Use of high-pressure processing and low-temperature storage to extend the performance shelf life of 2 types of string cheese

The manufacturing method of string cheese is similar to mozzarella, but the hot curd is extruded through narrow tubes or pipes, which align the protein fibers that provide the characteristic ability for consumers to pull strings from this cheese. Firmness is another important performance attribute for consumers who just bite into the string cheese without peeling off strings. There have only been a few studies on string cheese, but it is known that stringiness and firmness decrease during prolonged storage, which is a particular challenge for exporting string cheese. We explored 2 treatments to try to retain the stringiness and firmness of string cheese for longer storage periods. The techniques used were high-pressure processing (HPP; 600 MPa for 3 min) and reduced storage temperature (0°C). In other cheese varieties, these techniques have helped extend the performance shelf life. We tested these techniques using the 2 main types of commercial string cheese: direct acid string cheese (DASC) and cultured string cheese (CSC), which were obtained from 2 different manufacturing facilities. The DASC had higher fat (~2.2%) and higher pH values (~0.2 units) compared with the CSC. The CSC had higher protein content (~3.4%), higher insoluble calcium content (~8 mg insoluble Ca/g protein) and higher texture profile analysis (TPA) hardness values (~4 N) compared with the DASC. Due to the compositional differences, the 2 varieties were statistically analyzed separately for all other attributes. In both cheese types, HPP caused an immediate reduction in stringiness, some solubilization of insoluble calcium, and a slight increase in the cheese pH values. High-pressure processing also caused a slight increase in TPA hardness of the CSC samples until 14 d (possibly due to a slight increase in cheese pH). The use of the 0°C storage temperature reduced proteolysis and helped retain firmness during storage. Low-temperature storage could help extend the performance shelf life of string cheese by a couple of months, but HPP was not suitable, as the process caused an immediate reduction in stringiness due to the disruption of the matrix induced by the HPP treatment


INTRODUCTION
String cheese is a specific physical form of mozzarella cheese, usually similar to low-moisture part-skim (LMPS) mozzarella and is a popular snack cheese in North America and Europe (particularly for children).The worldwide string cheese market is valued at about $4 billion (Singh, 2022).The popularity of string cheese is growing globally, and some places, such as Mexico, South America, the Middle East, Southeast Asia, and Korea, are currently importing string cheese from the United States (Basu, 2014;Barrett, 2019).String cheese is made using a pasta filata (stretched curd)-style process (Taneya et al., 1992;Kosikowski and Mistry, 1997).String cheese can be made using starter cultures (CSC) in which the cheese milk is acidified by lactose fermentation, or by direct acidification (DASC) in which acid is added directly.Once the target pH is achieved, curds are fed through a cooker-stretcher filled with hot water, and the curd is heated to about 60°C.The hot curd is then extruded through small tubes, creating long, thin ropes.These cheese ropes are then typically cut into ~28-g pieces and brine-salted before being packaged into single-serving wrappers or in larger ~450-g packages.
String cheese is expected to have some degree of stringiness, which is defined as the ability for a consumer to be able to peel apart the cheese into thin fibers or strings (Taneya et al., 1992).Many consumers enjoy this fun aspect and peel apart the stick of string cheese as they eat it.Other consumers may prefer to forego peeling and just take bites out of the entire stick of string cheese.
There have been some studies on the texture and rheological properties of string cheese (Taneya et al., 1992;Oberg et al., 2015), although a quantitative method for evaluation of stringiness is still lacking.The 2 main attributes we considered important for performance quality during storage were the stringiness and firmness of string cheese.To quantify and characterize the stringiness of string cheese, we developed 2 new visual sensory attributes: feathering and fiber quantity.We conducted a focus group with cheese experts from the Wisconsin Center for Dairy Research (CDR; Madison, WI) and surveyed several string cheese brands from local grocery stores to determine the most important attributes.
There have not been many published studies on the properties of string cheese, particularly on how the firmness and stringiness attributes of string cheese change over time.Oberg et al. (2015) investigated low-fat string cheese and the effect of added polysaccharides (as fat replacers) on the textural properties.They noted that these low-fat cheeses became softer and lost stringiness with storage.Previous studies in LMPS mozzarella have shown a reduction in firmness and loss of fibrous character over refrigerated storage (Kiely et al., 1993;Kindstedt and Fox, 1993), and we expected that similar phenomena to occur in string cheese.These changes would create challenges for exporting string cheese because of the lengthy transportation and distribution time (>3 mo), which would result in reduced performance quality for overseas consumers.Many of the changes in the functional attributes of LMPS mozzarella during storage, including the reduction of fibrous character and firmness, have been attributed to proteolysis and shifts in the insoluble (INSOL) Ca content (Lucey et al., 2003).
One method that has been reported to reduce proteolysis in LMPS mozzarella cheese is the use of high-pressure processing (HPP; Ozturk et al., 2018).High-pressure processing is done by applying hydrostatic pressure for a selected length of time, and it is a commonly used nonthermal treatment in the food and beverage industry to reduce the number of pathogens and spoilage organisms in food products (Trujillo et al., 2000).Along with reducing bacterial numbers, which can reduce proteolysis caused by proteolytic bacterial enzymes, HPP can help reduce proteolysis due to the denaturation and partial inactivation of residual chymosin (Malone et al., 2003;Ozturk et al., 2018).Another strategy that can slow the rate of enzyme activity and bacterial growth in cheese is reducing the storage temperature.We hypothesized that using HPP would reduce proteolysis by denaturing residual chymosin and reducing bacterial numbers, and that using low storage temperatures should slow bacterial growth and enzyme activity, as well as slow the establishment of the quasi-equilibrium between soluble and INSOL Ca.These changes could help maintain the firmness and fibrous character in string cheese for longer storage periods.The objective of this study was to evaluate the effect of using HPP and low storage temperatures on proteolysis and INSOL Ca shifts postmanufacture and thus, determine the effect of these treatments on retaining the stringiness and firmness of string cheese during storage.

