Identification of Chemical Drivers of Coffee Bitterness Using FlavoromicsChengyu Gao and Devin Peterson*

*Corresponding author, peterson.892@osu.eduDepartment of Food Science and Technology, The Ohio State University, USA

Flavoromics Workflow

Coffee is one of the most popular beverages in the world with an estimated market of200 billion. As consumers are willing to pay more for specialty coffee, it is important tounderstand coffee flavor complexity to meet consumer demands. Thus, understanding themolecular basis of coffee bitterness is of high interest. A few bitter compounds in coffee havebeen identified; however, the sensory relevance of these compounds and their contribution tooverall coffee bitterness are not well established.[1,2]The taste-guided method, which is widelyused in literature, only focus on direct actives but is not applicable to identify flavormodulators that are not taste active.

The objective of this project was to apply an untargeted flavoromics approach toidentify chemical drivers of bitterness in roasted coffee brew samples and enable thediscovery of flavor modulator compounds.

Untargeted chemical profiling by UPLC/MS-(QToF) and sensory evaluation of 14coffee brews samples were carried out. Multivariate statistical analyses orthogonal projectionto latent structures (OPLS) was subsequently employed to select chemical markers highlypredictive of coffee bitterness. Isolation, identification, and sensory validation of selectedchemical markers are ongoing. This project will provide a new understanding of chemistrydriving overall coffee bitterness and allow the industry to tailor coffee products to differentconsumer preferences.

2. Multivariate statistical analysis and marker selection

• Untargeted flavoromics analysis was successfully applied to investigate markers that highly correlated to bitterness in coffee brew. Selected negative markers have been isolated and purified with more than 90% purity.

• Sensory recombination test of negative markers showed significant modulation effect on coffee bitterness. Sensory evaluation of individual negative markers is ongoing.

• Three negative markers have been identified as caffeoylquinic acid derivatives.• Future work will focus on isolation and identification of positive markers. Sensory recombination

of positive markers will be carried out to investigate their sensory relevance.

14 commercial beans (7 manufacturers)

Sensory evaluation Chemical profiling Data processing

Multivariate statistical analysis

Isolate and purify markers

Sensory recombination

1. Bitterness evaluation of roasted coffee brew

OPLS - coffee brew modeled with bitter intensity

S-plot of coffee brew with marker selection

Compound identification

Negative Markers VIP scoreMarker 8 10.7Marker 9 10.7

Marker 10 9.0Marker 11 8.0Marker 12 4.2Marker 13 4.0

6. Identification of negative markers

Comparison of RT (A) and MS/MS fragmentation of 4,5-diCQA standard (B) and Marker 13 (C)

UPLC-Mass spectrometry• MS/MS experiments were performed in

negative and positive ESI to obtaininformation on fragmentation patterns andelemental composition of the marker

• Three negative markers were identified ascaffeoyl quinic acid derivatives

Nuclear magnetic resonance (NMR)• Mono and bidimensional NMR will be

performed to obtain accurate structuralinformation of unknown compounds

Conclusion and Future Work

References

A

B

C

1. Frank, O., Zehentbauer, G., & Hofmann, T. (2006). Bioresponse-guided decomposition of roast coffee beverage and identification of key bitter taste compounds. European Food Research and Technology, 222(5-6), 492-508. doi:10.1007/s00217-005-0143-6

2. Kreppenhofer, S.; Frank, O.; Hofmann, T. Identification of (furan-2-yl)methylated benzene diols and triols as a novel class of bitter compounds in roasted coffee. Food Chem. 2011, 126, 441−449.

Abstract

1. Sensory evaluation of coffee brews• Caffeine solution as standards• 0-15 intensity scale • 8 trained panelists, 2 replicates

Materials and Methods

Ground coffee/water1:20 ratio

Sample clean upSPE 96 wellplates

C18 columnUPLC/MS(QToF) ESI- mode

14 commercial roasted beans

2. Chemical profiling of roasted coffee brew

3. Data processing and multivariate statistical analysis

Data transform Data filtration Multivariate analysis

Remove noise and inconsistent data• Signal intensity < 750 ion count• CV > 20% in QC samples

Transform chromatographic data into features

(retention time, m/z)PCA, OPLS, S-plot

Results and Discussion

Positive Markers VIP scoreMarker 1 5.0Marker 2 4.6Marker 3 4.5Marker 4 4.3Marker 5 3.7Marker 6 3.5Marker 7 3.4

3. Isolation and purification of markers

1st dimension fractionation(Extract from coffee beans)

2nd D fractionation>50% purity

3rd D fractionation> 90% purity

TIC of coffee brew extract1st dimension fractionation

TIC of purified marker(3rd D fractionation)

PCA – chemical variation of coffee brew

4. Quantification of negative markers in coffee brew

5. Sensory recombination test

Control (high bitter coffee)

negative markers(pH reduced 0.37)

adjusted pH

10.33 9.928.78

0

2

4

6

8

10

12

Control Control adjusted pH Control + negative markers

Biite

r Int

ensi

ty

*

• Anova, post-hoc Turkey, *p = 0.001

194.7 192.0

13.3 17.3

42.7

140.9

18.3

43.0

73.3

25.817.4 11.2 7.0 10.4 13.3

44.0

4.9 9.0

28.511.2

0

40

80

120

160

200

1 2 3 4 5 6 7 8 9 10

Conc

entr

atio

n (m

g/L

in d

rip c

offe

e)

Negative markers

Low Bitter Coffee

High Bitter Coffee

TIC of 4,5-diCQA STDTIC of Marker 13

MSMS of 4,5-diCQA at 20 V

MSMS of Marker 13 at 20 V