Worthy of note is the reduction of 4- chlorobenzonitrile with L. fermentum 3165, an indication of possible degradation of this pesticide by the strain. A similar observation of mycotoxin degradation was noted for this strain in an earlier study (Chapter 5) and such pesticide reduction by LABs reported in the literature (Ruediger et al., 2005; Cho et al., 2009; Trinder et al., 2016; Zhang et al., 2016). Different food processing techniques including fermentation have been suggested as suitable means that could aid in the reduction of pesticides in food, via the activities of starter cultures and absorption of the pesticide onto the bacterial cell walls (Abou-Arab, 2002; Ruediger et al., 2005; Kaushik et al., 2009). While such a reduction is desirable, the presence and levels of this pesticide in raw HT-sorghum (Table 6.1) affected the viability of the substrate for fermentation. The study of Pale and Taonda (2010) though on malting of sorghum, highlighted this as a major hinderance to the malting process. Accordingly, in addition to phenolic composition, levels of this pesticide could have equally altered the scanning electron microscopy (SEM) data earlier obtained (Chapter Three), with no considerable modification in morphology when spontaneous fermentation of HT-sorghum to process HT-ting was done (Figure 3.6).

The decreased observation recorded for other compounds may be due to the biochemical and physiological changes that occur during the fermentation process as seen in this study. On the other hand, increases in most of the fatty acids and their derivatives (Table 6.2) might be attributed to hydrolysis of fats embedded in the sorghum grain into other forms identified as being significant by the OPLS-DA model. A similar decrease in a phenol compound (phenol, 2,2’-methylenebis[6- (1,1-dimethyl)-4-methyl) can be attributed to degradation and hydrolysis of this compound to bioactive ones, which might have contributed to the observed higher levels of the quantified bioactive compounds recorded in Chapter 4. The high level of 3,3’-(p-phenylenedioxy)diphenol after fermentation by the two LAB strains (Table 6.2) could be an indication of their negative competitive action that resulted in lesser microbial degradation during WG-ting production (Chapter Four).

147

6.2.4 Comparison of metabolites in raw LT-sorghum and subsequently obtained LT-ting samples Similar to the HT-samples in Section 6.2.3, raw LT-sorghum sorghum sample was compared to its derived products (LT-ting) samples to decipher the metabolites contributing to the variation (difference) in both samples. An OPLS-DA model was equally applied for investigating these differences, which then revealed variations in the clustering of the samples (Supplementary Figures 6.5 - 6.8). Further investigation into these types of metabolites revealed that a total of 6, 12, 16 and 13 metabolites contributed to the differences in raw LT-sorghum samples and LT3165, LT3321, LT2STRAINS and LT3424 ting samples, respectively (Table 6.3).

The variation in the number of volatile metabolites equally demonstrates a similar trend of cluster separation (Figure 6.1) and fermentation rate in these samples. Spontaneously fermented LT-ting (LT3424) is characterized by higher fold change of ethanone, 1-(2-hydroxyl-5-methylphenyl) and specific presence of 2(5H)-furanone, 3,4-difluorobenzoic acid, 2,2,2-trichloroethyl and azepan-1- yl-acetic acid, acridin-9-ylmethylene-hydrazide. This peculiarity can also be speculated to have resulted in a distinct cluster of this sample as reflected on the PCA score plot (Figure 6.1). On the other hand, similar metabolism of the L. fermentum strains during fermentation might however, explain the similar compounds observed in LT3165 and LT3321 samples (Table 6.3) and closeness of their respective clusters on the PCA score plot (Figure 6.1).

Likewise, the HT-ting samples, a majority of the compounds, especially cyclic compounds, sugar derivative and some FAMEs were found to reduce after fermentation. Starches and sugars are known to be hydrolyzed by enzymes and used as carbon sources during food fermentation (Adebiyi et al., 2018). As such, the higher reduction in the content of glucopyranoside, methyl 2,3,4-tri-O-methyl during fermentation with the LABs, could be due to improved sugar hydrolysis. A similar occurrence in sugar-related compounds during the fermentation of kimchi (a Korean fermented food) was reported and attributed to the consumption of sugars by the rapid growth of inoculated LAB (Park et al., 2016). Accordingly, such differing abilities of sugar hydrolysis by these microorganisms could have an impact on taste, aroma (Baek et al., 2010) and other sensory properties of ting.

148

Table 6.3: Significant metabolites contributing to the differences in the raw LT- and subsequently obtained LT-ting samples

Rt Observe Fragments MF Metabolite MC p value FC d m/z LT0000 vs. LT3165 -5 8.48 150.0676 77, 107 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 2.56×10 0.18 12.07 191.143 41, 57 C14H22O 2,4-Di-tert-butylphenol Phenol 0.004 0.78 -4 18.17 88.0519 70, 101 C10H20O6 Glucopyranoside, methyl 2,3,4-tri-O-methyl Sugar derivative 3.95×10 0.28 19.22 82.0778 67, 95 C19H34O2 Methyl 9-cis, 11-trans-octadecadienoate Fatty acid derivative 0.049 0.01 20.12 67.0938 81, 109 C19H34O2 11,14-octadecadienoic acid, methyl ester FAME 0.023 0.33 22.43 71.0730 58, 7 C12H25NO2 Octanoic acid, 2-dimethylaminoethyl ester Fatty acid derivative 0.095 0.28 LT0000 vs. LT3321 -6 3.20 54.0102 42, 68 C5H4O2 Cyclopent-4-ene-1,3-dione Ketone 1.33×10 6.31 -6 8.48 150.0676 77, 107 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 4.54×10 0.20 -4 11.45 147.0657 73, 281 C14H42O7Si7 Tetradecamethylcycloheptasiloxane Cyclic 3.06×10 0.78 14.25 355.0714 73, 147 C16H48O8Si8 Hexadecamethylcyclooctasiloxane Cyclic 0.002 0.82 18.17 88.0519 70, 101 C10H20O6 Glucopyranoside, methyl 2,3,4-tri-O-methyl Sugar derivative 0.023 0.40 19.22 82.0778 67, 95 C19H34O2 Methyl 9-cis, 11-trans-octadecadienoate Fatty acid derivative 0.095 0.67 20.12 67.0938 81, 109 C19H34O2 11,14-octadecadienoic acid, methyl ester FAME 0.010 0.22 20.14 69.0699 55, 68 C16H30O2 9-Hexadecenoic acid Fatty acid 0.056 0.05 20.19 95.0857 55, 67 C18H32O2 9,12-octadecadienoic acid (Z,Z) Fatty acid 0.005 6.36 -4 22.42 71.073 58, 72 C12H25NO2 Octanoic acid, 2-dimethylaminoethylester Fatty acid derivative 2.05×10 0.18 23.03 74.0362 43, 87 C27H54O2 Hexacosanoic acid, methyl ester FAME 0.017 0.29 -4 24.33 74.0363 87, 143 C25H50O2 Tetracosanoic acid, methyl ester FAME 4.27×10 0.21 LT0000 vs. LT2STRAINS

8.46 73.0467 147, 324 C12H36O6Si6 Dodecamethylcyclohexasiloxane Cyclic 0.005 0.39 -5 8.48 150.0676 77, 107 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 1.36×10 0.13 11.45 147.0658 73, 281 C14H42O7Si7 Tetradecamethylcycloheptasiloxane Cyclic 0.005 0.45 14.25 355.0714 73, 281 C16H48O8Si8 Hexadecamethylcyclooctasioxane Cyclic 0.036 0.50 -4 14.31 45.0452 73 C10H20O2 1,3-dioxolane, 2-heptyl Cyclic 6.46×10 0.001 16.26 429.0908 147, 221 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 0.035 0.48 -4 17.36 143.1070 43, 75 C17H34O2 Hexadecanoic acid, methyl ester FAME 1.83×10 0.34 18.17 88.0519 70, 101 C10H20O6 Glucopyranoside, methyl 2,3,4-tri-O-methyl Sugar derivative 0.002 0.22 19.21 82.0778 67, 95 C19H34O2 Methyl 9-cis, 11-trans-octadecadienoate Fatty acid derivative 0.008 0.45 20.12 67.0938 81, 109 C19H34O2 11,14-octadecadienoic acid, methyl ester FAME 0.001 0.06 20.19 95.0857 55, 6 C18H32O2 9,12-octadecadienoic acid (Z,Z) Fatty acid 0.044 3.99 21.43 84.0571 43, 83 C23H44O2 9-octadecenoic acid (Z)-, pentyl ester FAPE 0.006 0.43 -4 22.20 161.096 149, 155 C23H32O2 Phenol, 2,2’-methylenebis[6-(1,1-dimethyl)-4- Phenol 6.58×10 0.16 methyl

