Evolution of VOC and Sensory Characteristics of Stracciatella Cheese as Affected by Different Preservatives

Undesired volatile organic compounds (VOCs) can negatively affect the flavor of fresh food products; especially those characterized by a mild and delicate aroma. Finding connections between chemical and sensory analyses is a useful way to better understand the arising of off-flavors. A study was conducted on stracciatella; a traditional Italian cream cheese that is emerging on international markets. Samples were prepared by adding two different preservatives (alone or combined): sorbic acid and an olive leaf extract. Their influence on flavor preservation during refrigerated storage was investigated by chemical, microbiological and sensory analyses. A strong change of the VOC profile was ascertained after 8 days in the control cheese and in the sample added with leaf extract alone. The samples containing sorbic acid, alone or in combination with leaf extract, gave the best chemical and sensory results, demonstrating a significant shelf-life extension. In particular, these samples had lower concentrations of undesired metabolites, such as organic acids and volatiles responsible for off-flavor, and received better scores for odor and taste. Ex and Ex-So samples had significantly higher antioxidant activity than Ctr and So throughout the entire storage period, and the color parameter shows no differences among samples taken on the same day. The use of the olive leaf extract, at the concentration tested, seemed to be interesting only in the presence of sorbic acid due to possible synergic effect that mainly acted against Enterobacteriaceae.


Introduction
One of the decay's symptoms of an expiring food is its smell [1,2]. Thus, the study of the volatile organic compound (VOC) profile can be very useful for establishing the organoleptic status of the product. The onset of off-flavor can be caused by many factors, such as food handling, processing, spoilage microorganism and lipid oxidation [3][4][5]. Many researchers have investigated the VOC responsible of off-flavor and have tried to link them to specific sensory descriptors [5][6][7][8][9][10]. The fastest way to analyze these compounds is using SPME-GC/MS (Solid Phase Micro Extraction-Gas Chromatography/Mass Spectrometry). It is widely used among researchers for milk and dairy products [11][12][13][14][15] and other foods [16][17][18]. It is used also in the flavoromics approach [19,20], because it is capable of ppb detection level for a wide range of molecular weight. Such a kind of studies is very useful for fresh foods, which are preferred by modern consumers and are highly perishable. For cheeses, too, the consumer preference is increasingly orienting towards the fresh types, whose production volume in Table 1. Ingredients and preservatives concentration used to produce stracciatella cheese.

Volatile Organic Compounds (VOCs)
VOCs were extracted at 37 • C for 15 min as reported in a previous paper [40] after addition of 3-pentanone (81.3 ng) as an internal standard for semi-quantitation. A Triplus RSH autosampler was used, equipped with a divinylbenzene/carboxen/polydimethylsiloxane 50/30 mm SPME fiber assembly (Supelco, Bellefonte, PA, USA). The fiber was desorbed at 220 • C for 2 min in the injection port of the gas chromatograph operating in the splitless mode. The GC-MS analysis was performed using a Trace 1300 chromatograph equipped with a capillary column VF-WAX MS (60 m, 0.25 mm) and connected to mass spectrometer ISQ Series 3.2 SP1 (Thermo Scientific, Waltham, MA, USA). The operating conditions were: oven temperatures, 50 • C for 0.1 min then 13 • C min −1 up to 180 • C and 18 • C min −1 up to 220 • C with an isothermal for 1.5 min. The mass detector was set at 1700 V voltage; source temperature, 250 • C; ionization energy 70 eV and scan range 33-200 amu. Peak identification was done by means of Xcalibur V2.0 Qual Browse software (Thermo Fisher Scientific, Waltham, MA, USA) by matching with the reference mass spectra of the NIST (National Institute of Standards and Technology, Gaithersburg, MD, USA) library.

Color
Colorimetric readings were carried out under D65 illuminant by using a spectro-colorimeter CM-700d (Konica Minolta Sensing, Osaka, Japan) equipped with a pulsed xenon lamp. The analysis was performed by placing the sample in a transparent quartz container. Lightness (L*), redness (a*, ±red-green) and yellowness (b*, ±yellow-blue) were determined coordinates in the CIE (Commission Internationale de l'Éclairage) color space [41].

