Experimental design

Chihuahua cheese samples were selected considering the following criteria: that the product had been made at the state of Chihuahua, vacuum packed, with labeling and manufacturer data, of easy acquisition for the consumer in convenience stores, and that its expiration date was greater than 15 days after the purchase. The Chihuahua-like cheese products were not included in the sampling. We assessed the physicochemical characteristics by triplicate and the sensory evaluations (descriptive analysis) by duplicate. The variables were: moisture content, protein, fat, ash, total carbohydrates, sodium, acidity, and pH. Sensory attributes included odor intensity, odor description, moistness, fat character and firmness in mouthfeel, and sweetness, saltiness, bitterness, and acid taste and acid-bitter, acid-sweet and sweet-bitter interactions, preference and acceptance tests.

Obtaining the samples

According to the inclusion criteria, we sampled ten brands of Chihuahua cheese obtained in four supermarkets at Ciudad Juárez, Chihuahua. From each, 3.3 kg were purchased in 0.55 or 1.10 kg presentations, all from the same batch. We transported the samples in a cooler, identified them with three-digit random codes and stored them in refrigeration (2 to 4 °C) until analysis.

Sensory characterization

Samples were sensory characterized with a descriptive analysis by a trained panel of 10 judges. The attributes in the olfactory phase were odor intensity, first impression and odor description. In the oral phase, mouthfeel characteristics were considered, such as moistness and fat character, mechanical, such as firmness, taste and interactions, such as sweet, salty, bitter and acidic, and acid-bitter, acid- and bitter-sweet (^{Chamorro and Losada, 2002}).

First, the judging panel trained in each attribute for 6 weeks, with two weekly sessions. To estimate the perception intensity of each attribute they used a 9-point linear scale, labeled at the end as “weak” and “strong”. The slices for each panelist were 10 g of cheese in ≈30 mL plastic glasses, identified with three-digit random numbers at room temperature. The samples were hand over in individual booths, in a balanced and randomized form, together with evaluation sheets. The judges rinsed their mouths with purified water (Aurrera^®) at the beginning and between samples. Each attribute evaluated in different sessions (^{Chamorro and Losada, 2002}; ^{Lawless and Heymann, 2010}).

Physicochemical characterization

The physicochemical determinations were carried out in 50 or 100 g samples; the total sample per trademark was 300 g. We sectioned the samples with a commercial cheese grater, homogenized and placed in identified plastic containers. The moisture, ash and protein content were determined following the ^{AOAC (2000)} standardized methods, 926.08, 935.42 and 920.123-1920, respectively. We then quantified the total fat content with the Gerber-Van Gulik method (^{DGN, 1984}). The total carbohydrates were determined by the difference of the total product composition (^{FAO, 2002}). The sodium content was estimated by stoichiometry of NaCl, with the Volhard method (935.43 AOAC). Acidity was determined via the volumetric method (920.124 AOAC) (AOAC, 2000) and expressed as a lactic acid percentage. The pH was determined with a potentiometer (^{DGN, 1970}).

The protein profiles of the cheese samples were evaluated by electrophoresis in polyacrylamide gels, in denatured and reduced conditions (SDS-PAGE). The gels were polyacrylamide, discontinuous with a 12 % separation gel (^{Laemmli, 1970}). We performed the protein separation with a 110 V constant voltage. The molecular mass markers were a broad range (Bio-Rad^®). The gels were then stained with Coomasie blue R-250 (Bio-Rad^®), following ^{Merril (1990)}. The protein profiles analysis were performed by densitometry using the 1D Image Analysis software (Kodak, USA) (^{Bahan, 1987}).

Consumer preferences

Data analysis

Intensity data from attributes in sensory descriptive analysis were analyzed by a repeated measure ANOVA. The physicochemical data and the acceptance test were analyzed with a one-way ANOVA, and Fisher’s multiple comparisons in both cases. For the ANOVA, the data were previously analyzed with the Levene test. When it was significant, they were compared with the t-Student for unequal variances test. Protein profile data were analyzed by k proportions comparisons with the Chi-square test and pairwise comparisons with the z test. The nonparametric data from the preference tests were analyzed with the Friedman test and comparisons by Nemenyi. The Pearson correlation coefficient test was used to analyze the relationship between the acceptability data and the preference magnitude. To estimate the relationship between the composition of the samples and the sensory analysis data, a linear regression analysis performed with an ANOVA (type III sum of squares) test (selecting the best model and adjusting R²) was also used. All the analyzes were done with the XLSTAT program, version 2015.1 (Addinsoft^® Paris, France). The results are presented in mean values ± standard deviation (SD). The criterion to set statistical significance was p≤0.05.

