| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* Les Fromageries Occitanes (LFO), Borde Blanche – Zone Industrielle, F-31290 Villefranche de Lauragais, France
UR410 Qualité des Produits Animaux, INRA Site de Clermont-Ferrand-Theix, F-63122 Saint Gènes-Champanelle, France
UMR1253, INRA, Agrocampus Rennes, 65 rue de Saint-Brieuc, F-35000 Rennes, France
1 Corresponding author: anne.thierry{at}rennes.inra.fr
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: Kluyveromyces lactis • Cantalet cheese • adjunct culture • odorous compound
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
In raw-milk Cantal cheese, a French Protected Denomination of Origin hard cheese that is manufactured in a similar way to Cheddar cheese, the population of yeasts enumerated reached 9 log10 cfu/g of curd (I. De Freitas and N. Pinon; unpublished data). Kluyveromyces lactis and P. fermentans were the 2 main species identified, accounting for 80 and 10% of the isolates identified, respectively (I. De Freitas and N. Pinon; unpublished data). Earlier studies showed that most of the lactose-fermenting yeasts isolated from Cantal cheese were identified as K. lactis (formerly Saccharomyces lactis and Torulopsis sphaerica), and the non lactose-fermenting yeast species being identified as Pichia spp. and Rhodotorula spp. (Millet et al., 1974).
Yeasts can play a role in the ripening process, either directly or via their interactions with the other cheese microorganisms. The most studied species are D. hansenii, K. lactis, Y. lipolytica, S. cerevisiae, and G. candidum. Yeasts can either positively contribute to the formation of cheese flavor, or be responsible for yeasty flavors and other off-flavors (Jakobsen and Narvhus, 1996). Other defects caused by the presence of spoilage yeasts are gas production, discoloration, and changes of texture. Kluyveromyces lactis is known for its ability to assimilate lactose and its enzymatic potential such as aminopeptidase activity (Lenoir, 1984). The interactions between bacteria and yeasts may result in the stimulation or the inhibition of the growth of one or both populations (Viljoen, 2006). Yeasts, however, are considered to stimulate the growth of lactic acid bacteria, via the synthesis of growth factors such as vitamins (Addis et al., 2001; Viljoen, 2006).
Yeasts are able to produce a variety of volatile odorous compounds in dairy products such as esters (mainly ethyl acetate), volatile organic sulfur compounds derived from the conversion of Met, various compounds of the catabolism of aromatic and branched-chain amino acids, and methylketones (Spinnler et al., 2001; Arfi et al., 2004). As a result, yeasts have the potential to contribute to the formation of cheese flavor (Lenoir, 1984). The potential of Pichia spp. to produce volatiles in cheese has not been studied.
The present work aimed to determine the interest of 2 yeasts, K. lactis and P. fermentans, isolated from the dominant yeast population of a typical raw milk Cantal cheese, as adjunct cultures to promote the formation of flavor of Cantalet cheese. We characterized the global microbiological, biochemical, and sensory changes induced by the presence of yeasts in microfiltered milk Cantalet cheese, with a particular focus on the formation of odorous compounds, determined by using a recently developed multi gas chromatography-olfactometry (GC-O) device (Berdagué et al., 2007).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
350 L) and pasteurized (80°C for 20 s) cream (40 L) were refrigerated and carried at 4°C to the manufacture plant. Skimmed milk was then heated at 40°C and microfiltered by microfiltration. The pilot installation Bactocatch type (MFS1, Alfa Laval, Les-Clayessous-Bois, France) was equipped Sterilox ceramic membrane with a porosity gradient of mean diameter 1.4 µm (1P1940, length 1.02 m, filtration area of 0.24 m2, Société des Céramiques Techniques, Tarbes, France). Pasteurized cream was added to the microfiltration permeate to adjust the fat content to 3.3%, constituting cheese milk. The milk was then divided into 3 vats of 100 L each and heated to 31°C. Lactic starters and yeast cultures were then added. Calf rennet (33 mL/100 L of milk) was added 10 min later. Coagulation took around 25 min, and the curd was cut for 10 min. The mixture