Organic Acids and pH
Organic acids were extracted as reported by Buffa et al. [42]. Separation was carried out on a Synergy Hydro RP column 80 Å, 4 µm, 250 mm × 4.6 mm (Phenomenex, Torrance, CA, USA), installed on a Waters HPLC composed of 600E pumps and a 996 diode array detector (Waters Corporation, Milford, CT, USA). Mobile phases were 0.1% orthophosphoric acid in water (eluent A) and acetonitrile (eluent B). The gradient was 0-18 min 100% A at 1 mL min −1 flow rate, then 18-18.3  100% A and 23-43 min isocratic. Detection was done at λ = 214 nm. A pH meter equipped with a dairy specific electrode (FC2020, Hanna instruments, Woonsocket, RI, USA) was used for pH measurement.

Sensory Analysis
Sensory analysis was assessed by a trained panel consisting of nine experts (4 women and 5 men, ranging in age from 25 to 60 years) selected according to the ISO 8586-1993 method. All of them were members of the Italian Association of Cheese Tasters (ONAF) and had attended a 20 h course about evaluation of cheese texture and flavor. Evaluation was carried by a quantitative descriptive analysis (QDA) and the samples were presented in a randomized and balanced way, in white disposable dishes coded by three-digit codes. The panelists described the cheese using sensory attributes chosen from the ONAF (Italian Organization of Cheese Tasters) vocabulary [43] and by scoring their intensity from 0 to 10. In the case of perception of an attribute not included in the vocabulary, it was only considered if the "weight percentage" (frequency of citations × perceived intensity) was more than 30%.

Cultivable Microbiota
Microbiological analyses were performed according to methods described in Minervini et al. [44], using culture media and supplements purchased from Oxoid (Oxoid Limited, Basingstoke, UK), starting from 20 g of stracciatella. The microbial groups counted were: total mesophilic aerobic (plate count agar at 30 • C for 48 h), presumptive mesophilic lactobacilli (MRS agar with 0.1 g L −1 cycloheximide at 30 • C for 48 h), presumptive mesophilic cocci (M17 agar with 0.1 g L −1 cycloheximide at 30 • C for 48 h), enterococci (Slanetz and Bartley agar at 37 • C for 48 h), staphylococci (Baird Parker agar supplemented with egg yolk tellurite at 37 • C for 48 h), Enterobacteriaceae (VRBGA at 37 • C for 24 h), Pseudomonas spp. (Pseudomonas agar with CFC supplement at 30 • C for 24 h), yeasts (wort agar supplemented with 0.1 g L −1 chloramphenicol at 30 • C for 48 h) and molds (potato dextrose agar supplemented with 0.1 g L −1 chloramphenicol at 25 • C for 5 days). The microbiological counts were confirmed by taking representative colonies for each medium, and checking them for morphology, motility, Gram staining reaction and catalase test.

Statistical Analysis
The data were statistically processed by one-way ANOVA followed by Tukey's HSD procedure at p < 0.05 using XLSTAT software (Addinsoft, NY, USA). Each sample was analyzed in triplicate. The data from the sensory analysis were also processed for generalized Procrustes analysis (GPA) with the same software. Figure 1 shows the total amounts of volatile compounds during storage of the samples. As expected, low quantities were detected in the early days, ranging from 200 to 470 µg kg −1 . Until day 8, the differences among samples were not very relevant, even though they were sometimes statistically significant. Huge differences started after this time, with Ex and Ctr samples showing an increase to about 1800 µg kg −1 , then to about 2700 and 3600 µg kg −1 (at day 16), respectively. Differently, the concentrations in Ex-So and So remained almost constant during time. Increase in VOC formation during storage was expected, considering the typical chemical and microbiological characteristics of the product [27,30,31], and from these quantitative data it clearly appeared that a significant inhibition of their formation was associated to the presence of sorbic acid. In order to better understand the mechanisms of formation, the entire VOC data set was grouped into seven chemical classes ( Table 2). The samples were rather similar at day 1 as to the most abundant classes (ketones, esters and aliphatic hydrocarbons), whereas some differences were found in some less represented groups such as aldehydes, alcohols and sulfur compounds, whose content is known to be influenced by manipulation (they are highly produced by spoilage microorganism). The main difference at T1 accounted on acids, which were more abundant in So and Ex-So, but it was mainly due to the presence of sorbic acid added as a preservative. As already observed under the quantitative point of view, a clear qualitative differentiation among samples started from day 8, and the pair Ctr/Ex and Ex-So/So behaving quite similarly until the end of storage. At T8 aldehydes and alcohols resulted to be higher in the pair Ctr/Ex, whereas esters and acids were more abundant in Ex-So/So. The control sample had the highest concentration of sulphur compounds and the lowest of aliphatic hydrocarbons. The higher concentrations of aldehydes and alcohols in Ctrl and Ex could be the consequence of two different phenomena: (i) lipid oxidation, which is a common source of aldehydes formation in fresh cheeses, from whose dehydrogenation the corresponding alcohols can successively derive [3,45] and (ii) microbial metabolism. At the end of storage (T16) almost all chemical groups were much higher in Ctr and Ex, whereas only sulphur compounds were more abundant in Ex-So. Finally, very low concentration of aliphatic hydrocarbons characterized the Ex-So sample.

