Effect of Type of Coagulant and Addition of Stored Curd on Chemical, Rheological and Microstructural Properties of Low-Moisture Mozzarella Cheese

Abstract

Low-moisture Mozzarella cheese (LMMC) was manufactured in a dairy factory by stretching fresh curd in hot water, with the addition of 0–30% commercial curd (stored curd) purchased as a semi-finished product. Two commercial fermentation-produced camel chymosins, CC-M and CC-S, were employed as coagulants. The chemical, rheological and microstructural properties of LMMC were assessed during storage. The results demonstrated that cheese composition was not significantly influenced by curd addition. The use of CC-S promoted a slight increase of fat matter with respect to the CC-M samples because of the higher proteolytic specificity and clotting activity of the CC-S enzyme. A higher extent of proteolysis was found in LMMC manufactured with CC-M. The textural properties evaluated during storage revealed an increase in meltability, adhesiveness and springiness over time. The amount of added curd had only a minor effect. The melting behaviour was significantly influenced by proteolysis during the 35-day storage period. Overall, the proteolysis during aging was the most impactful factor affecting the properties of LMMC.

1. Introduction

Low-moisture Mozzarella cheese (LMMC) is a pasta filata variety representing a worldwide strategic product for the global dairy sector. Low-moisture Mozzarella is recognised for its long shelf-life and versatility, especially in processed foods and pizza, making it a dominant force in the Mozzarella cheese market in 2023 [1]. Generally, any shape of block can be formed, with 3–4 kg blocks being the most common. LMMC cheese is characterised by a broad compositional range in terms of moisture (45–52%) and fat (30–50% fat-in-dry matter) [2]. It is classified as a rennet-coagulated semi-hard cheese characterised by a firm homogeneous structure without holes. Manufacturing protocols can vary noticeably as a function of the final application. The addition of stored curd to afresh one during stretching is commonly exploited in LMMC manufacturing to reduce processing costs and as a strategy for its customisation [3]. For instance, this practice is applied in Italy, where milk is preferentially used for traditional cheeses. Some producers from the Southeast market import LMMC from Europe, the United States, Australia and/or New Zealand to compensate for issues with local milk quality and supply [4].

The techno-functionality of LMMC is the major contributor to its end use. Composition modifications and proteolytic phenomena occurring during storage drive the changes in the functional properties of LMMC, promoting softening and modifying the flowability and apparent viscosity of the melted product, among others [5]. Casein hydrolysis in Mozzarella-type cheese has been shown to be less extensive than in Cheddar or Gouda cheese of the same age [6]. In addition to the thermal inactivation of chymosin, other parameters can affect the rate of primary proteolysis in LMMC, including the manufacturing procedure, starter culture type, type and amount of rennet, salt concentration, pH and ripening temperature [7,8,9,10,11].

The progress in recombinant DNA technology allowed for the calf chymosin gene to be cloned into microorganisms to make a “fermentation-produced chymosin” [12]. This technology has made commercially available camel chymosin (CC) produced by fermentation. The interest in the recombinant CC is ascribable to its higher clotting activity and lower non-specific proteolysis in comparison to bovine chymosin [13]. These effects were demonstrated for Cheddar and low-moisture part-skim Mozzarella cheeses, which exhibited higher hardness (due to the decreased primary proteolysis) and longer shelf life in comparison to cheeses manufactured using calf chymosin [12,14]. Through the years, a new recombinant CC with altered molecular structure was developed to further optimise cheesemaking efficiency. The resulting commercial enzyme (CHY-MAX^® Supreme, CC-S) obtained by Chr. Hansen Ltd. (Cork, Ireland) by application of bioinformatics, sequence function analysis and structure-based design shows a greater milk coagulation specificity and a further reduction of non-specific proteolytic activity, with respect to the original recombinant CC (CHY-MAX^® M, CC-M) [15]. This new generation of recombinant enzymes is declared to provide both a more structured and compact casein network and a significant reduction of non-specific proteolysis. Li et al., using this novel CC-S in the manufacture of Cheddar cheese, found that it promoted lower levels of primary proteolysis, higher hardness and lower meltability, as well as less sulphur and barny flavor in comparison with the use of calf chymosin [15]. Overall, the industrial interest in recombinant CC is due to the high ratio of milk clotting activity/general proteolytic activity compared to bovine chymosin. Thus, the recombinant CC has the potential to be used as an alternative for cheeses requiring low proteolysis during ripening.

The higher milk-clotting activity of CC in comparison to bovine chymosin, as well as its commercial availability, has gained the attention of producers. This solution is attractive for commercial LMMC applications and investigations are required to evaluate factors affecting its efficiency and versatility. This current study, carried out at an industrial level, aims to provide insights related to the effect of the type of chymosin used and the amount of stored curd on the compositional, microstructural and techno-functional properties of LMMC during storage at 4 °C for up to 35 days. In detail, two different recombinant CC coagulants (i.e., CC-M and CC-S) and the addition of three levels (i.e., 10, 20, 30%) of vacuum-stored curd to the fresh curd before stretching were investigated.