149

Rt Observe Fragments MF Metabolite MC p value FC d m/z -6 22.42 71.073 58, 72 C12H25NO2 Octanoic acid, 2-dimethylaminoethylester Fatty acid derivative 3.01×10 0.07 23.03 74.0362 43, 87 C27H54O2 Hexacosanoic acid, methyl ester FAME 0.005 0.18 -5 25.00 72.0444 59 C11H16FNO3 Benzeneethanamine, 2-fluoro-á,3,4-trihydroxy-N- Benzene 1.64×10 13.12 isopropyl LT0000 vs. LT3424

3.35 84.0207 37, 54 C4H4O2 2(5H)-furanone Furan 0.004 4.54 3.40 70.0288 42, 71 C6H9NO3 Ethyl 2-isocyanatopropionate Ester 0.007 4.8 4.55 112.052 84, 93 C22H24N4O Azepan-1-yl-acetic acid, acridin-9-ylmethylene- Acid 0.00 2.23 hydrazide 5.22 112.9998 61, 218 C9H5ClF2O2 3,4-difluorobenzoic acid, 2,2,2-trichloroethyl ester Ester 0.002 1.98 -4 8.48 150.0676 77, 107 C9H10O2 Ethanone, 1-(2-hydroxyl-5-methylphenyl) Ketone 3.34×10 0.29 12.07 191.143 41, 68 C14H22O 2,4-di-tert-butylphenol Phenol 0.060 1.16 -4 14.32 45.0452 73 C10H20O2 1,3-dioxolane, 2-heptyl Cyclic 6.54×10 0.003 16.26 429.0908 73, 281 C16H48O6Si7 Hexadecamethylheptasiloxane Cyclic 0.001 0.21 19.26 83.0856 55, 87 C19H36O2 cis-13-octadecenoic acid, methyl ester FAME 0.078 0.44 20.12 67.0938 81, 109 C19H34O2 11,14-octadecadienoic acid, methyl ester FAME 0.018 0.26 21.42 122.1093 54, 81 C20H36O2 Linoleic acid ethyl ester FAME 0.002 14.17 -8 21.44 264.2464 41, 55 C21H40O2 Cis-9-octadecenoic acid, propyl ester FAPE 3.45×10 20.63 -4 23.02 74.0362 43, 87 C27H54O2 Hexacosanoic acid, methyl ester FAME 1.18×10 0.56 FAE – fatty acid ester, FAEE – fatty acid ethyl ester, FAME – fatty acid methyl ester, FAPE – fatty acid pentyl ester, FC – fold change (change (value of the average peak area of derived ting sample/that of LT0000), LT0000 – raw high tannin sorghum, LT3165 – HT-sorghum fermented with L. fermentum FUA 3165, LT3321 – LT-sorghum fermented with L. fermentum FUA 3321, LT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321, LT3424 – naturally fermented ting from LT-sorghum, m/z – mass-to-charge ratio, MC – metabolite class, MF – molecular formula, Rt – retention time (min).

150

Higher fold changes for the acids indicates higher concentration and release of these acids during WG-ting-fermentation. Generation of acids during fermentation of cereals by LABs are peculiar and desirable as they influence taste and flavor as well as affect shelf life (Leroy and De Vuyst, 2004; Jay et al., 2005; Sekwati-Monang and Gänzle, 2011; Adebo et al., 2018). Higher acid and esters levels have also been correlated to higher antioxidant activity in a study on fermented durum wheat (Ferri et al., 2016), which could suggest the higher antioxidant levels obtained for samples fermented with the L. fermentum strains (Chapter 4).

Although the fat content of sorghum like that of other cereals is relatively low, the nutritional value of the fat is related to the content of the fatty acids it contains. These sorghum fatty acids are in high quantities when compared to those from other cereals (Adeyeye and Ajewole, 1992; Carr et al., 2005; Afify et al., 2012), but some cannot be synthesized de novo in humans. Not only do these compounds contribute to nutrition, they equally contribute to flavor and other sensory properties in food (Hu et al., 2018). A food processing technique such as fermentation that improves the composition of these fatty acids is thus desirable. Studies in this regard have particularly demonstrated this, indicating that LAB fermentation may lead to conversion of inherent fatty acids into bioactive and beneficial derivatives (Bergamo et al., 2014; Di Cagno et al., 2017). Accordingly, this is the first study reporting the presence of some biologically important fatty acids and their esters in ting.

Increase in the levels of fatty acids and FAMEs could be attributed to dissociation of lipid complexes and activities of lipolytic enzymes, leading to increased extractability and liberation of these fat-related constituents. On the contrary, the reduction in hexadecanoic acid, methyl ester levels in all the WG-ting samples (both HT and LT) (Tables 6.2 and 6.3) could suggest a selective reductive lipase activity on this FAME during WG-ting fermentation. A similar phenomenon has equally been reported previously in other studies (Coutron-Gambotti and Gandemer, 1999; Liquori et al., 2015).

The fold change for 2(5H)-furanone in Table 6.3 reflects an increase of this volatile ketone after fermentation of LT-sorghum into LT-ting. Furanones are generally known flavor components reported in different fermented foods (Ohata et al., 2016; Singracha et al., 2017) and the occurrence of 2(5H)-furanone as a significant (p ≤ 0.05) metabolite differentiating raw LT-

151 sorghum from the spontaneously fermented LT-ting necessitates further investigation in this regard. Not only are the furanones related to flavor, they are also reported to exhibit antioxidant and anticarcinogenic effects in fermented soy sauce (Kataoka, 2005).

6.2.5 Similarities between the raw-sorghum samples and the subsequently obtained WG- ting samples A comparison of the OPLS-DA identified metabolites among each sample treatment and sample type (Tables 6.2 and 6.3) was made using a Venn diagram (Figures 6.3A and B). As observed the metabolites showed certain similarities, which further complemented the relationship of clusters established in the PCA score plot (Figure 6.1A). As reflected in Figures 6.5A and B, while ethanone, 1-(2-hydroxyl-5-methylphenyl was observed as the common single metabolite among the LT-samples, three metabolites (ethanone, 1-(2-hydroxyl-5-methylphenyl, hexadecamethylcyclooctasiloxane and hexadecamethylheptasiloxane) were all recorded in the HT-samples. The number of latter metabolites could further suggest the relative closeness of clusters of HT-samples as observed on PCA score plot (Figure 6.1B). As also earlier demonstrated in this study (Chapters Three, Four and Five), reduced microbial activity was found during WG- ting production when the HT-sorghum samples were used.

Figure 6.3: Venn diagram showing comparison between the metabolites of (A) raw and fermented LT-sorghum, (B) raw and fermented HT-sorghum. HT0000 – raw high tannin-sorghum; HT2872 – naturally fermented ting from HT- sorghum; HT3165 – HT-sorghum fermented with L. fermentum FUA 3165; HT3321 – HT-sorghum fermented with L. fermentum FUA 3321; HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; LT0000 – raw low tannin-sorghum; LT3424 – naturally fermented ting from LT-sorghum; LT3165 – LT- sorghum fermented with L. fermentum FUA 3165; LT3321 – LT-sorghum fermented with L. fermentum FUA 3321; LT2STRAINS – LT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321; 3424 – naturally fermented ting from LT-sorghum.

