Introduction

Manchego cheese is a high-fat pressed cheese with protected denomination of origin status, and it is made from raw or pasteurized ewe’s milk from the Manchega breed in “La Mancha” region. Its production constitutes almost 50% of the total production of ewe’s milk cheese in Spain, which reveals the popularity of this kind of cheese. The production of artisan cheeses (made with raw milk) is very much appreciated in South Europe because the final product is unique, very typical and has a long tradition. There is also a general agreement that cheeses made from raw milk develop a stronger flavor than cheeses made from pasteurized milk. Pasteurization influences the biochemistry of cheese ripening altering the indigenous flora and inactivating, partially or completely, certain enzymes of the milk [1].

In recent years a number of studies have been performed to study the cheese flavor and the agents responsible for the production of sapid components. McSweeney and Sousa [2] have published an excellent review about the principal biochemical pathways for the production of flavor compounds in raw and pasteurized milk cheeses during ripening. Different authors have studied the volatile composition and sensory characteristics of different types of cheeses including Manchego cheese [3–8]. However, studies about the contribution of water-soluble compounds to the taste of different cheeses are limited. Engels and Visser [9] suggested that peptides and amino acids from the water-soluble fractions of different types of cheeses were responsible for the basic taste of the cheese, although the exact role of these compounds in the flavor was not clearly stated [10]. Molina et al. [11] studied the contribution of small peptides, free amino acids (FAA) and volatile compounds to the flavor of cheeses made from different milk species. These authors found differences based on the origin of the milk. It was suggested that bovine milk cheeses were mainly salty and sour, those made from ewe’s milk had predominantly umami taste, and caprine milk cheeses were umami, astringent and bitter. The highest flavor intensity was found in those fractions with the highest concentration of volatile compounds. More recently, Taborda et al. [12] reported that water-soluble fractions of different Spanish cheeses were characteristic of each variety of cheeses.

However, in spite of the published data on the importance of peptides, amino acids and volatile compounds have on cheese flavor, little is known about the biochemical differences between the pasteurized and raw ewe’s milk cheeses. To shed some light on this fact, we have investigated the relationship between chemical and sensory analysis of the compounds of the water-soluble extract with molecular weight lower than 1,000 Da (WSE < 1,000 Da) from Manchego cheese at different stages of ripening and the influence of the pasteurization process. For this purpose, fractions were first isolated by gel filtration from the water-soluble extracts of different samples of Manchego cheeses. Later these fractions were subjected to sensory evaluation and biochemical analysis in order to elucidate the contribution of small peptides, FAAs and volatile compounds to their sensory properties.

Materials and methods

Cheese samples

Protected denomination of origin Manchego cheese samples (artisan and industrial) with 4 and 8 months of ripening were purchased in a local market. Artisan cheeses were made of raw milk and industrial cheeses were made of pasteurized milk. The samples were analyzed in duplicate.

Isolation and fractionation of the water-soluble extract with molecular weight lower than 1,000 Da (WSE < 1,000)

The WSE < 1,000 was obtained according to Salles et al. [13]; briefly: 50 g of cheese were homogenized with 100 mL of water, and maintained in a water bath at 40°C for 1 h. After centrifugation (3,800×g; 20°C, 30 min) the supernatant (WSE) was filtered and ultracentrifugated (100,000×g; 20°C, 30 min). Ultrafiltration was carried out in an ultrafiltration cell (Millipore Corp., Bedford, MA, USA) at 4°C with a 1,000-Da cut-off cellulose membrane YM1 (Millipore Corp.) to obtain the WSE < 1,000 Da. It was then fractionated on a Superformance column (Pharmacia, Uppsala, Sweden) Sephadex G10. The separation was made with Milli-Q water (pH 5.5) as eluent with a constant flow of 2 mL/min. The detection was carried out at 280 nm in a 2138 Uvicords detector (LKB, Uppsala, Sweden). Five fractions, corresponding to the elution profile shown in Fig. 1, were automatically collected.