Sensory characterization

Cheese is a product whose sensory characteristics are the result of multiple factors involved in its production such as the raw material quality, the type of the process, the native and added microbial cultures and the maturation time of the product. The biochemical transformation of cheese components such as lactose, fat and proteins, by the action of coagulants, milk enzymes and added lactic acid bacteria (LAB) contribute to the sensory characteristics that identify each cheese (^{Marchesseau et al., 1997}; ^{McSweeney and Souza, 2000}). Six samples had high-intensity odor, three (Q2, Q4 and Q5) had an attenuated odor and only one (Q8) had a weak odor (p≤0.01) (Table 1).

Weak smell or its absence in a cheese can be related to ferments with low lipolytic action (^{Navarro, 2015}) or short maturation time (^{Gamboa et al., 2013}). Milk lipases and esterases action (lipoprotein lipase) and LAB in cheese fat appear to be the main agents involved in the catabolism of medium and short chain free fatty acids. This generates compounds such as methyl ketones, lactones, esters, alkanes and secondary alcohols, which contribute to the olfactory-taste set of the product (^{Chamorro and Losada, 2002}; ^{Collins et al., 2003}). The panel established nine odor descriptors; the odor of fresh milk (p=0.03) and cooked milk (p=0.03) stood out. Q3, Q6 and Q10 samples stood out for their melted butter odor (p=0.03) and Q7 for the acidified curd and yogurt odor (p≤0.01), the latter also being the most acidic compared to the others (p≤0.01). The metabolism of lactose by LAB produces L- and D- lactate, which contributes to the acid taste of cheese, particularly in those of short maturation (^{McSweeney and Souza, 2000}). The excessive acidity can be due to the fact that previously acidified milk was used, since it favors curd draining and allows the necessary acidity to obtain compact, smooth and elastic pastes (“cheddarization”) and decreases the time of the process (^{Villegas de Gante, 2012}). Other causes of increased acidity are uneven curd cuts, high curing temperature or low salt percentage (^{Keating, 2007}). The cheese obtained from unpasteurized milk is more acidic, and in some cases generates a prickle mouthfeel (^{Van Hekken et al., 2006}). The relevance of controlling the acidification of cheese is because it determines microorganism’s growth and enzymatic activity during maturation and indirectly affects the taste of cheese (^{Collins et al., 2003}).

The salty most taste corresponded to Q1 and Q9, which contrasted with that of Q2 and Q8 (p=0.03). This may be due to the amount of NaCl added or the salting method. Salt contributes to the cheese flavor, partly determines the quality and acceptance by the consumer. Saltiness, pH, water activity and redox potential help to minimize the deterioration and growth of pathogenic microorganisms in cheese (^{Ramírez-Navas et al., 2017}). In some cases, more salt is added to regulate the development of lactic acid and proteolytic microflora in the cheese (^{Tunik et al., 2008}). Significant reduction of salt in the cheeses can generate a bitter taste. In our study the samples were characterized by certain bitter notes, Q2 (p=0.02) was the lowest. Proteolysis directly contributes to the cheese taste, in both, desirable and undesirable sapid characteristics, such as bitterness. The influence of bacterial proteases from the cultures is relevant in the cheese elaboration. Bitter taste may be due to the peptides type (mainly hydrophobic) and free amino acids (FAA) from casein hydrolysis and other milk proteins, by enzymes of the coagulant (chymosin, pepsin or fungal or plant proteinases), from milk (plasmin, cathepsin D and other somatic cell proteinases), LAB and exogenous proteinases used to accelerate cheese maturation. The secondary metabolism of peptides, FAA and changes in the cheese matrix facilitates bitter compounds release during chewing (^{McSweeney and Souza, 2000}). The bitter taste by certain peptides is determined by their terminal amino acid; when its concentration exceeds the perception threshold, it may constitute a defect in the cheese (^{Chamorro and Losada, 2002}). Q4 and Q5 exhibited a higher fraction of peptides ranging from 12.0 to 23.9 kDa compared to Q2 (p≤0.01).

