Effect of Starter Cultures on the Ripening Properties of Yak Milk Cheese

Abstract

Three batches of yak milk cheese were made using lactococcal culture (L culture) (Batch L), sour yak milk culture (S culture) (Batch S), as well as the mixed L and S cultures (1:1) (LS culture) (Batch LS). The effects of these starter cultures on cheese ripening (90 d) properties, including proteolysis, texture, and volatile production, as well as the sensory characteristics of cheese were evaluated. The results showed that cheese composition was not significantly affected by the different cultures. Batch S and LS cheeses had higher proteolytic activity with a softer and less chewy structure than Batch L cheese. More volatile compounds were detected in Batch LS cheese than in the other two batches and the former was the only one to contain decanoic acid, 2-ethyl-1-hexanol, ethylbenzene, and styrene. However, 4-methylphenol with cowy aroma, which was observed in Batches L and S cheeses, was not detected in Batch LS cheese. Bacterial identification of S culture revealed a complex microbial composition, including 22 different strains from seven lactic acid bacterial species. The presence of several proteolytic and autolytic strains (Enterococcus durans etc.) in S culture might be important in the proteolytic activity, formation of texture, and characteristic volatiles of yak milk cheeses. In the sensory evaluation, Batch LS cheese was the most acceptable with the highest scores among the three batches. The results of this study suggested that a combination of commonly used lactococcal culture with traditional sour yak milk culture may be suitable for producing yak milk cheese with good acceptability.

Introduction

Compared with cow milk, yak milk has higher dry matter content and is richer in protein, fat, lactose, and minerals, with an abundance of colloidal and soluble calcium and phosphorus (Nikkhah, 2011). Yak milk can be a precious milk source for making cheese, given its increased accessibility in particular Asia regions. Cheese production with yak milk in Nepal has been considered as a viable commercial enterprise because of its high local demand and market value (Wiener et al., 2003). In India, yak milk is processed into low-fat paneer, with desirable health effects, such as decreasing the risk of arterial hypertension, coronary heart disease, and obesity (Kandeepan and Sangma, 2010). In China, yak milk has been traditionally used by Tibetan households to make milk tea, sour milk, butter, and “Qula,” a dry and grainy “cheese-like” product made with skimmed yak milk for more than 1000 years (Tan et al., 2010). Naturally fermented yak milk is particularly popular among local herdsmen and other residents probably because of its characteristic flavor, mouth-feel, and nutritional quality (Sun et al., 2010). Although yak milk cheese production remains scarce, this cheese has been considered as a highly healthy food because of its high conjugated linoleic acid (CLA) content with a potential function in preventing heart disease, cancer, and metabolic disorders (Jiang et al., 2007; Or-Rashid et al., 2008).

In cheese processing, starter cultures have important functions in acid development to hasten milk coagulation, assist in the expulsion of whey, and influence the texture and flavor of cheese products (Law, 2001; McSweeney and Sousa, 2000). Among commercially available starter cultures, Lactococcus lactis subsp. cremoris and Lc. lactis subsp. lactis are commonly used for cheese making (Flórez et al., 2008; Lane and Fox, 1996). Although commercial starter cultures might provide optimal conditions in the different stages of cheese processing, e.g., curd formation and cheese ripening, such cultures were sometimes considered less suitable for making new types of cheese that require the use of starter culture strains with different and novel properties (Delgado and Mayo, 2004; Psoni et al., 2007). Murtaza et al. (2013) studied the sensory properties of cow and buffalo milk cheddar cheese made with both commercial and indigenous lactococcal cultures. They found that milk sources, starter cultures, and ripening temperatures significantly influenced sensory characteristics. Cheeses made with indigenous cultures and ripened at higher temperatures generally received higher sensory scores than those made with commercial cultures and ripened at lower temperatures.

In this study, yak milk cheeses were made using different starter cultures, including the commonly used lactococcal culture and the traditional sour yak milk culture used for making naturally fermented sour yak milk. The effects of these cultures on the composition, proteolysis, texture, volatile compounds, and sensory characteristics of the cheeses were compared. Given that the flavoring characteristic of cheese was of particular importance in determining consumer acceptance of the product (Ziino et al., 2005), the volatile compounds formed during ripening of the yak milk cheese were analyzed. A few reports have been published on the identification of volatile compounds of yak milk cheese. In addition, the presence of microbes in the sour yak milk culture was identified for further understanding the functions of starter cultures in ripening of yak milk cheese.

Materials and Methods

Yak milk and starter cultures    Raw yak milk and sour yak milk cultures were obtained from a local herdsmen cooperative in Qinghai Province, China. Yak milk samples were analyzed in terms of their composition by using a milk analyzer (Foss 78110, Sweden). The lactococcal culture containing Lc. lactis subsp. lactis XZ3303 and Lc. lactis subsp. cremoris QH27-1 at a ratio of 1:1 was provided by Inner Mongolia Agricultural University, Hohhot, China. The sour yak milk culture was prepared under aseptic conditions by using sterile yak milk (65°C for 30 min) as the medium, which was inoculated with 1% (w/v) of the natural fermented yak milk provided by the local herdsmen cooperative and fermented at 42°C until the pH decreased to 4.7. The culture was then cooled to room temperature and stored at 4°C before use for making yak milk cheese as described below.

Manufacture of yak milk cheese    Yak milk cheese was made from raw yak milk by using a cheddar cheese procedure as described (Zhang et al., 2013). Three batches (two repetitive samples for each, 6 L/sample) of cheese were prepared with the following starter cultures at 3% (w/v): Batch L, lactococcal culture (L culture) 1:1); Batch S, sour yak milk culture (S culture); and Batch LS, mix of L and S cultures at 1:1 (LS culture). After milk coagulation and whey expelling, the curds were transferred into cheese molds and then pressed at 40 pounds overnight. After pressing, cheese blocks (about 0.50 kg) were vacuum-packaged and ripened at 4°C for 90 d at a relative humidity of 85%. The samples were periodically taken for analysis at 1, 30, 60, and 90 d of ripening.

Analysis of cheese composition and yield    Cheese samples taken at 90 d of ripening were analyzed for total protein via the Kjeldahl method (AOAC 920.123, 1990), fat via the Babcock method (AOAC 933.05, 1990), moisture via oven-drying at 102°C (AOAC 926.08, 1990), salt via the Volhard method (AOAC 975.20, 1990), and ash content via the method of Kindstedt and Kosikowski (1985). The pH of cheese slurry, prepared by blending 10 g of grated cheese in 12 mL of deionized water, was measured by using a calibrated pH meter (FE20 pH meter, Mettler Toledo, Switzerland). The actual cheese yield was calculated as weight of cheese × 100 / (weight of milk + starter culture + salt).