VOC
As to the single VOC, 41 molecules were identified in the entire set of samples, the most abundant compounds detected at the end of the storage period are reported in Table 3. Acetaldehyde and 3-methylbutanal were found only in Ex and Ctr. Acetaldehyde can be formed throughout different pathways: ethanol oxidation should be a highly probable pathway, considering that ethanol concentration was very high in these samples (713.6 and 556.4 µg kg −1 for Ex and Ctr respectively versus 7.8 and 4.2 µg kg −1 for Ex-So and So, respectively). The mechanism based on phenolic compounds oxidation could also have played a role in Ex, considering the presence of polyphenols from the leaf extract [3]. 3-methylbutanal in cheese often arise from the conversion of the corresponding amino acid (isoleucine) by the LAB activity [46], and its concentration suggests a more marked microbial activity with respect to the samples without sorbic acid. This hypothesis was supported by the huge increase of ketones in Ctr and Ex, mainly connected to the formation of acetoin (1839.57 and 1512.48 µg kg −1 respectively, versus 7.8 and 7.4 µg kg −1 in So and Ex-So) that commonly derives from LAB and yeasts metabolism [47]. Ethyl acetate and ethyl butanoate increased much more in Ex, in connection with a higher presence of ethanol, which represents the limiting factor for their formation [48]. This aspect was probably connected to the absence of sorbic acid acting against yeasts. On the other hand, the presence of this preservative in So and Ex-So gave rise to the formation of ethyl sorbate. Among alcohols, besides ethanol, also 3-methyl-1-butanol greatly characterized the samples without sorbic acid, suggesting a higher microbial activity in them. In some food matrices such as fruit juices, 3-methyl-1-butanol and phenylethyl alcohol have been related to yeast metabolism [3]. The former was reported by Morales et al. [49] as an important component of volatilome of Enterobacteriaceae strains of dairy origin. Other volatiles that could be connected to faster microbial growth in Ctr and Ex samples were sulphur compounds (i.e., dimethyl sulfide) that are formed, for instance, by some Enterobacteriaceae species [7], and organic acids such as acetic, butanoic and hexanoic. hydrocarbons. The higher concentrations of aldehydes and alcohols in Ctrl and Ex could be the consequence of two different phenomena: (i) lipid oxidation, which is a common source of aldehydes formation in fresh cheeses, from whose dehydrogenation the corresponding alcohols can successively derive [3,45] and (ii) microbial metabolism. At the end of storage (T16) almost all chemical groups were much higher in Ctr and Ex, whereas only sulphur compounds were more abundant in Ex-So. Finally, very low concentration of aliphatic hydrocarbons characterized the Ex-So sample.     Figure 2 shows the quantitative descriptive analysis (QDA) graphs divided into odor (A) and taste (B) perceptions. All fresh samples had almost the same olfactory characteristics until day 4, demonstrating the absence of any influence of the preservatives on the sensory profile. Starting from day 8, in perfect agreement with the evolution of VOC, a loss of the freshness characteristics was observed in Ctr and Ex. In particular, the overall intensity increased, and considering that stracciatella must present a mild and delicate aroma, it has to be considered a symptom of ongoing changes. This variation led to the loss of "frankness" defined as the characteristic aroma of the product; with time, undesired notes of sour milk, sourdough, banana and boiled cabbage were also perceived. These results matched well with the increase of some VOC compounds, such as butanoic acid, ethyl esters, ethanol, dimethyl sulphide and 3-methylbutanal. The same trend was also observed for Ex-So and So, but only starting from day 12. At day 16 Ctr and Ex samples were totally unacceptable, differently from Ex-So and So. This finding could be related to the strong increase of some VOC that can be considered as responsible of off-flavors. As to taste (B) no differences were observed among samples until day 8, then Ctr and Ex started to lose sweetness and to evidence bitterness and sourness; a slight increase in sourness was also perceive in So. The responsibility of these changes could be attributed to the formation of free hydrophobic amino acids deriving from proteolysis and/or increase of organic acids. At day 12, Ex and Ctr were rejected for excessive sourness and sweetness decreased in Ex-So and So, but they were still judged as acceptable. At day 16 also these latter samples were rejected for bad smell (they were not tasted). Figure 3 shows the GPA plot based on the dataset of odor sensory analysis. Here it is possible to observe the distance among samples on a bidimensional plot. The two dimensions account for 98.80% of the total variance. Most of the differences (93.08%) can be explained on F1, where So and Ex-So were positioned on the left side of the plot, correlated to fresh milk and frankness perceptions. Ex and Ctr were on the right side of the plot, very far between them and related to ripe fruit and sour yogurt respectively, and in both cases also with sourdough, sour milk and odor intensity. This information allows us to make connections between the VOC profile and sensory analyses: So and Ex-So resulted to be quite similar, while Ex and Ctr were far from them and strongly affected by off-flavors. Thus, using sorbic acid or a mixture of the two preservatives seems to not interfere negatively on the product.

