Flavor development in Cheese: a food-omics approach · 2019-06-25 · Flavor development in Cheese:...
Transcript of Flavor development in Cheese: a food-omics approach · 2019-06-25 · Flavor development in Cheese:...
Flavor development in Cheese: a food-omics approach
Baukje Folkertsma, Lizette Maljaars, Brenda Ammerlaan, Eduard Derks,
Rudi van Doorn, Bob Savage, Marijke van Vliet, Carry van Kleef, Tatjana
van den Tempel and Marcel van Tilborg
DSM Biotechnology Center
Delft, The Netherlands
November 12th , 2014
© 2014 DSM. All rights reserved.
Page Page 1
Chemical synthesis &
biotechnology
Energy, chemistry &
polymer technology
Biotechnology
Hoffman La Roche’s
Vitamins (1930s)
DSM (1902)
Gist-Brocades (1869)
Vitamins
Omega’s
Carotenoids
Premixes for food & feed
Enzymes
Minerals
Cultures & Yeasts
Nutraceuticals
Pharmaceuticals
Cellulosic bioethanol
Biomedical materials
Bio-plastics
High Performance Plastics
Polyamides and precursors
Resins for coatings and composites
Functional Materials
Solar – advanced surfaces
Life Sciences
Materials Sciences
DSM: Building on an impressive history
© 2014 DSM. All rights reserved.
Page
Combination of ripening
enzyme and DVC adjunct
cultures providing original
flavor profiles for
Continental/ Cheddar cheese.
Functional and clean kosher
lipase for natural, ingredient
and enzyme modified
cheeses.
FLAVOR SYSTEMS
TEXTURE TOOLBOX
YIELD IMPROVEMENT
Cultures that produce
creamier mouth feel,
especially in low fat
products.
Cultures with fat mimicking
properties.
Unique strain diversity that
develops better curd
integrity.
Fast acidifying cultures
for high volume cheese
producers.
Large rotation reducing
phage challenges.
Process Scan service with
our coagulant range.
DSM Business Unit Cultures Three focus areas
Page 2
© 2014 DSM. All rights reserved.
Page
Flavor analysis platform
A powerful tool to profile (flavor) metabolites in cheese to
support targeted product development in collaboration with
our customers
• Outline of the flavor analysis platform
• Example on Gouda cheese
• Summary and outlook
• Acknowledgements
3
© 2014 DSM. All rights reserved.
Page
Flavor analysis platform Outline
4
Sensory
BioIT
Analysis
Sample preparation: extraction of non-volatile key compounds
Analysis of volatile and non-volatile flavors and
metabolites using GC-MS, LC-MS (peptides) and
NMR spectroscopy
Statistical data
evaluation and
correlation of
sensory and
analysis
Taste panel
Integrated food-omics approach with Sensory and
BioIT maximizes the value of the results © 2014 DSM. All rights reserved.
Page
Flavor analysis platform Sensory analysis
• QDA method (QDA®, is one of the
main descriptive analysis techniques
in sensory evaluation)
• Non-biased and objective: trained
external panel
• Powerful: 14 panelists in one panel
• 39 attributes were assessed divided
over 6 categories
Discrimination
• Quantitative
Descriptive
Analysis
• Check-all-that-apply
Descriptive
• 2 or 3 Alternative
Forced Choice tests
• Triangle test
© 2014 DSM. All rights reserved.
Page
Flavor analysis platform Key analytical techniques
Sample preparation • Key to obtain analytical data with sufficient quality to use for
correlation to sensory
6
NMR spectroscopy • Detection of broadest possible range of compound classes in a
single measurement
• Easy to set up and automatize, high-throughput and cost-effective
GC-MS of non-volatiles • Very sensitive method for the detection of a broad range of
compound classes
GC-MS of volatiles • Sensitive method for profiling of volatile metabolites
LC-MS of peptides • General profiling of peptides of 2 to ~15 amino acids
© 2014 DSM. All rights reserved.
Page
Flavor analysis platform Chemical analysis: a food-omics approach
7
Initially apply untargeted
• Broadest list of metabolites
• Limited risk of overlooking relevant
compounds
• Identity of metabolites not
determined
• Minimum analytical data
interpretation
• Statistical modelling
Targeted assessment of results
• Select relevant peaks that correlate
with sensory attributes
• Identify metabolites using existing
data
• Analytical data interpretation only on
peaks of interest
bitter
sweet
1
3
4
5 6
7
8 9
10
2
11 13
14
15
16
12
3
5 6
16 14
bitter
sweet
1
4
7
8 9
10
2
11 13
15
12
3: propionic acid
5: cinnamic acid
6: isovaleric acid
14: benzaldehyde
16: phenol
© 2014 DSM. All rights reserved.