Commercially Manufactured String Cheeses
On 3 separate occasions, 2 types of commercial string cheese, DASC and CSC, as well as samples of pasteurized milk from each trial, were obtained from the 2 manufacturing facilities.A sample of the acidified milk from the plant that manufactured the DASC was also collected each time.Samples were stored in a cooler with ice packs during transportation.For each trial, the DASC and CSC were manufactured and collected on the same day.Because the cheeses were from 2 different facilities, there were differences in packaging.The CSC was formed into small, individually wrapped, 28-g sticks.The DASC were formed into long, thick ropes in 2.3-kg packages.Therefore, the DASC were converted into smaller pieces, similar in size to that of the CSC and repackaged in vacuum-sealed plastic bags with several pieces per bag but in a single layer.

HPP and Low-Temperature Storage
On the day after cheesemaking, half of the samples from both cheese types underwent HPP at American Pasteurization Company (Milwaukee, WI).The HPP parameters were 600 MPa for 3 min.Samples were stored in a cooler with ice packs during transportation.Then, the HPP and non-HPP samples from both cheese types were further subdivided by storage temperature at 2 d.Two incubators (Nor-Lake Inc., Hudson, WI), one at 4.0 ± 0.2°C as the control storage temperature, and the other at 0.0 ± 0.2°C were used for storage of the string cheese.At each sampling time point, the 0°C samples were tempered in the 4°C incubator for 1 wk before sampling.However, at the 2-d time point, only the 4°C samples were evaluated because the cheeses were held only at 4°C for the first 2 d of storage.
Rennet whey was prepared from the milks (Lucey and Fox, 1993), and inductively coupled argon plasma emission spectroscopy (ICP; Agilent 5100 ICP-OES, Agilent Technologies, Santa Clara, CA) was used to quantify the total Ca content in both the rennet whey and the cheese milks to calculate the percentage of INSOL Ca in the milk (Hassan et al., 2004).The samples were digested using the MARS 5 Express microwave digester (CEM Corporation, Matthews, NC) and diluted before ICP analysis (Govindasamy- Lucey et al., 2007).
Sample preparation methods for cheese analysis were mostly similar to those reported by Reale et al. (2020).The string cheese samples were analyzed after 14 d of storage for composition, including moisture (Marshall, 1992), fat (Mojonnier method, AOAC International, 2000), salt (Johnson and Olson, 1985), and protein (Kjeldahl method, AOAC International, 2000).Total Ca in the cheeses was measured by ICP at 14 d of storage, and acid-base titrations, as described by Hassan et al. (2004), were performed to determine the buffering capacity and to calculate the proportion of INSOL Ca in the cheeses at 2, 14, 60, 120, and 180 d of storage.It was assumed there would be no gross compositional differences between the storage temperatures, so only the 4°C storage samples were used for compositional analysis at 14 d of storage.It was also probable that there would be no differences in composition due to HPP, but the HPP cheeses were analyzed for all components except total Ca, for which only the control (non-HPP) samples were tested.
The microbiology of the cheese samples was analyzed at each time point.Due to the differences in acidification, the CSC were analyzed for both starter lactic acid bacteria (LAB) and nonstarter LAB (NSLAB), but the DASC were only analyzed for NSLAB.Starter LAB were evaluated on M17 agar (Difco, Becton, Dickinson and Co., Sparks, MD) and NSLAB on Rogosa SL agar (Difco, Becton, Dickinson and Co.) as described by Boor and Martin (2024).The plates were incubated anaerobically at 32°C for 48 h.
The pH of the cheeses was determined at each time point using a gel-filled spear-tip pH electrode (Thermo Fischer Scientific, AB15 Basic pH meter) after the ground-up cheeses were allowed to warm to room temperature.The lactose, galactose, and lactic acid concentrations of the cheeses were measured using HPLC at each time point (Zeppa et al., 2001).

Rheological and Texture Analyses
Rheological analysis was conducted at 14, 60, 120, and 180 d using dynamic small-amplitude oscillatory rheology (Lucey et al., 2003;Reale et al., 2020) on an Anton Paar MCR 301 rheometer (Ashland, VA).The sample size was modified based on the diameter of a typical string cheese stick.The samples were cut by slicing 3-mm cross sections from the string cheese with a knife, and a 12mm diameter corer was used to make uniform samples.A 12-mm diameter profiled parallel-plate geometry was used for rheological analysis.The frequency used was 0.08 Hz and the strain was 0.5%.The sample was heated from 5°C to 85°C, increasing at a rate of 1°C per minute.The storage modulus (G′, elastic), loss modulus (G″, viscous), and loss tangent (LT, ratio of G″ to G′) were the rheological measurements obtained during this heating profile.The crossover point, or melting point, was calculated at the point where the LT was equal to 1, or when G′ = G″, which indicated the transition from a solid-to liquid-like system.