Assessment of proteolysis    Cheese samples taken at days 1, 30, 60, and 90 during ripening were grated, and a portion (30 g) was immediately vacuum-packaged and stored at −20°C until analysis for primary and secondary proteolysis. The primary proteolysis of cheese during ripening can be defined as the changes in beta-, gamma-, and alpha_s-caseins, as well as peptides, which are assessed by determining water-soluble nitrogen (WSN) as a percentage of total nitrogen (TN) of the total cheese samples collected during ripening. The secondary proteolysis in cheese can be defined as the changes that occur during the further hydrolysis of products generated by primary proteolysis, including the peptides, proteins, and amino acids that are soluble in the aqueous phase of cheese and are extractable as the water-soluble fraction (Rank et al., 1985).

To determine the primary proteolysis of the cheese samples, the water-soluble extract (WSE) was prepared by making a cheese slurry comprising 20 g of cheese and 40 mL of distilled water (55°C) by using a Ultra-Turrax (XHF-D, Scientz, Ningbo, China) for 2 min, as described by Kuchroo and Fox (1982), with slight modifications. The slurry was maintained at 40°C for 1 h and centrifuged at 3000 × g for 15 min at 20°C (CR21G III, Hitachi, Tokyo, Japan), after which the supernatant was filtered through Whatman No. 42 filter paper. The nitrogen contents in the filtrate were determined in duplicate via the Kjeldahl method (IDF, 1993).

The degree of secondary proteolysis was quantified by measuring the levels of 70% (w/w) ethanol-soluble nitrogen (EtOH-SN), 5% (w/w) phosphotungstic acid-soluble nitrogen (PTA-SN), and free amino acids as described below.

EtOH-SN was measured using the method of Lynch et al. (1997) and expressed as a percentage of TN. First, 20 mL of WSE was mixed with 46.7 mL of absolute ethanol. Second, the mixture was maintained at room temperature for 1 h and centrifuged at 5000 × g for 15 min at 20°C. The supernatant was filtered through Whatman No. 42 filter paper, and ethanol was removed using a rotary evaporator (RV 10DS25, IKA Labortechnik, Staufen, Germany) at 50°C under vacuum. The nitrogen contents in the concentrated solution were determined in duplicate via the Kjeldahl method (IDF, 1993).

PTA-SN was measured using the method of Tavaria et al. (2003) and expressed as a percentage of TN. First, 20 mL of WSE was mixed with 14 mL of 3.95 M sulfuric acid and 6 mL of 33.3% (w/v) PTA. Second, the mixture was maintained at 4°C overnight and centrifuged at 3000 × g for 15 min at 20°C. The nitrogen contents in the supernatant were determined in duplicate via the Kjeldahl method (IDF, 1993).

The concentration of free amino acids in the 12% (w/w) trichloroacetic acid-soluble fraction of WSE was determined by an amino acid analyzer (L8900, Hitachi, Tokyo, Japan). A standard amino acid mixture was used to calibrate the column, and norleucine was added to all samples as an internal standard. The amino acids were detected at 570 nm with the exception of proline, which was detected at 440 nm.

Analysis of cheese texture    Texture profile analysis (TPA) was performed using a TA-XT plus texture analyzer (Texture Technologies Corp., Scarsdale, NY/Stable Microsystems, Godalming, U.K.) following the method of Dabour et al. (2006) with slight modifications. Cheese samples taken at days 1, 30, 60, and 90 during ripening were cut into 15 mm cubes with a sharp knife, wrapped in plastic to prevent water loss, and maintained at room temperature for 1 h before testing. In all cases, the samples were compressed by 50% by using two compression cycles at a constant compression speed of 0.4 mm s⁻¹. TPA parameters, namely, hardness, cohesiveness, springiness, and chewiness, were determined in triplicate.

Volatile analysis    Volatile compounds of the cheese samples taken at day 90 of ripening were extracted according to the method of Lee et al. (2003) with slight modifications. The external part of the cheese was removed (1 cm). About 5 g of the finely grated samples was placed in 20 mL solid phase microextraction (SPME) vials, after which 5 mL of NaH₂PO₄ (25%, w/v) was added. The mixtures were stirred for 30 min at 50°C to accelerate the equilibrium of headspace volatile compounds between the cheese matrix and the headspace. Volatile compounds were then extracted by placing a single 1 cm × 50/30 µm divinylbenzene/carboxen/polydimethylsiloxane SPME fiber (Supelco, Bellefonte, PA) into the vial and exposing it to the headspace above the samples for 30 min at 50°C. The absorbed volatile compounds were analyzed using a GC 7890A gas chromatograph coupled to a Triple Quad 7000B MS (both Agilent, Palo Alto, CA, USA) and equipped with a Sniffer 9000 Olfactometer (Brechbühler AG, Plan-les-Ouates, Switzerland). Separations in GC were performed on DB-WAX (30 m length × 0.25 mm inner diameter × 0.25 µm film thickness; J & W Scientific, Folsom, CA). The carrier gas used was ultra-high purity helium, and the column had a flow rate of 1.2 mL min⁻¹. The oven temperature was programmed at 40°C for 3 min, increased by 5°C min⁻¹ to 230°C, and then increased by 15°C min⁻¹ to 260°C and maintained for 3 min. The temperatures of the injector and the GC-MS transfer line were 250°C and 280°C, respectively. Electron-impact mass spectra were generated at 70 eV, with an m/z scan range from 35 amu to 550 amu. The ion source temperature was 230°C. Compounds were identified according to NIST 2.0 mass spectrum libraries installed in the GC-MS equipment. GC-O was performed by three experienced panelists.

Isolation, identification, proteolytic, and autolytic activities of the microbes from sour yak milk culture    The sour yak milk culture samples were transported to the laboratory under refrigerated conditions within 24 h. One gram of each sample was dissolved in 9 mL of sterile 0.9% normal saline and 10-fold serial dilutions were prepared. A volume of 0.1 mL of appropriate dilutions was spread plated in triplicate on MRS agar (Difco, USA) and M17 agar (Becton Dickinson, USA) plates and then incubated at 37°C for 48 h under anaerobic conditions. The pure isolates were then subjected to microscopic observation, gram staining, and catalase reactions. The gram-positive, catalase-negative rods or cocci were presumed to be lactic acid bacteria (LAB) and identified to the species level by 16S rDNA sequence analysis.