Page
Gouda cheese Cheese making
Conditions
• Acidifier CT110
• Milk type Full fat (48% FDM) and reduced fat (30% FDM)
• Rennet Maxiren
• Other additions Annatto, CaCl2 and sodium nitrate (allowed in the
Netherlands to prevent late blowing)
• Adjuncts 6 different adjunct blends with distinguished flavor
profiles
• Ripening 88% relative humidity; 13°C
8
Sampling
• Composition analysis at
2 wks
• Flavor platform at 6, 12
and 24 wks
© 2014 DSM. All rights reserved.
Page
Gouda cheese Reproducibility of cheese making
SDM (%) pH FDM (%) Moisture (%)
Control 2.3 5.25 51.3 41.5
RSD (%) 4.8 0.6 1.9 1.8
All cheeses 2.3 5.23 51.3 41.9
RSD (%) 7.4 0.8 1.7 2.0
9
Control = CT110 full fat
SDM = salt on dry matter
FDM = fat on dry matter
© 2014 DSM. All rights reserved.
Page
Gouda cheese Statistical data evaluation
• Raw data of three analytical methods and sensory
evaluation by the trained external panel
10
Analytical
data
data
processing
Modelling
NMR Non-volatiles
GC-MS Volatiles
GC-MS Non-volatiles
© 2014 DSM. All rights reserved.
Page
Gouda cheese Statistical data evaluation
• Allow combination of very different datasets
• Data processing addresses variability in
baselines, retention times, peak positions and
sample weights
• Peak tracking & integration and aggregation to
the same level as QDA
11
Analytical
data
data
processing
Modelling
14.5 14.6 14.7 14.8 14.9 15 15.1 15.2-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
Time (min)
Effect
OSGCMS - wk6 (pos) versus wk12 (neg)
14.5 14.6 14.7 14.8 14.9 15 15.1 15.2-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
Time (min)E
ffect
OSGCMS - wk6 (pos) versus aligned wk12 (neg)
6 wk
12 wk
© 2014 DSM. All rights reserved.
Page
Modelling
Gouda cheese Statistical data evaluation
• PLS – Partial Least Squares Regression: Multivariate
regression stabilized for collinear data using
SIMPLS from the Matlab PLS-Toolbox
• PLS-DA – PLS discriminant analysis that calculates the
scores as such that they discriminate maximally
between classes (ageing steps)
– Explore differences between cheeses during
ripening
– Identify sensory attributes or metabolites that
are linked to ripening
• PLS2 – Multivariate regression to model odor, mouth
feel and flavor sensory attributes as a function
of analytical profiles
12
Analytical
data
data
processing
© 2014 DSM. All rights reserved.
Page 13
Gouda cheese Complementary value of analytical techniques
Attribute NMR spectroscopy
Non-volatiles
OS-GC-MS
Non-volatiles
GC-MS
Volatiles
Salt-fl
Sweet-fl
Savoury-fl
Bitter-at
Creamy-mf
Rubbery-mf
The analytical methods*
• each describe a unique set of
sensory attributes
• overlap in describing some
sensory attributes
* This is for the models generated with the current set of Gouda cheeses and may be dependent on the type of cheese
© 2014 DSM. All rights reserved.
Page
Flavour analysis platform
• Powerful and complete overview of the broadest possible range of
non-volatile and volatile (flavor) metabolites in cheese
• Non targeted approach looks for more than the usual suspects
• More extensive platform (e.g. non-volatiles, volatiles and peptides,
NMR) than reported in literature
• Ongoing development of the platform to meet future industry needs
• Allows to distinguish our products based on their sensory and
analytical flavor development profile during ripening
• Expand Flavor Library for flavor prediction as targeted solution in
collaboration with our customers
14
© 2014 DSM. All rights reserved.
Page
Acknowledgements
18
Analysis
• Adriana Carvalho de Souza
• Leon Coulier
• Erwin Kaal
• Marjon Kok
• Raymond Ramaker
• Joep van Rijn
• Cock Tas
Microbiology
• Laurens Hanemaaijer
• Claire Price
Application
• Ronald Metselaar
© 2014 DSM. All rights reserved.
© 2014 DSM. All rights reserved.