Texture profile analysis (TPA) was done at 2, 14, 60, 120, and 180 d using the TA.XT2 Texture Analyzer (Texture Technologies Corp., Scarsdale, NY) as described by Bourne (1978) with a 30-kg load cell.A 19-mm wire cube cutter was used to slice the string cheese sticks cross sectionally, and a 12-mm diameter corer was used to produce uniform, cylindrical samples from the cut ends of the string cheese pieces.The samples were refrigerated overnight before analysis and were kept cool (~4°C) using a small Styrofoam cooler with ice packs during analysis.The Texture Analyzer attachment was a flat, cylindrical 50-mm diameter, aluminum probe.The samples were compressed by 20% of the original height in the double-bite TPA test at a speed of 0.8 mm/s.Cheese hardness was calculated as the maximum force recorded during compression.

Descriptive Sensory Analysis
Each member of the CDR descriptive sensory panel was trained for at least 20 h to evaluate the string cheese samples at 14, 60, 120, and 180 d of storage using sensory Spectrum and quantitative descriptive analysis (QDA; Meilgaard et al., 1999).At least 8 panelists were used for each test.The tests were split into a flavor/texture test to evaluate the basic tastes, flavors, and textures of a piece of string cheese and a visual test to evaluate stringiness.The samples for all sensory evaluations were given 3-digit blinding codes, were assessed in duplicate on the same day, and were tempered in an 11.0 ± 0.2°C incubator for at least 2 h before evaluation.
The string cheese samples for evaluation of flavor and tactile properties were prepared by cutting the cheese cross sectionally using a 19-mm wire cutter.The basic tastes (acid, salt, bitter), the milky and sour flavors, and the astringent mouthfeel were evaluated in the samples as described by Ibáñez et al. (2020).Buttery and cooked flavors, hand firmness, cohesiveness, adhesiveness, chewiness, and particle size were also evaluated similarly to previous studies, with some modifications to the descriptions and references used (Chen et al., 2009;Ibáñez et al., 2020).The numerical intensity scales ranged from 0 to 15 points, and the descriptions and updated references are listed in Table 1.
The stringiness of the string cheese samples was visually analyzed using an entire, original-length string cheese stick.Before evaluation, by peeling longitudinally, the stick of string cheese was pulled/split into thirds, and each one-third section was pulled/split in half.This created 6, separated sections of the cheese stick and thus exposed the fibrous nature of the string cheese.The 6 sections were laid close, but not touching, with the exposed sides facing up and the smooth outside of the stick was kept facing down.The 6 sections were placed in a circular 22.9-cm aluminum cake pan with a The total amount of energy required to masticate the sample to a state pending swallowing.Place cheese cube between molars, chew cheese cube at an even rate; both sides of mouth may be used.Measure total energy required.
Chewiness is a product of cohesiveness, hardness, and springiness.The longer time required to chew, the chewier the sample.
dark gray interior surface, and a photograph was taken of the peeled-apart string cheese sample.The panelists then evaluated 2 visual stringiness attributes: feathering and fiber quantity, by carefully analyzing all visible surfaces of each of the 6 sections of the sample in the pan.
Commercial string cheese samples from several national brands were used to both develop the 2 visual stringiness scales and to train panelists on evaluating the 2 visual stringiness attributes (Figure 1).Feathering was measured on a QDA scale from 0 to 15 and was evaluated based on the average intensity of feathers present on the exposed surfaces between the 6 sections of each sample.Feathers were visually defined as the small, thin (less than 1 mm in diameter, or the average diameter along the length of tapered strands) and wispy strands of cheese that became partially or completely detached from the string cheese stick when the 6 sections were peeled apart.
Fiber quantity (FQ) was measured on a QDA scale from 0 to 15, with a score of 0 being 0%, and 15 being 100%.Because the original stick was divided into only 6 sections for consistency, there was a high probability that most samples could be further subdivided, and the fiber quantity scale was developed to indicate the percentage of the original stick that could potentially be divided into individual, separable pieces.Fibers were defined as being either partially detached (similar to feathers, but larger than 1 mm in diameter) or attached (as seen by distinct grooves in the string cheese samples that appear where it may be possible to further subdivide the string cheese stick into more and smaller fibers).A visual aid to help illustrate the fiber quantity attribute could be a bundle of spaghetti pasta, where the spaghetti noodles represent the fibers in the string cheese, and the bundle is the entire stick (before peeling).The larger the diameter and the fewer pieces of visible fibers, the lower the fiber quantity score, whereas a smaller diameter and more numerous fibers in the bundle (or string cheese stick), the higher the fiber quantity score.Many types of string cheese had ropy, spaghetti-like fibers, but a variation that we observed between brands of string cheese was that some had fibers that peeled off in a flat, sheet-like orientation (similar to slate rock).A similar visual for these slatelike, or slatey, fibers could be fettuccini noodles stacked together in a bundle.Panelists considered the shape of the fibers and could comment whether the fibers were ropy or slatey in character.
Another element to be considered when scoring the fiber quantity was whether the original piece of string cheese was able to be fully peeled into 6, full-length pieces.It was moderately common when analyzing different commercial string cheese samples that one or a few sections out of the 6 would not peel down the entire length of the original stick, leaving it shorter than the other sections.This was typically observed when peeling apart older or firmer samples.This phenomenon indicated a string cheese that had a lower degree of stringiness, and therefore a panelist would reduce their fiber quantity score accordingly, based on the extent and number of incomplete sections.An incomplete section was always kept and evaluated as one of the 6 sections.For an extreme example, if a stick of Cheddar was split into 6 sections as described by this procedure, it would only break off into chunks, as Cheddar is not peelable and does not have the stringiness functionality (see "FQ score: 0" in Figure 1).Thus, it would receive a score of zero for both feathering and fiber quantity.