2. Materials and Methods

2.1. Starter, Coagulants and Stored Curd

The direct-vat-set starter STI-06 (Chr. Hansen Ltd., Cork, Ireland) consisting of a culture of Streptococcus thermophilus was used. Commercial fermentation-produced (Aspergillus niger, var. awamori) CC (CHY-MAX^® M, 200 IMCU/mL, benzoate-free, Chr. Hansen Ltd., Cork, Ireland) (CC-M) and commercial fermentation-produced CC with structural changes through amino acid substitution (CHY-MAX^® SUPREME, 200 IMCU/mL, benzoate-free, Chr. Hansen Ltd., Cork, Ireland) (CC-S) were used. Two commercial batches of curd (M and S) for Mozzarella cheese were purchased from Friesland Campina (Amersfoort, The Netherlands). Stored curd M was used for the trial, with CC-M coagulant and curd S for trial with CC-S coagulant. Each block (15 kg) was vacuum-sealed and stored under refrigerated conditions. The declared ingredients included pasteurised milk, salt, lactic ferments and microbial coagulant.

2.2. LMMC Manufacture

The coagulants CC-M and CC-S and four levels of curd addition (0%, 10%, 20% and 30%) were tested, for a total of eight cheesemaking trials. Cheesemaking trials were carried out in Caseificio Giordano S.r.l. (Oleggio, Italy) on two different days. Each cheesemaking trial used 5000 L milk. Bovine raw milk was standardised to a protein and fat content of 3.95 ± 0.02% and 4.00 ± 0.03% (w/v), respectively. Subsequently, milk was pasteurised at 73.5 °C for 20 s and cooled to 39 °C. The pasteurised milk was pumped into a stainless-steel vat (5000 L capacity) and pre-acidified with citric acid (approx. 3 kg) to a milk-in-vat pH of 6.20 ± 0.02. The direct-vat-set starter was used at a level of 0.05 g/L. Subsequently, 450 mL of coagulant was added (18 IMCU/L). The additional levels of CC were predetermined by Chr. Hansen through a series of rheological experiments to achieve coagula of similar gel strength. After a set period of 17 min, the curd was cut into cubes (approx. 2 cm side) and kept under stirring for a further 25 min to harden curd grains. Finally, stirring was stopped to allow for curd settling for 15 min. After whey drainage, the curd was discharged on drainage tables and allowed to acidify for 240 min up to a pH value of 5.10 ± 0.05. The acidified fresh curd was then added with 0%, 10%, 20% or 30% (w/w) of the commercial vacuum-stored curd (Friesland Campina), corresponding to the addition of 0, 80, 160 and 240 kg of curd, respectively. Both curds were cut into strips (10 cm width) by the milling equipment and then homogeneously mixed during mincing. Stretching in hot water (85 °C) was carried out with a horizontal double-screw system. The curd reached a core temperature of 58 °C. A subsequent compressor shaped the pasta filata into 1 kg cylindrical blocks using jets of cold water. The blocks were immersed in chilled water (8–12 °C) for 30 min and brine salted (30% NaCl brine solution) at 6–8 °C for 60 min. Finally, the LMMC blocks were vacuum-packed, stored at 4 ± 1 °C and periodically sampled for up to 35 days.

2.3. Sampling

Chemical, rheological and microstructural analyses were performed at 0, 15 and 35 days of refrigerated storage. One cheese block was used for each sampling time and type of analysis.

2.4. Chemical Analyses

A central slice (approx. 130 g) of the cylindrical cheese block was cut, finely ground and used for chemical analyses. Standard ISO methods were applied for dry matter [16], fat [17] and protein [18], as described by Masotti et al. [19]. pH was measured on a cheese slurry obtained from 10 g cheese and 100 mL distilled water [20]. Sugars were determined using high-performance liquid chromatography (HPLC) as described in the ISO method ISO 22662:2024 [21]. Briefly, ground cheese (5 g) was diluted with 10 mL of MilliQ water and the mixture was homogenised with UltraTurrax^® homogeniser (Type T25, Janke & Kunkel, Staufen, Germany). The suspension was centrifuged (8000× g at 4 °C for 30 min) and filtered on paper. The filtrate (2 mL) was added with MilliQ water (4 mL) and Biggs reagent (1 mL). After an hour, the mixture was centrifuged (6000 rpm for 10 min at 4 °C) and filtered on 0.22 μm membrane filter (Millex GV, Millipore, Milano, Italy). The samples were then subjected to HPLC analysis (Alliance, Waters, Milford, MA, USA) on a BioRad Aminex HPX-87P column under isocratic elution with MilliQ water as the mobile phase, with a refractive index detector (Waters, Milford, MA, USA). The quantification of sugars was carried out by external standardisation. Nitrogenous fractions of cheese samples, consisting of soluble nitrogen (SN) at pH 4.4 (pH 4.4 SN), SN in 12% trichloroacetic acid (TCA-SN) and SN in 5% phosphotungstic acid (PTA-SN) were measured using the Kjeldahl method as described in standard ISO 27871:2011 [22] using an automatic digestor K-439 and a distillation and titration unit K-375 (Büchi Labortechnick, Flawil, Switzerland). Casein fractions were determined by reversed-phase high-performance liquid chromatography as described in Visser et al. [23]. All analyses were carried out in triplicate.