The proteolytic activities of the LAB strains were determined via the o-phthaldialdehyde (OPA) method as described by Church et al. (1983) with slight modifications. A 1% (v/v) inoculum of each culture was inoculated in 10% (w/v) sterile reconstituted skim milk and incubated at 37°C for 24 h. The uninoculated 10% (w/v) sterile reconstituted skim milk was used as the control sample. A 2.5 mL sample from each tube was mixed with 0.5 mL of distilled water before 0.5 mL of 0.75 M trichloroacetic acid was added while tube contents were vortexed. The sample was held at room temperature for 10 min and then centrifuged at 3000 × g for 5 min at 4°C. Thereafter, 1 mL of supernatant was mixed with 3 mL of OPA reagent prepared by combining 25 mL of 100 mM sodium borate, 2.5 mL of 20% (w/w) SDS, 40 mg of OPA dissolved in 1 mL of methanol, and 100 µL of β-mercaptoethanol, and the volume was adjusted to 50 mL with distilled water. The absorbance was read at 340 nm using a UV spectrophotometer (U-3900, Hitachi, Tokyo, Japan). Absorbance of the control was subtracted from each reading. The results were calculated from a standard curve obtained from dilution of tyrosine in distilled water and expressed as milligrams of Tyr per liter of milk.

The autolytic activities of the LAB strains were assayed according to Ouzari et al. (2002) with slight modification. The overnight cultures of LAB cells were harvested via centrifugation at 8000 × g for 10 min at 4°C, and the obtained pellets were resuspended in sodium citrate buffer (50 mM, pH 5.5) containing 0.5 M NaCl and then diluted to OD₆₅₀ = 1.0. The autolytic activity was determined as the percentage decrease in the absorbance at 650 nm after 48 h of incubation at 37°C.

Sensory evaluation    Cheese samples from each batch were subjected to sensory evaluation after 90 d of ripening according to the method described by Madadlou et al. (2005) with slight modifications. Cheese samples were cut into 2 cm cubes and placed into identical plastic sample cups sealed with plastic lids. A random three-digit number was assigned to the samples, which were left at room temperature for 2 h and were served with spring water and unsalted crackers for palate cleansing. A trained sensory evaluation panel (n = 20, 15 females, 5 males, age from 21 to 45 years) comprised the laboratory staff and graduate students with over 150 h of training in cheese evaluation. The scorings of three attributes, including texture, flavor, and appearance, as well as the overall acceptability, were recorded on a five-point hedonic scale (1 = least liked to 5 = most liked).

Statistical analysis    Data analysis was performed using SPSS 16.0. Significant differences between treatments were tested by ANOVA. All measurements were performed in triplicate on duplicate trials. All data were presented as means ± standard deviation of means.

Results and Discussion

Yak milk cheese composition and yield    The composition of raw yak milk was determined as follows: 3.75% protein, 4.88% fat, 5.43% lactose, and 10.26% not-fat solids. The pH of the raw yak milk was 6.52 (Table 1).

Table 1. Composition of raw yak milk and yak milk cheddar cheeses

Composition	Raw yak milk	Yak milk cheddar cheese
		Batch L	Batch S	Batch LS
Fat (%, w/w)	4.88 ± 0.25	27.63 ± 3.61	26.35 ± 2.14	26.14 ± 1.62
Protein (%, w/w)	3.75 ± 0.38	24.83 ± 0.22	25.29 ± 1.04	25.72 ± 0.41
pH	6.52 ± 0.87	5.20 ± 0.04	5.24 ± 0.04	5.25 ± 0.01
Lactose (%, w/w)	5.43 ± 0.98
Solids-not-fat (%, w/w)	10.26 ± 1.12
Yield (%)		8.42 ± 0.13a	8.43 ± 0.13a	8.09 ± 0.12b
Salt (%, w/w)		3.90 ± 0.54	3.98 ± 0.61	4.23 ± 0.75
Ash (%, w/w)		5.98 ± 0.63	5.55 ± 0.64	6.05 ± 0.64
Moisture (%, w/w)		32.88 ± 1.10	35.68 ± 2.69	33.34 ± 1.42

Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials. Batch L cheese was made with L culture; Batch S cheese was made with S culture; Batch LS cheese was made with LS culture. ^ab Means in the same row followed by different superscripts are significantly different (P < 0.05).

The compositions of the three batches of cheese after 90 d of ripening were not significantly (P > 0.05) different (Table 1), despite the different starter cultures used, as reported earlier with other types of cheese (Awad et al., 2007; Randazzo et al., 2008). The pH values of all cheese samples were also not significantly (P > 0.05) different, indicating that the starter cultures possessed similar acidification capacities during cheese ripening. The yields of the three batches of cheese were 8.42%, 8.43%, and 8.09% for Batches L, S, and LS, respectively.

Assessment of proteolysis    The primary proteolysis during 90 d of ripening of yak milk cheeses in this study was found to increase significantly (P < 0.05) in all three batches, as measured by the WSN expressed as a percentage of TN (Fig. 1A). The cheese made with S culture (Batch S) showed the highest primary proteolytic activity with the ratio of WSN to TN increasing by 14.63% from day 1 (6.65%) to 90 (21.28%), followed by Batch LS with an increase of 13.94% by using the LS culture, and Batch L with an increase of 9.89% by using the L culture.

Fig. 1.

WSN/TN (A), 70% Ethanol-SN/TN (B), and 5% PTA-SN/TN (C) changes in the cheeses made with three different starter cultures: Batch L, Batch S, and Batch LS during 90 d of ripening at 4°C. Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials. ^abc Means with different lowercase letters are significantly different (P < 0.05) between each day for each type of cheese during the ripening periods. ^ABC Means with different uppercase letters are significantly different (P < 0.05) between each type of cheese for a particular day of ripening periods.