Experimental Design and Statistical Analyses
The trials were completed in triplicate and contained 4 treatments: HPP and stored at 4°C, HPP and stored at 0°C, non-HPP and stored at 4°C, and non-HPP and stored at 0°C.However, because there were 2 cheese types, DASC and CSC, there were a total of 8 treatments.Because the chemical composition of the DASC and CSC differed, the cheese types were separated for all other statistical analyses except for the cheese composition.SAS 9.4 software (SAS Institute Inc., Cary, NC) was used to calculate ANOVA and statistically analyze the data.When significant differences (P < 0.05) were found, the means of the different treatments were analyzed using Scheffe's multiple-comparison test.
A randomized complete block design was used with the 3 trials blocked and 2 treatment factors (HPP and storage temperature), each with 2 levels (HPP or no-HPP; 0°C or 4°C), for a total of 4 treatments for a cheese type.This design was used to evaluate the cheese composition and differences between treatments at individual time points for pH, INSOL Ca, primary and secondary proteolysis, hardness, rheological properties, and sensory characteristics.A 3 × 4 factorial design was also used to compare the 4 treatments without discriminating between HPP and storage temperature factors separately.The Tukey multiple comparisons test with a significance level of P < 0.05 was used to evaluate differences between treatments.A split-split plot design was used to determine the effects of the different treatments and the storage time, along with the interactions between the factors for pH, INSOL Ca, primary and secondary proteolysis, urea-PAGE, hardness, rheological properties, and sensory characteristics.

Composition of Cheese Milks and Cheeses
The milk used to make the CSC had slightly lower (P < 0.05) TS, lactose, and fat than the non-acidified milk used to make the DASC, however, the CN-to-fat ratio was ~0.3 higher (P < 0.05) in the CSC milk (Table 2).This indicated that the CSC milk was likely partially skimmed.
The CSC milk had lower (P < 0.05) buffering capacity (Lucey and Fox, 1993;Hassan et al., 2004), likely due to the slightly lower (P > 0.05) protein and CN contents, compared with the non-acidified DASC milk (Table 2).Acidification of milk results in increased buffering around pH 5 during acid-base titrations as the colloidal calcium phosphate (CCP) in casein micelles becomes solubilized.Thus, the higher the casein content of the milk, as seen in the DASC milk, the higher the buffering capacity becomes due to the presence of more CCP (Lucey and Fox, 1993).The total Ca contents were similar (P > 0.05), and the average INSOL Ca content of both milks was ~66% of the total Ca in the milk (results not shown).The acidified milk sample from the DASC facility had a much lower average INSOL Ca content of 23% of the total Ca in the milk (results not shown).This was expected due to the solubilization of INSOL Ca due to the extensive acidification of the directly acidified cheese milk (Choi et al., 2008) to a pH value near 5.46 for the DASC samples (pH of the DASC at 2 d, Table 3).

Becher et al.: PERFORMANCE SHELF LIFE OF STRING CHEESE
Table 3 shows the composition of the 2 types of string cheese samples measured at 14 d of storage.As expected, there were no significant differences (P > 0.05) in cheese composition due to HPP (similar to the results of Ozturk et al., 2018), but there were composition differences (P < 0.05) between the 2 cheese types (Table 3), which was expected from the observed differences in the cheese milk composition (Table 2).The CSC had lower fat (by ~2.2%) and higher protein content (by ~3.4%) than the DASC (Table 3).The CSC also had lower moisture in the nonfat substance (MNFS; by ~4%) and lower fat in DM (by ~6%) compared with the DASC.The CSC met the standard of identity for a LMPS mozzarella as listed in the Code of Federal Regulations (CFR), and the DASC met the CFR standard of identity for a low-moisture mozzarella (21CFR; 133.156, 133.158;FDA, 2022).
The total Ca content was also different between cheese types (Table 3).Small, white crystals (calcite, CaCO 3, and Mg-rich dolomite, CaMg(CO 3 ) 2 , identified by the Geoscience Department at the University of Wisconsin-Madison using X-ray diffraction) appeared in the acid digest of the DASC, and because these crystals contained Ca, this could have affected the total Ca readings for the DASC, as the total Ca content of the CSC was about 2 times higher (~330 mg/100 g higher) than the DASC.The total Ca content in our DASC cheeses were a little lower than the levels reported in other DA mozzarella studies (e.g., Choi et al., 2008), but these cheeses had higher protein contents than our samples.The CSC also had a higher protein content than the DASC cheeses, which could have provided a higher INSOL Ca content.  2 The difference between the area under the forward and back curves from pH 4.1 to the pH where the 2 curves intersect (Hassan et al., 2004).8.6 c 7.9 c 13.1 b 14.9 a 0.45 a-c Means within a row sharing the same letter are not significantly different (P < 0.05).
1 Moisture in the nonfat substance.
2 Fat in DM.
3 Salt in the moisture phase. 4Control and HPP assumed to have the same level of total Ca within the cheese type. 5Texture profile analysis hardness.
The DASC contained higher (P < 0.05) NSLAB counts in the non-HPP cheeses than the HPP cheeses at both storage temperatures in the first trial only (results not shown).The DASC in the second and third trials had undetectable levels (<10 cfu/g of cheese) of NSLAB.Postmanufacture contamination could have occurred during the conversion, handling, and repackaging of the original DASC ropes in the first trial, but this process was necessary for every trial and careful sanitation measures were implemented each time.A difference in the first trial was that the DASC came in bags that were not vacuum sealed whereas, for the next 2 trials, the DASC came in vacuum-sealed packaging.The DASC stored at 0°C had lower NSLAB numbers, although not significant (P = 0.054).The higher lactose content and higher pH values of the DASC may have provided slightly more optimal conditions for growth of NSLAB compared with the CSC sample.