The pH 4.4 insoluble fractions of the LMMCA were also measured by urea–polyacrylamide gel electrophoresis (urea–PAGE) performed on a Hoefer vertical slab gel unit according to the method of Andrews [24] with modifications by Veloso et al. [25] on a 10% resolving gel. Electrophoresis was performed on a mini vertical electrophoresis unit (SE250, Hoefer, Holliston, MA, USA) at a constant voltage of 60 V, as described by Masotti et al. [26].

2.5. Texture Profile Analysis

The texture of LMMC samples was determined with a Texture Profile Analysis (TPA) using a 3365, Instron Universal Testing Machine (Instron Division of ITW Test and Measurement Italia S.r.l., Trezzano sul Naviglio, Italy) equipped with a 100 N load cell and a flat metal plate (80 mm diameter). Three slices (25 mm thick) were obtained from the central part of the LMMC block by using a wire cheese cutter. Then, from each slice, two cylindrical portions (25 mm diameter) were excised from the central part using a hand-coring device. The cylinders were conditioned in an incubator (Memmert UFE500, Schwabach, Germany) at 20 °C for 10 min before TPA. Then, the cylindrical samples were compressed twice to 60% deformation with a cross-head speed of 1 mm/s [26]. The parameters of hardness (N), adhesiveness (mJ), cohesion energy (-), springiness (mm) and chewiness (mJ) were calculated from the force–time curves according to Bourne [27].

2.6. Meltability

The meltability of each LMMC sample was measured using a modified Schreiber test [28] and an Image Analysis procedure purposely developed. Three slices (5 mm thick) were cut from the central part of each LMMC block by using an electric slicer. From the center of each slice, a 25 mm cylinder was obtained with a hand-coring device and placed in a Petri dish. Each cylinder was conditioned at 20 °C for 10 min in an incubator (Memmert UFE500, Schwabach, Germany) before melting in a forced air oven (iXelium, Whirlpool Italia s.r.l., Milan, Italy) at 230 °C for 4 min and then cooled to room temperature for 30 min. A picture of each Petri dish with a cheese sample was taken before and after melting by using a flatbed scanner (Perfection V850 Pro, Epson, Suwa, Japan) at 24-bit depth and 600 dpi resolution and covering the Petri dish with a black box to avoid light dispersion. The digital images were processed with the Image Pro Plus software (v. 7.0, Media Cybernetics, Rockville, MD, USA) to calculate the area of the samples before and after melting. The melting capacity of LMMC was evaluated as the area increase of the spread cheese after melting in the oven. The meltability was calculated as the change in the area of the melted cheese compared with the original area and expressed as a percentage.

2.7. Colour

The colour of LMMC was measured on the cylindrical samples used for meltability by using a tristimulus colorimeter (Chroma Meter II, Konica Minolta, Tokyo, Japan), with the standard illuminant C, previously calibrated with a standard white tile. Results are expressed in the CIE L*a*b* space, where L* corresponds to brightness (from black, 0, to white, 100), a* to the red-green component and b∗ to the blue-yellow component. Colour was measured before and after melting.

2.8. Microstructure by Confocal Laser Scanning Microscopy (CLSM)

The microstructure of LMMC samples was analysed by CLSM. Cube samples (approximately 2 × 2 × 2 mm) were carefully excised from the outer surface (skin) and from the core region (core) of the cheese using a razor blade (Figure S1). The samples were carefully placed within embryo dishes (Electron Microscopy Sciences, Hatfield, PA, USA) for staining, as previously described by D’Incecco et al. [29].

Briefly, samples were stained using Nile Red (Sigma Aldrich, St. Louis, MO, USA) for labelling fat and Fast Green FCF (Sigma Aldrich) for labelling protein. Just before staining, both stock solutions of Nile Red (1 mg/mL in 80% dimethyl sulfoxide) and Fast Green (1 mg/mL in Millipore MilliQ water) were diluted tenfold in MilliQ water. Observations were carried out using a CLSM microscope (Nikon A1+, Minato, Japan) with a 60× oil immersion objective. The excitation wavelengths were 485 nm for Nile Red and 635 nm for Fast Green. The emission wavelengths were 530–590 nm for Nile Red and 660–720 nm for Fast Green. For each sample, 10 images (1024 × 1024 pixels) in different areas of the sample were acquired. The images were subsequently processed using ImageJ Fiji 2.0 software to calculate the area of fat and protein components as well as the circularity and the diameter of fat structures. Image analysis of the micrographs was carried out on three representative images for each sample.