In the taste interactions, Q9 showed a less acid-bitter taste (p=0.03), Q2 and Q4 had higher acid-sweet taste compared to Q1 and Q8 (p=0.04). Q3 had a sweet-bitter interaction greater than Q2, Q4, Q5, Q7, Q8, Q9 and Q10 (p≤0.01). The taste interactions allow to establishing the effect of one taste over another in food (^{Chamorro and Losada, 2002}). In our study, the judges perceived less bitter taste when they evaluated the interaction with acid and sweet, than when they evaluated this taste separately in the same sample. That is, the adequate balance of sweet and acid taste in Chihuahua cheese has a suppressive effect on the bitter taste and can contribute to generate its typical flavor.

Samples showed medium-low balance in mouthfeel attributes (Table 1). This is characteristic of young cheese or that in a young maturation stage (^{Moushumi et al., 2012}). Q1 showed the lowest firmness (p=0.02), Q5 and Q7 were perceived as those with the lowest moistness (p≤0.01) and Q7 and Q9 had the lowest fat character among the samples (p≤0.01). The protein matrix in cheese plays a major role in its characteristics and functional properties, controls the water content and the balance of components within the matrix and influences the firmness and homogeneity of the fat globules (^{Ramírez-López and Vélez-Ruiz, 2012}). The composition of fatty acids in milk fat can significantly affect firmness and cheese taste (^{Caro et al., 2014}).

Physicochemical characteristics

The sample’s moisture content was between 38.8 % (Q9) and 44.9 % (Q10) (p≤0.01) (Table 2). All cheeses had the established moisture for Chihuahua cheese (45 % max.) (^{COFOCALEC, 2011}). Protein content was 17.6 % (Q4) to 21.4 % (Q5) (p≤0.01). None of the sampled cheeses had the minimum protein content (23 %) established for Chihuahua cheese (^{COFOCALEC, 2011}). The protein material is one of the main components for its nutritional value and influence cheese yield. Protein content in milk can vary between seasonal periods, breeds and animal feed, or other factors. The milk used to produce cheese should be standardized in its casein / fat ratio, but standardization is usually only carried out by adjusting the fat content, on the assumption that the protein content does not change between successive or seasonal milking; this, causes variability in the product quality. The moisture, fat and protein content influence the sensory characteristics from cheese, mainly its texture (^{Caro et al., 2014}). In our study, a linear regression analysis indicated significance between the firmness perceived by the sensory panel and the protein, fat and mineral content (p=0.02, adjusted R²=0.72).

During the cheese manufacture, fat globules are trapped in a protein network. This prevents the protein from forming aggregates and determines the cohesion and elasticity of the gel, which in turn modulates the pH and ionic strength of cations, such as Na⁺ and Ca⁺². Small changes in these conditions can greatly affect the structure of the cheese and consequently affect its sensorial properties (^{Marchesseau et al., 1997}). The fat content was between 26.2 % (Q9) and 34.2 % (Q8) (p≤0.01). The gustatory perception of the fat character was closely related to the fat, protein and mineral content of the cheese (p=0.03, R² adjusted = 0.71). The minimum fat content for Chihuahua cheese is 28.0 % (^{COFOCALEC, 2011}), due to its fatty matter it has a great influence on the development and organoleptic quality of cheese; this because it dissolves components that release odor and modifies the perception thresholds and interferes in the balance of the dissociated and non-dissociated forms of fatty acids (^{Chamorro and Losada, 2002}). In contrast, the high fat content in a weak protein matrix (Q8) affected the sensory properties. Therefore, this feature can be a limiting factor in the final product acceptance. Higher fat-protein proportion affects the interaction between proteins, causes weak texture, which can be a defect on the product (^{Jonhson et al., 2009}).

