Flavor Release - American Chemical Society


Flavor Release - American Chemical Societypubs.acs.org/doi/pdf/10.1021/bk-2000-0763.ch026SimilarNestlé Research Center,...

1 downloads 31 Views 892KB Size

Chapter 26

Relative Influence of Milk Components on Flavor Compound Volatility Deborah D. Roberts and Philippe Pollien

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

Nestlé Research Center, Vers-Chez-les-Blanc, CP 44, 1000 Lausanne 26, Switzerland

Although specific interactions between flavor compounds and proteins or carbohydrates are often characterized in simplified model systems, the importance of these effects in more complex fat-containing food is less well-documented. This study determined the relative effects of milk components on flavor compound volatility. Within an experimental design, milk samples with different levels of milk-solids-non-fat (MSNF; milk proteins and lactose) and milk fat were compared. The relative amounts of added flavor compounds in the headspace were determined using Solid-Phase Microextraction GC-FID, using a 1 min adsorption time. Three categories of compound behavior were identified: (1) not influenced by milk component addition (diacetyl, 2,3-pentanedione, guaiacol), (2) reduced in volatility with milk fat but not with M S N F (3-methyl butanal, 2-methylpropanal, 4-ethylguaiacol), (3) reduced in volatility with M S N F and much greater reductions with fat (β-damascenone, l-octen-3-one). This last group is the only one that showed flavor compound interactions with milk proteins or lactose. However, when fat was present in the system, the level of M S N F did not further influence flavor compound release. This shows that in complex liquid milk systems, compound absorption by fat is the primary retaining mechanism. In addition, measured lipophilicity of the compounds showed a good correlation (R = 0.9) with retention by milk for compounds with k > 0.7. 2

w

In our eating experience, flavor compounds exist in the realm of a food product. The final flavor perception is influenced by any chemical interactions with the food components. For this reason, a certain amount of flavor research in recent years has concentrated on a better understanding of these interactions. Proteins, for example, have been shown to bind to certain flavor compounds with measurable binding

© 2000 American Chemical Society

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

321

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

322 constants (7). Instances of reversible and non-reversible binding have been demonstrated (2). One protein in particular, β-lactoglobulin, has been extensively studied because its structure is very well described. Competition phenomena were observed between two compounds presumably binding at the same site. The interactions were investigated in-depth using IR spectroscopy where conformational changes by the protein were seen upon aroma addition. Modelling, headspace, and sensory analysis confirmed that the binding of β-lactoglobulin to compounds leads to lower release rates, lower perception, and a lower headspace concentration for compounds with higher affinity constants (5). Some of these binding effects have been illustrated sensorially as well as chemically. Casein and whey protein added at 0.5 % caused a perceptual decrease for some compounds such as vanillin and limonene (45). Carbohydrates also have the ability to form complexes with aroma compounds, such as with rx-amylose from starch (6) or specific binding with hydrocolloids (7). Additionally, hydrocolloids can markedly influence flavor release by changes in viscosity and compound diffusion: up to a 60% decrease in release has been found with 2% C M C added (8). Simple sugars can decrease or increase compound volatility as well but these effects are generally present at rather high concentrations, above 20% w/w (9). Lipids absorb and solubilize lipophilic flavor compounds, as explained by mathematical models (77-72), headspace analysis (75), and sensory analysis (9, 14). Limonene and ethyl heptanoate showed 80% decreases in volatility from water with the inclusion of only 1% vegetable oil (75). The fact that protein - flavor interactions and oil - flavor interactions are both hydrophobic in nature brings an interesting question: In systems containing both fat and protein, are both interaction effects present or is one effect predominant? While basic model systems are necessary to show specific binding to proteins or carbohydrates, further testing on more complex systems will indicate if the binding is significant in real food products. This work seeks to take the next step in determining the relative importance of milk components (fat vs. protein/carbohydrates) in the binding and release of flavor compounds.

Experimental

Experimental Design Two independent factors were assessed for their ability to influence flavor release: fat and milk-solids-non-fat (MSNF). Both factors were varied independently, as seen in the experimental design (Figure 1). The extreme points of M S N F and fat content were studied as well as a middle point. The double point at the high level represents samples prepared with two different milk sources. One was a commercial U H T whole milk and the other was the in-house preparation. The zero point is water.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

323

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

1.5

0.0 • 0.0

!

0.5



I

1

1.0

1.5

1

2.0

1

2.5

3.0

% milk-solids-non-fat

Figure L Experimental design for samples studies showing the choice of fat and milk-solids-non-fat levels.