of curd and whey was then directed toward a pressing vat in which whey drainage and pressing took place at room temperature. The curd was turned and pressed 6 times in pressure cycles with an increase from 1.8 x 105 to 6.0 x 105 Pa for 60 min. Then, the pressed curd was matured overnight at 17°C to a pH of about 5.1 to 5.2. The matured curd was then ground and mixed with 2.0% of dry salt. One hour later, the salted curd was molded and pressed for 48 h (pressure was increased from 0.9 x 105 to 5.0 x 105 Pa) at 16°C. After removal from the molds, the cheeses were ripened under controlled conditions (10°C and 95% relative humidity). The control cheeses were made by adding only commercial lactic starters (F-DVS DCC-260 and CHN19, Chr. Hansen, St-Germain Les Arpajon, France), to obtain a global population of 105 cfu/mL in cheese milk. Starters contained lactococci, leuconostocs, Streptococcus thermophilus, Lactobacillus helveticus, and Lactobacillus casei. The 2 experimental cheeses were made by adding the same lactic starters and the yeasts K. lactis F41 no. 1 (Kl) or P. fermentans F31 no.2 (Pf) (2 strains isolated from raw milk Cantal cheese, from LFO, Villefranche de Lauragais, France) inoculated at 105 cfu/mL in cheese milk. For each cheese trial, 5 samples were analyzed: microfiltered milk and cream mixture (T0), curd after maturation (T14h), cheese after demolding (T3d), and cheeses after 45 and 95 d of ripening. Samples were stored at –20°C before analyses, except those for volatile compound and FFA analyses, for which samples were kept at –80°C. Preliminary results obtained by enumerating yeasts in the same Cantalet cheese sample either freshly sampled or after being stored for 15 d at –20°C showed that counts did not differ by more than 0.5 log10 cfu/g.
Microbiological Analyses
Microbial Counts.
Cheese samples were prepared as previously described (De Freitas et al., 2005). Mesophilic aerobic microflora were enumerated on PCA (Biokar Diagnostics, Beauvais, France), lactococci on M17 agar (Difico, BD Diagnostics, Pont de Claix, France) incubated for 7 d at 15°C, thermophilic streptococci on M17 agar (Difco) incubated for 1 d at 43°C, thermophilic lactobacilli on de Man, Rogosa, and Sharpe, pH 5.4, incubated for 2 d at 43°C under anaerobiosis, and facultative heterofermentative lactobacilli on FH agar, incubated for 3 d at 37°C under anaerobiosis, as previously described (De Freitas et al., 2005). The use of M17 at 15°C for lactococci enumeration was adapted from the IDF Standard to improve the selective recovery of lactococci from mixed lactic acid bacteria populations, because it is known that M17 is poorly selective. Yeasts were enumerated on oxy-tetracycline glucose agar (Biokar Diagnostics) after incubation for 5 d at 24°C. Mean values resulting from duplicate plates are given on the log10 basis.
Compositional Analyses
Sampling.
For all analyses, an inner cheese sample was grated, thoroughly mixed, and used directly for pH, TS content, and fat determinations. pH was measured by using a pH meter (HI 9025, Hanna Instruments, Vila do Conde, Portugal); TS was determined in triplicate by drying cheese samples (2 g) mixed with sand at 102°C for 7 h (FIL-IDF, 1987b); fat was determined in duplicate by the gravimetric method (FIL-IDF, 1987a). Thirty grams of grated cheese was homogenized with 60 g of distilled water by mixing in an ice bath for 2 x 1 min at 20,500 rpm using an Ultraturrax blender (Janke & Kunkel, Staufen, Germany). The homogenates were frozen at –20°C before analysis of acidity, ethanol, and lactose. Lactose contents were determined by an enzymatic method through the use of Boehringer kits (Ref. 10176303, R-Biopharm, Darmstadt, Germany).
Free AA.
Individual free AA (FAA) were determined in duplicate from a homogenate of 5 g of grated cheese with 10 g of lithium citrate buffer, 0.5 M, pH 7.0. An aliquot of the homogenate was deproteinized by addition of a 10% sulfosalicylic acid solution (1:4, wt/wt), incubated for 1 h at 4°C, and centrifuged at 1,000 x g at 4°C for 15 min. The supernatant was filtered on a 0.2-µm pore diameter membrane, and diluted (1:4, vol/vol) with lithium citrate buffer, 0.2 M, pH 2.2. Amino acids were analyzed using an amino acid analyzer (AlphaPlus series 2, Pharmacia, Uppsala, Sweden).
Individual FFA.
