Application of Animal- and Plant-Derived Coagulant in Artisanal Italian Caciotta Cheesemaking: Comparison of Sensory, Biochemical, and Rheological Parameters

Abstract

Consumer interest in vegetarian, ethical, and clean-label foods is reviving the use of plant-derived milk coagulants. Cardosins from Cynara cardunculus (“thistle”) are aspartic proteases with strong clotting activity, yet their technological impact in cheese remains under-explored. This study compared a commercial thistle extract (PC) with traditional bovine rennet rich in chymosin (AC) during manufacture and 60-day ripening of Caciotta cheese. Classical compositional assays (ripening index, texture profile, color, solubility) were integrated with scanning electron microscopy, three-dimensional surface reconstruction, and descriptive sensory analysis. AC cheeses displayed slower but sustained proteolysis, yielding a higher and more linear ripening index, softer body, greater solubility, and brighter, more yellow appearance. Imaging revealed a continuous protein matrix with uniformly distributed, larger pores, consistent with a dairy-like sensory profile dominated by milky and umami notes. Conversely, PC cheeses underwent rapid early proteolysis that plateaued, producing firmer, chewier curds with lower solubility and darker color. Micrographs showed a fragmented matrix with smaller, heterogeneous pores; sensory evaluation highlighted vegetal, bitter, and astringent attributes. The data demonstrate that thistle coagulant can successfully replace animal rennet but generates cheeses with distinct structural and sensory fingerprints. The optimization of process parameters is therefore required when targeting specific product styles.

1. Introduction

Milk coagulation is a crucial biochemical transformation in cheesemaking, triggered by the enzymatic cleavage of κ-casein in milk micelles, leading to casein aggregation and curd formation. Traditionally, this process has relied on animal-derived rennets extracted from the abomasum of unweaned calves, with chymosin being the predominant active enzyme. Chymosin exhibits a high degree of specificity, hydrolyzing the Phe105–Met106 bond of κ-casein, which initiates the destabilization of casein micelles and promotes a cohesive gel matrix. Pepsin, a secondary enzyme in animal rennet preparations, contributes to further proteolysis during ripening, but its broader specificity may also lead to the formation of bitter peptides [1,2]. The increasing consumer demand for vegetarian, ethical, and sustainable food products has spurred interest in non-animal coagulants.

The application of plant-derived coagulants in cheesemaking is an ancient practice, particularly prominent in Mediterranean regions, where both climatic conditions and cultural traditions fostered the use of non-animal rennet alternatives. Among these, plant-based coagulants, especially aspartic proteases such as cardosins extracted from Cynara cardunculus (commonly known as wild cardoon), have attracted substantial scientific interest. These enzymes have long been employed in traditional Mediterranean cheese production, notably in Portugal, as well as regions of Italy and Spain [3].

Cardosins, derived from Cynara cardunculus L., exhibit milk-clotting activity comparable to that of chymosin but differ in their substrate specificity. Unlike chymosin, which primarily targets κ-casein, cardosins act on a broader range of casein fractions, including αs- and β-caseins. This broader proteolytic activity may result in a more heterogeneous curd structure and can lead to the generation of off-flavors, such as bitterness and astringency [4]. Among the cardosins, Cardosin A and Cardosin B are the most extensively characterized. These aspartic proteases are predominantly localized in the floral tissues and exhibit high milk-clotting activity (MCA). Cardosin A demonstrates relatively high specificity for κ-casein, similar to chymosin, whereas Cardosin B exhibits a broader proteolytic spectrum, effectively cleaving αs- and β-caseins. The proteolytic profile of Cardosin B contributes to increased levels of free peptides, which can alter curd texture and contribute to undesirable sensory attributes such as bitterness [5].

Despite their proteolytic drawbacks, cardosins are gaining renewed interest due to their alignment with vegetarian, religious, and sustainability-based dietary trends. Moreover, cardoon plants are hardy, drought-resistant perennials that thrive in marginal lands, enhancing their value from an agroecological perspective [3]. Beyond cardosins, several other plant-derived coagulants have been investigated. For instance, fig latex (Ficus carica) contains ficin, a cysteine protease with significant milk-clotting ability. Although effective, ficin tends to exhibit high proteolytic activity, often resulting in bitterness and rapid cheese softening [6,7]. Similarly, papain from Carica papaya and bromelain from Ananas comosus have been studied. These enzymes can coagulate milk but typically produce fragile and overly proteolyzed curds, limiting their application to specific cheese types or in low concentrations [8]. Another promising enzyme is actinidin from kiwi fruit (Actinidia deliciosa), which has shown potential for controlled milk clotting with moderate proteolytic side effects [9]. Extracts from Moringa oleifera (moringa) [10] and Withania coagulans (panirband) [11] are also under investigation due to their reported clotting activity and traditional use in local dairy products in parts of Asia and Africa. To mitigate the limitations of plant coagulants, research has focused on enzyme purification, dose optimization, and recombinant expression. For example, the heterologous expression of cardosin A in Kluyveromyces lactis has enabled the production of recombinant enzymes with enhanced purity and reduced non-specific activity [12]. In conclusion, plant-based coagulants represent a promising but complex alternative to animal rennets. While cardosins offer high potential, their proteolytic profile requires careful management. The broader class of plant coagulants continues to expand, with ongoing efforts to adapt enzymatic characteristics to meet technological, sensory, and consumer demands in dairy production.

While studies have reported the successful application of plant coagulants in the manufacture of ovine and caprine cheeses [3,13,14], their use in cow milk cheeses such as Caciotta has not been extensively characterized in terms of both technological and organoleptic performance. The term Caciotta (or “Caciofiore”) is a generic designation applied to various types of cheese, both fresh and aged, produced from cow, sheep, or mixed milk. In particular, the name “Caciofiore” is traditionally associated with cheeses coagulated using floral extracts of Cynara cardunculus (wild cardoon), historically employed as a plant rennet.

Caciotta is one of the oldest expressions of Italian cheesemaking tradition. Its name, derived from the Latin caseus (cheese), attests to its ancient origin. Among Roman authors, Martial, in the 1st century AD, mentions a “cheese from Tuscany,” whose description closely resembles that of a traditional Caciotta. During the Middle Ages, the Tuscan variety known as Marzolina was already well known and highly appreciated.