Similar to the trend for WSN, the levels of EtOH-SN and PTA-SN in all cheese samples significantly increased (P < 0.05) as ripening progressed (Fig. 1B, 1C), and significant (P < 0.05) differences were observed in the secondary proteolysis pattern among the different starter culture treatments of cheese. Batch S cheese had the highest levels of EtOH-SN and PTA-SN at days 1 (EtOH-SN 3.37%, PTA-SN 1.55%) and 90 (EtOH-SN 8.65%, PTA-SN 2.99%), followed by Batch LS cheese (EtOH-SN 2.68% to 6.58%, PTA-SN 1.11% to 2.95%) and Batch L cheese (EtOH-SN 1.11% to 3.80%, PTA-SN 0.54% to 1.58%).

The concentration of free amino acids detected in Batch S cheese was similar to that in Batch LS cheese, and both were higher than the levels detected in Batch L cheese after 90 d of ripening (Fig. 2). Glu, Leu, and Phe were found at higher levels in all three batches of cheese.

Fig. 2.

Profiles of individual free amino acids (FAAs) in cheeses made with three different starter cultures: Batch L (□), Batch S (▨), and Batch LS (▩) after 90 d of ripening at 4°C. Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials.

Texture analysis    Table 2 shows the changes in hardness, cohesiveness, springiness, and chewiness of the cheeses during 90 d of ripening at 4°C. The magnitude of these four texture parameters significantly decreased (P < 0.05) during ripening. During the first 30 d of ripening, the hardness, cohesiveness, and chewiness exhibited a fast decrease, but the springiness decreased slightly. Subsequently, all values showed a slow decrease up to the end of the ripening period. A similar trend of decrease in hardness, cohesiveness, and springiness during ripening was reported in other types of cheese (Sallami et al., 2004; Kheadr et al., 2002). In terms of hardness, no significant difference (P > 0.05) was observed among all cheese samples at day 1 of ripening. However, Batch L and LS cheeses were significantly (P < 0.05) harder than Batch S cheese after 90 d of ripening (Table 2). As regards cohesiveness and springiness, no significant (P > 0.05) difference was observed among the three batches of cheese during the whole ripening period. The trend of change in cheese chewiness during ripening was similar to that of hardness (Table 2). Beal and Mittal (2000) reported that cheese hardness was positive correlated with chewiness. Hence, the cheeses were harder and therefore chewier.

Table 2. Changes of hardness, cohesiveness, springiness and chewiness of the yak milk cheddar cheeses during 90 d of ripening at 4 °C

Texture parameter	Ripening time (d)	Yak milk cheddar cheese
		Batch L	Batch S	Batch LS
Hardness (N)	1	70.30 ± 6.82aA	74.20 ± 5.35aA	75.13 ± 10.46aA
	30	44.70 ± 3.75aB	44.63 ± 3.59aB	49.37 ± 3.50aB
	60	36.50 ± 4.07aB	40.30 ± 1.41aB	39.31 ± 6.53aB
	90	35.55 ± 2.54aB	29.18 ± 2.40bC	32.80 ± 2.22abB
Cohesiveness	1	0.80 ± 0.01aA	0.78 ± 0.02aA	0.80 ± 0.01aA
	30	0.74 ± 0.02aB	0.74 ± 0.02aAB	0.77 ± 0.01aB
	60	0.73 ± 0.01aB	0.73 ± 0.02aAB	0.74 ± 0.02aBC
	90	0.72 ± 0.01aB	0.70 ± 0.02aB	0.70 ± 0.04aC
Springiness	1	0.88 ± 0.02aA	0.88 ± 0.02aA	0.90 ± 0.01aA
	30	0.86 ± 0.01aA	0.85 ± 0.01aAB	0.88 ± 0.03aAB
	60	0.80 ± 0.03aB	0.84 ± 0.02aAB	0.85 ± 0.01aAB
	90	0.78 ± 0.02aB	0.82 ± 0.02aB	0.83 ± 0.03aB
Chewiness (N)	1	41.60 ± 4.73aA	46.68 ± 9.40aA	47.48 ± 5.91aA
	30	28.54 ± 2.47abB	26.37 ± 2.34bB	34.01 ± 2.51aB
	60	23.35 ± 3.12aB	25.57 ± 1.78aB	22.84 ± 1.47aC
	90	22.97 ± 1.84aB	16.86 ± 1.06aB	20.17 ± 1.51aC

Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials.Batch L cheese was made with L culture; Batch S cheese was made with S culture; Batch LS cheese was made with LS culture. ^ab Means in the same row followed by different superscripts are significantly different (P < 0.05). ^ABC Means in the same column followed by different superscripts are significantly different (P < 0.05).

The texture of cheese largely depends on proteolysis during ripening. As shown in Fig. 1, over the course of ripening, all three batches of cheese showed the greatest increases in proteolysis during the first 30 d, after which the process slowed down. In addition, the Batch S cheese with the lowest hardness and chewiness showed the highest proteolytic activity, but the Batch L cheese with the highest hardness and chewiness showed the lowest proteolytic activity after 90 d of ripening. The Batch LS cheese showed a medium level of proteolytic activity (Fig. 1 and Table 2). The development of cow cheese texture during ripening was suggested to include two phases: the initial softening of cheese texture attributed to the transfer of casein-bound calcium phosphate to the aqueous phase of the cheese during the early stages of ripening (two weeks to four weeks) and the subsequent and more gradual changes in cheese texture attributed to the rate of proteolysis and increase in pH (O'Mahony et al., 2005). The change in texture during the ripening of yak milk cheese in the present study seemed to follow a similar pattern as described above.

Volatile analysis    Consumers often select cheese primarily based on flavor characteristics. Volatile compounds of a variety of cheese types, such as cheddar, gouda, camembert, and parmigiano, were characterized (Burbank and Qian, 2005; Bellesia et al., 2003; Jung et al., 2013; Pionnier et al., 2002). However, few reports are available on the analysis of volatile compounds in yak milk cheese. In this study, analyses via solid phase microextraction (SPME) and GC-O-MS methods of the volatile compounds from the ripened yak milk cheese samples showed that a total of 25 volatile compounds were detected in Batch LS cheese, whereas 19 and 17 compounds were detected for Batch L and S cheeses, respectively. These volatile compounds belong to different chemical families, including five free fatty acids (FFAs), four ketones, six alcohols, eight aromatic compounds, two aldehydes, three hydrocarbons, and one sulfur compound (Table 3 and Fig. 3).