Free fatty acid contents were determined at 95 d of ripening after extraction according to the method of De Jong and Badings (1990). Two independent extractions of cheese lipids were performed using aminopropyl solid-phase extraction columns (500 mg/3 mL; Phenomenex, Torrance, CA) and a vacuum manifold (Macherey Nagel, 52313 Duren, Germany). Individual FFA (from C2:0 to C18:3) were quantified by GC using a Varian 3800 gas chromatograph with flame-ionization detection, a Varian 1079 universal capillary injector, a Varian 8410 liquid auto sampler, and Varian Star operating software (Varian Analytical Instruments, Harbor City, CA). A fused-silica semicapillary column BP21 (SGE Europe, Courtaboeuf, France; 25 m x 0.53 mm x 0.5 µm film thickness) was used. Direct cold on-column injection took place at 65°C, the oven temperature increased from 65°C to 240°C at a rate of 10°C/min, and was then held at 240°C for 20 min. The flame-ionization detection temperature was 250°C. The flow rate of the carrier gas (hydrogen) was 9.7 mL/min at 65°C.
Quantification of Volatile Compounds.
Twenty-one neutral volatile compounds were preselected on the basis of 1) previous analyses of volatile compounds in Cantal cheese (De Freitas et al., 2007) and 2) volatile compounds that were reported to be produced in cheese by yeasts, such as products from the catabolism of methionine and of branched-chain AA, ethanol, and ethyl esters (Spinnler et al., 2001; Arfi et al., 2004; Leclercq-Perlat et al., 2004). Selected compounds included 6 alcohols (ethanol, propanol, 2-butanol, 2-methyl-1-propanol, 3-methyl-butanol, and 2-methyl-butanol), 7 ketones (2,3-butanedione, 2-butanone, 2-pentanone, 2,3-pentanedione, 3-methyl-2-pentanone, 2-heptanone, and 2-nonanone), 5 esters (ethyl acetate, ethyl butanoate, ethyl hexanoate, ethyl octanoate, and 3-methylbutyl acetate), 2 aldehydes (ethanal and 3-methyl-butanal), and 1 sulfur compound (dimethyldisulfide). These selected neutral volatile compounds were extracted, identified, and quantified using a first dynamic headspace (DH)-GC-MS method. The DH-GC-MS equipment comprised a Tekmar 3000 headspace device (Tekmar Inc., Cincinnati, OH) coupled to a GC HP5890A (Agilent Technologies, Massy, France) and MS HP 7972A (Agilent Technologies). A 20-g sample was homogenized with 80 g of refrigerated solution of boiled distilled water by mixing for 2 min at 20,500 rpm using an Ultraturrax blender. A 7-g sample of this cheese homogenate (± 0.05 g) was used for each headspace GC-MS analysis, performed in duplicate. Samples were purged with ultrapure helium gas (35 mL/min) at 65°C for 15 min to isolate headspace volatiles, which were adsorbed on a Vocarb 3000 trap (Supelco, Bellefonte, PA). The trapped compounds were thermally desorbed at 250°C over 4 min and cryofocused at –100°C before being injected by heating at 270°C. They were separated on an Agilent Technologies HP5 capillary column (60 m x 0.32 mm x 1.0 µm film thickness) as previously described (Thierry et al., 2005b). Peaks were quantified by the areas of the total ion current (TIC). To avoid the approximation related to the commonly used internal standard calibration, volatiles were quantified (ng/g) using an external calibration method based on the addition of standards to the cheese homogenate (Thierry et al., 2005b). Volatiles were identified by comparison of mass spectra and retention times with those of authentic standards purchased from Sigma-Aldrich (St. Quentin Fallavier, France). Ethanol was detected by DH-GC-MS, but it was quantified by an enzymatic method through the use of Boehringer kits (Ref. 10176290).
Determination of Odorous Compounds
Extraction of the Volatile Compounds.
For 8-way GC-olfactometry (8W-GC-O) analyses, volatile compounds were extracted using a second DH-GC method. This DH-GC method was used in parallel to identify the candidate chemical entities to account for the odorous zones previously detected by 8W-GC-O, by using an independent single GC-O/MS set-up. A 3-g aliquot of grated cheese was placed on glass wool at the bottom of a Pyrex extractor (M3, Maillères, Aubière, France) of a Tekmar headspace device (Tekmar Inc.). The extractor was maintained at ambient temperature and purged for 60 min with a helium stream at a flow rate of 60 mL/min (Air Liquide, He/U purity: 99.995%). The trap (Tenax TA 60/80, Supelco) was operated at 30°C. The dry purge step was set at 3 min. The volatile components were then desorbed from the trap at 180°C for 10 min using helium (Air Liquide, He/N55 purity: 99.9995%) and cryofocused at –150°C before splitless injection at 220°C.
8W-GC-O.