Fig. 1
figure 1

Sephadex G-10 gel permeation chromatogram of the water-soluble extracts with molecular weight lower than 1,000 Da (WSE < 1,000) of a 4-month-old pasteurized Manchego cheese. Column Superformance (1.6 × 60 cm) flow rate, 1 mL min−1 solvent, Milli-Q water temperature, 20°C detection at 280 nm

Full size image

Determination of ions

Inorganic phosphorus, chloride, and calcium ions were estimated in each fraction by diagnostic kits (Sigma chemical, St Louis, MO, USA) using a spectrophotometer UV-120-01 (Shimadzu corp., Kyoto, Japan) at 340, 460 and 660 nm, respectively.

Volatile compounds analysis by purge- and trap gas chromatography–mass spectrometry (P&T-GC/MS)

Separation and identification of volatile components from WSE < 1,000 Da of cheeses were performed using a purge- and trap gas chromatography–mass spectrometry procedure (P&T-GC/MS) [14], using an automatic P&T concentrator 7975A (Hewlett Packard, Palo Alto, CA, USA). The concentrator was coupled by means of fused silica line to a GC (HP 5890), equipped with a quadrupole mass detector HP-5971A operating in electron impact mode at 70 eV. Helium was used as carrier gas through all the system. Ethyl pentanoate was used as internal standard. Peaks were identified from retention times and mass spectral data from the Wiley Library [15] and confirmed by using standard compounds when available. Semi quantitative values expressed as arbitrary units were calculated from peak areas of volatile compounds and internal standard. Since the aim of these analyses was to compare samples, differences in response factor and recovery were not taken into account.

Free amino acids analysis

Analyses of FAA were carried out by HPLC of the o-phthalaldehyde (OPA) derivatives [16]. Derivative formation was performed automatically and separations were carried out on a Nova Pack C18 precolumn (Waters Corp., Mildford, MA, USA) of 60 Å, 4 μm (20 mm × 3.9 mm) and a Nova Pack C18 column (Waters) of 60 Å, 4 μm (150 mm × 3.9 mm). Quantitative analysis was performed using a calibration curve for each amino acid obtained from a master solution of amino acids (Sigma, St Louis, MO, USA) to which glutamic acid (Glu), asparagine (Asn), β-alanine (β-Ala), α-alanine (α-Ala), γ-amino butyric (GABA), tryptophan (Trp), ornithine, and histamine were added.

Peptide analysis by HPLC

Separation of peptides was performed following the method described by González de Llano et al. [17]. Separations were performed at room temperature on a C18 Novapack, 4 μm, 60 Å column (150 mm × 3.9 mm) (Waters Corp., Mildford, MA, USA). For the partial identification of the peptides, the protocol proposed by Bartolomé et al. [18) was used. Commercial standards of the aromatic amino acids tyrosine (Tyr), phenylalanine (Phe) and tryptophan (Trp) (Sigma Chemical Company, St Louis, MO, USA) were run on the HPLC system to identify their retention times. Total content of hydrophobic and hydrophilic peptides was estimated as the sum of the peak areas eluted after and before Trp, taking the retention time of Trp as criteria to differentiate hydrophobic (eluted after Trp) and hydrophilic (eluted before Trp) peptides [11, 12, 17].

Sensory analysis

A panel of six experienced members who were previously trained to recognize the basic tastes conducted the sensory analysis. Training sessions were carried out with the following standard solutions prepared in mineral water: 2 g L−1 NaCl (salty), 0.6 g L−1 monosodium glutamate (umami), 0.4 g L−1 aluminum potassium sulfate (astringent), 1.65 g L−1 lactic acid (sour), 35 g L−1 lactose (sweet) and 5 g L−1 l-leucine (bitter). The standard solutions were presented, as such and in mixtures, to the panelists, who were requested to identify the different tastes. The results were discussed among the members of the panel.