Sensory Analysis and Possible Connections with VOC
The potentially involved molecules in stracciatella flavor were hypothesized by calculating the odor active value (OAV) of VOC according to Qian et al. [50], on the basis of the odor thresholds in air taken from the literature. The VOC having OAV > 1 are shown in Table 4. Ethyl-acetate exceeded a value of 1 in all samples at any time, and should be one of the key odorants in this cheese. It presents an ethereal and green note and was found at the highest level in Ex, where also its precursors, ethanol and acetic acid, were at the highest level than all samples. Hexanal (green grass odor) should only play a role in fresh cheese, since OAV > 1 was only found at day 1. Surprisingly, ethanol was only present at day 1 with OAV = 1.3 in So and Ex-So, but the reason was unclear. As expected, it was not newly formed in these samples due to the inhibition of yeasts, whereas it strongly increased in Ctr and Ex, reaching a maximum OAV value of 69.6 and 89.2 at day 16 respectively. Other two potentially aroma-active esters, ethyl butanoate and ethyl hexanoate responsible of banana and apple notes [51], increased faster in Ctr and Ex. In these two samples, also acetoin (buttery and creamy odor) could contribute to the aroma, whereas in So and Ex-So it remained under the perception threshold. At day 8, also 3-methylbutanal (ethereal and fruity odor, but acrid when present at high concentration) exceeded a value of 1 in Ex and Ctr.    T1  T4  T8  T12  T16  T1  T4  T8  T12  T16  T1  T4  T8  T12  T16  T1  T4  T8  T12  T16 3-Methylbutanal    The potentially involved molecules in stracciatella flavor were hypothesized by calculating the odor active value (OAV) of VOC according to Qian et al. [50], on the basis of the odor thresholds in air taken from the literature. The VOC having OAV > 1 are shown in Table 4. Ethyl-acetate exceeded a value of 1 in all samples at any time, and should be one of the key odorants in this cheese. It presents an ethereal and green note and was found at the highest level in Ex, where also its precursors, ethanol and acetic acid, were at the highest level than all samples. Hexanal (green grass odor) should only play a role in fresh cheese, since OAV > 1 was only found at day 1. Surprisingly, ethanol was only present at day 1 with OAV = 1.3 in So and Ex-So, but the reason was unclear. As expected, it was not newly formed in these samples due to the inhibition of yeasts, whereas it strongly increased in Ctr and Ex, reaching a maximum OAV value of 69.6 and 89.2 at day 16 respectively. Other two potentially aroma-active esters, ethyl butanoate and ethyl hexanoate responsible of banana and apple notes [51], increased faster in Ctr and Ex. In these two samples, also acetoin (buttery and creamy odor) could contribute to the aroma, whereas in So and Ex-So it remained under the perception threshold. At day 8, also 3-methylbutanal (ethereal and fruity odor, but acrid when present at high concentration) exceeded a value of 1 in Ex and Ctr.