Table 3. Volatile compounds of the yak milk cheddar cheeses after 90 d of ripening

No.	Compounds	RT1	RI2	Aroma3	Yak milk cheddar cheese
					Batch L	Batch S	Batch LS	ANOVA
	Free fatty acids
1	Acetic acid	19.97	1435		Vinegar	8.980 ± 1.351a	14.443 ± 2.048b	7.660 ± 0.541a	**
2	Butanoic acid	24.39	1630	Cheesy	6.899 ± 0.332b	15.541 ± 1.653a	11.587 ± 2.367a	**
3	Hexanoic acid	29.33	1833	Pungent	16.408 ± 2.347a	15.989 ± 1.478a	16.160 ± 1.654a	NS
4	Octanoic acid	33.79	2075	Sour fruit	1.716 ± 0.123a	1.098 ± 0.237b	1.729 ± 0.145a	**
5	Decanoic acid	37.50	2300	Sour fruit	-	-	0.145 ± 0.006a	***
	Ketones
6	2-Pentanone	6.86	975	Orange peel	5.311 ± 0.156a	7.282 ± 2.647a	6.076 ± 2.034a	NS
7	2-Heptanone	12.31	1162	Fruity	4.818 ± 0.365a	-	4.056 ± 0.585a	NS
8	3-Hydroxy-2-butanone	15.85	1286	Sour milk	1.643 ± 0.054b	2.766 ± 0.087a	2.602 ± 0.651a	*
9	2-Nonanone	18.33	1387	Fruity, oral	-	1.546 ± 0.216a	1.681 ± 0.056a	NS
	Alcohols
10	Ethanol	6.02	925	Bouquet	18.252 ± 4.543a	-	8.209 ± 1.586b	**
11	3-Methyl-1-butanol	13.61	1206	Fruity	15.185 ± 3.652a	4.272 ± 0.524b	4.541 ± 1.256b	**
12	1-Pentanol	14.73	1247	Fruity	1.770 ± 0.032a	3.377 ± 0.354b	1.875 ± 0.257a	***
13	1-Hexanol	17.47	1345	Green	0.767 ± 0.017b	0.895 ± 0.058a	0.921 ± 0.045a	*
14	2-Ethyl-1-hexanol	21.01	1494	Fruity	-	-	1.147 ± 0.120a	***
15	Phenylethanol	30.86	1903	Floral	0.730 ± 0.056a	0.488 ± 0.034b	0.533 ± 0.038b	**
	Aromatic compounds
16	Benzene	5.97	924	Fragrance	-	12.042 ± 4.254a	-	***
17	Toluene	8.37	1028	Nutty, bitter	10.239 ± 3.562a	15.175 ± 1.954a	11.199 ± 1.657a	NS
18	Ethylbenzene	10.48	1122	Aromatic odor	-	-	1.632 ± 0.259a	***
19	Styrene	14.23	1243	Fragrance	-	-	1.164 ± 0.026a	***
20	1,3,5-Trimethylbenzene	14.90	1250	Fragrance	-	-	0.630 ± 0.023a	***
21	Benzaldehyde	21.84	1515	Almond	1.697 ± 0.096a	-	-	***
22	Naphthalene	27.12	1751	Faint scent	0.566 ± 0.089b	0.845 ± 0.156a	0.576 ± 0.032b	*
23	4-Methylphenol	34.16	2079	Cowy	0.383 ± 0.038a	0.285 ± 0.014b	-	**
	Aldehydes
24	Octanal	15.32	1287	Herbaceous	1.478 ± 0.984a	-	6.125 ± 1.657b	*
25	Nonanal	18.31	1385	Green	1.332 ± 0.328a	-	-	***
	Hydrocarbons
26	Undecane	9.77	1100	No	1.825 ± 0.327b	3.458 ± 0.359a	4.638 ± 0.988a	**
27	Dodecane	11.93	1200	Dirt	-	-	3.992 ± 0.255a	***
28	Tridecane	15.15	1300	No	-	-	0.856 ± 0.066a	***
	Sulphur compound
29	Dimethylsulphone	30.54	1273	Sulfur, burnt	-	0.488 ± 0.040a	0.275 ± 0.056b	**

Values (relative content, %) presented are means ± standard deviation on duplicate trials . Compounds not detected are indicated by “-”. Batch L cheese was made with L culture; Batch S cheese was made with S culture; Batch LS cheese was made with LS culture. ¹ Retention time; ² Retention index; ³ Odor description at the GC-sniffing port. ^abcd Means in the same row followed by the same letter are not significantly different (P > 0.05). NS, not significant; *P < 0.05; **P < 0.01; ***P < 0.001.

Fig. 3.

Total ion chromatogram of volatile compounds in cheeses made with three different starter cultures: Batch L (A), Batch S (B), and Batch LS (C) after 90 d of ripening at 4°C. Peak numbers correspond to those in Table 3.

The FFAs were the major volatile compounds detected in yak milk cheese samples, with higher concentrations in Batch S and LS cheeses than in Batch L. Decanoic acid, which gave a slight sour fruit odor, was found only in Batch LS cheese. Hexanoic acid (pungent aroma) had the highest concentration (∼16%) in all cheese samples and might contribute to the typical aroma of yak milk cheese in this study. Previously, hexanoic acid was identified as a characteristic odorant of Ibores cheese (Delgado et al., 2011) and Italian cheese (Innocente et al., 2013). Among the four ketones detected, 2-pentanone (orange peel aroma) and 3-hydroxy-2-butanone (sour milk aroma) were present in all the three batches of yak milk cheese, with 2-pentanone being the most abundant. However, the two ketones were not detected in the cow milk cheddar cheese made with the same lactococcal culture as that used by Wang et al. (2012). This finding suggests the possibility of these two ketone compounds comprising the typical flavor of yak milk cheese. Among the six alcohols detected, 2-ethyl-1-hexanol (fruity aroma) was detected only in Batch LS cheese. Ethanol (bouquet aroma), 3-methyl-1-butanol (fruity aroma), and phenylethanol (floral aroma) were found with relatively high concentrations in both Batch L and LS cheeses probably because of the presence of a large population of lactococci that metabolized other amino acids to produce branched-chain alcohols. As regards the eight aromatic compounds detected, ethylbenzene (aromatic odor) was detected only in Batch LS cheese, whereas 4-methylphenol was detected only in Batch L and S cheeses. The compound 4-methylphenol was considered as a main cheese odorant of British farmhouse Cheddar cheese being responsible for the “cowy-barny” note (Suriyaphan et al., 2001). The relatively low level of aldehydes found in the cheese samples of this study indicated an optimal maturation of cheese (Moio and Addeo, 1998).