Then, the panelists were requested to assess the taste of the gel permeation fractions of the WSE < 1,000 Da, reconstituted in mineral water, by evaluating the presence and intensity of each basic taste (salty, umami, astringent, sour, sweet and bitter) in a scale ranging from 0 to 100. As a reference, the taste intensity scores of the standard solutions were taken as 80. The general “cheese taste” in these fractions, taking as a reference the taste of the cheese before any fractionation, was also considered. Panelists were also requested to evaluate the general flavor in the fractions, taking as a reference the guidelines of Issanchou et al. [19] and Berodier et al. [20].

Results and discussion

Elution profile

Following gel permeation chromatography (GPC) of the WSE < 1,000 Da of raw and pasteurized Manchego cheeses with 4- and 8-month-old, five different fractions (F1–F5) were collected for each cheese. The gel filtration profiles were similar and only slight variations among them were observed. As an example, Fig. 1 shows that of the 4-month-old pasteurized milk cheese. The profiles obtained in this work showed remarkable similarity with those obtained by Molina et al. [11] in cheeses made of cow’s, ewe’s and goat’s milk, and by Cliffe et al. [21] in Cheddar cheese. The poor separation observed may be attributed to the fact that cheese extraction was performed with water. Although an acidic extraction procedure could be a most promising approach for low-molecular-mass nitrogen component isolation [22], water was preferred to allow one to taste the hydro-soluble fraction.

Mineral content in the fractions

Minerals were found in fractions 1, 2 and 3, and their distribution is shown in Fig. 2. No significant differences were found between the different ripening stages, and only a slight increase in the mineral content was detected in the industrial cheeses when compared to the artisan ones. Calcium was around 0.04 and 0.07 g/100 g of dry matter (DM), phosphorus inorganic was within the range of 0.2 and 0.3 g/100 g of DM and chloride within the range of 1 and 1.4 g/100 g of DM. Ionic calcium and inorganic phosphorus were low since the most of them are in insoluble forms. Moreover, the recovery of minerals, particularly calcium, has been reported to be low after GPC due to an interaction with the gel matrix [23]. On the contrary, the main part of chloride should be recovered in the soluble fraction of cheese. Fraction 2 was the richest fraction, with a total content of ions ranging from 1,500 to 1,750 mg/100 g of DM in artisan cheeses and from 1,000 to 1,500 mg/100 g DM in artisan cheeses. F3 was also very rich in ions, mainly chloride, with all the values obtained for the different samples between 1,000 and 1,300 mg/100 g of DM.

Fig. 2
figure 2

Chloride (open square), calcium (filled square) and inorganic phosphorous (shaded square) ions (mg/100 g of dry matter) detected at 4 and 8 months of ripening in artisan (a and b, respectively) and industrial (c and d, respectively) Manchego cheeses and their distribution in each fraction

Full size image

Free amino acids analyses

The absolute amounts of FAA in the WSE < 1,000 Da and the gel permeation fractions of raw and pasteurized Manchego cheeses largely varied with the ripening time and the milk treatment (supporting information available). The proportion of each present amino acid was roughly constant. Glutamic and aspartic acid, leucine, lysine, phenylalanine, valine and isoleucine showed the highest concentrations in all the cheeses. These seven amino acids accounted for the 73.6 and 80.7% of the total amount of amino acids in raw milk cheeses at 4 and 8 months of ripening, respectively. The same amino acids were also the most abundant in pasteurized milk cheeses, although their percentages were lower than those found in raw milk cheeses (64.2 and 63.9% for 4- and 8-month-old cheeses, respectively).

Glutamic acid was the most ubiquitous FAA. The explanation for this behavior may be on the easy removal of N-terminal residues by amino peptidases as indicated by Gómez et al. [24].