The assessed samples had sodium contents from 288.9 mg per 100 g in sample (Q1) to 610.8 mg per 100 g in sample (Q5) (p≤0.01). The sodium content is related to the salting process of the paste; 70 to 85 % is attributed to Na⁺ and 15 to 30 % to Cl^- (^{Villegas de Gante, 2012}; ^{Ramírez-Navas et al., 2017}). The samples identified as the saltiest were Q1 and Q9, but the sodium content in Q9 was one of the lowest. Linear regression analysis showed no significant relationship between both variables. The difference between sensory perception and sodium content could be related to the cheese composition; particularly, Q9 had the lowest fat content and the highest moisture and pH. In reduced fat cheeses, greater taste has been detected when the salt-moisture ratio increases (^{Jonhson et al., 2009}). The pH higher than 5.8 promotes excess Ca⁺² ions linked to the paracasein molecules, which causes excessive Na⁺ incorporation in the molecule and affects its microstructure (Ramírez-Navas et al., 2017). Thus, salting can cause undesirable effects on flavor (^{Hernández-Morales et al., 2010}) and stop the natural maturation process of the cheese (^{Villegas de Gante, 2012}).

The titratable acidity was between 2.7 % (Q9) and 9.1 % (Q6 and Q7) from lactic acid (p≤0.01). In Chihuahua cheese, the acidification of the paste allows to obtain an adequate texture and maturation process. The products derived from this process are responsible for some of the changes in the appearance, texture, odor, taste and aroma of cheese (^{Chamorro and Losada, 2002}). During lactose fermentation by LAB, some pyruvate intermediates transformed into compounds that contribute to flavor, such as diacetyl, acetoin, acetaldehyde and acetic acid (^{Smith et al., 2005}). Acidity increases to the maximum in the first hours or days and then decreases, because the lactic acid combines with calcium and other salts in the cheese. The high acidity of the samples could relate to young cheeses or short maturation periods (^{Villegas de Gante, 2012}). Only Q7 was rated as the most acidic. However, the relation between the sensory perception in acidity and the composition of the cheese was not significant; although, a tendency of the mineral and carbohydrate content of the cheese was observed (p=0.09). The low acidity of the cheese may be due to the very low salt content, among other factors (^{Keating, 2007}). This could relate to the low sodium content of Q7. Salt reduction can also cause negative changes in LAB growth and risk of increased pathogenic microorganisms in cheese (^{Ramírez-Navas, 2017}).

The pH of the samples ranged between 5.8 and 6.5 (Q9). The pH of all the samples was higher than the interval for Chihuahua cheese (5.0 to 5.5) (^{COFOCALEC, 2011}) and similar to that of fresh, young cheeses (^{Ramírez-López and Vélez-Ruiz, 2012}). If the curd is not properly acidified during cheese making process (~pH 6.0) and pressing time is short, the cheese pH is not sufficiently reduced and its melting capacity is affected (^{Marchesseau et al., 1997}). The suggested maturation for Chihuahua cheese is at least 7 days; although, this type of cheese develops better organoleptic characteristics with greater than a month maturation lapses (^{Villegas de Gante, 2012}). In the practice, cheese is immediately packed after pressing (^{Cervantes et al., 2006}), this avoids the aeration phase, which is important to reduce moisture, increase acidity and reduce pH.

These factors contribute to the development of the characteristics that distinguish this cheese.

Thirteen protein bands were identified and corresponded to estimated masses from 6.7 up to 34.7 kDa (Figure 1).

Figure 1 Profiles and peptide intervals in Chihuahua cheese. St: standard. Peptide intervals bands according to their estimated mass. Comparison between samples of the same range bands interval. Different letters indicate significant difference (p≤0.05). 

Three intervals of the peptide sizes as a function of their estimated mass were identified: small mass peptides, up to 11.9 kDa (A), medium mass peptides, from 12.0 to 23.9 kDa (B) and high mass peptides, from 24.0 to 36.0 kDa (C). Overall, the analyzed samples showed higher peptide content with estimated mass between 24.0 and 36.0 kDa. These results agree with those found in one-week maturation Chihuahua cheese samples (^{Olson et al., 2011}) and corresponds to typical fresh cheeses (^{Tunik et al., 2008}, ^{Moushumi et al., 2012}). Q2, Q4 and Q5 had the lowest small peptides proportion and Q1, Q3, Q6 and Q10 the highest (p≤0.01). Q2, Q6 Q7, Q9 and Q10 showed the lowest medium mass peptides proportion and Q5 the highest (p≤0.01). Q5 and Q1 had the lowest high mass peptides proportion; Q2 had the highest proportion of these peptides (p≤0.01).