Over the centuries, Caciotta has become a regional specialty in several parts of Italy. From Tuscany, which still produces some of the most renowned varieties, its production spread to Lazio, Abruzzo, Marche, and other regions. Today, Caciotta is manufactured nationwide, using both traditional and modern techniques that vary according to local practices. In industrial and artisanal dairies of Northern Italy, sweet Caciottas made with cow’s milk and characterized by a soft texture are predominant. In Central and Southern Italy, production is still largely based on sheep’s milk, or a mixture of sheep’s and cow’s milk.

The typical shape of a Caciotta is cylindrical and flattened, with a diameter ranging from 8 to 10 up to 16 cm, a height of 4 to 8 cm, and a weight varying between 800 g and approximately 2 kg. The internal texture also exhibits notable variability: it can be very soft, white or chalky, crumbly, sweet, or slightly yellowish, depending on the type of milk used, the technological process, and the degree of ripening, ranging from one to several weeks [15].

Its relatively short ripening time makes it sensitive to early proteolytic activity, thereby providing a suitable model for evaluating the effects of coagulant type on maturation dynamics. Cheesemaking performance and the biochemical pathways of ripening are influenced not only by the primary structure of milk proteins and fat content but also by the action of residual coagulant enzymes and the microstructure of the curd matrix [1,2,16]. Chymosin’s highly specific cleavage pattern tends to produce fine, uniform casein networks with limited non-specific proteolysis, resulting in stable and smooth-textured cheeses. In contrast, cardosins’ broader proteolytic activity may cause fragmentation of the protein matrix and formation of heterogeneous microdomains, influencing water retention, solubility, and mechanical strength [14]. Moreover, sensory science approaches, particularly Quantitative Descriptive Analysis (QDA), allow for the comprehensive profiling of aroma, taste, and mouthfeel attributes associated with different enzymatic treatments [4,16]. Our goal is to elucidate how enzymatic specificity influences cheese quality, and to assess the feasibility of plant-based coagulants as sustainable alternatives in modern dairy production.

2. Materials and Methods

2.1. Artisanal Production of Caciotta Cheese with Animal and Plant Coagulants

For the production of Caciotta cheese using animal-derived coagulant, 100 L of fresh cow’s milk was used. The milk was obtained from the evening milking of Bruna Alpina cows raised in a dairy farm located in Trentino-Alto Adige, Italy. The cows were fed a diet based on locally sourced forages. Immediately after milking, the milk was cooled to 4 °C.

For the production of Caciotta cheese using an animal-based coagulant (AC), raw milk was then heated to 38 °C, and a commercial starter culture (Lyofast Y 082 D, Sacco SRL, Cadorago, Italy) containing Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricus was added (1.0 UC/100 L). After 30 min of incubation, calf rennet, called animal coagulant (AC) (Ceska^® Kalase 150, 150 IMCU/mL, Česká, Czech Republic; CSK Food Enrichment, Leeuwarden, The Netherlands), containing 75% chymosin and 25% bovine pepsin, was added to the milk. Calf rennet was diluted to 10% (v/v) in distilled water to obtain a final concentration of 50 IMCU per liter of milk. The milk-clotting activity (International Milk Clotting Units, IMCU) was determined according to IDF Standard 157:2007 [17].

Coagulation was allowed to proceed for approximately 15 min, after which the curd was initially cut coarsely with a knife. After an additional 5–10 min, it was further broken into smaller pieces using a curd cutter to improve whey drainage.

For the production of Caciotta using a plant-based coagulant (PC), the enzymatic preparation Galium was used. The coagulant used is a commercially available enzymatic preparation. Patented by Laboratorio Prodor in 1994, it is extracted from Cynara cardunculus (organic wild cardoon). According to the manufacturer, its milk-clotting activity is 1:6000. On average, 10 g of plant coagulant was used per 10 L of milk (equivalent to 100 g per 100 L). Milk was heated to 38 °C, starter cultures (Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricus) were added, and the plant coagulant was introduced. Compared to the traditional rennet-based process, curd formation occurred more slowly; therefore, coagulation was allowed to proceed for 25 min before cutting and breaking the curd. The curd was considered ready for cutting when it reached a firm and cohesive texture, with smooth and well-defined edges upon gentle pressing with a flat spatula. At this point, the gel also began to release clear whey upon slight surface pressure, indicating sufficient syneresis and structural integrity for cutting. These criteria were consistently applied across all trials to ensure comparable gel firmness regardless of coagulation time.

Molding was carried out by allowing the curd to flow into the molds through the opening of a valve, while the whey was simultaneously drained off. In the case of the Caciotta with AC, 13 cheese wheels were obtained. In contrast, production with the plant coagulant (PC) yielded 11 cheese wheels, with each Caciotta weighing around 800 g. The average dimensions of the cheese were approximately 13–14 cm in diameter and 5–6 cm in height.

After molding, the cheeses underwent a series of turns, at 21 °C, as follow: the first immediately after placement in the mold, followed by additional rotations at 20 min, 45 min, 3 h, and finally after 12 h. These rotations aimed to improve the shape, promote curd compaction, and homogenize air distribution within the cheese. Upon completion of the turning steps, on the day following production, dry salting was performed. The cheeses were then transferred to a ripening room maintained at approximately 16 °C with 90% relative humidity. Ripening was carried out over a period of 60 days, during which the cheeses were turned daily to promote uniform drying.