2.9. Statistical Analysis

Data analysis was performed using the Statgraphics Centurion software (v. 18.1.02, Statgraphics Technologies, Inc., The Plains, VA, USA) and JMP Pro 17.2.0 (JMP Statistical Discovery LLC, Cary, NC, USA). The chemical, rheological and structural analyses of each sample were carried out along three lines, one for each analysis. The results were analysed by multifactorial analysis of variance (MANOVA) to evaluate whether the type of coagulant, the percentage of curd added and the storage time had a significant effect (p < 0.05) on the parameters under study; the interaction “coagulant x added curd” was also studied. Fisher’s least significant difference (LSD) test was performed to compare the averages of the samples. Principal Component Analysis (PCA) was performed with all the available mean data after standardisation to evaluate possible sample patterns based on the experimental factors, as well as the variable weights.

3. Results and Discussion

3.1. Chemical Analyses

3.1.1. Chemical Composition and Proteolysis of the Stored Curd

Stored curds M and S (91 days aged) used for cheesemaking trials were evaluated in terms of composition and degree of proteolysis (

Table 1. Composition and soluble nitrogen (SN) fractions of stored curds M and S used as ingredients in cheesemaking trials. FDM: fat-in-dry matter; pH 4.4 SN: SN at pH 4.4. TCA-SN: SN in 12% trichloroacetic acid. PTA-SN: SN in 5% phosphotungstic acid. Curd M: stored curd added in the trial with Chy-Max^® M coagulant; Curd S: stored curd added in the trial with Chy-Max^® Supreme coagulant. Means within a row with different superscripts differ (p < 0.05). Means of three replicates ± standard deviation.

Parameter	Curd M	Curd S
Dry matter (%, w/w)	54.2 ± 0.3 a	55.2 ± 0.4 a
Protein (%, w/w)	24.5 ± 0.2 a	24.6 ± 0.1 a
Fat (%, w/w)	24.3 ± 0.1 a	25.0 ± 0.2 b
FDM (%, w/w)	44.7 ± 0.1 a	45.1 ± 0.2 a
Lactose (%, w/w)	<0.01 a	<0.01 a
Galactose (%, w/w)	0.6 ± 0.1 a	0.6 ± 0.1 a
pH 4.4 SN (% TN)	16.5 ± 0.7 a	15.5 ± 0.6 a
TCA-SN (% TN)	6.7 ± 0.4 a	5.9 ± 0.5 a
PTA-SN (% TN)	1.2 ± 0.3 a	0.8 ± 0.3 a
pH	6.23 ± 0.04 a	6.27 ± 0.03 a

). Curd S was characterised by a higher level of dry matter as a consequence of the slightly higher fat content. Lactose was absent in both samples, whereas a similar residual amount of galactose was detected. Nitrogenous fraction levels were similar in the two curds and suggested that the hydrolysis of CN was in an advanced stage in comparison to a fresh rennet curd [30]. These data were confirmed by observing the urea–PAGE electrophoretic patterns of curds M and S, which almost overlapped and showed intense bands of γ-CNs and α_s1-I-CN peptides (Figure 1).

3.1.2. Chemical Composition and Proteolysis of LMMC Samples

Physicochemical and biochemical properties of LMMC were studied to evaluate the effect of the three experimental factors under investigation, i.e., the type of coagulant (CC-M and CC-S), the addition rate (0–30%, w/w) of stored curd prior to the stretching step, and the storage time (up to 35 days). Compositional parameters fitted well with the data reported in the literature for LMMC [2,31].

CC-M and CC-S cheese slurries showed overlapping pH values at t0 (5.73 vs. 5.72, respectively). An increase of 0.2 pH units was recorded after 35 days of cold storage (

Table 2. Gross composition and soluble nitrogen (SN) fractions of low-moisture Mozzarella cheese samples. The p-values of the main factors and the interaction “coagulant x curd” are reported at the bottom. n.a.: not analysed; CC-M: Chy-Max^® M coagulant; CC-S Chy-Max^® Supreme coagulant; TN: total nitrogen; pH4.4-SN: SN at pH 4.4; TCA-SN: SN in 12% TCA; PTA-SN: SN in 5% PTA.