152

In all the investigated samples, ethanone, 1-(2-hydroxyl-5-methylphenyl), a ketone was noted as the only OPLS-DA identified metabolite common, thus demonstrating this compound as an important volatile compound present. Although the production of WG-ting using both substrates led to a reduction in the concentration of this compound, this ketone has been identified as a vital compound contributing to the flavor of fermented foods and other fermented foods (Kim and Lee, 2004; Wilson et al., 2015; Baek, 2017).

6.3 Conclusion Through the application of metabolomics and appropriate multivariate data analysis, differential volatile metabolites and changes due to fermentation and sample sources were unraveled in this study. Accordingly, the types of significant metabolites and varying levels of modification of these metabolites largely indicated that volatile metabolites in whole grain ting is dependent on the sorghum type as well as LAB strain used. These thus suggest that different whole grain ting can be produced based on the source of sorghum and LAB used as starter culture. Pharmacologically important compounds found in whole grain sorghum also position this crop as a natural potential source of these compounds. Not only was the reduction of a pesticide residue of significance, the presence of a wide range of fatty acids, their esters and other relevant compounds suggest that LAB-derived WG-ting may effectively convey a potentially safe, nutritious and health beneficial food to potential consumers of this product. Such exceptional qualities thus make whole grain sorghum a suitable substrate for the development of novel functional foods, dietary supplements or pharmaceutical preparations through lactic acid bacteria fermentation. Finally, the metabolomic study adopted in this study can thus assist in subsequent selection of an appropriate starter culture for whole grain ting production with uniform quality.

References Abou-Arab, A. A. K. (2002). Degradation of organochlorine pesticides by meat starter in liquid media and fermented sausage. Food and Chemistry Toxicology, 40, 33-41. Adebiyi, J. A., Obadina, A.O., Adebo, O. A., & Kayitesi, E. (2018). Fermented and malted millet products in Africa: Expedition from traditional/ethnic foods to industrial value added products. Critical Reviews in Food Science and Nutrition, 58, 463-474. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., Phoku, J. Z., & Kayitesi, E. (2017a). Fermented pulse-based foods in developing nations as sources of functional foods. In:

153

Functional Food - Improve Health through Adequate Food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 77-109. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., & Kayitesi, E. (2017b). Food metabolomics (Foodomics), a new frontier in food analysis and its potential in understanding fermented foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 211-234. Adebo, O. A., Njobeh, P. B., Mulaba-Bafubiandi, A. F., Adebiyi, J. A., Desobgo, S. C. Z., & Kayitesi, E. (2018). Optimization of fermentation conditions for ting production using response surface methodology. Journal of Food Processing and Preservation, 42, e13381, (1-10). Adeyeye, A., & Ajewole, K. (1992). Chemical composition and fatty acid profiles of cereals in Nigeria. Food Chemistry, 44, 41-44. Afify, A. M. R., El-Beltagi, H. S., Abd El-Salam, S. M., & Omran, A. A. (2012). Oil and fatty acid contents of white sorghum varieties under soaking, cooking, germination and fermentation processing for improving cereal quality. Notulae Botanicae Horti Agrobotamici, 40, 86-92. Al-Zahraa, M. D., Soumia, M. D., & Fathy, E. E. (2016). Analysis of organochlorine and organophosphorus pesticide residues in dairy products and baby foods from Egyptian markets. Journal of Environmental and Analytical Toxicology, 6, 1-5. Alabdulkarim, B., Bakeet, Z. A. N., & Arzoo, S. (2012). Role of some functional lipids in preventing diseases and promoting health. Journal of King Saud University of Sciences, 24, 319-329. Ayash, M., Al-Nuaimi, A. K., Al-Mahadin, S., & Liu, S. Q. (2018). In vitro investigation of anticancer and ACE-inhibiting activity, α-amylase and α-glucosidase inhibition, and antioxidant activity of camel milk fermented with camel milk probiotic: A comparative study with fermented bovine milk. Food Chemistry, 15, 588-597. Baek, H. H. (2017). Compilation of volatile flavor compounds in cheonggukjang and doenjang. Food Science and Industry, 12, 24-49. Baek, J. G., Shim, S. M., Kwon, D. Y., Choi, H. K., Lee, C. H., & Kim, Y. S. (2010). Metabolite profiling of cheonggukjang, a fermented soybean paste, inoculated with various Bacillus strains during fermentation. Bioscience Biotechnology and Biochemistry, 74, 1860-1868. Bergamo, P., Luongo, D., Miyamoto, J., Cocca, E., Kishino, S., Ogawa, J., Tanabe, S., & Rossi, M. (2014). Immunomodulatory activity of a gut microbial metabolite of dietary linoleic acid,

154

10-hydroxy-cis-12-octadecenoic acid, associated with improved antioxidant/detoxifying defences. Journal of Functional Foods, 11, 192-202. Bender, D. M., Peterson, J. A., McCarthy, J. R., Gunaydin, H., Takano, Y., & Houk, K. N. (2008). Cyclopropanecarboxylic acid esters as potential prodrugs with enhanced hydrolytic stability. Organic Letters, 10, 509-511. Bonnaud, B., Cousse, H., Mouzin, G., Briley, M., Stenger, A., Fauran, F., & Couzinier, J. P. (1987). 1-Aryl-2-(aminomethyl)cyclopropanecarboxylic acid derivatives. A new series of potential antidepressants. Journal of Medicinal Chemistry, 30, 318-325. Brits, M., Gorst-Allman, P., Rohwer, E. R., De Vos, J., de Boer, J., & Weiss, J. M. (2018). Comprehensive two-dimensional gas chromatography coupled to high resolution time-of-flight mass spectrometry for screening of organohalogenated compounds in cat hair. Journal of Chromatography A, 1536, 151-162. Carr, T. P., Weller, C. L., Schlegel, V. L., Cuppett, S. L., Guderian, D. M., & Johnson, K. R. (2005). Grain sorghum lipid extract reduces cholesterol absorption and plasma non-HDL cholesterol concentration in hamsters. The Journal of Nutrition, 135, 2236-2240. Cevallos-Cevallos, J. M., Reyes-De-Corcuera, J. I., Etxeberria, E., Danyluk, M. D., & Rodrick, G. E. (2009). Metabolomic analysis in food science: A review. Trends in Food Science and Technology, 20, 557-566. Cho, K. M., Math, R. K., Islam, S. M. A., Lim, W. J., Hong, S. Y., Kim, J. M., & Yun, H. D. (2009). Biodegradation of chlorpyrifos by lactic acid bacteria during kimchi fermentation. Journal of Agricultural and Food Chemistry, 57, 1882-1889. Coutron-Gambotti, C., & Gandemer, G. (1999). Lipolysis and oxidation in subcutaneous adipose tissue during dry-cured ham processing. Food Chemistry, 64, 95-101. Dharni, S., Sanchita, Maurya, A., Samad, A., Srivastava, S. K., Sharma, A., & Patra, D. D. (2014). Purification, characterization, and in vitro activity of 2,4-Di-tert-butylphenol from Pseudomonas monteilii PsF84: Conformational and molecular docking studies. Journal of Agricultural and Food Chemistry, 62, 6138-6146. Di Cagno, R., Filannino, P., Cavoski, I., Lanera, A., Mamdouh, B. M., & Gobbetti, M. (2017). Bioprocessing technology to exploit organic palm date (Phoenix dactylifera L. cultivar Siwi) fruit as a functional dietary supplement. Journal of Functional Foods, 31, 9-19.