The coagulant and the lactic acid, sodium chloride and milk enzymes carry out the first proteolysis. During cheese maturation, peptides are generated and amino acids released by the action of endopeptidases, aminopeptidases and carboxypeptidases. The peptides and generated FAA contribute to the development of the basic olfactory-tasting characteristics of the product. The bitter taste in cheese correlates with the hydrophobicity of peptides generated in the cheese matrix (^{Ramírez-López and Vélez-Ruiz, 2012}), specifically in the proteolysis of α_s1-casein and β-casein (^{Moushumi et al., 2012}). This characteristic was mainly identified at Q4 and Q5, which had the highest peptide content, between 12.0 and 23.9 kDa. This could be because the cheeses did not have a greater enzymatic hydrolysis phase of these peptides, due to insufficient stage of maturation, which eliminated them along with their bitter taste (^{Olson et al., 2011}). Another cause of intensive proteolysis may be the excessive addition of coagulant together with sodium chloride or lactic cultures (^{Keating, 2007}). Q4 and Q5 had particularly higher sodium content. The fraction of smaller mass peptides was homogeneous among the cheeses; in this fraction, free amino acids can be broken down into amines, ammonia, carbon dioxide, methanethiol, alcohols and other compounds that influence the cheese flavor and color (^{Chamorro and Losada, 2002}). Several short peptides, less than 1000 Da, give umami taste to cheese (^{Moushumi et al., 2012}).

Consumer preferences

At the first preference test Q2, Q3 and Q5 were the most preferred (p≤0.01) (Figure 2a); in the second test Q6 and Q10 (p≤0.01) were the most preferred (Figure 2b). A joint analysis of the ten samples indicated that Q8 was intermediate between the most preferred and the least preferred cheese (p≤0.01); therefore, this sample was included in the last preference test. Thus, we confirmed that, out of the six samples, Q8 was the least preferred one (p≤0.01) (Figure 2c). The acceptance of these samples by the consumers relates to their protein, fat, minerals and carbohydrates content (p≤0.01, adjusted R²=0.77).

Figure 2 Preference-ranking test for Chihuahua cheese samples. a) First preference test. b) Second preference test. c) Third preference test on the most preferred samples in a) and b). Means and medians of position in the ranking (ranks). Different letters indicate significant difference (p≤0.05). 

The analysis of preference magnitude between the samples allowed identifying the differences between the samples with greater preference (Figure 3a). In this test, Q2 and Q3 were the most preferred and Q8 the least (p≤0.01). The acceptance test showed that Q2, Q3 and Q6 were most accepted, and Q8 was the least (p≤0.01) (Figure 3b). The preference magnitude of the six samples significantly correlated with the consumer acceptance test (r=0.85, p=0.03).

Figure 3 Preference and acceptance (±SD) of Chihuahua cheese. a) Estimation of preference magnitude with a 15-points linear scale. b) Acceptance test with a 9-point hedonic scale. 

In our study, the use of a 15-points linear scale to estimate the magnitude of the preferences and the 9-point hedonic scale to measure the acceptability of the samples, allowed to determine among the samples the most preferred by the consumers. The comparison of both methods has shown that the hedonic scale is superior to the magnitude estimation (^{Moskowitz and Sidel, 1971}; ^{Warren et al., 1982}). In our study both tests had an important correlation and confirmed consumer preferences by two different methods.

The descriptive analysis (trained judges) and consumer tests (affective tests) together allowed identifying that the variability in the cheese attributes, perceived by the judges, was related to the acceptance or rejection of the product by the consumers. Thus, high acidity (Q7), salty taste (Q1, Q9), weak odor, high fat content (Q8) or bitter taste (Q4, Q5) were factors that limited the consumer preferences. With the information on consumer preferences, physicochemical (Table 3) and sensory profile (Table 4) models characterizing the cheeses of this study were elaborated.

These models can be the basis for standardizing Chihuahua cheese manufacturing based on consumer preferences and current Mexican regulations. The proposed models allow identifying how close or far a product is respect to the product preferred by the consumers (^{Ojeda et al., 2015}). The information on the product position helps direct the producers to reformulate or modify the cheese making process (^{López-Velázquez et al., 2012}; ^{Almanza-Rubio et al., 2013}). The sensory and physicochemical profile models of the product also provide reference characteristics that Chihuahua cheese must meet to be within the consumer’s preferences.