Main Factors			Chemical Parameters
A: Coagulant	B: Curd Addition (% w/w)	C: Storage Time (Days)	pH	Dry Matter (%)	Protein (%)	Fat (%)	FDM (%)	Galactose (% w/w)	pH 4.4-SN (% TN)	TCA-SN (% TN)	PTA-SN (% TN)
CC-M	0	0	5.73	50.9	22.8	23.8	46.8	0.48	3.0	1.1	0.4
		15	5.88	51.2	22.7	24.1	47.1	0.44	4.2	1.2	0.5
		35	5.90	50.8	22.5	23.9	47.1	0.07	9.2	2.3	0.4
	10	0	5.72	50.3	22.1	23.7	47.0	0.53	5.6	1.5	0.5
		15	5.86	50.8	22.7	23.0	45.2	0.44	5.7	2.1	0.4
		35	5.90	50.8	22.0	23.8	46.8	n.a.	10.1	2.9	0.5
	20	0	5.78	50.4	22.0	23.8	47.2	0.53	5.3	1.9	0.5
		15	5.85	50.4	21.5	23.6	46.8	0.51	6.1	2.3	0.5
		35	5.90	50.2	21.6	23.6	47.1	n.a.	7.9	2.5	0.5
	30	0	5.87	50.7	22.3	23.5	46.3	0.46	7.1	2.4	0.6
		15	5.90	51.0	22.1	23.9	46.8	0.45	7.8	2.5	0.5
		35	5.93	50.9	22.3	23.9	47.1	n.a.	10.5	3.5	0.5
CC-S	0	0	5.72	51.9	22.6	25.5	49.1	0.65	2.8	1.0	0.4
		15	5.82	52.4	22.7	25.5	48.7	0.53	3.6	1.2	0.5
		35	5.96	51.5	23.0	25.0	48.5	0.06	6.0	2.0	0.5
	10	0	5.84	52.2	22.2	24.8	47.6	0.53	3.9	1.4	0.4
		15	5.92	53.0	23.2	25.4	47.9	0.51	5.3	1.8	0.5
		35	5.90	51.9	23.1	24.9	47.9	n.a.	8.1	2.2	0.5
	20	0	5.79	52.0	21.8	24.5	47.1	0.60	6.2	1.6	0.5
		15	5.90	52.5	22.5	24.5	46.6	0.51	7.9	2.1	0.6
		35	5.95	52.1	22.0	25.8	49.5	n.a.	8.9	2.7	0.5
	30	0	5.87	52.7	24.0	24.8	47.0	0.55	6.6	2.1	0.4
		15	5.94	52.8	23.0	25.2	47.8	0.52	7.4	2.4	0.6
		35	6.02	52.3	23.0	25.0	47.7	n.a.	9.8	3.4	0.5
Main factors (p-value)
A			0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0013	0.0000	0.0996
B			0.0000	0.0000	0.0000	0.2862	0.1012	0.0194	0.0000	0.0000	0.0333
C			0.0000	0.0000	0.9300	0.4296	0.3497	0.0000	0.0000	0.0000	0.0009
Interaction (p-value)
AxB			0.0494	0.0000	0.0001	0.8255	0.6285	0.0023	0.0000	0.1018	0.7464

). The addition of stored curd contributed to the increase of pH in LMMC as a function of the addition rate. This effect was due to the remarkably high acidity levels of the stored curds (mean pH value = 6.25).

CC-S cheese samples were characterised by higher values in dry matter and fat contents with respect to the CC-M counterparts. These data could be explained by the higher proteolytic specificity of the CC-S enzyme and the subsequent formation of a compact coagulum capable of entrapping more fat than CC-M, as reported in the technical sheet of the manufacturer. Lactose and glucose were absent in all cheese samples. Only residual levels of galactose were recorded in LMMC samples. This outcome was in line with data from the literature for Mozzarella cheese obtained using Streptococcus thermophilus as the starter lactic acid bacterium [32].

Proteolysis, assessed by means of urea–PAGE electrophoresis and Kjeldahl nitrogen fractions, increased during storage in all samples. The patterns of 1-day-old CC-M and CC-S samples overlapped and were characterised for the major CN fractions bands (α_s1-CN and β-CN). The plasmin-associated hydrolysis of β-CN to γ-CN during ripening was weak and no significant decreases in β-CN fraction were revealed by HPLC (Figure 1). The major activity of chymosin on α_s1-CN is the cleavage of the Phe₂₃–Phe₂₄ bond, yielding the α_s1-I-CN peptide. In this study, the temperature of the stretched curd reached about 58 °C, causing an incomplete thermal inactivation of this enzyme, which in turn resulted in an increase of the release of this peptide upon storage. This phenomenon fits well with data reported in the literature [6]. In particular, the contribution of the residual coagulant activity to α_s1-CN hydrolysis of LMMC was slightly more intense in CC-M than in CC-S samples obtained using fresh curd, as revealed by the HPLC analysis of CN fractions (Figure 2). The relative chromatographic area of α_s2-CN and β-CN did not change during storage with both coagulants. Overall, the differences in the hydrolysis of the CN fractions were not statistically significant. The slightly lower proteolytic activity of CC-S enzyme was confirmed by the Kjeldahl maturation index.

The pH 4.4-SN is a heterogeneous nitrogen fraction representing an accepted index of the degree of primary proteolysis. This fraction sensitively increased as a function of the cheese age and the amount of added curd. The use of the CC-M coagulant promotes a higher level of this nitrogen fraction, too (Table 2). The TCA-SN contains small peptides (approx. < 20 amino acids length), as well as low-molar-mass nitrogen compounds [22]. It represents an index of secondary proteolysis, which takes place through the action of indigenous and microbial enzymes. Values of the TCA-SN fraction increased through storage up to 35 days with both types of coagulants. Nevertheless, the increase in absolute levels remained low in comparison to the steep increase observed for the pH 4.4-SN fraction. The values of PTA-SN, representing small peptides with fewer than four amino acid residues, did not modify sensitively in all cheese samples during storage.