Analysis of covariance was used with time between sample vial preparation and analysis as the covariant. This time showed a significant effect for 3 compounds: 2,3pentanedione, 3-methyl-2-butenal, and l-octen-3-one where the longer the time, the smaller the headspace peak. A multiple comparison test (Fisher's L S D , α = 0.05) was used to determine significant differences among the samples.

Sample Preparation The ingredients (Table I) were mixed together (except the butterfat) and heated to 60 °C. The mixture was pre-emulsified at 60 °C using a hand-held Ultraturrax for 3 min at 8000 min" using the medium size dispersing head. During this preemulsification, the butterfat was added in small portions. With stirring, the mixture was then homogenised with three passes using a Buchi homogeniser at 65°C. The sample was cooled on ice and then stored in the refrigerator until use. The fat globule size of the samples was checked microscopically and found to be of a similar size to pasteurized milk. The Malvern Mastersizer (Malvern, U K ) was also used and 85% of the fat globules were found to be under 1 micron. 1

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

324

Table I. Recipes for Milk Preparations that Represent the Experimental Design Points in Figure 1.

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

1

Skim Milk

Butterfat

(g)

(g)

Whole Milk UHT(g)

Water (g)

% MSNF

2

2

%Fat

48

0

0

32

2.87

0.02

12.8

2.176

0

65

0.77

1.36

12.8

0

0

67.2

0.77

0

24

1.088

0

54.9

1.43

0.69

48

2.176

0

29.8

2.87

1.38

0

0

54.4

25.6

2.95

1.36

0

0

0

80

0

0

1

Skim milk was prepared from low heat skim milk powder by adding 20 g to 180 g of water. % MSNF and % Fat were determined based on a proximate analysis of the skim milk powder and whole milk UHT.

2

Table II shows the flavor compounds chosen for this study. The compounds were analyzed in two batches of compounds that were shown to be stable together. The concentrations of aroma compounds were chosen so that they were in the linear quantification range of the S P M E fiber. The aroma compounds were dissolved in water with extended vial shaking. They were prepared at double concentration as in Table II. M i l k solutions (400 mg) and aqueous aroma solutions (400 mg) were added to silylated 2 mL glass vials (7 total preparations) and mixed without inverting the vial. A minimum time of 2 hours was determined for equilibration. The vials were kept at room temperature between 2 and 20 hours before analysis for batch 1 and 2 to 12 hours for batch 2.

Table II. Information about Aroma Compounds. Compound Diacetyl 2,3-Pentanedione 3-Methylbutanal 2-Methylpropanal 3-Methyl-2-butenal Guaiacol 4-Ethylguaiacol l-Octen-3-one β-Damascenone

Lipophilicity (K) -0.3 0.21 1.24 0.67 0.75 0.97 1.87 2.09 2.79

(ref. 16) -0.25 0.24

Supplier Cone, used Batch # (ppm) 1 Fluka 10 1 Fluka 10 1 Fluka 10 1 Fluka 10 1 Aldrich 10 2 St-Fons Chemie 10 2 Oxford 10 2 Oxford 2 Firmenich 2 2

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

325

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

Headspace Analysis by SPME GC-FID After a 1 hour equilibration at room temperature in the S P M E carousel (29°C), the headspace of the samples was sampled using a Varian 8200 autosampler. A S P M E (Solid phase microextraction) fiber was inserted into the headspace and allowed to equilibrate for 1 min exactly. This time was chosen so that the extraction would be primarily from the headspace and not from the sample. The fiber used was polydimethylsiloxane with 100 pm thickness. It was placed into the injection port of the G C for 5 minutes at 250 °C containing a 0.75 mm ID liner. During the first three minutes of desorption, the purge was off and the last two minutes with purge on further cleaned the fiber. G C separation with FID detection was used for quantification of the aroma compounds ( D B W A X , J&W, 30 m; 0.32 mm ID, 0.25 pm film, 10 psi helium). M i l k blanks were checked and no contaminating peaks over 1 % area were found.

Aroma Compound Lipophilicity Measurement Compound lipophilicity was determined based on its retention time on a reversed phase H P L C column (75). The conditions were the same as those used previously to measure 96 different compounds (16) so the values obtained can be directly compared. The H P L C used was a Hewlett Packard series 1100 with diode array and HP1097A refractive index detection (Avondale, P A , U D A ) . The column used (250 mm χ 4 mm) was packed with Nucleosil 50-5 CI8, particle size 5 p M (MachereyNagel, Oensinger, Switzerland). The mobile phase was made up volumetrically from various combinations (30-70%) of methanol and a solution containing 3morpholinopropane sulphonic acid buffer (0.01 M ) plus n-decylamine (0.2% v/v). The p H of the aqueous solution was adjusted beforehand to 7.4 (4.5 for aldehydes) by addition of HC1. Retention times (t ) were measured at room temperature with a 1.0 mL/min flow rate. The column dead time (to) was determined with uracil. The capacity factor was defined as k = (t - to) / to. Log k for 100 % water (low k ) was linearly extrapolated from results obtained for different mobile phase compositions. Table II shows the lipophilicity values obtained. r