2.2. Chemical Analyses

Chemical analyses were performed on milk, curd, and cheese samples. Moisture content in the milk used for the production of both cheese types (AC and PC) was determined according to AOAC Official Method 945.46 (17th ed., 2003) [18]. Ash content was assessed using the same AOAC method [18]. Protein content was analyzed following AOAC Method 991.20 (17th ed., 2003) [18] using the Kjeldahl method with a Kjeltec 2300 (Foss A/S Tecator^TM system, Hilleroed, Denmark), based on nitrogen mineralization. Lipid content was determined using the German official Röse-Gottlieb method, with fat extraction performed via a SOXTET 255 (Foss A/S Tecator^TM system, Hilleroed, Denmark) [18]. Carbohydrate content was calculated indirectly by subtracting the percentage values of moisture, ash, protein, and lipid from 100, according to the following formula:

$$\mathrm{\%}\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{h}\mathrm{y}\mathrm{d}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{s}=[100-(\mathrm{m}\mathrm{o}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}+\mathrm{a}\mathrm{s}\mathrm{h}+\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{i}\mathrm{n}+\mathrm{l}\mathrm{i}\mathrm{p}\mathrm{i}\mathrm{d}\mathrm{s}\left)\right].$$

For each production batch, samples of curd and cheese were analyzed at three stages: one week, one month, and two months of ripening. The following parameters were assessed, using the same methods applied for milk analysis: moisture, ash, protein, lipids, carbohydrates, water-soluble protein, and ripening index [18]. For each cheese wheel, two radial cuts were made from the center to extract a wedge. The rind and any superficial mold were removed, and the edible portion was homogenized using a GRINDOMIX GM 200 Retsch (GmbH, Haan, Germany) homogenizer for 10 s at 4500 rpm. Water-soluble protein content was determined using the Kjeltec 2300 FOSS TECATOR system, based on the cold-water extraction method. The extract was filtered, and the filtrate was subjected to Kjeldahl digestion, according to the International Dairy Federation (IDF), Brussels, Belgium [19]. The ripening index was then calculated as the ratio of water-soluble protein to total protein [20].

$$\mathrm{R}\mathrm{i}\mathrm{p}\mathrm{e}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{g} \mathrm{I}\mathrm{n}\mathrm{d}\mathrm{e}\mathrm{x} \mathrm{\%}={\displaystyle \frac{\mathrm{W}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{r} \mathrm{s}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e} \mathrm{p}\mathrm{r}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{i}\mathrm{n}}{\mathrm{T}\mathrm{o}\mathrm{t}\mathrm{a}\mathrm{l} \mathrm{p}\mathrm{r}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{i}\mathrm{n}}}\times 100$$

2.3. Color Measurement

Color analysis was performed on Caciotta cheese (PC and AC) cylinders measuring 1 cm in diameter and 1 cm in height, after exposure to air for 1 h at 21 ± 1 °C. Measurements were taken from five consecutive points spanning from the rind to the center of the cheese. A spectrophotometer (CM-600, Minolta Corp., Ramsey, NJ, USA) equipped with a D65 standard illuminant (daylight simulation) and a 10° standard observer was used for color evaluation. The results were expressed according to the CIE Lab* color space system (CIE, 1986) [21], where L* indicates lightness (ranging from 0 = black to 100 = white), a* represents the green–red axis (negative values = green, positive values = red), and b* corresponds to the blue–yellow axis (negative values = blue, positive values = yellow).

2.4. Texture Profile Analysis (TPA)

Texture profile analysis was performed using a TA.XTplus Texture Analyzer (Stable Micro Systems, London, UK) equipped with a 500 N load cell (Stable Micro Systems, London, UK) and controlled by the Texture Exponent software (Stable Micro Systems, London, UK TA.XTPlus 100 Connect). Cylindrical cheese samples were obtained using a corer with a diameter of 1.15 cm (Stable Micro Systems, London, UK), yielding cylinders with a base area of approximately 1 cm². The rind was removed, and each sample was trimmed with a knife to a uniform height of 2 cm. To obtain representative measurements, five replicates were taken from each cheese type (animal coagulant, AC, and plant coagulant, PC), with sampling sites distributed from the center to the outer regions of the cheese wheel, avoiding the rind in all cases. Each sample was centrally positioned under a 35 mm diameter cylindrical probe (Stable Micro Systems, London, UK). The test was carried out with a pre-load of 20 g and a crosshead speed of 2 mm/s. The probe compressed the sample to 30% of its original height, then returned to the initial position before executing a second compression cycle.

Data were collected at a rate of 250 points per second and were used to generate force–time curves from which the following TPA parameters were calculated: hardness (peak force during the first compression), cohesiveness (ratio of the area under the second compression curve to that of the first, A2/A1), elasticity index (ratio of induced deformation to recovered deformation), gumminess (product of hardness and cohesiveness), chewiness (product of gumminess and springiness), and resilience (ratio of the area during recovery to the area of the first compression, ∆A1/A1), as described by Peleg (2019) [22].

2.5. Microscopic and Image Analyses

For the analysis of the cheese samples, the SEM (Hitachi High-Technologies Corporation TM-1000 scanning electron microscope, Tokio, Japan) was used. Both types of cheese were halved using a clean knife, and a small fragment (approximately 1 cm × 1 cm, thickness 0.2–0.3 mm) was excised from the central region of each sample using a sterile scalpel. The sample was then placed onto a specimen holder, and the desired magnification was selected. Focusing was adjusted manually via cursor navigation on the screen until a sharp image was obtained. For each cheese type, three observations were performed at magnifications of 2000×. The resulting micrographs were analyzed using the ImageJ bundled with 64-bit Java 8 software (National Institutes of Health, Bethesda, MD, USA). To enhance the detection of structural differences, images were first converted to 8-bit grayscale (356 gray levels), then transformed into 16-color mode and subsequently into RGB format to better visualize color-related structural variations. The threshold function was applied to isolate specific regions of interest. This function identifies pixel intensity thresholds, rendering pixels within the selected range in black and all others in white. The size of the identified structures was measured by considering the maximum distance between parallel tangents to the projection area (Feret’s diameter, measured in µm) of the particles; this parameter was chosen because of the irregular shape of the targeted structures. For each of the two treatments (PC and AC cheese), 30 images were analyzed using ImageJ.

For 3D surface reconstruction and image binarization, two-dimensional scanning electron microscopy (SEM) images were transformed into three-dimensional surface representations using the interactive 3D surface plot function available in ImageJ software. To ensure optimal rendering of topographical features, the following parameters were applied: lighting intensity set at 0.2, smoothing at 12.0, scale factor at 1.34, Z-axis ratio at 0.18, and a grid resolution of 512. Subsequently, image binarization was performed through dedicated ImageJ plugins to enhance contrast and facilitate the identification and segmentation of microstructural elements. This approach was adapted from the methodologies reported by Karami et al. (2009) [23].