3.2. Texture Profile Analysis

Results of TPA showed that the type of coagulant, the addition of stored curd and cheese storage altered the texture of LMMC samples (

Table 3. Mean values of texture parameters of low-moisture Mozzarella cheese samples. The p-values of the main factors and the interaction “coagulant x curd” are reported at the bottom (CC-M: Chy-Max^® M coagulant; CC-S Chy-Max^® Supreme coagulant).

Main Factors			Texture Parameters
A: Coagulant	B: Curd Addition (% w/w)	C: Storage Time (Days)	Hardness (N)	Adhesiveness (mJ)	Cohesiveness (-)	Springiness (mm)	Chewiness (mJ)
CC-M	0	0	67.5	1.4	0.52	13.48	526
		15	53.1	4.7	0.44	13.72	349
		35	44.9	5.0	0.36	14.14	245
	10	0	47.5	1.9	0.50	13.76	356
		15	42.6	3.9	0.42	13.57	269
		35	34.6	4.6	0.42	14.15	216
	20	0	52.7	1.5	0.46	13.83	365
		15	46.5	3.8	0.44	13.88	306
		35	44.7	4.9	0.39	14.26	264
	30	0	48.6	3.0	0.47	13.94	340
		15	47.2	4.1	0.41	13.71	291
		35	42.8	5.5	0.42	14.21	272
CC-S	0	0	60.1	0.8	0.51	13.23	460
		15	42.2	3.1	0.38	13.45	242
		35	49.2	5.1	0.41	14.42	306
	10	0	70.9	0.9	0.49	13.76	517
		15	54.8	3.9	0.41	13.59	337
		35	43.4	5.9	0.40	13.81	260
	20	0	61.3	1.0	0.51	13.67	466
		15	48.8	3.1	0.42	13.60	303
		35	40.0	4.8	0.42	14.06	251
	30	0	64.6	0.9	0.54	13.75	525
		15	52.6	3.2	0.42	13.59	335
		35	47.5	5.1	0.42	13.92	298
Main factors (p-value)
A			0.0000	0.0002	0.2414	0.0326	0.0000
B			0.0017	0.1380	0.4145	0.4101	0.0339
C			0.0000	0.0000	0.0000	0.0000	0.0000
Interaction (p-value)
AxB			0.0000	0.0287	0.0332	0.8840	0.0000

). Hardness is a measure of the force required to compress the sample during the first cycle and is related to the strength of the cheese matrix. Samples manufactured with CC-S showed higher values of hardness, possibly related to the lower moisture content (more specifically, a lower moisture-to-protein ratio) compared to CC-M samples. This result confirms the formation of a more compact coagulum in CC-S samples, able to entrap more fat and thus being stronger than CC-M samples. For the same reason, CC-S resulted in significantly weaker elasticity (lower values of springiness) and adhesiveness. Similarly, chewiness, which is the work required to masticate the sample to be swallowed, was higher in CC-S.

The addition of stored curd decreased LMMC hardness and chewiness (Figure S2). Independently of the CC used, the textural parameters hardness and chewiness were reduced at 10% and 20% addition rates. This was not surprising as a consequence of higher CN hydrolysis characterising stored curds. The breakdown products of CN are largely water soluble and cannot contribute to the protein network [33]. Storage time significantly affected all the texture parameters, decreasing hardness, cohesiveness and chewiness, while increasing adhesiveness and springiness (Figure S3). This behaviour was in line with data reported in the literature and ascribed to proteolytic phenomena promoting the reorganisation of major components [5]. In the first stage of storage, the coarse protein matrix of cheese was rapidly converted to a smoother and more homogeneous structure due to the hydrolysis of α_s1-CN moiety [33]. A major decrease in hardness and chewiness was observed in CC-M samples, which were characterised by a higher extent of proteolysis. Cold storage also promoted a reduction in cohesiveness (Table 3), probably related to the pH increase. Actually, as reported by Lawrence et al. [33], cohesion increased from pH 5.8 to pH 5.2, while during storage, the LMMC pH increased from 5.80 to 5.93. Springiness significantly increased from 15 to 35 days of storage independently of the coagulant used. This is in line with results by To et al., which demonstrated a direct correlation between pH and springiness during storage of low-moisture part-skim mozzarella [34]. Moreover, it has to be noted that the average percentage increase in springiness was very low (3.3%). Other than springiness, adhesiveness also significantly increased, but to a significantly higher extent, especially in LMMC produced with CC-S, which had average values five times higher at the end of storage, compared to 1.8 times higher values observed in LMMC obtained with CC-M. The higher adhesiveness can represent a problem during slicing operation, favouring the sticking of the product to the slicer [5]. An increase in adhesiveness with storage time was also observed by Zisu and Shah [35]. Curd addition did not significantly affect adhesiveness. Nevertheless, the “coagulant x curd addition” interaction was significant, showing that when CC-M is used, there is an increase in adhesiveness at 30% curd addition (i.e., 4.18 vs. 3.70 mJ), which is not observed in the cheese samples produced with CC-S (i.e., 3.07 vs. 3.00 mJ) (Figure S4).