The CSC had undetectable levels of NSLAB throughout storage and had consistent packaging between trials that did not require additional handling to prepare individual sticks.The CSC had more hurdles than the DASC to prevent NSLAB growth, such as lower pH, lower lactose content by 14 d, and competition with the added starters for residual lactose in the cheese.The starter LAB numbers in the CSC were significantly (P < 0.05) reduced by about 3 log cfu due to HPP (results not shown).Other HPP studies (Ozturk et al., 2013(Ozturk et al., , 2018) ) have also found a similar reduction in starter numbers but to varying degrees based on pressure level and application time.
Thermophiles, such as those used in LMPS mozzarella (i.e., Streptococcus thermophilus), are known to have greatly reduced growth at low storage temperatures, because their optimal growing temperatures are much higher (around 40°C).However, neither the numbers of starters nor NSLAB in the CSC were affected (P > 0.05) by using 0°C compared with 4°C (i.e., regular refrigeration storage temperature).Neither HPP, storage temperature, nor the storage time had an effect (P > 0.05) on the lactose, galactose, and lactic acid contents for the DASC (results not shown).
At 2 d of storage, the pH of the DASC was higher (P < 0.05) than CSC (Table 3; Figure 2a and 2b), which was expected due to the higher pH values at stretching used in DA mozzarella-type cheese (Lucey et al., 2003).When the cheese types were analyzed separately, both had a significant effect (P < 0.05) on the pH values from the storage time, but not the storage temperature (P > 0.05, Table 4).Pressure treatment had a significant (P < 0.05) impact on pH for both types of cheese, which was consistent with several other HPP studies (Johnston and Darcy, 2000;Rynne et al., 2008;Ozturk et al., 2013Ozturk et al., , 2018)).This initial pH increase is thought to be a buffering effect from some immediate solubilization of the INSOL Ca associated with CN (i.e., CCP) under high pressure (Ozturk et al., 2018).There was a small but significant decrease (P < 0.05) in the initial INSOL Ca content of the HPP samples for both cheese types (Figure 2c and 2d).
During the initial phase of storage, the pH value increased (P < 0.05) for both the DASC and CSC, but after 60 d, the pH values remained consistent (Figure 2a and  2b).The pH of the DASC increased by ~0.1 units compared with a ~0.2-unit increase in the CSC.There is typically a slow solubilization of INSOL Ca during cheese storage (Figure 2c and 2d), and this creates buffering that is known to be the main contributing factor to the increase in pH of cheese with time (Hassan et al., 2004;Johnson and Lucey, 2006).There was a lower INSOL Ca content in the DASC, and the pH was already higher, which could explain why the pH of the DASC did not buffer up as much as the CSC during storage.
The INSOL Ca levels per gram of protein were much lower in the DASC samples compared with CSC throughout storage (Figure 2c and 2d).Storage temperature did not have an effect (P > 0.05) on the INSOL Ca content in the cheeses, but storage time was significantly impacted (Table 4).Storage time caused an overall decrease in the INSOL Ca content in both cheese types (Figure 2c and 2d), due to the ongoing solubilization of INSOL Ca until the establishment of a quasi-equilibrium in the cheese matrix (Hassan et al., 2004).

Proteolysis
In both types of string cheese, primary proteolysis, as measured by the percentage of pH4.6-SN of the total N in the cheese, was significantly affected (P < 0.05) by the pressure treatment, storage temperature, storage time, and the interaction between storage temperature × time (Table 4).Figure 3 shows the changes in pH4.6-SN in both types of string cheese during storage, and there was an overall increase (P < 0.05) in pH4.6-SN during storage.Higher rates of proteolysis were observed for the DASC (Figure 3a) than the CSC (Figure 3b), possibly due to its higher MNFS and lower protein levels (Table 3).At ≥60 d of storage, in both cheese types, the samples stored at 0°C had lower (P < 0.05) levels of pH4.6-SN compared with the cheese samples stored at 4°C (Figure 3).The level of pH4.6-SN is typically attributed to residual chymosin activity (Rank et al., 1985;Sousa et al., 2001), therefore, these results suggested that HPP and the 0°C storage temperature both reduced the activity of residual chymosin; however, lowering the storage temperature seemed to have a greater effect on reducing the level of pH4.6-SN levels during storage than the use of HPP (Figure 3).
Secondary proteolysis, measured by the 12%TCA-SN fraction, is typically attributed to hydrolysis of the pep- tides produced in primary proteolysis by bacterial peptidases from starters or NSLAB (Rank et al., 1985;Sousa et al., 2001).The 12%TCA-SN levels were very low in both cheeses, probably due to the high curd temperature used during stretching, which reduced bacterial numbers.There was a significant effect (P < 0.05) of storage temperature, time, and the interaction between these 2 factors in the DASC (Table 4).The same factors significantly (P < 0.05) affected the 12%TCA-SN in the CSC, in addition to pressure treatment (Table 4).There was ~2% increase (P < 0.05) in 12%TCA-SN during 180 d of storage in both cheese types (results not shown).A reduction in secondary proteolysis with lower storage temperatures was in agreement with previous studies on mozzarella cheese (Feeney et al., 2001;Sheehan et al., 2004).