The volatile components were separated on an RTX-5 capillary column (60 m x 0.53 mm x 1.5 µm film thickness, Restek, Evry, France). The chromatograph oven temperature was programmed as follows: 5 min isothermal at 40°C, 4°C/min increase to 205°C, and 5 min isothermal at 205°C. The duration of the GC-O analyses was 2,100 s; the helium flow rate was 8 mL/min. The 8W-GC-O device previously described in detail was designed to distribute the volatile components separated by chromatography to 8 sniffing ports synchronously and evenly (Berdagué et al., 2007). Thus, 8 sniffers can detect the substances eluted in a single chromatographic separation simultaneously and under identical conditions, thereby improving the quality of the data obtained. The compounds separated by GC (Hewlett Packard 4890D, Agilent Technologies) were transferred via a transfer line heated at 210°C from the end of the capillary column to the effluent divider of the 8W-GC-O device. The total olfactory signal (TOS), corresponding to the overall response of the sniffer panel, was expressed by the sum of the intensities of each individual aromagram from sniffer 1 to sniffer 8 against time. To study the information in the vocabulary items, the VIDEO-Sniff (Berdagué and Tournayre, 2004) method was used, which breaks up the TOS into olfactory classes. The deconvolution of the olfactory signal to classes facilitates a consensual description of the odor of the olfactory peaks.
Identification of the Odorous Compounds.
To identify the candidate chemical entities to account for odorous zones, an independent single DH-GC-O/MS setup composed of a chromatograph (GC 6890, Agilent Technologies), a mass detector (MSD 5973 Inert, Agilent Technologies), and a homemade sniffing port was used. The split between sniffing port and MS was 1:1. The purge and trap and the chromatographic conditions were the same as those used for 8W-GC-O, except for the capillary column used (RTX-5, 60 m x 0.32 mm x 1.0 µm film thickness, Restek) and the helium flow rate (2 mL/min). The volatile components were ionized by electronic impact at 70 eV. Ions were scanned over the range 18 < m/z < 220 amu and identified using databases of spectra (NIST/EPA/NIH 2005 V2.0d http://www.nist.gov/srd/nist1a.htm; Masslib 1999, MSP Köfel, Koeniz, Switzerland), relative retention indices, and odors (Flavor-Base 2004, Leffingwell and Associates, Canton, GA; Flavornet 2006, www.flavornet.org; LRI & Odor Database 2006, www.odour.org.uk).
The computer coupling between the 8W-GC-O set-up and the independent GC-O/MS system made it possible to obtain an alignment that was sufficiently precise for candidate structures to be matched with olfactive peaks. This coupling was obtained by importing the aromagram of the TOS (in the netCDF exchange format) into the mass spectrometry software MSDChem Agilent D.01.02 (Agilent Technologies), to place the TOS in parallel with the TIC or with specific ions. Before export, the 8W-GC-O aromagram retention time was resynchronized with the independent GC-O/MS aromagram/chromatogram retention time using olfactory standards (occurring naturally in the mixture or added specifically). To localize the standards accurately by mass spectrometry, a search based on specific ions was necessary, because of the frequent occurrence of coelutions. The best alignment was calculated using a third-order polynomial regression model between the start of the odor-active peaks and the start of the mass spectrometry peaks.
Flavor Assessment
Descriptive Evaluation by Trained Panel.
Sensory evaluations of ripened cheeses were performed by Les Maisons du Goût (Bourgen-Bresse, France) as previously described (Thierry et al., 2005a). A panel of 18 highly experienced trained assessors, selected for their ability to memorize various odor and flavor attributes, participated in the sensory analysis. The cheeses were evaluated in duplicate by each assessor, at an interval of 2 d. On the day of assessment, the core cheese samples were cut into 30-g sticks, placed in disposable flasks, and allowed to warm to 15°C for 2 h before evaluation. Cheese odor (nasal perception) and flavor (defined as the global perception of aroma and taste when samples are put in the mouth) intensities were rated from 0 (none) to 10 (intense) by each panelist. Four attributes of taste (salty, sweet, acidic, and bitter) and 53 attributes of odor (nasal perception) and aroma (retronasal perception) were used to describe Cantalet cheeses. For each of these attributes, the assessors indicated whether the corresponding odor, aroma, and taste were perceived or not, allowing a frequency of perception (number of "presence" responses out of the total number of responses; i.e., 18 assessors x 2 replicates = 36) to be calculated and expressed as a percentage. The numbers of attributes cited to describe odor and flavor were also determined. This method of sensory evaluation was devised to show a maximum of odor and aroma attributes, whereas classical sensory evaluation by scaling the intensity of odor/aroma attributes considers only a limited number of attributes. This method proved useful to compare the impact of different starters or to show the sensory particularities of Protected Denomination of Origin cheeses (J. F. Clément, Les Maisons du Goût, Bourgen-Bresse, France; personal communication). Data were collected and analyzed using FIZZ software (Biosystèmes, Couternon, France).