3.3. Meltability

Meltability is an important property of LMMC because it is used as an ingredient in pizza, toast and other similar food products. It is defined as the ability of cheese to flow and spread and is related to the number and strength of the interactions among casein micelles. The effect of coagulant type on LMMC meltability was not significant, while a slight but significant increase as a function of curd addition was observed, maybe due to the higher CN hydrolysis observed in stored curds. However, the stronger effect was exerted by storage time, which caused an increase in meltability (

Table 4. Mean values of meltability and colour parameters (L*, a*, b*) of low-moisture Mozzarella cheese samples before and after melting as a function of type of coagulant, added stored curd and storage time. The p-values of main factors and the coagulant-curd interaction are reported at the bottom.

Main Factors			Meltability	Colour Before Melting			Colour After Melting
A: Coagulant	B: Curd Addition (% w/w)	C: Storage Time (Days)	(%)	L*	a*	b*	L*	a*	b*
CC-M	0	0	183	88.3	−5.2	12.2	68.7	−6.6	10.3
		15	292	84.0	−5.5	13.6	64.0	−4.8	6.8
		35	249	81	−5.9	13.9	65.2	−5.1	6.9
	10	0	239	84.2	−5.0	13.6	63.0	−5.0	8.5
		15	272	83.0	−5.5	14.7	65.3	−4.5	7.4
		35	228	82.8	−6.2	14.5	67.7	−5.9	8.3
	20	0	285	87.3	−5.7	13.6	63.4	−4.5	7.6
		15	228	84.8	−5.8	15.6	69.1	−5.3	8.8
		35	246	82.6	−6.0	14.1	64.3	−4.8	6.9
	30	0	289	83.5	−5.4	15.0	61.2	−4.3	7.6
		15	299	84.3	−6.1	17.1	64.7	−4.7	7.7
		35	235	82.1	−6.5	16.4	68.7	−6.1	9.5
CC-S	0	0	250	88.8	−5.0	12.4	69.9	−6.4	11.9
		15	275	83.3	−6.0	14.2	62.9	−4.8	6.6
		35	289	83.8	−6.3	14.6	65.1	−4.7	5.7
	10	0	126	86.0	−5.3	13.2	70.7	−7.2	13.2
		15	283	82.4	−6.0	15.4	64.6	−4.5	7.5
		35	257	82.3	−6.3	15.9	66.5	−5.4	8.1
	20	0	200	88.5	−5.4	13.8	65.35	−6.9	12.8
		15	270	83.2	−6.4	16.7	65.4	−5.5	8.8
		35	282	81.4	−6.6	15.9	64.4	−4.9	6.5
	30	0	231	87.5	−4.8	12.9	62.8	−5.4	8.9
		15	290	83.2	−6.7	18.2	62.5	−4.9	8.2
		35	278	80.6	−6.6	16.4	63.8	−5.2	7.6
Main factors (p-value)
A			0.7279	0.4410	0.0086	0.0174	0.4380	0.2269	0.1898
B			0.0489	0.0043	0.0003	0.0000	0.0995	0.5797	0.5744
C			0.0000	0.0000	0.0000	0.0000	0.4640	0.0030	0.0000
Interaction (p-value)
AxB			0.2368	0.5089	0.4739	0.0576	0.3141	0.3790	0.4092

, Figure S5), especially in the first period (15 days), in agreement with data reported by Guinee et al. [5]. This behaviour may be attributed to the age-related proteolysis of the para-CN network. Proteolysis indeed results in weaker protein–protein and protein–fat interactions, making the protein network looser and increasing melting properties [36]. Water state can also affect Mozzarella cheese meltability during the first days of storage, as demonstrated by McMahon et al. [37]. During the initial days of storage, the protein matrix becomes more hydrated, thus allowing the proteins to slip more easily one on the other. This effect, combined with the lubricating properties of fat, results in improved meltability.

The addition of stored cured was reflected in a slight increase in the spread area of melted cheese. Nevertheless, the meltability percentages of 0% and 30% curd addition almost overlapped.

3.4. Colour

The colour of LMMC samples was investigated before and after melting (Table 3). Before melting, LMMC colour was mainly affected by storage time and curd addition, while the type of coagulant had a minor effect only on a* and b*. In a recent study, Adachi et al. [38] clarified factors affecting the browning in a model processed cheese during storage. The authors reported that the browning index (delta-E value) strongly correlated with the concentration of galactose and pH value. During storage, brightness (L*) slightly decreased (5% on average), while the green and yellow components increased an average of 20% and 14%, respectively. The same trend was observed also as a function of stored curd addition, especially affecting the yellow index (average variation of 19%); this result is due to the colour of the stored curd, which had a significantly more intense yellow component compared to LMMC (b* = 22.0 ± 0.4 vs. 12–18).

After melting, the most affected parameter was again the yellow index, which decreased by 42% on average, followed by the brightness (average decrease of 22%). A less clear effect was observed for a*, even if a general increase was observed after melting, indicating a redder colour. Cheese browning due to melting is well known and affected by several factors, including residual reducing sugars, proteolysis, free amino acids and pH value, among others [39].