Antioxidant Activity and Color
As shown in Figure 4A, Ex and Ex-So samples had significantly higher antioxidant activity than Ctr and So throughout the entire storage period. Similar results were reported in studies on different food matrices [38,52]. The oxidative stability test carried out by RapidOxy gave different results ( Figure 4B). No significant differences between samples were found at day 1, the So samples had the lowest induction time values after 4 and 8 days, and the Ex-So samples showed a higher value than Ctr at day 16, in accordance with antioxidant activity. These results could be due to the OLE effect, but the synergy with sorbic acid and interaction with the food matrix cannot be excluded. Overall, the results indicated that OLE, both added alone and together with sorbic acid, exerted an antioxidant activity in cheese. Different authors showed the antioxidant effect exerted by OLE in dairy products and according to literature the antioxidant power was lower in Ctr in a range of 40-80% and decreased during storage [53,54]. From this outcome it can be hypothesized that the higher presence of aldehydes in the VOC profile of OLE samples was mostly due to microbial activity rather than to fat oxidation.
the results indicated that OLE, both added alone and together with sorbic acid, exerted an antioxidant activity in cheese. Different authors showed the antioxidant effect exerted by OLE in dairy products and according to literature the antioxidant power was lower in Ctr in a range of 40-80% and decreased during storage [53,54]. From this outcome it can be hypothesized that the higher presence of aldehydes in the VOC profile of OLE samples was mostly due to microbial activity rather than to fat oxidation.  As to color (Table 5), the values of some indices changed with time, but no significant differences were observed among samples taken at the same day. L* index decreased, in accordance with other authors [31,36], and according to García-Pérez et al. [55] it should be connected to the acidification of the product. Differently, the a* index slightly increased during storage, whereas b* index showed a non-defined trend. The increase of a* value could originate from polymerization of polyphenols and related tendency to browning [56].  As to color (Table 5), the values of some indices changed with time, but no significant differences were observed among samples taken at the same day. L* index decreased, in accordance with other authors [31,36], and according to García-Pérez et al. [55] it should be connected to the acidification of the product. Differently, the a* index slightly increased during storage, whereas b* index showed a non-defined trend. The increase of a* value could originate from polymerization of polyphenols and related tendency to browning [56].

Organic Acids and pH
The changes in organic acids in dairy products are highly connected to microbial activities. In bottled milk and non-fermented products, as is stracciatella, they are undesired since they are responsible of off-flavors. From Figure 5, it can be observed that changes started early in Ctr and Ex samples: being a fermentable substrate, citric acid decreased from 0.91 and 0.93 mg kg −1 to 0.09 and 0.26 mg kg −1 , respectively; lactic acid increased from 0.12 and 0.13 mg kg −1 to 0.65 and to 1.05; acetic acid increased from 0.01 to 0.11 and to 0.24 mg kg −1 . There is no information concerning organic acids content of stracciatella cheese, but looking at other dairy products it is possible to confirm our results. In fact, citric acid is often involved in the Krebs cycle, and this justifies its reduction. Moreover, the increase in other organic acids (i.e., acetic and citric) is explained by higher microbial activities during storage [57][58][59]. According to Adda et al. [60] citric acid is used as a substrate by the microorganism to produce pyruvic and acetic acid, this finding clearly suits with our results. Differently, only slight variations were observed in the other samples containing sorbic acid or a mixture of the two preservatives. These results matched well with the decreasing trend of pH that went from 6.34 to 5.89 in Ctr and from 6.29 to 5.83 in Ex; in contrast, pH remained almost unchanged in So and Ex-So. The lower values in these latter samples found already at day 1 (6.02 and 6.14, respectively) were caused by the acidic properties of sorbic acid (pKa 4.76). These findings highlight a possible antimicrobial effect of olive leaf extract in synergy with sorbic acid.