2.6. Sensory Analysis

A descriptive sensory analysis was conducted to evaluate the organoleptic characteristics of Caciotta cheeses produced with animal coagulant (AC) and plant coagulant (PC). Sensory evaluations were performed by a panel of 10 trained assessors (6 women and 4 men; mean age 37 ± 2 years) with experience in dairy product evaluation. The assessments took place in a sensory testing room compliant with ISO 8589 (ISO, 1988) [24], equipped with individual booths and standard D65 lighting conditions, as defined by the CIE (Commission Internationale de l’Éclairage, Vienna, Austria) (Pinho et al., 2004) [25]. Panelists were trained on their ability to recognize and quantify the four basic tastes (sweet, bitter, salty, sour) as well as characteristic aroma and flavor compounds. Prior to the evaluation sessions, all panelists participated in four training sessions using commercial full-fat milk and Caciotta cheese (aged 50 days) to familiarize themselves with the sensory properties of the products and to refine the descriptive vocabulary, following the methodology described by Meilgaard et al. (1991) [26]. A list of 14 sensory descriptors, smell intensity, milk, yogurt, hay, garlic, dry fruit, astringent, spicy, sour, umami, salty, bitter, sweet, and overall odor intensity, was established, in accordance with recommendations by Pagliarini et al. (1991) [27].

Attributes were rated on a structured scale from 0 (no perception) to 10 (very intense), using standard food references [28], specifically commercial milk and Caciotta cheese. A total of four evaluation sessions were carried out, in accordance with ISO 6658 [29]. Ripened cheese samples were prepared following the procedure of Bàrcenas et al. (2007) [30]; each cheese wheel was halved, vacuum-packed, and stored at 4 °C until analysis. Prior to evaluation, cheese samples (1.5 cm thick × 5 cm wide × 6–8 cm long) were cut to represent the internal portion of the cheese without the rind. All samples were equilibrated to room temperature (21 ± 1 °C) and served to panelists using randomized three-digit codes in a randomized presentation order [31]. Water and apple slices were provided to cleanse the palate between samples.

The evaluation was conducted under controlled conditions to minimize external and psychological variability. The described sensory analysis was performed for each of the three independent cheesemaking trials. Sensory evaluations were performed in duplicate for each cheesemaking. The results were expressed as mean intensity scores for each descriptor and presented using radar plots to visualize sensory differences between the two cheese types.

2.7. Statistical Analysis

The cheesemaking procedures, described for the artisanal production of Caciotta, were replicated three times for each type of coagulant, (plant coagulant, PC, and animal coagulant, AC). For each analysis, three cheeses were used with about 6 samples per cheese type (samples processed were 3 × 3 × 6). Statistical analysis was carried out using Origin 2018 Pro, Statgraphic CenturionXVII (19.2.02 64-bit) software and Excel (2021). Data sets were statistically processed by descriptive and inferential tests (Tukey test, p < 0.05).

3. Results and Discussions

This study investigates the physicochemical characteristics of two Caciotta cheeses, a traditional Italian dairy product, produced using either an animal-derived or a plant-derived coagulant. The analysis was conducted on milk and cheese over a 60-day ripening period to assess potential differences in selected maturation parameters.

The chemical composition of the milk sample used for cheesemaking reveals a high moisture content (85.76%), typical of fresh bovine milk.

Protein and lipid contents were 3.91% and 4.74%, respectively, aligning with standard values for whole milk. The ash content (0.80%) reflects the presence of minerals, while the carbohydrate fraction (4.79%) is primarily attributed to lactose. These compositional parameters provide a meaningful baseline for evaluating the milk’s transformation during cheesemaking and ripening.

Over the 60-day ripening period, both PC and AC cheeses exhibited changes typical of moisture loss and biochemical evolution. As shown in

Table 1. Summary of cheese composition (mean ± SD). Data were collected after 2, 7, 30, and 60 days of ripening. Statistical analysis was performed using Tukey’s test on four replicates. CHO refers to carbohydrates (carbon, hydrogen, and oxygen). Statistical comparisons are made across columns and refer, for each parameter, to the direct comparison between PC and AC samples. Means with different letters indicate statistically significant differences (p < 0.05).

Parameter	1	7	30	60
Ash PC	1.544 ± 0.065 b	3.468 ± 0.012 b	4.004 ± 0.001 b	5.031 ± 0.009 a
Ash AC	1.818 ± 0.061 a	3.863 ± 0.004 a	4.167 ± 0.003 a	4.807 ± 0.028 b
Protein PC	11.994 ± 0.504 b	21.073 ± 0.762 b	22.011 ± 0.47 b	24.667 ± 0.414 a
Protein AC	13.465 ± 1.086 a	22.431 ± 0.212 a	22.822 ± 0.039 a	24.591 ± 0.325 a
Lipids PC	13.156 ± 0.232 b	29.323 ± 0.132 b	30.601 ± 0.552 a	33.478 ± 0.224 a
Lipids AC	15.853 ± 0.324 a	30.665 ± 0.15 a	30.83 ± 0.578 a	33.203 ± 0.405 a
CHO PC	4.591 ± 2.042 a	2.981 ± 0.801 a	2.131 ± 0.131 a	3.342 ± 0.712 a
CHO AC	3.888 ± 1.666 a	3.023 ± 0.075 a	2.706 ± 0.469 a	3.162 ± 0.75 a
Soluble Protein PC	1.465 ± 0.076 a	2.294 ± 0.08 b	3.381 ± 0.077 b	4.148 ± 0.111 b
Soluble Protein AC	1.359 ± 0.135 a	2.608 ± 0.14a	3.885 ± 0.102 a	5.452 ± 0.062 a
Moisture PC	68.715 ± 3.249 a	43.155 ± 3.105 a	41.254 ± 0.95 a	33.482 ± 0.684 a
Moisture AC	64.975 ± 3.842 a	40.018 ± 3.283 a	39.476 ± 0.967 a	34.236 ± 0.849 a