Statistical Analyses
Averaged concentration data of duplicate analyses of each compound were used for statistical analysis. One-way ANOVA were performed by using the GLM procedure of Statgraphics Plus (Statistical Graphic Corp., Englewood Cliffs, NJ) to determine the effect of the addition of yeasts on the concentration data of each compound. The Fisher’s least significant difference (LSD) test was used to determine which means were significantly different at a 95% confidence level.
In the same way, the Fisher test was used to determine which of the olfactory zones of the 3 TOS corresponding to the 3 cheese treatments were significantly influenced by the addition of yeasts. An ANOVA was performed in each of the 2,100 scans (Si) of the aromagrams within this model: Si = µ + yeast effectp + judge effectj + 
i, p, j, n (with µ = average effect, p = 3, j = 8, and n = 2). The 2,100 ANOVA thus calculated made it possible to build an "ANOVAgram" to identify the zones of aromagrams that were most influenced by the treatment (here the addition of adjunct yeasts).
Sensory data were analyzed by using FIZZ software (Biosystèmes). The ability of odor and flavor intensity scores and richness to discriminate between cheeses was investigated using 1-way ANOVA. The Newman Keuls test was used to determine which means were significantly different at a 95% confidence level. Only odor/aroma attributes having a perception frequency over 10% for at least 1 cheese, were subjected to statistical analyses. Perception frequencies lower than 10% (less than 4 presence responses out of 36) can be considered as the noise of the method. The statistical significance of differences between the frequencies of this attribute in the 3 cheeses was investigated using a 
2 test, by using STATITCF version 5 (ITCF, Paris, France).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
shows the changes occurring in the starter and adjunct populations. Total mesophilic bacteria counts were 2.9 and 2.7 log10 cfu/mL in pasteurized creams, and 1.1 and 1.4 log10 cfu/mL in microfiltered milks on the 2 d of manufacture, respectively. Yeasts were not detected in microfiltered milks or in pasteurized creams. As expected, after addition of the starter and adjuncts, the population of yeasts was 5.11 ± 0.09 and 5.07 ± 0.03 log10 cfu/mL in Kl and Pf cheese milks, respectively, whereas yeasts were not detected in the control cheese milks. Yeasts counts were stable during manufacture (5.87 ± 0.04 and 5.80 ± 0.07 after maturation at T14h, and 5.99 ± 0.19 and 5.94 ± 0.08 log10 cfu/g at T3d in Kl and Pf cheeses, respectively). During the ripening period, yeast populations decreased to 4.1 and 4.7 log10 cfu/g in Kl and Pf cheeses at 45 d, respectively, and then increased to 5.4 and 5.0 log10 cfu/g at 95 d. All the colonies of yeasts on agar plates prepared from Kl and Pf cheese samples had the same morphotype, showed the same morphology under microscopic observation, and corresponded to those of K. lactis and P. fermentans, respectively. The dominant populations of yeasts in Kl and Pf cheeses thus represented the added yeast species throughout the ripening period. In contrast, colonies of different sizes, colors, and aspects were observed on the oxytetracycline glucose agar plates made from the control cheeses. These contaminating yeasts probably originated from the plant environment, and increased in cheese during acidification to reach 2.7 log10 cfu/g at T3d. Thereafter, yeast counts of the control cheeses decreased to 1.8 log10 cfu/g at 45 d. Some significant differences of the populations of lactococci and thermophilic lactobacilli (P = 0.02 and P = 0.01, respectively) in the 3 cheeses were observed at 45 d of ripening (Figure 1). Hence, the cultivable populations of lactococci decreased by 81% in the control cheeses during the 45-d ripening period, vs. 16 and 57% decreases in Kl and Pf cheeses, respectively.