r

w

Results and Discussion Figure 2 illustrates the theory behind flavor compound interactions with milk, based on compound lipophilicity and interactions with fat and proteins. Explanations of these three categories of compounds are complemented by Figures 3-6 which show the detailed results of each compound. The headspace concentration of each compound was determined when dissolved in water and in 6 milks of different composition. The placement of the points represents the particular milk studied. The values shown are the headspace concentration relative to water (100) and the different letters correspond to statistically significant differences.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

326

No Effect Low Compound Lipophilicity

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

(KJ -0.3

Compound Absorption by Milkfat ( f withf lipophilicity) Binding by MSNF in no fat systems

A

c

Β

0.7

High Compound Lipophilicity

2.0

4.0

Figure 2. Three categories of compound changes in volatility from water upon milk addition, dependent on compound lipophilicity and milk composition.

(A) No Effect Figure 3 shows results of compounds from this category. Statistically, the samples were not significantly different in their headspace concentration from water. Another study confirmed this finding: milk fat up to 12% did not influence the partition coefficient of diacetyl at 30 °C and 50 °C (7 7).

(B) Reduction in Volatility due to Fat but No Effect of Protein/Carbohydrates As seen in Figure 4, increasing M S N F at a constant fat level (0 and 1.4%) did not change the headspace concentration. However, increasing fat level at a constant 2.9% M S N F level did show reductions in headspace concentration. This category contains moderately lipophilic compounds. 3-Methylbutanal and 2-methylpropanal were slightly affected by the milk fat, with up to 30% reductions in headspace concentration. 4-Ethyl guaiacol was moderately affected by milk fat, with up to 55% reductions in headspace concentration. Although 3-methyl-2-butenal (Figure 5) belongs to this category by its lipophilicity, it is the only compound that has shown low levels of specific binding to M S N F , even in the presence of 1.4% fat. (C) Large reduction in Volatility due to Fat. Protein/Carbohydrate Effect Only at No Fat. As seen in Figure 6, the samples that contained fat were markedly different from those fat-free samples. As interactions with proteins are also hydrophobic in nature, βdamascenone and l-octen-3-one showed effects of M S N F , but only at zero fat content. The hydrophobic nature of fat overrides the protein effect in milks containing both ingredients. These flavor compounds had decreases of up to 94% in the headspace upon milk addition. A protein/carbohydrate binding effect was seen only in the samples without fat. Once fat is present, flavor compound absorption is the primary retaining mechanism.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

327

2,3-pentanedione

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

1.5



105 #100 φ3

LOO

101

100



0.0



1.0

2.0

3.0

4.0

3.0

4.0

% rniïk solids non fat

diacetyl

0.0* 0.0

1.0

2.0 % rnilk solids non fat

Figure 3. Relative headspace concentrations (water=100) for compounds in the presence of milk of different compositions showing no statistically significant differences from water.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

328

guaiacol

0.0

1.0

2.0

3.0

4.0

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

% milk solids non fat 3-methylbutanal

0.0

1.0

2.0

3.0

4.0

% milk solids non fat 2-methylpropanal

0.0

1.0

2.0

4.0

3.0

% milk solids non fat

b -42

o.o

1.0

ethylguaiacol b·

2.0

b-

3.0

4.0

% milk solids non fat

Figure 4. Relative headspace concentrations (water=100) for compounds in the presence of milk of different compositions showing statistically significant effects of fat.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

329

0.0

1.0

2.0

3.0

4.0

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

% milk solids non fat Figure 5. Relative headspace concentrations (water=100) for3-methyl-2-butenal in the presence of milk of different compositions showing a slight effect of MSNF.

l-octen-3-one

32.