Organic Acids and pH
The changes in organic acids in dairy products are highly connected to microbial activities. In bottled milk and non-fermented products, as is stracciatella, they are undesired since they are responsible of off-flavors. From Figure 5, it can be observed that changes started early in Ctr and Ex samples: being a fermentable substrate, citric acid decreased from 0.91 and 0.93 mg kg −1 to 0.09 and 0.26 mg kg −1 , respectively; lactic acid increased from 0.12 and 0.13 mg kg −1 to 0.65 and to 1.05; acetic acid increased from 0.01 to 0.11 and to 0.24 mg kg −1 . There is no information concerning organic acids content of stracciatella cheese, but looking at other dairy products it is possible to confirm our results. In fact, citric acid is often involved in the Krebs cycle, and this justifies its reduction. Moreover, the increase in other organic acids (i.e., acetic and citric) is explained by higher microbial activities during storage [57][58][59]. According to Adda et al. [60] citric acid is used as a substrate by the microorganism to produce pyruvic and acetic acid, this finding clearly suits with our results. Differently, only slight variations were observed in the other samples containing sorbic acid or a mixture of the two preservatives. These results matched well with the decreasing trend of pH that went from 6.34 to 5.89 in Ctr and from 6.29 to 5.83 in Ex; in contrast, pH remained almost unchanged in So and Ex-So. The lower values in these latter samples found already at day 1 (6.02 and 6.14, respectively) were caused by the acidic properties of sorbic acid (pKa 4.76). These findings highlight a possible antimicrobial effect of olive leaf extract in synergy with sorbic acid.

Cultivable Microbiota
The evolution of cultivable microbiota during cheese storage matched well with the results of the chemical and sensory analyses. It changed with different magnitudes in samples during storage, depending on the storage time and microbial group ( Figure 6). The cell densities of mesophilic aerobic microorganisms ranged from about 4.1 (Ex and Ex-So) to 4.81 (Ctr) log CFU g −1 after 1 day of storage. At the end of shelf life, the values were higher than 7.0 log CFU g −1 in all samples, without significant differences (p > 0.05) between experimental and control. However, preservatives exerted an effect during storage, since the highest cell load was reached more rapidly (at 8 days) in Ctr than in Ex and Ex-So samples (at 16 days). Thus, the probability to find earlier VOC responsible for off-flavor in these samples was higher. Lactic acid bacteria (LAB) in stracciatella are considered altering agents, just like in bottled milk. The average cell densities of mesophilic lactobacilli and cocci at day 1 were 4.7 and 5.1 log CFU g −1 , respectively, while enterococci were about 2.6 log CFU g −1 , without significant  Lactic acid Acetic acid Citric acid