155

Fernando, S., Jobst, K. J., Taguchi, V. Y., Helm, P. A., Reiner, E. J., & McCarry, B. E. (2014). Identification of the halogenated compounds resulting from the 1997 Plastimet Inc. fire in Hamilton, Ontario, using comprehensive two-dimensional gas chromatography and (ultra)high resolution mass spectrometry. Environmental Science and Technology, 48, 10656-10663. Ferri, M., Serrazanetti, D. I., Tassoni, A., Baldissarri, M., & Gianotti, A. (2016). Improving the functional and sensorial profile of cereal-based fermented foods by selecting Lactobacillus plantarum strains via a metabolomics approach. Food Research International, 89, 1095-1105. Giusti, L., Gabriele, M., Penno, G., Garofolo, M., Longo, V., Del Parto, S., Lucchesi, D., & Pucci, L. (2017). A fermented whole grain prevents lipopolysaccharides-induced dysfunction in human endothelial progenitor cells. Oxidative Medicine and Cellular Longevity, 1-13. Article ID 1026268. González-Curbelo, M. A., Asensio-Ramos, M., Herrera-Herrera, A. V., & Hernández-Borges, J. (2012). Pesticide residue analysis in cereal-based baby foods using multi-walled carbon nanotubes dispersive solid-phase extraction. Analytical and Bioanalytical Chemistry, 404, 183- 196. Grover, N., & Patni, V. (2013). Phytochemical characterization using various solvent extracts and GC-MS analysis of methanolic extract of Woodfordia fruticosa (L.) Kurz leaves. International Journal of Pharmacy and Pharmaceutical Sciences, 5, 291-295. Hadbaoui, Z., Djeridane, A., Yousfi, M., Saidi, M., & Nadjemi, B. (2010). Fatty acid, tocopherol composition and the antioxidant activity of the lipid extract from the sorghum grains growing in Algeria. Mediterranean Journal of Nutrition and Metabolism, 3, 215-220. Hassan, S., Imran, M., Ahmad, N., & Khan, M. K. (2017). Lipids characterization of ultrasound and microwave processed germinated sorghum. Lipids in Health and Disease, 16, 1-11. Hernández, F., Portolés, T., Pitarch, E., & López, F. J. (2011). Gas chromatography coupled to high-resolution time-of-flight mass spectrometry to analyze trace-level organic compounds in the environment, food safety and toxicology. Trends in Analytical Chemistry, 30, 388-400. Holtze, M. S., Sørensen, S. R., Sørensen, J., & Aamand, J. (2008). Microbial degradation of the benzonitrile herbicides dichlobenil, bromoxynil and ioxynil in soil and subsurface environments – Insights into degradation pathways, persistent metabolites and involved degrader organisms. Environmental Pollution, 154, 155-168.

156

Hu, K., Jin, G. J., Mei, W. C., Li, T., & Tao, Y. S. (2018). Increase of medium-chain fatty acid ethyl ester content in mixed H. uvarum/S. cerevisiae fermentation leads to wine fruity aroma enhancement. Food Chemistry, 239, 495-501. Ibánez, M., Sancho, J. V., Bijlsma, L., van Nuijs, A. L. N., Covaci, A., & Hernández, F. (2014). Comprehensive analytical strategies based on high-resolution time-of-flight mass spectrometry to identify new psychoactive substances. Trends in Analytical Chemistry, 57, 107-117. Jay, J. M., Loessner, M. J., & Golden, D. A. (2005). Modern Food Microbiology (7th ed). Springer, USA. Kataoka, S. (2005). Functional effects of Japanese style fermented soy sauce (shoyu) and its components. Journal of Bioscience and Bioengineering, 100, 227-234. Kaushik, G., Satya, S., & Naik, S. N. (2009). Food processing a tool to pesticide residue dissipation – A review. Food Reviews International, 42, 26-40. Kikafunda, J. K., Abenakyo, L., & Lukwago, F. B. (2006). Nutritional and sensory properties of high energy/nutrient dense composite flour porridges from germinated maize and roasted beans for child-weaning in developing countries: a case for Uganda. Ecology of Food and Nutrition, 45, 279-294. Kim, J. K., & Lee, E. J. (2004). Characteristic flavor compounds of traditional Korean chongkukjang. Food Science and Biotechnology, 13, 566-571. Lee, S. Y., Lee, S., Lee, S., Oh, J. Y., Jeon, E. J., Ryu, H. S., & Lee, H. S. (2014a). Primary and secondary metabolite profiling of doenjang, a fermented soybean paste during industrial processing. Food Chemistry, 165, 157-166. Lee, B. H., Carr, T. P., Weller, C. L., Cuppett, S., Dweikat, I. M., & Schlegel, V. (2014b). Grain sorghum whole kernel oil lowers plasma and liver cholesterol in male hamsters with minimal wax involvement. Journal of Functional Foods, 7, 709-718. Lee, D. E., Lee, S., Jang, E. S., Shin, H. W., Moon, B. S., & Lee, C. H. (2016). Metabolomic profiles of Aspergillus oryzae and Bacillus amyloliquefaciens during rice koji fermentation. Molecules, 21, 1-15. Leroy, F., & De Vuyst, L. (2004). Lactic acid bacteria as functional starter cultures for the food fermentation industry. Trends in Food Science and Technology, 15, 67-78.

157

Liquori, A., Belsito, E. L., Di Gioia, M. L., Leggio, A., Malagrino, F., Romio, E., Siciliano, C., & Tagarelli, A. (2015). GC/MS analysis of fatty acids in Italian dry fermented sausages. The Open Food Science Journal, 9, 5-13. Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Mehmood, S., Orhan, I., Ahsan, Z., Aslan, S., & Gulfraz, M. (2008). Fatty acid composition of seed oil of different Sorghum bicolor varieties. Food Chemistry, 109, 855-859. Mozzi, F., Ortiz, M. E., Bleckwedel, J., De Vuyst, L., & Pescuma, M. (2013). Metabolomics as a tool for the comprehensive understanding of fermented and functional foods with lactic acid bacteria. Food Research International, 54, 1152-1161. Muturi, E. J., Donthu, R. K., Fields, C. J., Moise, I. K., & Kim, C. H. (2017). Effect of pesticides on microbial communities in container aquatic habitats. Scientific Reports, 7, 1-10. Nagao, K., & Yanagita, T. (2008). Bioactive lipids in metabolic syndrome. Progress in Lipid Research, 47, 127-146. Ohata, M., Zhou, L., Higuchi, K., Nagai, T., Kasamatsu, H., & Arihara, K. (2016). Investigation of volatile components and identification of the most potent odour‐active component in fermented meat sauce. Flavour and Fragrance Journal, 32, 171-177. Pale, S., & Taonda, S. J. B. (2010). Sorghum malt and traditional beer (dolo) quality assessment in Burkina Faso. Ecology of Food and Nutrition, 49, 129-141. Park, M. K., Cho, I. H., Lee, S., Choi, H. K., Kwon, D. Y., & Kim, Y. S. (2010). Metabolite profiling of cheonggukjang, a fermented soybean paste, during fermentation by gas chromatography mass spectrometry and principal component analysis. Food Chemistry, 122, 1313-1319. Park, S. E., Yoo, S. A., Seo, S. H., Lee, K. I., Na, C. S., & Son, H. S. (2016). GC-MS based metabolomics approach of Kimchi for the understanding of Lactobacillus plantarum fermentation characteristics. LWT-Food Science and Technology, 68, 313-321. Ruediger, G. A., Pardon, K. H., Sas, A. N., Godden, P. W., & Pollnitz, A. P. (2005). Fate of pesticides during the winemaking process in relation to malolactic fermentation. Journal of Agricultural and Food Chemistry, 53, 3023–3026.