Isolation, identification, proteolytic, and autolytic activities of the microbes from sour yak milk culture    A total of 22 bacterial strains were isolated and identified from the fermented sour yak milk culture based on their gram-positive and catalase-negative properties, as well as on the 16S rRNA sequence analysis. The strains include six Lactobacillus casei strains, namely, QHD71, QHD72, QHA3, QHA10, QHB2, and QHB3; two L. plantarum strains, namely, QHC42 and QHC9; five Enterococcus durans strains, namely, QHA12, QHA13, QHB32, QHB34, and QHB37; three E. faecium strains, namely, QHB31, QHB33, and QHB36; four Pediococcus acidilactici strains, namely, QHC22, QHC31, QHC52, and QHC53; one P. pentosaceus strain, namely, QHA16; and one Lc. lactis strain, namely, QHC27. The obtained sequences were deposited in GenBank and assigned with the following accession numbers: KJ629287-KJ629308.

Determination of the proteolytic activity of these strains (Fig. 4) revealed significant variations ranging from 0.17 mg of Tyr/L to 17.80 mg of Tyr/L of milk. E. durans strains QHA12, QHB34, and QHB37 showed the highest proteolytic activities of 17.80, 17.06, and 16.82 mg of Tyr/L of milk, respectively. These values were significantly (P < 0.05) higher than those of Lc. lactis subsp. lactis XZ3303 and Lc. lactis subsp. cremoris QH27-1 used as the lactococcal cultures for cheese making in this study. The autolytic activity of the isolated strains ranged from 22.56% to 81.11% (Fig. 4). E. durans strains QHB32 and QHA13 had the highest autolytic activity at 81.11% and 75.06%, respectively, followed by L. casei QHA10, E. faecium QHB33, L. plantarum QHC9, L. plantarum QHC42, and L. casei QHD72, and all values were significantly (P < 0.05) higher than those of Lc. lactis subsp. lactis XZ3303 and Lc. lactis subsp. cremoris QH27-1.

Fig. 4.

Proteolytic and autolytic activities of LAB strains isolated from fermented sour yak milk. Proteolytic activity was expressed as milligrams of Tyr per liter of milk after 24 h of incubation in 10% (w/v) sterile reconstituted skim milk at 37°C. Autolytic activity was determined as the percentage decrease in absorbance at 650 nm after 48 h of incubation in sodium citrate buffer (50 mM, pH 5.5) containing 0.5 M NaCl at 37°C. Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials.

The presence of several highly proteolytic and autolytic strains (E. durans etc.) in S culture could explain why cheese samples (Batches S and LS) made using this culture had higher primary and secondary proteolytic activities (Fig. 1) as well as higher levels of free amino acids (Fig. 2), than the cheese sample (Batch L) made only with L culture. The production of free amino acids in cheese was found to be affected by the starter culture used with respect to peptidase activity, degrees of autolysis, and permeability of starter and non-starter lactic acid bacteria (Doolan and Wilkinson, 2009; Gobbetti et al., 1999). In addition, the presence of enterococci in S culture might also have a function in increasing the concentration of FFAs in Batch S and LS cheeses (Table 3). Enterococci were earlier reported to promote the growth of lactic acid bacteria and have potential metabolic traits involved in aroma and flavor development during cheese ripening (Morandi et al., 2006).

Sensory evaluation    The average scores for the sensory evaluation of the yak milk cheeses after 90 d of ripening are shown in Table 4. The results showed that Batch LS cheese was the most acceptable with the highest score in overall acceptability among the three batches of cheese. Batch LS cheese also had higher sensory scores in terms of flavor, texture, and appearance than Batch L and S cheeses, although no statistical differences (P > 0.05) were observed in these parameters (except appearance).

Table 4. Sensory evaluation of yak milk cheddar cheeses after 90 d of ripening

Yak milk Cheddar cheese	Appearance	Texture	Flavour	Overall acceptance
Batch L	3.6 ± 0.3a	3.5 ± 0.3a	3.2 ± 0.6a	3.6 ± 0.3a
Batch S	3.5 ± 0.2a	3.6 ± 0.6a	3.5 ± 0.3a	3.8 ± 0.2a
Batch LS	4.2 ± 0.2b	3.8 ± 0.7a	3.8 ± 0.4a	4.5 ± 0.5b

Values presented are means ± standard deviation of data from triplicate analysis on duplicate trials. Batch L cheese was made with L culture; Batch S cheese was made with S culture; Batch LS cheese was made with LS culture. ^ab Means in the same column followed by different superscripts are significantly different (P < 0.05).

The sensory characteristics of cheese could be affected by many factors, including degree of primary proteolysis, concentrations of free amino acids and fat, milk sources, type of starter culture, and technological conditions (Ahmed et al., 2005; Hou et al., 2014; Drake et al., 2010). The use of the adjunct cultures of L. casei I90, and L. plantarum I91 with peptidolytic activity was shown to improve the sensory characteristics of soft and semi-hard cheeses (Milesi et al., 2009). The supplementation of cheeses with probiotic bacteria affected secondary proteolysis to increase the total free amino acid content and the formation of flavor and aromatic compounds (Cruz et al. 2009). In the present study, the combined use of L and S cultures to make yak milk cheese (Batch LS) was shown to be better than the use of L or S culture alone (Batch L or Batch S) in terms of the improvement of the sensory properties of cheese.

Conclusion

In this study, yak milk cheeses were made with commonly used lactococcal culture and the traditional sour yak milk culture. A combination of these two starter cultures was shown to be better than either of them used alone for making yak milk cheese in terms of ripening properties and sensory quality of cheese. The complex microbes (e.g., enterococci) present in the traditional sour yak milk culture apparently have important functions in promoting proteolysis and possible lipolysis as well as in forming the characteristic flavor of yak milk cheese. In addition, the volatile compounds present in the ripened yak milk cheeses were identified. More volatile compounds were found in the cheese made with the mixed LS culture than that made with the single L or S culture, which suggests possible interactions of these two cultures. Further studies will focus on the functions of bacterial strains from the sour milk culture and their possible interactions with lactococcal culture strains during cheese ripening. These interactions are important to obtain good-quality yak milk cheese.

Acknowledgments    The financial support for this work from the Natural Science Foundation of China (31371804), National Public Benefit Research (Agriculture) Foundation (201303085), and The Key Project of the Educational Committee of Beijing City (KZ201310011011) is kindly acknowledged.