|
), demonstrating that the addition of yeasts did not modify the characteristics of the cheese. All cheeses showed the same changes over the ripening: increase in TS and fat contents due to loss of moisture, and increase in pH, acidity, and FAA concentrations due to proteolysis (Table 1). Five individual FAA (Pro, Arg, Orn, Phe, and Leu) showed slight but significant differences of concentrations between the control and the experimental cheeses at 45 d of ripening only. The control and Kl cheeses had 11 to 16% greater concentrations of Pro, Phe, and Leu than did Pf cheeses (P < 0.05). Arginine was at 3- to 5-fold greater concentrations in the control cheeses than in Kl and Pf cheeses (P < 0.01), and Orn was at a 14% greater concentration in Kl cheeses than in the other cheeses (P = 0.02; data not shown). The Cantalet cheese composition values were in the range of those of Cantal cheese, as expected (De Freitas et al., 2005).
|
). The 3 Cantalet cheeses also contained VFA derived from carbohydrate (acetic acid) and branched-chain AA (methylpropanoic and methylbutanoic acids) catabolism. Methyl-branched acid concentrations were about 2-fold greater in both experimental cheeses but these differences were not statistically significant (Table 2).
|
) were present at significantly greater concentrations in the Kl cheeses at 95 d compared with the control and Pf cheeses. Ethanol was produced early during manufacture, whereas most of the other compounds increased in concentrations throughout ripening. Esters showed the greatest change in concentration.
Odorous Compounds Analysis
Thirty-three odors were perceived by using 8W-GC-O. Half of the odors perceived (16 of 33) could be classified in "fruity-floral" and "lactic-cheesy" odor families (9 and 7 occurrences, respectively). In total, 30 of 33 odorous compounds were identified with a high reliability, and their odor attributes, as described in literature and databases, were in agreement with 8W-GC-O data (Table 3). The total olfactory signal (TOS, sum of the 16 individual aromagrams) of the control and the 2 experimental cheeses were compared to investigate the effect of yeast adjuncts on the olfactory characteristics of Cantalet cheese (Figure 2). The Fisher test values revealed that 12 zones of the TOS differed significantly (P < 0.01) between the 3 cheeses. In most cases, the Kl cheeses showed a more intense olfactory signal than the other cheeses (Figure 2). Ten of these 12 olfactory signals were related to 3 acids, 4 aldehydes, 2 alcohols, and 1 ester (Table 3). Regarding the 2 remaining olfactory zones, the fruity odor (peak 17) likely corresponded to one or several (coeluted) esters, and the empyreumatical odor (peak 32), described as caramel/toasted bread/roasted, could not be identified. To further confirm the link between the olfactory peaks and the 10 identified volatile compounds, peak areas of the corresponding DH-GC-MS chromatograms were compared for the 3 cheeses (Table 4). Eight of the 10 volatile compounds were effectively detected at greater concentrations in the Kl cheeses than in the other cheeses (Table 4), as expected from the analysis of olfactometry data (Figure 2).
|
|
|
), but more attributes were used to described the odor of Kl cheeses (P = 0.060), and the flavor of both Kl and Pf cheeses (P = 0.053; Table 5). Twenty-five and 26 out of the 53 odor and aroma attributes used to describe the flavor of Cantalet cheeses were cited at a frequency > 10% (shown in Figure 3). The attributes used frequently to describe the odor of cheeses were "butter", "boiled milk", "whey", "lemon", and "fresh milk", which represented more than 30% of the total citations for at least 1 trial. The attributes used frequently to describe the taste and aroma of cheeses were "acidic", "boiled milk", "salty", "butter", "bitter", "whey" and "lemon", which represented more than 38% of the total citations. Cheeses significantly differed in the frequency of only few attributes (Figure 3). The control cheeses were characterized by "apricot/peach" (P = 0.02) odor and "hazel-nut" odor (P = 0.02) and aroma (P = 0.08). The Kl cheeses were characterized by "acetaldehyde" and "alcohol" odors (P < 0.001). At lower levels of significance, Kl cheeses differed from the other cheeses by "yeast" (P = 0.12) odor. The Pf cheeses had a higher "acetic" (P = 0.13) and "potato" (P = 0.12) aromas and a lesser "malty" (P = 0.04) aroma than the control and Kl cheeses (Figure 3).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The populations of K. lactis and P. fermentans yeasts were stable at approximately 6 log cfu/g during cheese manufacture and showed variations of 1.5 log cfu/g over the ripening. The 2 yeast strains used as adjunct cultures in the present study were isolated from the dominant yeast population of a Cantal cheese containing 5 x 104 yeasts/g. Therefore, their lack of growth was unexpected and may result from their inability to adapt their metabolism to cheese conditions after being grown in a glucose-yeast extract medium and directly inoculated in cheese milk. The growth of yeasts in cheese is highly dependent on the environmental conditions and on the species used. In curd slurries, for example, K. lactis grew from 5 x 105 to 1 x 108 cfu/g over 5 d of incubation (Arfi et al., 2004), whereas in Raclette cheese, the population of viable cells of 4 species of adjunct yeasts (Galactomyces geotrichum, Pichia jardinii, Y. lipolytica, and D. hansenii) decreased from 105 to 106 cfu/g at the molding step to < 102 cfu/g after 90 d (Wyder et al., 1999). In contrast, indigenous yeasts grew in the control cheeses during the manufacture and reached 2.7 log cfu/g, as reported in Cheddar cheeses, where populations up to 6 log10 cfu/g in 37-d-old Cheddar were detected (Welthagen and Viljoen, 1999).