158

Schaffer-Lequart, C., Lehmann, U., Ross, A. B., Roger, O., Eldridge, A. L., Ananta, E., Bietry, M. F., King, L. R., Moroni, A. V., Srichuwong, S., Wavreille, A. S., Redgwell, R., Labat, E., & Robin, F. (2017). Whole grain in manufactured foods: Current use, challenges and the way forward. Critical Reviews in Food Science and Nutrition, 57, 1562-1568. Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Seo, S. H., Park, S. E., Yoo, S. A., Lee, K. I., Na, C. S., & Son, H. S. (2016). Metabolite profiling of Makgeolli for the understanding of yeast fermentation characteristics during fermentation and aging. Process Biochemistry, 51, 1363-1373. Singracha, P., Niamsiri, N., Visessanguan, W., Lertsiri, S., & Assavanig, A. (2017). Application of lactic acid bacteria and yeasts as starter cultures for reduced-salt soy sauce (moromi) fermentation. LWT-Food Science and Technology, 78, 181-188. Song, Y. W., Lim, Y., & Cho. S. K. (2018). 2,4-Di-tert-butylphenol, a potential HDAC6 inhibitor, induces senescence and mitotic catastrophe in human gastric adenocarcinoma AGS cells. BBA – Molecular Cell Research, 1865, 675-683. Sun, Y., Zhou, D., & Shadidi, F. (2018). Antioxidant properties of tyrosol and hydroxytyrosol saturated fatty acid esters. Food Chemistry, 245, 1262-1268. Tapiero, H., Ba, G. N., Couvreur, P., & Tew, K. D. (2002). Polyunsaturated fatty acids (PUFA) and eicosanoids in human health and pathologies. Biomedical Pharmacotherapy, 56, 215-222. Trinder, M., McDowell, T. W., Daisley, B. A., Ali, S. N., Leong, H. S., Sumarah, M. W., & Reid, G. (2016). Probiotic Lactobacillus rhamnosus reduces organophosphate pesticide absorption and toxicity to Drosophila melanogaster. Applied and Environmental Microbiology, 82, 6204- 6213. Tugizimana, F., Paiter, L., & Dubery, I. (2013). Plant metabolomics: a new frontier in phytochemical analysis. South African Journal of Science, 109, 1-11. Ubukata, M., Jobst, K. J., Reiner, E. J., Reichenbach, S. E., Tao, Q., Hang, J., Wu, Z., Dane, A. J., & Cody, R. B. (2015). Non-targeted analysis of electronics waste by comprehensive two- dimensional gas chromatography combined with high-resolution mass spectrometry: using accurate mass information and mass defect analysis to explore the data. Journal of Chromatography A, 1395, 152-159.

159

Villa, B., Calbresi, L., Chiesa, G., Rise, P., Galli, C., & Sirtori, C. (2002). Omega-3 fatty acid ethyl esters increase heart rate variability in patients with coronary disease. Pharmacology Research, 45, 475-478. Wilson, E. M., Johanningsmeier, S. D., & Osborne, J. A. (2015). Consumer acceptability of cucumber pickles produced by fermentation in calcium chloride brine for reduced environmental impact. Journal of Food Science, 80, 1360-1370. Wishart, D. S. (2008). Metabolomics: Applications to food science and nutrition research. Trends in Food Science and Technology, 19, 482-493. Yang, A., Abd El-Aty, A. M., Park, J. H., Goudah, A., Rahman, M. M., Do, J. A., Choi, O. J., & Shim, J. H. (2014). Analysis of 10 systemic pesticide residues in various baby foods using liquid chromatography–tandem mass spectrometry. Biomedical Chromatography, 28, 735-741. Yaqoob, P. (2002). Monounsaturated fatty acids and immune function. European Journal Clinical Nutrition, 56, 9-13. Yoon, M. A., Jeong, T. S., Park, D. S., Xu, M. Z., Oh, H. W., Song, K. B., Lee, W. S., & Park, H. Y. (2006). Antioxidant effects of quinoline alkaloids and 2,4-di-tert-butylphenol isolated from Scolopendra subspinipes. Biological and Pharmaceutical Bulletin, 29, 735-739. Zhang, Y. H., Xu, D., Zhao, X. H., Song, Y., Liu, Y. L., & Li, H. N. (2016). Biodegradation of two organophosphorus pesticides in whole corn silage as affected by the cultured Lactobacillus plantarum. 3 Biotech, 6, 1-6.

160

Supplementary Figures

Supplementary Figure 6.1: (A) OPLS-DA score plot separating raw HT0000 and HT3165 samples (R2X = 0.693 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; HT3165 – HT-sorghum fermented with L. fermentum FUA 3165.

Supplementary Figure 6.2: (A) OPLS-DA score plot separating raw HT0000 and HT3321 samples (R2X = 0.787 and Q2 = 0.999), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; HT3321 – HT-sorghum fermented with L. fermentum FUA 3321.

161

Supplementary Figure 6.3: (A) OPLS-DA score plot separating raw HT0000 and HT2STRAINS samples (R2X = 0.797 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; HT2STRAINS – HT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321.

Supplementary Figure 6.4: (A) OPLS-DA score plot separating raw HT0000 and raw HT2872 (R2X = 0.797 and Q2 = 0.994), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. HT0000 – raw high tannin-sorghum; HT2872 – naturally fermented HT-ting.

162

Supplementary Figure 6.5: (A) OPLS-DA score plot separating raw LT0000 and LT3165 (R2X = 0.581 and Q2 = 0.954), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. LT0000 – raw low tannin-sorghum; LT3165 – LT-sorghum fermented with L. fermentum FUA 3165.

Supplementary Figure 6.6: (A) OPLS-DA score plot separating raw LT0000 and LT3321 (R2X = 0.585 and Q2 = 0.966), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. LT0000 – raw low tannin-sorghum; LT3321 – LT-sorghum fermented with L. fermentum FUA 3321.

163

Supplementary Figure 6.7: (A) OPLS-DA score plot separating raw LT0000 and LT2STRAINS (R2X = 0.721 and Q2 = 0.984), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. LT0000 – raw low tannin-sorghum; LT2STRAINS – LT-sorghum fermented with L. fermentum FUA 3165 and L. fermentum FUA 3321.

Supplementary Figure 6.8: (A) OPLS-DA score plot separating raw LT0000 and LT3424 (R2X = 0.446 and Q2 = 0.947), (B) ROC curve for the computed OPLS-DA model, (C) OPLS-DA loadings S-plot, (D) VIP plot for the same model. LT0000 – raw low tannin-sorghum; LT342 – naturally fermented LT-ting.

164

CHAPTER SEVEN 7.0 GENERAL DISCUSSION, CRITIQUE OF METHODS, CONCLUSION AND RECOMMENDATION 7.1 GENERAL DISCUSSION This study in part was aimed at investigating the biochemical and physicochemical properties of whole grain (WG) ting obtained through natural and controlled fermentation. Variations in volatile metabolite profiles and safety (mycotoxins) of the WG-ting samples were also evaluated. This section provides an overview of the study and a juxtaposed description of the findings presented in Chapters Three to Six.

Optimization of food processes is essential not only to understand underlying processes and effects of processing parameters, but also to select and identify optimal conditions for obtaining a product of a better quality as well as maximize the use of limited resources. As earlier hypothesized in Chapter One, the different regimes of fermentation time and temperature caused modifications to physicochemical properties, generally being pronounced in low tannin (LT) sorghum type as compared to high tannin (HT) type (Chapter Three). Subsequent results obtained after numerical optimization suggest that fermenting LT- and HT-sorghum types for 34 oC, 24 h and 28 oC, 72 h, respectively, yielded WG-ting with better biochemical composition and maximal health promoting constituents. The differences in these optimal conditions achieved could largely be attributed to differences in composition of the grains. Multi response numerical optimization (MRNO) was however observed to have compensated for this, by suggesting a longer fermentation time for HT- ting samples (72 h) than for LT-ting samples (24 h), although at a slightly higher fermentation temperature (34 oC) for the latter. Knowledge about these optimal conditions obtained (Chapter Three) is relevant for subsequent production of WG-ting from the respective sorghum grain types. This also indicates that inherent composition could largely influence the fermentation process and subsequent WG-ting obtained. Accordingly, these optimal conditions were further adopted to produce WG-ting using Lactobacillus fermentum strains during subsequent studies (Chapters Four – Six).