, a progressive decrease in moisture content was observed in both treatments, even though no statistically significant differences were detected between the two samples in terms of moisture content, with PC cheeses retaining slightly more water than AC cheeses throughout ripening. This trend is indicative of gradual dehydration, curd contraction, and structural remodeling during maturation. As reported by Bornaz et al. (2010) [32] in a study comparing whey composition from cheeses produced using Cynara cardunculus extract and traditional chymosin, plant-derived coagulants were associated with distinct proteolytic profiles and whey characteristics. In our study, despite the cheese produced with C. cardunculus showing a more compact and firmer structure compared to the chymosin-coagulated counterpart, it also exhibited higher moisture content. This apparent contradiction may be explained by the specific proteolytic pattern of cardosins, which initially hydrolyze αs- and β-caseins in a manner that promotes the formation of hydrophilic peptides. These peptides are capable of binding water within the curd matrix, limiting syneresis despite the denser structure. Therefore, the increased water retention observed in the PC cheese is not necessarily the result of a looser matrix, as commonly reported, but may instead derive from the water-holding capacity of specific peptide fragments produced during the early stages of proteolysis. This highlights the complex interplay between enzymatic specificity, curd microstructure, and moisture dynamics during cheese ripening.

Ash, protein, and lipid contents increased progressively over time, primarily due to moisture reduction and concentration effects [33]. Soluble protein content, a marker of proteolysis, increased markedly over time in both cheese types (Table 1), with AC cheeses displaying higher levels at each timepoint. This suggests more extensive proteolytic activity in AC cheeses, potentially due to the specificity and efficiency of animal-derived enzymes [16]. Carbohydrate content (Table 1) (CHO) declined rapidly in the early stages and remained low.

Particular attention was given to the evaluation of the ripening index, an indicator of proteolytic activity in cheese, which reflects the extent of casein hydrolysis and nitrogen solubilization [1,34] and overall cheese maturation. In fact, soluble proteins, which increase as a result of casein degradation, serve as a reliable indicator of proteolytic activity, a key biochemical process that drives cheese maturation and affects both texture and flavor development [1,2].

Figure 1 shows the temporal evolution of soluble protein percentage during cheese ripening, comparing samples treated with plant coagulant (PC) and animal coagulant (AC). The data were fitted using second-order polynomial regression models, with high coefficients of determination for both coagulants (R² = 0.9994 for PC and R² = 0.9972 for AC), confirming the suitability of the quadratic model for describing the experimental trends and observations.

The fitted equation shows the form

$$y=\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{c}\mathrm{e}\mathrm{p}\mathrm{t}+{\mathrm{B}}_1\cdot \mathrm{x}+{\mathrm{B}}_2\cdot {\mathrm{x}}^2$$

In particular, y represents the ripening index (%) and represents x the ripening time in days.

The ripening index data (Figure 1) reveal significant differences in proteolytic behavior between cheeses produced using a plant-derived coagulant (PC) containing cardosins and an animal-derived coagulant (AC) composed primarily of chymosin (75%) and pepsin (25%). These differences are reflected in the distinct curvature of the ripening index over the 60-day maturation period, as modeled by second-order polynomial regression.

The cheese manufactured with AC exhibited a faster and more sustained increase in the ripening index over time. The regression parameters (

Table 2. Parameters of ripening index of Cacciota cheese produced using plant coagulant (PC) and animal coagulant (AC).

Parameter	Plot: PC	Plot: AC
Equation	y = Intercept + B1·x1 + B2·x2	y = Intercept + B1·x1 + B2·x2
Intercept	9.94 ± 7.44 × 10−16	9.81 ± 1.82 × 10−15
B1	0.25 ± 5.62 × 10−17	0.27 ± 1.38 × 10−16
B2	−0.0022 ± 8.93 × 10−19	−0.00114 ± 2.19 × 10−18
R-Square (COD)	0.999	0.997
Adjusted R-Square	0.998	0.991

) for this curve included a linear coefficient (B₁) of 0.27428 and a moderately negative quadratic coefficient (B₂) of −0.00114. These values indicate that the rate of proteolysis was initially high and remained active across the entire ripening period, with only a gradual decline in activity. This sustained trend suggests a high proteolytic system, potentially due to higher enzyme stability or a broader spectrum of proteolytic action. Chymosin is used in traditional cheese-making due to its high specificity and slow, controlled proteolytic action. Pepsin, while less specific, complements chymosin by targeting other casein fractions at later ripening stages [35,36]. The regression parameters (Table 2) for coagulant AC, with a higher B₁ value (0.27428), and a less negative B₂ (−0.00114), indicate a more rapid and sustained increase in the ripening index, with a more linear progression over time and delayed plateau. This suggests that the chymosin/pepsin system maintains effective enzymatic activity throughout ripening, likely due to greater stability and proteolytic balance between primary and secondary enzymes [2,36].

In contrast, the cheese produced with PC displayed (Table 2) a slightly higher initial intercept (9.94) but a lower linear coefficient (B₁ = 0.25), indicating a slower initial rate of proteolysis. Moreover, the quadratic coefficient (B₂ = −0.0022) was more negative than that observed for AC, suggesting that the rate of proteolysis declined more sharply as ripening progressed. These findings reflect a slower initial proteolytic rate, which stabilizes earlier during the maturation process. This behavior is consistent with previous literature, where cardosin-rich coagulants, such as PC, are associated with a faster onset but shorter duration of proteolytic activity. The plant coagulant, extracted from Cynara cardunculus, has been described as responsible for primary proteolysis [5] with a lower proteolytic activity towards α-casein and β-casein, as ripening progressed [37,38,39,40].