1.5 $ 1-0

16*20 SI

0.5

a

100

0.0 •

0.0

b

3^5

51 1.0

2.0

3.0

4.0

% milk solids non fat

^ β-damascenone

0.0 • 0.0

1.0

2.0

3.0

4.0

% milk solids non fat Figure 6. Relative headspace concentrations (water=100) for compounds in the presence of milk of different compositions showing statistically significant effects of fat in the presence of MSNF but not the inverse.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

330 A general observation is, that for most compounds, the mid- and high- fat levels had similar values. The change in headspace concentration from water to a 0.7% fat milk was much larger than doubling the milk amount (to 1.4% fat milk). This indicates a non-linear relationship between fat concentration and headspace amount. Figure 7 depicts the relationship between compounds' lipophilicity and retention by milk. There appears to be a minimum value of compound lipophilicity (k around 0.7) for retention by milk. For a compound with k over 0.7, the more the compound is lipophilic, the more it was absorbed by milk fat. A similar relationship between k and compound retention by triolein in fresh cheese was also found within a chemical family (18). w

w

w

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

»100

Compounds' headspace concentration in whole milk relative to the headspace concentration in water

ι

Φ—

• •

2

for k^O.7, R = 0.9



75"

\

50-



25"



i -0.5

0.5

1

1

1

I

2.5

1.5

compound lipophilicity (k ) w

Figure 7. Relationship between compound lipophilicity and relative headspace (water=100) amount when in whole milk containing 0.7% fat and 1.4% MSNF. In addition to compounds from Table II, other compounds were also analyzed to complete the range of lipophilicity.

Figure 8 shows the difference caused by milk fat in the headspace concentrations of these 9 compounds. In the low fat sample, only the most lipophilic compounds showed a significant headspace reduction as compared to water. The higher fat sample showed greater reductions for the highly as well as the moderately lipophilic compounds.

Conclusions As many food products contain fats, carbohydrates, and proteins, the relative flavor-binding capacity of these constituents was shown using milk as a model. In emulsion systems, milk fat had a much larger effect than binding to milk proteins or

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

331

Figure 8. Comparison of flavor compound retention between whole and skim milk.

carbohydrates. M i l k with a fat content of only 0.7 % showed substantial flavor absorption (up to 91%). The model based on compound lipophilicity can be used to predict which compounds will change in volatility when going from a non-fat to a fat system as well as which compounds have a potential for hydrophobic protein interactions.

Literature Cited 1. 2. 3. 4. 5. 6. 7. 8.

Pelletier, Ε.; Sostmann, Κ.; Guichard, Ε. J. Agric. Food Chem. 1998, 46, 15061509. Overbosch, P.; Afterof, W.G.M.; Haring, P.G.M. Food Rev. Int. 1991, 7, 137184. COST 96: Interacton of food matrix with small ligands influencing flavor and texture. Volumes 1-3. 1998. European Communities. Conference Proceeding. Hansen, A.P.; Heinis, J.J. J. Dairy Sci. 1992, 75, 1211-1215. Hansen, A.P.; Heinis, J.J. J. Dairy Sci. 1991, 74, 2936-2940. Nuessli, J.; Sigg, B.; Conde-Petit, B.; Escher, F. Food Hydrocolloids 1997, 11, 27-34. Yven, C.; Guichard, E.; Giboreau, Α.; Roberts, D. J.Agric.Food Chem. 1998, 46, 1510-1514. Roberts, D.D.; Elmore, J.S.; Langley, K.R.; Bakker, J. J. Agric. Food Chem. 1996, 44, 1321-1326.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.

332 9. 10. 11. 12. 13. 14.

Downloaded by CORNELL UNIV on June 6, 2017 | http://pubs.acs.org Publication Date: September 7, 2000 | doi: 10.1021/bk-2000-0763.ch026

15. 16. 17. 18.

Ebeler, S.E.; Pangborn, R.M.; Jennings, W.G. J. Agric. Food Chem. 1988, 36, 791-796. Nawar, W.W. J. Agric. Food Chem. 1971, 19, 1057-1059. de Roos, K . B . ; Wolswinkel, K.In Trends in Flavor Research; Maarse, H., van der Heij, D.G., Eds.; Elsevier: Amsterdam, 1994; pp 15-32. Harrison, M.; Hills, B.P.; Bakker, J.; Clothier, T. J. Food Sci. 1997, 62, 653658, 664. Schirle-Keller, J.P.; Reineccius, G.A.; Hatchwell, L . C . J. Food Sci. 1994, 59, 813-815. Guyot, C.; Bonnafont, C.; Lesschaeve, I.; Issanchou, S.; Voilley, Α.; Spinnler, H.E. J. Agric. Food Chem. 1996, 44, 2348 E l Tayar, N.; Van de Waterbeemd, H . ; Testa, B . J. Chromatogr. 1985, 320, 305-312. Piraprez, G.; Herent, M.-F.; Collin, S. Flavor Fragr.J. 1998, 13, 400-408. Lee, K . D . ; Lo, C.G.; Richter, R.L.; Dill, C.W. J. Dairy Sci. 1995, 78, 26662674. Piraprez, G.; Herent, M.-F.; Collin, S. Food Chem. 1998, 61, 119-125.

Roberts and Taylor; Flavor Release ACS Symposium Series; American Chemical Society: Washington, DC, 2000.