The presence of lactic acid bacteria (LAB) and their role in cereal fermentation have been reported with favourable modifications in constituents and food properties (Adebo et al., 2018). L. fermentum has also been noted as a dominant LAB during the fermentation of ting (Madoroba et

165 al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). The study presented in this Thesis thus utilized two L. fermentum strains, i.e., (L. fermentum FUA 3165 and L. fermentum FUA 3321), singly and in combination, earlier isolated from ting (Sekwati-Monang and Gänzle, 2011) for the fermentation of WG-ting from HT- and LT-sorghum types (Chapter Four). As envisaged and corroborating earlier observations (Chapter Three), fermentation with LAB strains resulted in better acidification and desirable modifications of titratable acidity, microbial load, health promoting components and antioxidant activity (AA). Reduced total phenolic content (TPC), total flavonoid content (TFC) and tannin content (TNC) values in L. fermentum WG-ting samples translated to increased amounts of specific bioactive compounds quantified via liquid chromatography tandem mass spectrometry (LC-MS/MS). This might have contributed to an increase in AA values, catechin, quercetin and gallic acid levels. Although a similar trend of desirable modification was observed for LAB strains, a better result was noted when L. fermentum FUA 3321 was used alone. This could suggest a better adaptation of this strain to both LT- and HT-sorghum substrates, yielding better and desirable modifications to WG-ting samples. The retention of AA, particularly in HT-ting samples, indicates that WG-sorghum can be processed into functional foods with potential health benefits.

The safety of foods is of immense concern to intended consumers. Of paramount importance are mycotoxins known to elicit negative effects on health and the economy. This challenge has led to proposals on effective strategies to implement in reducing or at best eliminating the presence of these fungal toxins in food. In this regard, fermentation is considered a safe, viable, cheap and effective alternative to ameliorate these toxins (Ezekiel et al., 2015; Adebo et al., 2017a). Accordingly, Chapter Five presented in this Thesis investigated the possibility of reducing mycotoxins via fermentation during WG-ting production from WG-sorghum. Initial mycotoxin levels observed in both LT- and HT-sorghum grains were below limits of mycotoxins regulated in South Africa, although the co-occurrence of fumonisins, zearalenone and its derivatives may still potentiate some additive or synergistic effects on human health, as they could toxicologically interact with one another. Although identification of the fungal strains was not performed in this study, presence of fungi (Chapters Three and Four) and occurrence of mycotoxins (Chapter Five) gives an indication that some of the fungal strains therein were mycotoxigenic. The mycotoxin levels were nonetheless significantly (p ≤ 0.05) reduced after fermentation, validating the hypothesis that mycotoxins in the raw sorghum may be modified during fermentation,

166 corroborating data from previous studies (Ezekiel et al., 2015; Adebo et al., 2017a; Okeke et al., 2018). Similar to results earlier obtained in Chapter Four, better mycotoxin reduction was also observed when L. fermentum FUA 3321 was used in isolation. Although fermentation was noted to reduce these mycotoxin levels, concerted efforts should still intensively be geared towards ensuring that sorghum used is mycotoxin-free. This is critical as the extent of mycotoxin reduction via fermentation is largely dependent on the initial mycotoxin content present in the raw material.

Adopting a metabolomics approach, an understanding of variations in volatile metabolites of the raw HT- and LT-sorghum types and subsequently obtained WG-ting was done. While all metabolites in a sample cannot possibly be extracted at once and quantified on a single instrument, an informed compromise was reached and extraction using 80% MeOH was eventually adopted. Subsequent data obtained from the system indicated that significant metabolites influencing the variation in the samples were acids, pesticides, ketones, cyclic compounds, fatty acids and their esters (Chapter Six), some of which are known to possess health promoting properties and vital for nutrition. Multivariate data analysis (MVDA) on the metabolomic dataset showed the different compounds that significantly contributed to the difference in raw HT- and LT-sorghum, which caused the earlier observation during MRNO (Chapter Three) and subsequently investigated parameters (Chapter Four). Some of these metabolites might have equally shielded the sorghum grain from fungal proliferation resulting in low mycotoxin levels (Chapter Five). Orthogonal partial least square discriminant analysis (OPLS-DA) equally separated the different obtained WG- ting samples with some overlapping compounds contributing to this observation. This was reflective of results obtained in other Chapters of this Thesis, indicative of differences due to sample source (sorghum type) and fermentation form (natural, single strain and mixed strain). Data obtained also confirmed the hypothesis suggesting that metabolites may be modified after fermentation, as some metabolites observed in the raw HT and LT-sorghum increased/decreased in the WG-ting samples.

Rising to the challenge of food insecurity, mal- and under-nutrition require concerted efforts and harnessing available indigenous knowledge on food/diets. Fermented foods have been consumed for ages and their significant role and importance in African cultures cannot be undermined. Although these fermented foods are complex metabolomes that require adequate and thorough

167 investigation of their underlying processes and composition, their role as an important part of African diets is worthy of note.

Though not investigated in this study, the perception and consumer acceptability of WG-ting still need to be ascertained. This in fact necessitated the term “modified” as reflected in the Executive Summary of this study, indicative of the change in the conventional refined sorghum usually adopted as the substrate for this product. Acceptability studies conducted on regular consumers of WG-ting and the general populace is still much needed to provide the necessary insight into the acceptability of this product. This is particularly important as studies in the literature have reported mixed observations with varying levels of acceptability for WG-products (Adams and Engstrom, 2000; Kahlon et al., 2015; Laureati et al., 2016; Mkandawire et al., 2015; Dovi et al., 2018), attributing this to the gritty mouth feel, perceived poor palatability and appearance of the product. Equally important is also the deep connection to the culturally accepted and habitual products from refined grains since this is regarded as “norm”. While this may not be totally dissuaded, effective persuasive communication and sensitization of intending consumers on the health beneficial potential of WG-products will hugely assist with increasing acceptance of WG-foods. Not only would this contribute to increased consumption of WG-ting, it will also open doors for the introduction of WG-sorghum in a broad spectrum of products such as snacks, meals and beverages.

Further to the acceptance of a WG-ting product is an envisaged challenge of a modification in the “customary” technique of processing ting. As described in Chapter Two, this is usually done in pots and calabash using traditional knowledge of “guesstimating” time and temperature combinations. While the use of an incubator for traditional use may not be feasible, would the use of starter cultures for traditional fermentation of ting in rural communities be accepted? The latter as well would probably not due to reasons not only limited to cultural beliefs, but to challenges of storage and distribution as well. Studies in this regard have however, suggested that dehydrated fermented products (with starter cultures) still maintain viability of LABs over long periods (Nche et al., 1994; Holzapfel, 2002). The study of Sekwati-Monang and Gänzle (2011) demonstrated the effectiveness of such technique with comparable acidification of ting by these dried cultures in contrast to fresh cultures.

168

7.2 CRITIQUE OF METHODS An evaluation of the methodologies used in this study warrant some discussions subsequently in this chapter. Therefore, a summarized critique of the experimental approach is presented in Section 7.2.1, while an appraisal of the analytical methods is provided thereto in Section 7.2.2.

7.2.1 Experimental approach Raw sorghum samples (Avenger and Titan cultivars) were obtained from a local supplier in Potchefstroom, South Africa and subsequent findings were thus based on this. As demonstrated in this study, differences in the composition of these sorghum grains influenced WG-ting composition. The presence of seed coat also contributed to results presented in this study as compared to other studies on ting from refined grains (Taylor and Taylor, 2002; Madoroba et al., 2009; Madoroba et al., 2011; Sekwati-Monang and Gänzle, 2011). Not only did the bioactive components present in the coat limit microbial action, they equally did influence the rate of mycotoxin and pesticide degradation (observed in Chapters Five and Six). Although the primary focus of this study was on the use of WG-sorghum for ting production, future studies can also look at the effect of decorticating these sorghum cultivars on ting quality and other parameters investigated in this study. This could further contribute to indigenous knowledge and provide a better comparison with earlier studies on ting from refined sorghum grains.