The urea-PAGE analyses showed that the degradation of all CN components in both cheese types were significantly (P < 0.05) affected by storage time (Supplemental Figure S1, see Notes; Table 5), which corresponded to the results from primary and secondary proteolysis.The contents of intact α S1 -CN and β-CN significantly (P < 0.05) decreased with a concomitant increase (P < 0.05) of 6 casein key breakdown products throughout storage (Figure 4, Table 5).Even though there was a decrease in intact α S1 -CN levels, as expected, this decrease was much smaller compared with Cheddar (O' Mahony et al., 2005;Ozturk et al., 2015), especially in the CSC samples, due to the partial heat inactivation of some enzymes, such as chymosin, during the stretching step of mozzarella manufacture.Usually during the refrigerated storage of mozzarella cheese, higher degradation of α S1 -CN than β-CN has been reported (Kindstedt and Fox, 1993;Yun et al., 1993;Sheehan et al., 2004).In our CSC, the extent of degradation was similar between the 2 types of caseins, but there appeared to be slightly more degradation in the intact α S1 -CN than β-CN in the DASC, especially in samples stored at 4°C (Figure 4).
For the DASC, pressure treatment significantly (P < 0.05) accelerated β-CN degradation at both storage tem-peratures (Figure 4c) but did not significantly (P > 0.05) affect the degradation of α S1 -CN (Table 5).The increase of β-CN degradation could possibly be due to the pressure treatment in the DASC samples activating some PA or inactivating PA inhibitors.It was surprising that HPP did not affect α S1 -CN levels in the DASC, because we saw that HPP slightly reduced primary proteolysis, both of which are linked to residual chymosin activity.There was a slight trend of HPP cheeses having slightly higher intact α S1 -CN levels at 180 d (Figure 4a).Perhaps the higher moisture and pH values in the DASC cheeses (Figure 2) led to more proteolysis by plasmin activity (which has an alkaline pH optimum), which can also contribute to primary proteolysis in some cheeses (O'Mahony et al., 2005).For the CSC, pressure treatment significantly (P < 0.05) reduced α S1 -CN degradation at 180 d of storage but did not significantly (P > 0.05) affect β-CN hydrolysis (Table 5, Figure 4b and 4d).The HPP likely denatured some chymosin, thus reducing the α S1 -CN breakdown during storage, which also corresponded to the reduction in primary proteolysis with HPP in the CSC.
Pressure treatment significantly (P < 0.05) reduced the formation of β-CN f(1-189/192) and α S1 -CN f(102-199) fragments in both cheese types (Table 5), both of which are formed by residual chymosin (Grappin et al., 1985;Sousa et al., 2001).The α S1 -CN f(102-199) was the only fragment in both cheese types to be significantly (P < 0.05) affected by the interaction between pressure treatment × storage time.Figure 4e and 4f shows the α S1 -CN f(102-199) formation during storage, and as the storage time increased, residual chymosin activity was evidently reduced by both HPP and the 0°C storage temperature.
Storage temperature significantly (P < 0.05) affected intact α S1 -CN and β-CN degradation for both types of string cheese (Table 5).There was more degradation of intact α S1 -CN and β-CN at the 4°C storage temperature in both cheese types, which corresponded to both   the primary and secondary proteolysis results and was consistent with Feeney et al. (2001).The interaction between storage temperature × storage time significantly (P < 0.05) affected both α S1 -CN and β-CN degradation in the DASC, as well as the β-CN degradation in the CSC (Table 5).

Rheological and Texture Properties
The melting point (Gʹ = Gʺ) and the maximum loss tangent (LT max ) value, measured during heating, were significantly (P < 0.05) affected by storage time in both cheese types (Table 4).The melting point significantly decreased (P < 0.05), whereas LT max values increased (P < 0.05) during storage in both cheese types (results not shown).Ongoing proteolysis and reduction of INSOL Ca as cheese ages weakens the cheese matrix and typically decreases the melting point, while increasing the LT max values (Lucey et al., 2003).
Pressure treatment significantly (P < 0.05) increased the LT max value in DASC but did not affect it in the CSC (P > 0.05; Table 4).Intact β-CN levels were affected by pressure treatment only in the DASC but not CSC (Table 5), so increased hydrolysis of β-CN during storage (Fig- ure 4c) may have led to an increase in the LT max of the DASC.The CSC cheese already exhibited lower levels of proteolysis than DASC (Figures 3 and 4), which could have limited the impact that HPP was able to have on reducing the LT max values.The compositional differences between cheese types (Table 3) also may have led to a smaller effect of HPP on proteolysis in the CSC than in the DASC.Storage temperature did not affect (P > 0.05) the melting point nor LT max values of either cheese type (Table 4).
The TPA hardness of the CSC was significantly (P < 0.001) higher than the DASC sample throughout storage (Figure 5).The TPA hardness of both cheese types slowly decreased (Figure 5) with storage time (Table 4), which was in agreement with many other mozzarella studies (e.g., Kindstedt and Fox, 1993;Guinee et al., 2001) and was likely related to the gradual degradation of intact CN and the shift from INSOL to soluble Ca (Lucey et al., 2003).Pressure treatment did not significantly (P > 0.05) affect the TPA hardness of the DASC but did significantly (P < 0.05) affect TPA hardness of the CSC (Table 4).The interaction between pressure treatment × storage time was another significant (P < 0.05) factor for TPA hardness of the CSC.The HPP CSC were slightly firmer (P < 0.05) than the non-HPP CSC through 14 d of storage (Figure 5).The increase in firmness initially after HPP in the CSC was surprising, because there was also a small reduction in INSOL Ca, which usually produces a weaker matrix.The initial increase in firmness in our CSC cheeses could be related to a slight increase in pH.Storage temperature significantly (P < 0.05) affected TPA hardness for the DASC but was not a significant factor (P > 0.05) for CSC.At 180 d, in both cheese types, the samples stored at 0°C were slightly firmer (P < 0.05) than those stored at 4°C (Figure 5).The slight reduction in proteolysis due to the 0°C storage temperature (Table 4) likely helped that sample to retain slightly more firmness by 180 d.