References

•  Ahmed,  N.H.,  El Soda,  M.,  Hassan,  A.N., and  Frank,  J. (2005). Improving the textural properties of an acid-coagulated (Karish) cheese using exopolysaccharide producing cultures. LWT-Food Sci. Technol., 38, 843-847.
• AOAC (Association of Official Analytical Chemists) (1990). Official Methods of Analysis, 15th ed. AOAC, Arlington.
•  Awad,  S.,  Ahmed,  N., and  Soda,  M.E. (2007). Evaluation of isolated starter lactic acid bacteria in Ras cheese ripening and flavour development. Food Chem., 104, 1192-1199.
•  Beal,  P. and  Mittal,  G.S. (2000). Vibration and compression responses of Cheddar cheese at different fat content and age. Milchwissenschaft, 55, 139-142.
•  Bellesia,  F.,  Pinetti,  A.,  Pagnoni,  U.,  Rinaldi,  R.,  Zucchi,  C.,  Caglioti,  L., and  Palyi,  G. (2003). Volatile components of Grana Parmigiano-Reggiano type hard cheese. Food Chem., 83, 55-61.
•  Burbank,  H.M. and  Qian,  M.C. (2005). Volatile sulfur compounds in Cheddar cheese determined by headspace solid-phase microextraction and gas chromatograph-pulsed flame photometric detection. J. Chromatogr. A, 1066, 149-157.
•  Church,  F.C.,  Swaisgood,  H.E.,  Porter,  D.H., and  Catignani,  G.L. (1983). Spectrophotometric assay using o-phthaldialdehyde for determination of proteolysis in milk and isolated milk proteins. J. Dairy Sci., 66, 1219-1227.
•  Cruz,  A.G.,  Buriti,  F.C.A.,  Souza,  C.B.H.,  Faria,  J.A.F., and  Saad,  S.M.I. (2009). Probiotic cheese: Health benefits, technological and stability aspects. Trends Food Sci. Technol., 20, 344-354.
•  Dabour,  N.,  Kheadr,  E.,  Fliss,  I., and  LaPointe,  G. (2006). Improvement of texture and structure of reduced-fat Cheddar cheese by exopolysaccharide-producing lactococci. J. Dairy Sci., 89, 95-110.
•  Delgado,  F.J.,  González-Crespo,  J.,  Cava,  R., and  Ramírez,  R. (2011). Formation of the aroma of a raw goat milk cheese during maturation analysed by SPME-GC-MS. Food Chem., 129, 1156-1163.
•  Delgado,  S. and  Mayo,  B. (2004). Phenotypic and genetic diversity of Lactococcus lactis and Enterococcus spp. strains isolated from Northern Spain starter-free farmhouse cheeses. Int. J. Food Microbiol., 90, 309-319.
•  Doolan,  I. A. and  Wilkinson,  M.G. (2009). Comparison of the effects of various attenuation methods on cell permeability and accessibility of intracellular enzymes in Lactococcus lactis strains. Int. Dairy J., 19, 215-221.
•  Drake,  M.A.,  Miracle,  R.E., and  McMahon,  D.J. (2010). Impact of fat reduction on flavor and flavor chemistry of Cheddar cheeses. J. Dairy Sci., 93, 5069-5081.
•  Flórez,  A.B.,  Ammor,  M.S., and  Mayo,  B. (2008). Identification of tet(M) in two Lactococcus lactis strains isolated from a Spanish traditional starter-free cheese made of raw milk and conjugative transfer of tetracycline resistance to lactococci and enterococci. Int. J. Food Microbiol., 121, 189-194.
•  Gobbetti,  M.,  Lanciotti,  R.,  De Angelis,  M.,  Rosaria Corbo,  M.,  Massini,  R., and  Fox,  P. (1999). Study of the effects of temperature, pH, NaCl, and aw on the proteolytic and lipolytic activities of cheese-related lactic acid bacteria by quadratic response surface methodology. Enzyme Microb Tech., 25, 795-809.
•  Hou,  J.,  Hannon,  J.A.,  McSweeney,  P.L.H.,  Beresford,  T.P., and  Guinee,  T.P. (2014). Effect of curd washing on cheese proteolysis, texture, volatile compounds, and sensory grading in full fat Cheddar cheese. Int. Dairy J., 34, 190-198.
• IDF (International Dairy Federation) (1993). Milk: Determination of Nitrogen Content (Kjeldahl Method), (Standard 20B). IDF, Brussels.
•  Innocente,  N.,  Munari,  M., and  Biasutti,  M. (2013). Characterization by solid-phase microextraction-gas chromatography of the volatile profile of protected designation of origin Montasio cheese during ripening. J. Dairy Sci., 96, 26-32.
•  Jiang,  J.,  Chen,  S.,  Ren,  F.,  Luo,  Z., and  Zeng,  S.S. (2007). Yak milk casein as a functional ingredient: preparation and identification of angiotensin-i-converting enzyme inhibitory peptides. J. Dairy Res., 74, 18-25.
•  Jung,  H.J.,  Ganesan,  P.,  Lee,  S.J., and  Kwak,  H.S. (2013). Comparative study of flavor in cholesterol-removed Gouda cheese and Gouda cheese during ripening. J. Dairy Sci., 96, 1972-1983.
•  Kandeepan,  G. and  Sangma,  S. (2010). Comparison of quality characteristics of full fat and low fat paneer developed from yak milk. Int. J. Dairy Technol., 64, 117-120.
•  Kheadr,  E.E.,  Vachona,  J.F.,  Paquina,  P., and  Flissa,  I. (2002). Effect of dynamic high pressure on microbiological, rheological and microstructural quality of Cheddar cheese. Int. Dairy J., 12, 435-446.
•  Kindstedt,  P.S. and  Kosikowski,  F.V. (1985). Alternative test for phosphorous in cheese. J. Dairy Sci., 68, 1859-1864.
•  Kuchroo,  C.N. and  Fox,  P.F. (1982). Soluble nitrogen in Cheddar cheese: comparison of extraction procedures. Milkchwissenschaft, 37, 331-335.
•  Lane,  C.N. and  Fox,  P.F. (1996). Contribution of starter and adjunct lactobacilli to proteolysis in Cheddar cheese during ripening. Int. Dairy J., 6, 115-728.