Starter cultures may have benefitted from the presence of yeasts at the beginning of ripening, because cultivable lactococcal populations remained greater in both Kl and Pf cheeses than in the control cheeses at 45 d of ripening, in agreement with results reported in Cheddar cheese with D. hansenii or Y. lipolytica added as adjunct cultures (Ferreira and Viljoen, 2003). A synergistic effect between lactic acid bacteria and many species of yeasts including K. lactis, which resulted in a prolonged survival of lactic acid bacteria, has also been observed in cheese slurries (Arfi et al., 2004).
The presence of yeasts did not modify gross cheese composition (Table 1), which can be explained by the fact that the adjunct yeasts did not grow in Cantalet cheeses, and thus did not modify the rate of acidification. Moreover, the presence of yeasts did not modify the level and the specificity of lipolysis, and had only a limited impact on the concentrations of FAA. In contrast, the presence of K. lactis in Cantalet cheese induced large variations in the concentrations of some volatile odorous compounds. This effect was particularly marked for ethanol, ethyl esters, some aldehydes, branched-chain alcohols (Table 2), branched-chain acids, and 1-octen-3-ol (Figure 2, Table 4). Ethanol was formed at concentrations 200 µg/g in Kl cheeses [vs. 40 µg/g in the control cheeses and in raw milk Cantal cheese (De Freitas et al., 2005)]. Such high values are observed in Emmental cheese containing obligatory heterofermentative lactobacilli as adjuncts (Chamba, 2000). Ethyl esters were formed at concentrations approximately 5-fold greater in the presence of K. lactis, and up to 30-fold greater for ethyl acetate (Table 2). Moreover, a branched-chain ester (3-methylbutyl acetate) was also detected in Kl cheeses only. Only one strain of K. lactis was tested in the present study, but the results observed were in general agreement with previous reports regarding the use of K. lactis as adjunct. Hence, high levels of ethyl acetate were detected in cheese slurries containing K. lactis (Arfi et al., 2004). Various esters including ethyl acetate, ethyl butanoate, and 3-methylbutyl acetate were also produced by K. marxianus in cheese slurries (Leclercq-Perlat et al., 2004). The formation of esters in Kl cheeses can result directly from the ester-forming activity of K. lactis, because yeasts, and particularly K. lactis, are known to produce esters by the acylation of an alcohol by an acylCoA catalyzed by alcohol acyltransferases (Mason and Dufour, 1985). Alternatively, the formation of greater levels of ethyl and 3-methylbutyl esters in Kl cheeses could have simply resulted from the presence of greater concentrations of the corresponding alcohols (ethanol and 3-methylbutanol), because the concentration of alcohols can be a limiting factor of ester synthesis in cheese (Thierry et al., 2006). Branched-chain compounds were detected at concentrations from 2- (acids) to 20-fold (alcohols) greater in Kl cheeses than in the 2 other cheeses, in agreement with previously reported results for K. lactis or D. hansenii added in model cheese media (Arfi et al., 2004; Leclercq-Perlat et al., 2004). Branched-chain compounds likely originate from the catabolism of branched-chain AA. Several saturated (hexanal, octanal) and unsaturated (2-pentenal, 2-hexenal) aldehydes were also detected in the present study at approximately 4-fold-greater concentrations in the presence of K. lactis, in agreement with previous observations in cheese slurries inoculated with D. hansenii (Leclercq-Perlat et al., 2004). These results were unexpected because K. lactis possess alcohol dehydrogenase activity (Bozzi et al., 1997), which could have enhanced the rate of reduction of aldehydes in Kl cheeses, as generally observed in raw milk cheeses compared with pasteurized or microfiltered milk cheeses (Beuvier and Buchin, 2004). Regarding the formation of sulfur compounds, only dimethyl disulfide was detected in the present study, and its concentration did not vary with the presence of yeasts (data not shown), and no sulfur odors were detected by GC-O. This result is in contrast with previous studies reporting that many yeast species including K. lactis are able to produce volatile sulfur compounds from Met (Spinnler et al., 2001). The ability to produce volatile compounds can also depend on the strain, although most of the effects of K. lactis on cheese volatiles shown in the present study are in agreement with previously reported results regarding the species K. lactis. The effects observed were largely dependent on the species. Hence, at the same populations of viable K. lactis and P. fermentans over ripening, only the former had an effect on cheese flavor under the experimental conditions used.