Cultivable Microbiota
The evolution of cultivable microbiota during cheese storage matched well with the results of the chemical and sensory analyses. It changed with different magnitudes in samples during storage, depending on the storage time and microbial group ( Figure 6). The cell densities of mesophilic aerobic microorganisms ranged from about 4.1 (Ex and Ex-So) to 4.81 (Ctr) log CFU g −1 after 1 day of storage. At the end of shelf life, the values were higher than 7.0 log CFU g −1 in all samples, without significant differences (p > 0.05) between experimental and control. However, preservatives exerted an effect during storage, since the highest cell load was reached more rapidly (at 8 days) in Ctr than in Ex and Ex-So samples (at 16 days). Thus, the probability to find earlier VOC responsible for off-flavor in these samples was higher. Lactic acid bacteria (LAB) in stracciatella are considered altering agents, just like in bottled milk. The average cell densities of mesophilic lactobacilli and cocci at day 1 were 4.7 and 5.1 log CFU g −1 , respectively, while enterococci were about 2.6 log CFU g −1 , without significant differences among samples (p > 0.05). Lactobacilli and cocci remained almost constant during storage in So or in Ex-So samples, whereas their number increased up to 1 log cycle in Ctr and Ex cheeses at 16 days. Additionally, enterococci increased up to 1 log cycle at the end of storage, but only in Ctr (about 3.6 log CFU g −1 ), whereas it decreased in all other samples. These findings suggest that OLE was effective in controlling enterococci when used alone, whereas it only had an effect on the other two LAB groups when associated to sorbic acid. Our results corroborate the findings of Roila et al. [61] and Servili et al. [62], who verified that the inhibitory effects of a polyphenol extract from an olive oil byproduct against lactic acid bacteria in mozzarella and a fermented milk beverage were scarce and dose-dependent. According to Hurtado et al. [63], some LAB strains can degrade oleuropein and metabolize specific compounds of OLE. In addition, the antimicrobial activity of the leaf extract in dairy products may be partially inhibited by the chemical interaction between phenolic hydroxyl groups and milk proteins [64]. Pseudomonas spp. are the most feared spoilage bacteria in fresh cheeses, since they can replicate at refrigeration conditions [65,66]. After 1 day of storage, the two cheeses containing OLE showed a lower (p < 0.05) cell density of Pseudomonas spp. than control. Successively, the counts increased in all samples reaching the maximum cell densities of about 6.3 log CFU g −1 . Despite the effect observed at the early stage, OLE did not work against these bacteria, both alone and in the presence of sorbic acid. Inefficacy could be due to two main reasons: (1) OLE concentration was too low to have an effect. In fact, antimicrobial activity of olive polyphenols against Gram-negative bacteria, such as Escherichia coli and Pseudomonas, was reported to be dose-dependent [61,67] and (2) pH of the cheese (6.0-6.2) was too high for allowing efficacy of sorbic acid: in cottage cheese it acted against Pseudomonas at pH 4.6-5.1 [68]. As for Enterobacteriaceae, after the first day of storage significant differences were observed between Ctr (3.8 log CFU g −1 ) and So (3.7 log CFU g −1 ) compared to Ex (1.7 log CFU g −1 ) and Ex-So (<1 log CFU g −1 ) samples. Successively, they disappeared more rapidly in the cheeses added with antimicrobials. This finding is of particular relevance, because Enterobacteriaceae are involved in spoilage phenomena in the dairy product. In accordance with VOC results, Enterobacteriaceae are known as producers of some VOC as 3-methyl-1-butanol, found only in samples without sorbic acid [49,69]. Staphylococci and molds were not detected, differently from yeasts. Overall, effectiveness of OLE in controlling yeasts was ascertained at early phases (day 4), successively it had an effect only in the presence of sorbic acid, confirming the usefulness of sorbic acid in controlling mold and yeast.  T1  T4  T8  T12  T16  T1  T4  T8  T12  T16  T1  T4  T8  T12  T16  T1  T4  T8  T12 T1  T4  T8  T12  T16  T1  T4  T8  T12  T16  T1  T4  T8  T12

Conclusions
The results of the present work pointed out the importance of relating the VOC profile and sensory results as a decay's symptom in fresh cheeses. In particular, VOC analysis could be a fundamental tool for predicting the shelf life of stracciatella, together with the classic microbiological and sensory assays. In addition, it could help to monitor the formation of some key volatiles responsible for an off-flavor and prevent rejection of the product by consumers in large-scale retails. The study allowed us to suitably evaluate the influence of the preservatives on the sensory characteristics. The samples added with sorbic acid or a combination of sorbic acid and olive leaf extract gave the best result as to the evolution of the VOC profile. It was reflected in better flavor preservation during storage, allowing us to better preserve both stracciatella's delicate aroma and taste. The color parameter results did not show any relevant difference among products taken at the same day, so the addition of preservatives does not influence the product aspect. Although stracciatella OLE-enriched samples reached higher antioxidant activity than Ctr and So, such performances were mostly connected to the well-known antimicrobial effect of sorbic acid, also further to a possible synergic effect with OLE, in particular against Enterococcus spp. and Enterobacteriaceae. Possibly better results might be reached also against other spoilage microorganisms by increasing the extract concentration, but the negative impact on the sensory characteristics appears to be a main limit. Nevertheless, the possible synergic effect could help producers to reduce the concentration of chemical preservatives. A further synergic effect could be obtained by slight lowering of pH of the cheese, compatibly with maintaining the flavor acceptable.