The highest concentration of FAAs in 8-month-old cheeses was found in F2. It is noteworthy to say that this fraction possesses 91 and 86,5% of the glutamic acid in artisan cheeses (in the 4- and 8-month-old cheeses, respectively). Fraction 3 had values similar to those of F2. The major components of F4 were Tyr and Phe.

Peptides analyses

The reverse-phase high-performance liquid chromatography (RP-HPLC) peptide profiles showed that during the ripening process the content of peptides with molecular weight lesser than 1,000 Da increased. This increase was higher in artisan than in industrial Manchego cheeses (data not shown). The increase of the proteolytic activity during the ripening process and the formation of low-molecular-weight peptides results from the action of enzymes from milk and lactic acid bacteria have been already reported [25].

Hydrophilic peptides were the most abundant in all cheese samples, but the proportion of hydrophobic peptides increased with the ripening process being this proportion higher in pasteurized than in raw Manchego cheeses (Fig. 3).

Fig. 3
figure 3

Hydrophilic (open square) and hydrophobic (shaded square) ratio of peptides in artisan cheeses at 4 months (a) and 8 months (b) of ripening, and in industrial cheeses at 4 months (c) and 8 months (d) of ripening

Full size image

RP-HPLC profiles of the fractions showed that F2 and F3 were the richest in amino acids and peptides, while fractions F4 and F5 were poor in peptides and contained mainly amino acids Tyr and Phe. These results were later confirmed in a study carried out in our laboratory by mass spectrometric analysis [26].

Non-volatile compounds and its influence on the taste

The panelists detected higher taste intensity in the overall WSE < 1,000 Da fraction of artisan cheeses than in that of industrial cheeses. Salty taste was predominant in both of them. Umami and sour tastes were also detected while bitter and astringent tastes were perceived with lower intensity. The taste in the WSE < 1,000 Da of the cheese has been mainly attributed to the non-volatile fraction (i.e., amino acids, peptides and ions), whereas volatile compounds were responsible of the odor and flavor [12].

Tables 1 and 2 show the relationship between the taste and the content in amino acids, peptides and ions in each of the five fractions obtained from the WSE < 1,000 Da of artisan and industrial Manchego cheese at 4 and 8 months of ripening. Umami taste was predominant, being distributed in almost all the analyzed fractions. Umami taste can be attributed to high contents in Asp and Glu residues [27], and these were the main FAAs in F1 and F2 (accounting around 70–100% of the total FAA content). Nevertheless, F3 was described as umami and the concentration of Glu and Asp residues was not relevant. In this case, umami taste can arise from the peptidic fraction. Hydrophilic peptides, mainly dipeptides, and some tripeptides with Glu in the N-terminal position and peptides containing Glu–Glu in their sequences taste umami [28]. Recently, 107 different peptides were identified by HPLC–MS/MS in these fractions of Manchego cheese [25]. Some of them as EQEEL, QEEL, EINEL, containing a high number of glutamic residues in their sequences were related to an intense umami taste; however, these authors reported that in several fractions with umami characteristics no peptides responsible for this taste were detected [25].

Table 1 Tastes and scores (in brackets) and presence of the main amino acids, peptides, and minerals of the Sephadex G-10 gel permeation fractions F1–F5 from the water-soluble extract with molecular weight lower than 1,000 Da (WSE < 1,000) of raw and pasteurized Manchego cheeses with 4 months of ripening time
Full size table
Table 2 Tastes and scores (in brackets) and presence of the main amino acids, peptides, and minerals of the Sephadex G-10 gel permeation fractions F1–F5 from the water-soluble extract with molecular weight lower than 1,000 Da (WSE < 1,000) of raw and pasteurized Manchego cheeses with 8 months of ripening time
Full size table

Therefore, it is difficult to assure the real impact of FAA and peptides on the umami properties of the WSE < 1,000 Da of Manchego cheese. These results are in agreement with those reported by Salles et al. [23] in the evaluation of taste compounds in the water soluble fraction of goat cheeses in which the direct impact of the low-molecular-weight compounds on the taste of cheeses was minimised.