The use of the recently developed multicanal olfactometry device (8W-GC-O), in parallel with GC-MS, allowed us to identify most (29 of 33) of the odorous compounds responsible for the olfactory zones detected in Cantalet cheese. Most of the compounds are common in cheese (Maarse et al., 1994) and were previously identified in cheeses by GC-O/MS (Curioni and Bosset, 2002). Three original compounds, however, were detected: 2-pentenal, 2-hexenal, and 2-ethyl-3-methoxypyrazine. 2-Pentenal and 2-hexenal, associated with the fruity odors perceived in peaks 12 and 18, were previously detected by GC/MS in cheese and in butter, respectively (Maarse et al., 1994), but their role in cheese flavor was not previously investigated, to our knowledge. 2-Ethyl-3-methoxypyra-zine was tentatively identified as being responsible for the empyreumatic odor perceived in peak 28. Various alkyl-pyrazines and alkyl-methoxy-pyrazines have been detected in cheese, including Cheddar cheese (Curioni and Bosset, 2002; Suriyaphan et al., 2001), but not 2-ethyl-3-methoxypyrazine, to the authors’ knowledge.
The main notes of Cantalet cheese flavor appeared to be "fruity" and "lactic-cheesy", from the results of both olfactory profiles and cheese qualitative sensory evaluations. Hence, 16 of the 33 olfactory zones identified in 8W-GC-O profiles were classified in 1 of these 2 families (Table 3), whereas the main attributes of Cantalet cheese odor and aroma were "boiled milk", "butter", "lemon", and "whey" (Figure 3). Regarding the impact of K. lactis as an adjunct culture, divergent results were given by 8W-GC-O profiles and sensory evaluations of cheeses. Hence, the 8W-GC-O profiles of Kl cheeses differed very significantly from those of the control and Pf cheeses by a greater intensity of perception of 8 olfactory zones: 1 alcohol odor (ethanol), 4 fruity odors associated with esters and unsaturated aldehydes, 2 cheesy odors associated with branched-chain VFA, and 1 empyreumatical odor (peak 32; Figure 2). The ability of K. lactis to develop strong fruity olfactory notes was previously reported (Martin et al., 2002). Regarding sensory evaluations, however, K. lactis had only a low impact on Cantalet cheese flavor in the present study. The Kl cheeses differed from the other cheeses by the perception of only 2 odor attributes, "acetaldehyde" and "alcohol" (Figure 3), in agreement with the data of concentrations of ethanol and acetaldehyde (ethanal). The other differences detected by DH-GC-MS or GC-O between Kl cheeses and the other cheeses did not result in detectable differences in cheese odor and flavor. This result could be because instrumental analyses such as DH-GC-MS have, in most cases, better sensitivity and accuracy than sensory evaluations. Regarding GC-O, this method gives only a first indication of the potent flavor-active compounds of cheese, because it does not make allowance for the effects of combinations of odorous compounds (Delahunty and Piggott, 1995). Moreover, the volatility of compounds is modulated by their interactions with cheese fat and proteins, which differ in the mouth during mastication and in the headspace device. Regarding VFA, their volatility and thus their perception depends on the pH. It could explain why VFA, which were the most potent odorous compounds and differed in CG-O profiles (Figure 2), did not play a role in differentiating the flavor profiles of the whole cheeses. These results also highlight the fact that sensory evaluations are required to characterize the characteristics of cheese flavor.
In conclusion, yeasts added as adjunct cultures can influence the formation of odorous compounds and, even if the overall effects on cheese flavor were quite small in the present study, yeasts could have the potential to impact cheese flavor.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication February 21, 2007. Accepted for publication October 4, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. J. Schingoethe, K. F. Kalscheur, A. R. Hippen, and A. D. Garcia Invited review: The use of distillers products in dairy cattle diets J Dairy Sci, December 1, 2009; 92(12): 5802 - 5813. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