•  Law,  B.A. (2001). Controlled and accelerated cheese ripening: the research base for new technology. Int. Dairy J., 11, 283-396.
•  Lee,  J.H.,  Diono,  R.,  Kim,  G.Y., and  Min,  D. (2003). Optimization of solid phase microextraction analysis for the headspace volatile compounds of Parmesan cheese. J. Agr. Food Chem., 51, 1136-1140.
•  Lynch,  C.M.,  McSweeney,  P.L.H.,  Fox,  P.F.,  Cogan,  T.M., and  Drinan,  F. (1997). Contribution of starter lactococci and non-starter lactobacilli to proteolysis in Cheddar cheese with a controlled microflora. Le Lait, 77, 441-459.
•  Madadlou,  A.,  Khosroshahi,  A., and  Mousavi,  M.E. (2005). Rheology, microstructure, and functionality of low-fat Iranian white cheese made with different concentrations of rennet. J. Dairy Sci., 88, 3052-3062.
•  McSweeney,  P.L.H. and  Sousa,  M.J. (2000). Biochemical pathways for the production of flavour compounds in cheese during ripening: A review. Le Lait, 80, 293-324.
•  Milesi,  M.M.,  Vinderola,  G.,  Sabbag,  N.,  Meinardi,  C.A., and  Hynes,  E. (2009). Influence on cheese proteolysis and sensory characteristics of non-starter lactobacilli strains with probiotic potential. Food Res. Int., 42, 1186-1196.
•  Moio,  L. and  Addeo,  F. (1998). Grana Padano cheese aroma. J. Dairy Res., 65, 317-333.
•  Morandi,  S.,  Brasca,  M.,  Andrighetto,  C.,  Lombardi,  A., and  Lodi,  R. (2006). Technological and molecular characterisation of enterococci isolated from north-west Italian dairy products. Int. Dairy J., 16, 867-875.
•  Murtaza,  M.A.,  Rehman,  S.U.,  Anjum,  F.M., and  Huma,  N. (2013). Descriptive sensory profile of cow and buffalo milk Cheddar cheese prepared using indigenous cultures. J. Dairy Sci., 96, 1380-1386.
•  Nikkhah,  A. (2011). Science of camel and yak milks: human nutrition and health perspectives. Food Nutr. Sci., 2, 667-673.
•  Or-Rashid,  M.M.,  Odongo,  N.E.,  Subedi,  B.,  Karki,  P., and  Mcbride,  B.W. (2008). Fatty acid composition of yak (Bos grunniens) cheese including conjugated linoleic acid and trans-18:1 fatty acids. J. Agr. Food Chem., 56, 1654-1660.
•  O'Mahony,  J.A.,  Lucey,  J.A., and  McSweeney,  P.L.H. (2005). Chymosin-mediated proteolysis, calcium solubilization, and texture development during the ripening of Cheddar cheese. J. Dairy Sci., 88, 3101-3114.
•  Ouzari,  H.,  Cherif,  A., and  Mora,  D. (2002). Autolytic phenotype of Lactococcus lactis strains isolated from traditional Tunisian dairy products. J. Appl. Microbiol., 92, 812-820.
•  Pionnier,  E.,  Engel,  E.,  Salles,  C., and  Le Quere,  J.L. (2002). Interaction between non-volatile water-soluble molecules and aroma compounds in Camembert cheese. Food Chem., 76, 13-20.
•  Randazzo,  C.L.,  Pitino,  I.,  De Luca,  S.,  Scifò,  G.O., and  Caggia,  C. (2008). Effect of wild strains used as starter cultures and adjunct cultures on the volatile compounds of the Pecorino Siciliano cheese. Int. J. Food Microbiol., 122, 269-278.
•  Rank,  T.C.,  Grappin,  R., and  Olson,  N.F. (1985). Secondary proteolysis of cheese during ripening: A review. J. Dairy Sci., 68, 801-805.
•  Sallami,  L.,  Kheadr,  E.E.,  Fliss,  I., and  Vuillemard,  J.C. (2004). Impact of autolytic, proteolytic, and nisin-producing adjunct cultures on biochemical and textural properties of Cheddar cheese. J. Dairy Sci., 87, 1585-1594.
•  Sun,  Z.H.,  Liu,  W. J.,  Gao,  W.,  Yang,  M.,  Zhang,  J.C.,  Wu,  L.,  Wang,  J.G.,  Menghe,  B.,  Sun,  T.S., and  Zhang,  H.P. (2010). Identification and characterization of the dominant lactic acid bacteria from kurut: The naturally fermented yak milk in Qinghai, China. J. Gen. Appl. Microbiol., 56, 1-10.
•  Suriyaphan,  O.,  Drake,  M.,  Chen,  X.Q., and  Cadwallader,  K.R. (2001). Characteristic aroma components of British farmhouse Cheddar cheese. J. Agr. Food Chem., 49, 1382-1387.
•  Tan,  Z.F.,  Pang,  H.L.,  Duan,  Y.H.,  Qin,  G.Y., and  Cai,  Y.M. (2010). 16S ribosomal DNA analysis and characterization of lactic acid bacteria associated with traditional Tibetan Qula cheese made from yak milk. Anim. Sci. J., 81, 706-713.
•  Tavaria,  F.K.,  Franco,  I.,  Carballo,  F.J., and  Malcata,  F.X. (2003). Amino acid and soluble nitrogen evolution throughout ripening of Serra da Estrela cheese. Int. Dairy J., 13, 537-545.
•  Wang,  C.,  Liang,  Q.,  Yang,  M.,  Qiao,  H., and  Zhang,  Y. (2012): Determination of volatile components in yak milk harda cheese by solid phase microextraction. J. Gansu Agr. Univ., 2, 141-146 (in Chinese).
•  Wiener,  G.,  Han  J. L., and  Long,  R.J. (2003). “The Yak.” Bangkok: RAP Publication.
•  Zhang,  L.,  Zhang,  X.,  Liu,  C.H.,  Li,  C. Y.,  Li,  S.Y.,  Li,  T.Z.,  Li,  D.,  Zhao,  Y.J., and  Yang,  Z.N. (2013). Manufacture of Cheddar cheese using probiotic Lactobacillus plantarum K25 and its cholesterol-lowering effects in a mice model. World J. Microb. Biot., 29, 127-135.
•  Ziino,  M.,  Condurso,  C.,  Romeo,  V.,  Giuffrida,  D., and  Verzera,  A. (2005). Characterization of “Provola dei Nebrodi”, a typical Sicilian cheese, by volatiles analysis using SPME-GC/MS. Int. Dairy J., 15, 585-593.