Since different fermentation conditions exist for ting production, optimization of these conditions is essential to identify optimal conditions that could be followed in order to obtain a product with a better quality. Although different optimization techniques exist such as central composite design (CCD), factorial design and box Behnken (Granato and Ares, 2013; Candioti et al., 2014), the Doehlert design used herein gives fewer experiments allowing for effective maximization of available resources as proposed by Ferreira et al. (2004) and Yolmeh and Jafari (2017). The Doehlert design has also been demonstrated as an effective optimization technique for food fermentation in other studies (Taragano and Pilosof, 1999; Li et al., 2007; Karanam and Medicherla, 2010; Makebe et al., 2017). The multi response numerical optimization (MRNO) technique used in this study, eliminates bias when selecting optimal conditions in optimization experiments. However, it can at times predict optimal conditions outside of the experimental runs as found in this study. This could be a challenge when handling larger experiments with other optimization models such as CCD. Accordingly, the MNRO predictions are based on results

169 emanating from investigated parameters specific to the sample in question. As such, “a one size fits all” approach might not be applicable for all sorghum types as inherent composition would affect the levels of these investigated parameters in the WG-ting. This would also mean that optimal fermentation conditions for ting from refined sorghum grains or other whole grain sorghum substrates would be different from those reported in this study.

The fermentation process followed in this study was under laboratory conditions using an incubator. While this is vital for repeatability and other scientific reasons, it does not however, translate to what is applicable in rural communities, where a majority of this product is produced and consumed. Selection of the starter cultures used in this study were subject to availability from other earlier authors as more than one lactic acid bacteria (LAB) could have been used. Challenges were initially faced in this regard as cells of both Lactobacillus plantarum strains intended for use were unfortunately dormant upon reception. Though the intention was to use both L. plantarum and L. fermentum (singly and in combination), challenges with the L. plantarum strains informed a slight change in the experimental design to accommodate this, informing the use of two different Lactobacillus fermentum strains. While a separate experiment to isolate the microorganisms from WG-ting could have been done, this was not the primary focus of the study. Such an isolation and subsequent identification of the microbiota of WG-ting could have possibly revealed the dominant LABs that are better adapted to WG-sorghum substrate during the production of WG-ting. Subsequent application of such LAB(s) might have resulted in better acidification and microbial activity when used as a fermentation starter culture(s) for WG-ting production.

7.2.2 Analytical methods The major analytical techniques employed in this study were colorimetric techniques, microscopy and chromatography (gas and liquid). For the colorimetric analytical technique used, it basically relies on the formation of coloured products and redox reactions that depend on time and temperature, which has previously been reported for different sorghum-based foods. It is however, worthy to mention herein that the acidified methanolic sample extracts may contain other components that might interfere with recovery of the analytes of interest. Nonetheless, these assays are still effective in showing trends and could adequately provide an estimate of the levels of total phenols, total tannins, total flavonoids and antioxidant activity investigated in this study.

170

Image analysis using scanning electron microscopy (SEM) reveals changes in the general morphology of the samples studied, but other objective response variables in the images such as granular size, coarseness and compactness could not be obtained. SEM usually provides surface details that could not at times be fully explained and explored. Further detailed information on these images can however, be obtained using additional computer algorithms, digital image analysis and other microscopy techniques including transmission electron microscopy (TEM), light microscopy and fluorescence microscopy, which were beyond the scope of this study.

The liquid chromatography triple-quadrupole tandem mass spectrometry (LC-MS/MS) system adopted for the quantification of bioactive components and mycotoxins is probably the best choice of instrument when standards of analytes of interest are available for use as reference materials. Due to this limited availability of standards, only four specific phenolic standards were quantified. As a result, phenolic compounds such as phenolic acids, flavonoids, tannins and additional bioactive compounds that might have contributed to the antioxidant levels and other observations in this study could not be investigated. Accordingly, the LC-MS/MS system utilized adequately quantified compounds based on retention time, parent nominal mass (m/z) and fragmentation patterns (daughter ions) of the available standards. These parameters were subsequently scanned, identified and quantified in both raw and processed samples. Although not the particular focus of this study, an identification of other compounds (peaks) would have been possible if the much- required chromatographic parameters such as UV spectra and corresponding wavelength, accurate masses, mass accuracy, elemental composition and molecular formula were available from the system used. These important chromatographic parameters are not available in the triple- quadrupole LC-MS/MS system used and as such, identification of other peaks could not be achieved. Other better suited LC platforms such as the LC-quad-time of flight mass spectrometry system (LC-Q-TOF-MS) and the LC-orbitrap-MS systems could be adopted for this in future studies.

Although mycotoxins are of paramount importance in cereals and their derived products, other food safety hazards in cereals of concern include biotoxins, food-borne pathogens and chemical residues, which can further be investigated in future studies on WG-ting. The occurrence of pesticide residues is of significance in future studies as it was observed as a significant metabolite as reported in Chapter Six when describing the metabolomics of raw sorghum and WG-ting

171 samples. Although the 14 quantified mycotoxins are vital and important ones from a health and economic point of view (Bryla et al., 2016; Tola and Kebede, 2016; Lee and Ryu, 2017), some emerging and masked mycotoxins, which were not profiled in this study due to lack of reference materials are equally vital and can be considered in future studies.

The challenge of low equipment sensitivity for unknown compounds (without standards) was not the case with the GC-HRTOF-MS system, which adequately identified these components in WG- ting samples. Exploration of metabolomics dataset identified major (significant) metabolites that contributed to differences/similarities in the samples investigated. This is not unusual, since the metabolomics pipeline as reviewed in Chapter Two utilizes bioinformatic and MVDA tools that operate on developed algorithms and in doing so, identified major (significant) metabolites. Without metabolomics, understanding the effect of variations in this study, i.e., sample variety (HT and LT), fermentation conditions (34 oC, 24h and 28 oC, 72h) and the type of fermentation (natural and LAB) could in fact be quite complex and difficult to effectively understand and describe. It should also be noted that no single chemometric software/MVDA method can extract all the available information from an untargeted metabolomic data set (Scalbert et al., 2009; Tugizimana, 2017) as different algorithms and vendor specific software are available for this purpose. Irrespective of the chemometric/data extraction approach used, there is still a “significant loss in data”, as some compounds that might correlate to certain aspects of a study might be omitted along the chemometric process. This is in fact the norm in GC-MS metabolomics with different available studies on fermented foods reporting few significant compounds (Mozzi et al., 2013; Adebo et al., 2017b) in comparison to the number of analytes a GC-MS system can detect at once.

Furthermore, due to the poor volatility of other compounds such as amino acids, sugars, some organic acids and other polar and mid-polar compounds, they could not be identified in this study. Accordingly, adequate derivatization targeting specific and/or all functional groups related to these compounds is needed to identify these compounds when using GC-MS. This can also be utilized in addition to other analytical platforms such as nuclear magnetic resonance (NMR), capillary electrophoresis mass spectrometry (CE-MS) and high resolution LC-MS, for a broader understanding of variations in metabolite profile of raw HT- and LT-sorghum and derived WG- ting samples. It should also be noted that application of multiple analytical platforms for metabolomics studies might be quite expensive as each sample must be adequately replicated. As

172 a general rule, at least three biological replicates and two analytical replicates (totaling six replicates) per sample must be utilized in a metabolomics experiment, which was performed in this study.

7.3 CONCLUSION AND RECOMMENDATION With the prevailing hunger, malnutrition and food insecurity in sub-Saharan Africa, which are further aggravated by climate change and political instability, there is a never ending need to ensure food security. As such, addressing this challenge would require a multifaceted approach, not only in developing food products for adequate nutrition, but those that are safe, affordable and value- added that will provide nutritional and health benefits to consumers. Whole grain sorghum is adequately positioned in this regard, considering the beneficial components embedded in it. This present study has provided the basis and understanding of “modified” ting from whole grain sorghum and to some extent, whole grain sorghum fermented foods. The data generated also confirm that whole grain ting produced with Lactobacillus fermentum is a safe and potential functional food product. Not only did the use of L. fermentum as a starter in fermentation produce a better ting product, it was accompanied by a desirable reduction in mycotoxin contamination and pesticide residue levels as well as led to desirable formation of relevant metabolites. This is typical for low tannin-ting samples fermented with L. fermentum FUA 3321 that can be recommended for subsequent adoption for whole grain ting production. This choice is based on shorter fermentation time (though at higher temperatures) coupled with rapid acidification, significantly higher reduction in mycotoxin contents, comparable bioactive components and presence of relevant health and nutritionally important metabolites. These qualities also position this product as a potential source of these beneficial compounds.