Sensory Analysis
Storage time was a significant factor (P < 0.05) affecting each of the measured sensory texture attributes for the DASC, but only significantly affected the hand firmness and cohesiveness of the CSC (Table 6).For both cheese types, the hand firmness decreased, and the cohesiveness increased with long storage times (results not shown).In the DASC, the chewiness decreased, while the adhesiveness and particle size increased, with long storage times (results not shown).Typically, softer cheeses tend to have higher cohesiveness values.
Hand firmness of DASC and cohesiveness of CSC were the only texture attributes significantly (P < 0.05) affected by pressure treatment (Table 6).It was surprising that more textural attributes were not affected by pressure, because primary proteolysis was significantly reduced by pressure for both types of cheese (Table 4).The HPP slightly increased (P < 0.05) sensory hand firmness of DASC and initially reduced (P < 0.05) the cohesiveness of the CSC (results not shown).The slightly higher sensory hand firmness in the HPP DASC sample was not reflected in the instrumental TPA hardness results, which showed no significant effect of HPP (Table 4).
Storage temperature was a significant factor (P < 0.05) affecting particle size of DASC and adhesiveness of both DASC and CSC (Table 6).Particle size was larger (P < 0.05) in the DASC stored at 0°C, and in both cheese types, adhesiveness was lower (P < 0.05) at 0°C (results not shown).Presumably, the softening of the cheese due to ongoing proteolysis and reduction of INSOL Ca likely  led to faster breakdown of the cheese during chewing because of the weakened matrix.This could have led to smaller sized particles and a higher degree of adhesiveness.Therefore, the reduction of proteolysis due to the lower storage temperature may have helped to retain more of the original particle size and adhesive characteristics of the string cheese.
The descriptive sensory scores for bitter, cooked, sour, and astringent were around the threshold level (1.5 points out of 15 points) or lower, and as such, were not prominent in the overall flavor of the cheese and were not included in the analysis (results not shown).
The HPP did not affect (P > 0.05) the flavors and basic tastes in either cheese type (results not shown).Storage temperature was not a significant factor (P > 0.05) for the flavors in the DASC but was a significant factor (P < 0.05) for the milky note in CSC (results not shown).The milky scores were slightly higher in the CSC at 4°C, where the 4°C sample had milky scores close to a "very slight" score (2.5 points out of 15) and the 0°C samples were about 0.3 points lower (results not shown).Storage time significantly (P < 0.05) affected acid, milky, and buttery notes in both cheese types (results not shown).By the later time points, the acid taste and the milky and buttery flavors of both cheese types slightly decreased (P < 0.05, results not shown).
The initial feathering scores in the DASC samples were lower compared with the CSC (Figure 6a and 6b).Pressure treatment and storage time significantly (P < 0.05) affected feathering for both DASC and CSC (Table 6).In both cheese types, there was a reduction (P < 0.05) in feathering scores due to HPP and storage time (Figure 6a and 6b).Storage temperature did not significantly affect feathering scores for both cheese types (Table 6).At several time points the non-HPP-0°C cheeses had higher feathering scores than the corresponding HPP-4°C cheeses.The feathering of both cheese types was significantly (P < 0.05) affected by the interaction of pressure treatment × storage time (Table 6), The fiber quantity scores were also lower in the DASC compared with the CSC, especially in young cheeses (Figure 6c and 6d).Pressure treatment and storage time significantly (P < 0.05) affected fiber quantity in both cheese types (Table 6).Storage temperature did not significantly affect fiber quantity (Table 6).The DASC fiber quantity was also significantly (P < 0.05) affected by the interaction of pressure treatment × storage time.Similar to feathering, the fiber quantity was reduced by the initial HPP treatment, and the non-HPP cheeses from both types had significantly higher fiber quantity at 14 d (Figure 6c and 6d).The CSC fiber quantity significantly (P < 0.05) decreased during the first 60 d, but the initial fiber quantity of the DASC was so low that the scores did not significantly decrease until 180 d.

DISCUSSION
The initial TPA hardness values (Figure 5) and both sensory stringiness scores (Figure 6) were much lower for the DASC cheese compared with the CSC samples.The difference between the 2 types of string cheese was likely due to the much higher INSOL Ca contents, higher protein and total Ca contents, and lower moisture and fat contents for CSC compared with DASC (Table 3).Higher protein contents and INSOL Ca levels are associated with a more crosslinked and denser protein matrix (Lucey et al., 2003).Because these were 2 commercial samples, presumably there are different types of consumers for these kinds of products, with some liking string cheese with a softer texture that has lower stringiness.
The initial INSOL Ca level was much lower in the DASC samples compared with CSC (Figure 2).This indicated that there was much greater demineralization in the DA type of acidification.The DA method dissolves more Ca upfront when the milk is acidified in the vat, because there is greater Ca solubility in the liquid milk state than in curd (Lucey and Fox, 1993;Lucey et al., 2003).Solubilized Ca can be removed during whey drainage, but once most of the serum (whey) is removed and acidification continues, the solubilized Ca accumulates in the serum phase of the curd, which slows further solubilization of INSOL Ca once the solubility is exceeded in the serum phase (Lucey et al., 2003;Johnson and Lucey, 2006).