These results imply that PC supports a less sustained proteolytic process, which may be due to reduced enzymatic stability or narrower substrate specificity over time [37]. According to Silva et al. (2003) [41], cardosins (proteolytic enzyme present in PC) are aspartic proteases derived from Cynara cardunculus. Unlike chymosin, which exhibits high specificity for the Phe105-Met106 bond in κ-casein, cardosins have broader proteolytic activity, targeting a wider range of peptide bonds across κ-, αs-, and β-caseins with the subsequent release of soluble nitrogen compounds. Other studies [41,42] have shown that cheeses produced using Cinara cardunculus extracts as milk coagulants exhibit a higher ripening index compared to those made with chymosin or calf rennet. In contrast to the findings reported by Silva et al. (2003) [41], the data observed here suggests that the proteolysis is less sustained over time. The observed contrasting behaviors could be attributed to the intrinsic enzymatic properties of the coagulants. It could be stated that cardosin-based coagulants (PC) in the initial stages exhibit proteolytic activity similar to that of AC; thereafter, its proteolytic activity follows a much slower trend compared to AC [43]. Chymosin/pepsin coagulants (AC), in contrast, drive a more controlled and extended proteolytic process [2,36], aligning with a longer-lasting ripening effect.

3.1. Texture Analysis

The instrumental texture profile analysis (TPA) (

Table 3. Textural, colorimetric, and sensory properties of Cacciota cheese produced using plant coagulant (PC) and animal coagulant (AC).

TPA Parameters	PC	AC
hardness (N)	19.09 a ± 1.58	12.42 b ± 1.99
cohesiveness	0.79 a ± 0.01	0.79 a ± 0.01
elasticity (mm)	0.52 a ± 0.17	0.51a ± 0.06
resilience	0.38 a ± 0.02	0.39 a ± 0.02
elasticity index	0.88 a ± 0.04	0.89 a ± 0.01
chewiness (J × 10−3)	7.72 a ± 0.21	4.81 b ± 0.67
chewiness (N)	12.96 a ± 1.28	8.75 b ± 1.47
gumminess (N)	14.64 a ± 1.60	9.79 b ± 1.58
Colorimeter Parameters	PC	AC
L	76.75 b ± 1.32	79.67 a ± 0.96
a	0.31 b ± 0.12	1.38 a ± 0.08
b	29.61 b ± 1.78	31.68 a ± 0.18
Texture and Kinesthetic-Related Sensory Attributes	PC	AC
color	4.5 b ± 0.5	6.6 a ± 0.5
hardness	8.4 a ± 0.5	5.3 b ± 0.7
gumminess	7.6 a ± 0.6	5.2 b ± 0.6
solubility	3.2 b ± 0.6	5.9 a ± 1.2

Values are expressed as mean ± standard deviation. Statistical differences were determined using Tukey’s test; different superscript letters within same row indicate significant differences (p < 0.05).

) revealed significant differences in the mechanical behavior of cheeses manufactured with PC, compared to those produced with AC. These differences in texture parameters can be directly linked to the proteolytic characteristics and specificity of the enzymes involved. Means with different letters indicate statistically significant differences (p < 0.05).

As shown in Table 3, cheeses produced with PC exhibited significantly higher values for hardness (19.09 ^a ± 1.58 N), gumminess (14.64 ^a ± 1.60 N), and chewiness (12.96 ^a ± 1.28 N) compared to those produced with AC (hardness: 12.42 ^b ± 1.99 N; gumminess: 9.79 ^b ± 1.58 N; chewiness: 8.75 ^b ± 1.47 N). These findings suggest that the plant coagulant results in a firmer, denser matrix, likely due to lower or less targeted proteolytic activity over time. Cardosins, as aspartic proteases, tend to act broadly but may have less specific caseinolytic activity compared to chymosin, leading to slower or more surface-limited degradation of the casein network [37,38,39,40].

In particular, aspartic proteases (such as cardosins) are characterized by high milk-clotting activity and generally low proteolytic activity. These enzymes typically cleave the peptide bond between Phe105 and Met106, mirroring the catalytic behavior of chymosin [38,39,40]. Nevertheless, different aspartic proteases may display distinct catalytic properties. For example, cardosin A and cardosin B, both isolated from the stigmas of Cynara cardunculus L. by Veríssimo et al. (1996) [44], exhibit divergent substrate specificities and enzymatic activities. Cardosin A cleaves the same peptide bond as chymosin (Phe105–Met106), whereas cardosin B demonstrates a specificity and activity profile more akin to that of pepsin [45]. In this study, it could be inferred that the plant-derived coagulant obtained from cardoon proteases (PC) is likely rich in cardosin A, while the aspartic protease cardosin B may be present at lower levels. This composition could account for the higher values observed in texture-related parameters. The PC cheese exhibits greater firmness and cohesiveness, indicative of reduced proteolytic activity [44].

In contrast, AC, dominated by chymosin, promotes a more efficient and controlled breakdown of κ-casein, and the presence of pepsin (25%) further enhances the degradation of αs- and β-caseins during ripening [16]. This likely results in a more open and softer structure, reflected in the lower hardness and chewiness values of AC cheeses.

No statistically significant differences were observed (Table 3) between the two treatments for cohesiveness (0.79 ^a), elasticity (~0.52 ^a mm), or resilience (~0.38–0.39 ^a). This suggests that, while the force required to deform the samples and the energy involved in chewing differ significantly, the internal bonding and recovery properties of the matrices are relatively not affected by the used coagulant.

3.2. Instrumental Color Analysis of Cheeses

Cheeses produced with AC exhibited significantly higher L* values (79.67 ^a ± 0.96), indicating a brighter and more luminous appearance (Table 3). This may result from the more uniform casein matrix and finer curd structure (see below) achieved through chymosin’s high substrate specificity. The enhanced light reflectance associated with such a homogeneous matrix contributes to a visually brighter surface.

In contrast (Table 3), PC cheeses showed lower L* values (76.75 ^b ± 1.32), suggesting a less bright appearance. This may be attributed to the proteolytic activity of cardosins, which could cause heterogeneous curd formation and a more irregular microstructure that scatters light unevenly, thus reducing overall brightness [46].

The a* parameter, representing the red–green axis, typically remains close to zero in dairy matrices but can shift due to pigment interactions or proteolytic byproducts (Table 3). In this study, cheeses from both treatments remained within the expected range; however, slightly lower a* values (0.31 ^b ± 0.12) were present in PC cheeses. Conversely, higher a* values (1.38 ^a ± 0.08) in AC samples suggest a pigment uniformity, consistent with a more refined matrix [42].