To adequately validate the suggestions made in this study, additional studies focused on consumer evaluation for acceptance of the product, comprehensive investigation of different bioactive compounds embedded in whole grain ting; nutritional composition including bioavailability and understanding possible effects of consuming this product using both in vitro and in vivo models are required. Further to these, are relevant studies that involve the use of innovative technologies and other novel food processing technologies for the production of value-added products from whole grain sorghum. While this study will contribute to scientific and indigenous knowledge in relation to food security, it should be noted that achieving an Africa without hunger and

173 malnutrition still requires sustained commitments and collaborative efforts among concerned stakeholders.

References Adams, J., & Engstrom, A. (2000). Helping consumers achieve recommended intakes of whole grain foods. Journal of American College of Nutrition, 19, 339-344. Adebo, O. A., Njobeh, P. B., Gbashi, S., Nwinyi, O. C., & Mavumengwana, V. (2017c). Review on microbial degradation of aflatoxin. Critical Reviews in Food Science and Nutrition, 57, 3208-3217. Adebo, O. A., Njobeh, P. B., Adebiyi, J. A., Gbashi, S., & Kayitesi, E. (2017b). Food metabolomics (Foodomics), a new frontier in food analysis and its potential in understanding fermented foods. In: Functional Food - Improve Health through Adequate Food, Hueda, M. C. (Ed.). InTech, Croatia. pp. 211-234. Adebo, O.A., Njobeh, P.B., Adeboye, A.S., Adebiyi, J.A., Sobowale, S.S., Ogundele, O.M., Kayitesi, E. (2018). Advances in fermentation technology for novel food products. In: Innovations in Technologies for Food and Beverage Industries, Panda, S., Shetty, P. (Eds.). Springer Publishers, USA. pp. 71-87. Bryla, M., Waskiewicz, A., Podolska, G., Szymczyk, K., Jedrzejczak, R., Damaziak, K., & Sulek, A. (2016). Occurrence of 26 mycotoxins in the grain of cereals cultivated in Poland. Toxins, 8, 1-20. Dovi, K. A., Chiremba, C., Taylor, J. R., & de Kock, H. L. (2018). Rapid sensory profiling and hedonic rating of whole grain sorghum-cowpea composite biscuits by low income consumers. Journal of the Science of Food and Agriculture, 98, 905-913. Ezekiel, C. N., Abia, W. A., Ogara, I. M., Sulyok, M., Warth, B., & Krska, R. (2015). Fate of mycotoxins in two popular traditional cereal-based beverages (kunu-zaki and pito) from rural Nigeria. LWT-Food Science and Technology, 60, 137-141. Ferreira, S. L. C., Dos Santos, W. N. L., Quintella, C. M., Neto, B. B., & Bosque-Sendra, J. M. (2004). Doehlert matrix: a chemometric tool for analytic chemistry–review. Talanta, 63, 1061- 1067. Granato, D., & Ares, G. (2013). Mathematical and statistical methods in food science and technology. Wiley, UK.

174

Holzapfel, W. (2002). Appropriate starter culture technologies for small-scale fermentation in developing countries. International Journal of Food Microbiology, 75, 197-212. Kahlon, T. S., Avena-Bustillos, R. J., & Chiu, M. M. (2016). Sensory evaluation of gluten-free quinoa whole grain snacks. Heliyon, 2, e00213. Karanam, S. K., & Medicherla, N. R. (2010). Application of Doehlert experimental design for the optimization of medium constituents for the production of L-asparaginase from palm kernal cake (Elaeis guineensis). Journal of Microbial and Biochemical Technology, 2, 7-12. Laureati, M., Conte, A., Padalino, L., Del Nobile, M. A., & Pagliarini, E. (2016). Effect of fiber information on consumer’s expectation and liking of wheat bran enriched pasta. Journal of Sensory Studies, 31, 348-359. Lee, H. J., & Ryu, D. (2017). Worldwide occurence of mycotoxins in cereals and cereal-derived food products: Public health perspectives of their co-occurence. Journal of Agricultural and Food Chemistry, 65, 7034-7051. Li, Y., Liu, Z., Cui, F., Liu, Z., & Zhao, H. (2007). Application of Plackett-Burman experimental design and Doehlert design to evaluate nutritional requirements for xylanase production by Alternaria mali ND-16. Applied Microbiology and Biotechnology, 77, 285-291. Madoroba, E., Steenkamp, E. T., Theron, J., Huys, G., Scheirlinck, I., & Cloete, T. E. (2009). Polyphasic taxonomic characterization of lactic acid bacteria isolated from spontaneous sorghum fermentations used to produce ting, a traditional South African food. African Journal of Biotechnology, 8, 458-463. Madoroba, E., Steenkamp, E. T., Theron, J., Scheirlinck, I., Cloete, T. E., & Huys, G. (2011). Diversity and dynamics of bacterial populations during spontaneous sorghum fermentations used to produce ting, a South African food. Systematic and Applied Microbiology, 34, 227-234. Makebe, C. W., Desobgo, Z. S. C., & Nso, E. J. (2017). Optimization of the juice extraction process and investigation on must fermentation of overripe giant horn plantains. Beverages, 3, 1-18. Mkandawire, N. L., Weier, S. A., Weller, C. L., Jackson, D. S., & Rose, D. J. (2015). Composition, in vitro digestibility, and sensory evaluation of extruded whole grain sorghum breakfast cereals. LWT-Food Science and Technology, 62, 662-667. Mozzi, F., Ortiz, M. E., Bleckwedel, J., De Vuyst, L., & Pescuma, M. (2013). Metabolomics as a tool for the comprehensive understanding of fermented and functional foods with lactic acid bacteria. Food Research International, 54, 1152-1161.

175

Nche, P. F., Nout, M. J. R., & Rombouts, F. M. (1994). The effect of cowpea-supplementation on the quality of kenkey, a traditional Ghanaian fermented maize. Journal of Cereal Science, 19, 191-197. Okeke, C. A., Ezekiel, C. N., Sulyok, M., Ogunremi, O. R., Ezeamagu, C. O., Sarkanj, B., Warth, B., & Krska, R. (2018). Traditional processing impacts mycotoxin levels and nutritional value of ogi – A maize-based complementary food. Food Control, 86, 224-233. Scalbert, A., Brennan, L., Fiehn, O., Hankemeier, T., Kristal, B. S., van Ommen, B., Pujos-Guillot, E., Verheij, E., Wishart, D., & Wopereis, S. (2009). Mass-spectrometry-based-metabolomics: limitations and recommendations for future progress with particular focus on nutrition research. Metabolomics, 5, 435-458. Sekwati-Monang, B., & Gänzle, M. G. (2011). Microbiological and chemical characterization of ting, a sorghum-based sourdough product from Botswana. International Journal of Food Microbiology, 150, 115-121. Taragano, V. M., & Pilosof, A. M. R. (1999). Application of Doehlert designs for water activity, pH and fermentation time optimization for Aspergillus niger pectinolytic activities production in solid-state and submerged fermentation. Enzyme and Microbial Technology, 25, 411-419. Tugizimana, F. (2017). Metabolomic studies of induced defense-related changes in Sorghum bicolor in response to the pathogen Colletotrichum sublineolum. A PhD Thesis submitted to the Faculty of Science, University of Johannesburg, South Africa. Yolmeh, M., & Jafari, S. M. (2017). Applications of response surface methodology in the food industry processes. Food Bioprocess Technology, 10, 413-433.

176