Salty taste was perceived with a high intensity in many of the fractions, mainly in F2 followed by F3. This perception was well correlated with the amount of ions determined in the GPC fractions. This is consistent with previous results obtained in our laboratory [11].

Bitter taste was mainly detected in F4 and F5 and it was attributed to the presence of Tyr, Phe and Trp residues in these fractions. F3 was also described to be with a bitter taste. The presence of some FAAs such as Val, Ile, Leu, Phe, Tyr and Trp may contribute to the bitter taste detected in this fraction [29]. Roudot-Algaron [28] reported that the savory characteristic of the peptides depends on their amino acid sequence. Moreover, Lemieux and Simard [30] have related the presence of hydrophobic peptides to the bitter taste, and the low proportion of this kind of peptides in the WSE < 1,000 Da of our samples may be responsible for the low intensity detected.

Sour taste was detected in some fractions and it was not specifically related to any of the observed compounds. It is possible that the presence of Asp and Glu residues or the presence of other soluble compounds such as lactic acid, that are obtained during the extraction of the WSE < 1,000 Da, may contribute to the sour taste.

Volatile compounds and its influence in the flavor

About 35 peaks were identified in the chromatographic profiles obtained for volatile compounds of WSE < 1,000. They were mainly ketones (methylketones from 3 to 7 carbon atoms), alcohols (ethanol, 2-butanol, 3-methyl-propanol, 1-butanol, 3-methyl-butanol), aldehydes (from C2 to C5, mainly 2-methyl-butanol and 3-methyl-butanol), esters (methyl and ethyl esters from 3 to 8 carbon atoms), sulfur compounds (dimethyldisulfure, DMDS), diacetyl, and others. The distribution of volatile compounds through the five GPC fractions of each cheese is presented in Tables 3 and 4. The flavors were mainly those corresponding to the lactic family. The concentration of volatiles was clearly higher in the cheeses made from raw milk than in those made from pasteurized milk; alcohols were especially abundant in the former. Volatiles were also higher in the cheeses ripened for 8 months than in those ripened for 4 months. Ketones and esters increased more than alcohols during the ripening time; diacetyl and DMDS also noticeably increased.

Table 3 Main groups of volatile compounds in WSE < 1,000 Da fractions (F1–F5) of cheeses at 4 months of ripening time prepared from raw and pasteurized milk
Full size table
Table 4 Main groups of volatile compounds in WSE < 1,000 Da fractions (F1–F5) of cheeses at 8 months of ripening time prepared from raw and pasteurized milk
Full size table

The scores obtained for each descriptor in every fraction of 4-month-old cheeses are shown in Table 3. Notes of fresh milk and floral were clearly perceived in these cheeses, along with other from heated milk and cheesy. Fresh notes could be related to the relative abundance of alcohols in these fractions, especially in those of artisan cheese. It has been described that 3-methyl-butanol presents a pleasant note of fresh cheese [11, 31].

All panelists detected very intense floral flavor (which some of them specified as rose-like flavor) in fraction 5 of artisan cheese and also in fraction 4 of industrial cheeses. Although this flavour could be, at least partially, related to the presence of some esters, additional work was undertaken in our laboratory to identify the compounds responsible for this flavor [32]. Some other flavors as bitter, pungent, chocolate and animal-like were found in the tasted fractions.

In 8-month-old cheeses flavor was also more intense and more complex than that of cheeses ripened for only 4 months (Table 4). Fresh and floral notes decreased and the scores of descriptors as acid, pungent, cheesy or heated milk were higher. These results are probably related to the higher complexity of the volatile fraction detected in 8-month-old cheeses compared to that of 4-month-old cheeses. Methylketones and DMDS were especially abundant in some fractions (for example, F2 in raw milk cheeses). They could contribute to the increase of cheesy notes. Methylketones probably contribute to background aroma, without standout notes.