The b* value corresponds to yellowness (positive) versus blueness (negative). Cheeses produced with AC demonstrated significantly higher b* values (31.68 ^a ± 0.18), reflecting a more pronounced yellow tone, which could be attributed to the higher ripening index of AC cheese, as discussed for ripening index (Table 3). PC cheeses, on the other hand, presented lower b* values (29.61 ^b ± 1.78), suggesting a less saturated yellow hue.

3.3. Texture and Kinesthetic-Related Sensory Attributes

The sensory evaluation of cheeses produced using two distinct coagulants revealed statistically significant differences (p < 0.05) in all tested attributes: color, hardness, gumminess, and solubility (Table 3). These differences could be attributed to the distinct enzymatic profiles and proteolytic behavior associated with each coagulant type. It can be stated that the sensory texture parameters are consistent with the instrumental measurements and support the influence of the two different coagulants in modulating cheese texture.

Panelists perceived cheeses made with PC as significantly harder (8.4 ^a ± 0.5) and gummier (7.6 ^a ± 0.6) than those produced with AC (hardness: 5.3 ^b ± 0.7; gumminess: 5.2 ^b ± 0.6) (Table 3). These findings are consistent with the instrumental texture results and can be explained by the less specific proteolytic activity of cardosins [37,38,39,40]. Cardosins, as plant aspartic proteases, may lack the precision and sustained activity of chymosin. As a result, the casein matrix in PC cheeses is less degraded, preserving structural rigidity and resulting in a firmer, denser texture (see above).

In contrast, the chymosin/pepsin mixture in AC facilitates targeted and progressive casein breakdown, leading to weaker protein networks and a softer cheese, perceived as less hard and gummy by the panel.

Cheeses made with AC had significantly higher solubility scores (5.9 ^a ± 1.2) compared to PC (3.2 ^b ± 0.6) (Table 3). This parameter reflects the ease with which the cheese breaks down and disperses in the mouth, often linked to the extent of proteolysis. Chymosin, being highly specific to κ-casein, initiates curd formation, while pepsin further contributes to the hydrolysis of αs- and β-caseins during ripening [35].

In synthesis, the differences in sensory parameters underscore the distinct functional roles of plant vs. animal coagulants, as follow: (a) the cardosin-based coagulant (PC) imparts a firmer, gummier texture and less solubility, (b) the chymosin/pepsin-based coagulant (AC) promotes a more tender, soluble, and yellowish cheese.

3.4. SEM Microstructural Analysis of AC and PC Cheeses

Scanning electron microscopy (SEM) was employed to characterize the microstructural organization of cheeses produced using the two different enzymatic coagulants PC and AC. The micrographs (Figure 2) provide evidence of how enzymatic specificity and proteolytic activity influence curd formation and protein network architecture.

The SEM image of the cheese produced with AC shows a compact, continuous protein matrix with fine, evenly distributed pores.

The limited pepsin fraction (25%) contributes to controlled secondary proteolysis, without disrupting the structural integrity of the protein matrix. The fine porosity observed could suggest efficient moisture entrapment and uniform fat dispersion, properties typically associated with improved texture [1,2,47].

In contrast, the SEM image of the cheese coagulated with PC shows an irregular surface, characterized by areas where no clear porosity is visible and others where it is present with smaller pore sizes, though with an uneven, fragmented, and less organized network.

The plant coagulant could be responsible for irregular hydrolysis during and after coagulation [43], impairing the development of a compact casein gel. The resulting matrix exhibits reduced coherence and stability, which may explain the greater firmness and gumminess observed in sensory and instrumental evaluations of cardosin-based cheeses.

The observed microstructure of PC cheese could also affect optical properties, likely contributing to lower L* values and diminished brightness [3].

3.5. Three-Dimensional Image of Cheese

To evaluate the effects of different milk coagulants on the surface microstructure of cheese, three-dimensional (3D) surface plots were generated from scanning electron microscopy (SEM) images and analyzed using ImageJ software. The resulting 3D reconstructions (Figure 2) illustrate significant differences in surface topography, which reflect underlying variations in casein micelle aggregation and proteolytic behavior.

The surface plot of the AC-coagulated cheese (Figure 2) reveals a homogeneously rough and finely textured microstructure, characterized by densely packed, uniformly distributed peaks. This regular surface pattern is indicative of a cohesive and well-organized protein matrix, attributable to the proteolytic effect of chymosin and pepsin cleavage promoting curd formation, resulting in a regular and continuous network [2,47]. The limited presence of pepsin (25%) may further enhance proteolysis during ripening without compromising matrix integrity. The uniformity observed in the AC matrix is consistent with the smaller pore sizes, reduced heterogeneity, and enhanced smoothness previously reported in similar systems [1].

Conversely, the PC-coagulated cheese (Figure 2) presents a markedly heterogeneous surface topography, characterized by large plateaus interspersed with irregular, isolated protrusions. The surface exhibits greater variability in vertical height (z-axis), with less continuity in the protein network. This irregularity can be attributed to the effect of cardosin that is exhibited in this case with lower substrate specificity [41]. This enzymatic behavior results in disrupted micellar interactions and heterogeneous gelation. Such features have been associated with increased firmness and gumminess and reduced solubility in both instrumental and sensory texture assessments [3].

3.6. Microstructural Pore Size Distribution

As the pores shapes of SEM images were irregular, Feret diameter was considered the most suitable for this measure [48]. The results are visualized through a box plot and a density plot, providing complementary statistical and distributional insights.

The box plot (Figure 2) shows a clear distinction between the two coagulants; in particular, AC displays larger mean and median pore sizes, centered around 8 μm, with a relatively narrow interquartile range (IQR), indicating homogeneous pore size distribution. The maximum values approach 10 μm. On the contrary, PC exhibits smaller and more variable pore sizes, with the majority of values between 1 μm and 4 μm, and a median closer to 3 μm. The distribution is broader, indicating a more heterogeneous microstructure.

This difference suggests that the enzymatic specificity of the coagulant plays a fundamental role in shaping the microarchitecture of the curd matrix.