As for the experimental factors studied, only storage time significantly affected a* and b* after LMMC melting, with an increase of the red component (9% on average) and a decrease of the yellow index (26% on average), indicating a less brown colour of the stored melted cheese compared to the fresh one. This result may be attributed to the lower amount of galactose available in the stored LMMC, which is reflected in a lower rate of a Maillard reaction during melting [39].

3.5. Microstructure Analysis

The stretching phase in hot water transforms the curd into a plastic consistency and promotes the formation of a protein network oriented in the stretching direction, incorporating pockets of fat globules, whey and bacteria [8]. The microstructure of samples was investigated by CLSM within the skin layer as well as in the core region of the cylindrical blocks of LMMC samples (Figure S1). The orientation of the fibrous protein along the stretching direction was particularly defined in the skin layer of both CC-M and CC-S samples at T0 (Figure 3A,C). The fat in the skin layer appeared as discrete fat globules, often in chains. This organisation remained partially visible also in the core region of CC-S (Figure 3C’) but it disappeared in CC-M LMMC, where the aggregation and partial coalescence of fat droplets was observed. However, after 35 days, this structure was not evident in the core of CC-S samples as well. The effect of the 30% addition of stored curd slightly changed the fibrous protein orientation that was still perceived in the skin layer, although it showed a denser protein structure (Figure 3). The different structural organisation between the skin and core layer agrees with a previous observation of Mozzarella cheese carried out by CLSM and cryo-scanning electron microscopy [40].

Proteolysis as well as modifications in the water state of proteins are drivers of modifications in the microstructure of cheese [7]. Major changes were observed in the protein matrix in the core region and consisted of small protein fragments appearing in the serum pores surrounding fat globules (Figure 4B,E). These fragments were observed in 35-day-old LMMC samples, compared to LMMC at time 0, regardless of the type of coagulant used. The presence of small protein fragments in 35-day-old LMMC is possibly an effect of proteolysis in accordance with HPLC data on the αs₁-CN breakdown. The fragmentation of protein matrix is also in agreement with the textural behaviour of samples showing a decrease in hardness and cohesion during storage, as well as an increase of meltability. By day 35 of ripening, the fat showed aggregation and partial coalescence among globules (Figure 4B’,E’), especially in the core region, showing larger fat patch areas compared to LMMC at T0. An image analysis showed Feret’s diameter and area of fat patches to increase during storage as an effect of partial coalescence, regardless of the type of coagulant used. We can speculate that the fragmentation of proteins around fat globules could have indirectly contributed to the size increase of fat patches. It has been previously reported that fat content and its physical state directly affect the meltability of low-moisture Mozzarella cheese [41,42]. As a consequence, we can hypothesise that the intense fat coalescence contributed, together with proteolysis, to an increased meltability at T35 compared to T0. No differences were perceived in the microstructure of LMMC at T35 regardless of the type of coagulant and the curd addition.

3.6. Principal Component Analysis

A Principal Component Analysis (PCA) was conducted to evaluate the contribution of the factors of variability considered in this study to the pattern of samples within the vector space (Figure 5). The first two PCs accounted for 63.9% of the total variability, with the largest contribution attributed to PC1. This component primarily explained the distribution of LMMC samples as a function of storage time. Specifically, fresh cheeses were predominantly located in the negative region of PC1, whereas LMMC stored for 15 or 35 days was mostly in the positive region of PC1. The factor storage time, irrespective of the type of coagulant used, was associated with the increased meltability, adhesiveness, springiness, TCA-SN, PTA-SN, pH 4.4-SN and pH of LMMC, while resulting in decreased hardness, cohesiveness, chewiness and colour brightness. The factor type of coagulant was primarily associated with the distribution of samples along PC2. LMMC produced using CC-S was predominantly positioned in the positive region of PC2, regardless of curd addition or storage time. This distribution was mainly driven by higher values of dry matter, fat, protein and fat-in-dry-matter (FDM) compared to LMMC produced using CC-M, mainly located in the negative region of PC2. Finally, no significant effect of curd addition was observed in the distribution of LMMC within the vector space of the two first PCs.

4. Conclusions

The chemical properties of LMMC were affected by the type of coagulant. Cheese samples made with CC-S were characterised by higher fat content and a concomitant higher dry matter level with respect to CC-M samples. Upon storage at 4 °C up to 35 days, the rate of proteolysis in CC-S samples was lower. This is likely due to the specific milk clotting activity and high proteolytic specificity promoting the formation of a compact protein-to-protein network. This trend was reflected in slight differences in microstructure and textural parameters. The type of coagulant did not exert a specific effect on the melting behaviour. The addition of stored curd before the stretching step in LMMC making was demonstrated to be a feasible technological practice promoting only minor changes in terms of chemical, microstructural and texture properties under the conditions adopted. These properties were mainly modified by refrigerated storage.

Overall, these results could have practical applications because manufacturers, by properly selecting ingredients, can control functionalities and suit the end-use applications of LMMC, thus conforming to customer specifications or consumer acceptability. Given the main use of this type of cheese as an ingredient for pizza, a post-cooking characterisation for sensory and functional properties of the LMMC thus produced would be of interest.