The density plot (Figure 2) confirms and expands upon these observations, in fact, the AC curve is unimodal and sharply peaked, indicating a high degree of uniformity in larger pores, which is characteristic of chymosin’s controlled cleavage of κ-casein, resulting in pretty uniform gel porosity. In contrast, the PC curve is bimodal, with peaks around 2.5 μm and 4 μm, suggesting dual population sizes of pores. This can be attributed to the proteolytic activity of cardosins, which could lead to the formation of heterogeneous porosity.

These differences are consistent with SEM image interpretations and align with instrumental and sensory evaluations showing that PC cheese is firmer, more fragmented, and less soluble, while AC cheese has a smoother, more continuous matrix.

The observed microstructural differences are directly linked to the proteolytic characteristics of the coagulants; chymosin (AC) is highly specific and cleaves κ-casein at a single site, initiating precise and uniform micelle aggregation [2,47]. The addition of pepsin (25%) allows a secondary proteolysis during ripening without compromising structural integrity. Cardosins (PC), by contrast, exhibit proteolysis, which produces irregular micelle interaction, matrix discontinuity, and larger variability in pore formation [42,46].

The mechanical behavior observed in the TPA results can be directly linked to the microstructural differences revealed by SEM, 3D imaging, and pore size analysis. The higher hardness, gumminess, and chewiness values recorded in PC cheeses are consistent with the heterogeneous protein matrix observed under SEM and in 3D reconstructions. Specifically, the presence of smaller and more variable pores, along with irregular topography, suggests cohesive micellar interactions during coagulation with Cynara cardunculus extract. This irregular microstructure, caused by the proteolytic action of cardosins, results in a protein network that resists deformation more strongly, thus increasing perceived firmness and chewiness. Conversely, the AC cheese, coagulated with chymosin and pepsin, exhibited a smoother, more continuous microstructure with larger and more evenly distributed pores. This well-organized matrix correlates with its lower mechanical resistance and greater solubility, as confirmed by both instrumental and sensory assessments. These findings demonstrate a clear interdependence between enzymatic specificity, protein network architecture, and textural attributes in cheese.

3.7. Sensory Analysis

A Quantitative Descriptive Analysis (QDA) was conducted to compare the sensory characteristics of cheeses produced with the two different milk coagulants. The results, presented in a radar plot (Figure 3), show distinct differences in flavor, aroma, and basic taste attributes, reflecting the influence of enzymatic composition on cheese sensory properties. The solid line represents PC samples, while the dotted line represents AC samples.

Cheese made with AC exhibited higher scores in overall smell intensity, milky aroma, and yogurt-like notes, suggesting a cleaner, more dairy-forward profile. These attributes are typically associated with controlled primary proteolysis and minimal generation of off-flavor compounds, as promoted by the high specificity of chymosin and moderate secondary action of pepsin [1,2]. In contrast, cheeses coagulated with PC displayed stronger intensities for hay, garlic, dry fruit, spicy, and astringency, indicating a more complex flavor profile.

With regard to basic tastes, the AC cheese showed elevated levels of sweetness, umami, and milkiness, statistically higher than PC. Conversely, PC cheese showed significantly higher intensities in astringent, bitter, and sour attributes, sensory markers often associated with hydrolysis of caseins and the resulting accumulation of hydrophobic peptides and astringent compounds, especially in plant coagulant systems [4].

The overall flavor fingerprint of PC cheese was more intense and multidimensional, characterized by strong non-dairy notes, which may be appreciated in artisanal or traditional cheeses. The AC cheese, by contrast, maintained a more dairy-oriented sensory profile.

These findings agree with prior studies comparing chymosin–pepsin and cardosin systems, where plant-derived coagulants often introduce bitter notes [3,49], due to non-targeted protein degradation.

The QDA results reveal that the choice of coagulant significantly shapes the sensory profile of cheese. Cheeses made with chymosin/pepsin (AC) exhibit more conventional, dairy-forward attributes, while cardosin-based (PC) cheeses display greater complexity and the presence of non-dairy sensory dimensions. These results emphasize the importance of tailoring coagulant selection to the target flavor profile and market positioning of the final cheese product.

4. Conclusions

This study evaluated the effects of two different milk coagulants, an animal-derived rennet (AC; primarily chymosin with pepsin) and a plant-derived extract from Cynara cardunculus (PC; rich in cardosin-type aspartic proteases), on the physicochemical, microstructural, and sensory characteristics of Caciotta cheese. Proteolytic activity, as measured by the ripening index, was more sustained and linear in AC cheeses, while PC cheeses exhibited an earlier but less progressive proteolysis. These enzymatic differences significantly impacted the texture, solubility, and moisture retention of the final products. AC cheeses were softer and more soluble, with lower firmness and gumminess values, while PC cheeses were significantly firmer and gummier, suggesting limited and less targeted proteolysis over time. Colorimetric analysis revealed that AC cheeses had higher L* and b* values, corresponding to a brighter and more yellow appearance, whereas PC cheeses appeared darker and less saturated, possibly due to structural irregularities and differences in moisture distribution. Microstructural analysis through SEM and 3D surface reconstruction confirmed these findings. AC cheeses exhibited a regular protein network with larger pores. Pore size distribution analysis further supported these observations: AC samples had a narrower distribution centered around larger pore sizes, while PC samples showed a broader, bimodal distribution, reflecting heterogeneous curd formation. Sensory evaluation revealed that AC cheeses were characterized by milky and umami notes, consistent with a smooth and uniform matrix. In contrast, PC cheeses exhibited more complex, vegetal, and bitter notes such as hay, garlic, and astringency.

In summary, while both coagulants were effective in producing Caciotta cheese, they resulted in products with markedly different structural, functional, and sensory properties. The plant-derived coagulant from C. cardunculus demonstrated its suitability as a rennet alternative, particularly for applications targeting distinctive texture and flavor profiles. However, the cheeses produced with PC cannot be considered direct analogs of those made with animal rennet. Further optimization of formulation and process parameters may be needed to tailor the application of plant coagulants to specific cheese types and desired quality outcomes.