Pressure treatment significantly reduced the 2 key sensory stringiness attributes, feathering and fiber quantity (Table 6).This effect was observed for both types of string cheese (Figure 6).The pressure treatment likely disrupted some bonds and forced other interactions within the cheese matrix causing curd fusion.This type of bond disruption and altered orientation of casein molecules within the fiber structure likely led to the reduction in the visual stringiness attributes.This disruption of the matrix negatively affected string cheese, which relies on the heterogeneous structure of aligned protein fibers for its characteristic stringiness.The HPP treatment likely physically forced the protein fibers closer together, allowing for more immediate fusion of the protein strands and likely closed or compressed some gaps between the fibers.Previous research (Johnston and Darcy, 2000;Ozturk et al., 2018) found that HPP treatment greatly reduced the void areas in mozzarella, which may include channels of fat and serum phase, which may act to help separate the protein fibers.
The sensory stringiness (Figure 6) and TPA hardness values (Figure 5) decreased during storage for both types of string cheese.In particular, the feathering scores for the CSC cheese exhibited a rapid decline over the first 60 d of storage.There was a significant increase in the pH of the CSC samples over the first 60 d (Figure 2b), whereas the DASC had a much higher pH (Figure 2a) and had lower stringiness.Proteolysis reduced the levels of intact caseins during storage (Table 5), which could disrupt the protein fibers.The rate of primary proteolysis was lower in the CSC cheese compared with the DASC, which would not explain the more rapid decline in stringiness for the CSC samples.Both types of string cheese exhibited a decline in the INSOL Ca levels during storage (Table 4, Figure 2c and 2d), which likely also contributed to the decrease in stringiness or hardness with the loss of protein crosslinking.Mozzarella cheeses exhibit an increase in the water holding capacity during the first couple of weeks of storage (as measured by techniques such as a decrease in the amount of serum that can be expressed by centrifugation; Guo and Kindstedt, 1995).Lucey et al. (2003) explained that the ongoing solubilization of INSOL Ca during the first few weeks after mozzarella cheese manufacture, releases phosphate ions, which become protonated, affecting water mobility.This INSOL Ca solubilization also weakens protein crosslinking and increases repulsion between casein molecules causing swelling (absorption of serum) of the protein matrix, which has been visually observed during the storage of mozzarella cheese (McMahon et al., 1999;Auty et al., 2001).Metzger et al. (2001) reported that for mozzarella made with acidification to pH 5.8 with citric acid (before addition of starter cultures to continue acidification), no expressible serum could be generated during storage.The preacidification of milk to low pH values with citric acid significantly reduced the Ca content (Metzger et al., 2001) similar to the DASC in our study.We presume that low expressible serum in cheeses made by direct acid addition to the milk is a consequence of greater swelling (absorption of serum by the casein network) of the matrix at d 1 due to the greater solubilization of INSOL Ca during the manufacturing phase of this type of cheese.It appears that swelling and rearrangements of the protein matrix in cheese reduces the stringiness attributes, which explains why these attributes were much lower initially in our DASC samples compared with the CSC.Unfortunately, as the CSC undergoes solubilization of INSOL during the first few weeks of storage (Figure 2d) its stringiness also deteriorated (Figure 6d).
The presence of fat between the casein strings can prolong stringiness but eventually it has little effect due to loss of Ca and proteolysis.The addition of polysaccharides to the cheese serum can help prevent rapid fusion of casein molecules (Oberg et al., 2015) but eventually fusion occurs due to the ongoing loss of Ca and proteolysis.
Storage temperature had no effect on rheological properties (LT max and melting point), most sensory attributes (e.g., hand firmness and cohesiveness), and the stringiness attributes.Storage temperature did not affect INSOL Ca levels (Table 4) but did affect proteolysis (Tables 4  and 5).Enzyme activity is slower with the reduction in temperature although the difference in temperatures in our study was relatively small (only 4°C).Other mozzarella studies that used larger differences in storage temperatures did report significant effects on texture and flowability (Guinee et al., 2001;Sheehan et al., 2004).
For the non-HPP CSC, the reduced storage temperature helped to retain a feathering score of ~2 and a fiber quantity score of ~4 for about 120 d, whereas at 4°C, this cheese could only retain these scores until around 60 d.We considered those scores to be near the lower end of consumer acceptability for the performance attributes of string cheese.We are currently conducting a complete consumer evaluation of the performance attributes and acceptability/likeability of commercial string cheeses.Perhaps a colder storage temperature, even freezing, may be able to retain a greater degree of stringiness and firmness during longer storage periods.In mozzarella cheese, during the initial storage period there is swelling of the casein network and microstructural changes, and it is likely that a similar change occurs in the CSC and contributes to the initial change (loss) in stringiness.
Evidently some consumers of string cheese prefer its snacking properties and are not concerned that it lacks stringiness.
This study developed novel methods for the evaluation of string cheese, which should be helpful for use in in-novation projects by cheese manufacturers.String cheese is growing in popularity and is versatile as a snack.This study identified approaches that could partially extend the performance shelf life of this type of cheese, which could be used by the dairy industry for new markets, including exports.

CONCLUSIONS
High-pressure processing was able to lower bacterial numbers and reduce proteolysis; however, it was not a suitable postmanufacturing intervention to extend the performance shelf life of string cheese, because it caused an immediate loss of stringiness due to the disruption of some bonds in the delicate fiber matrix.The string cheese made by direct acidification had lower stringiness and firmness compared with the CSC.This difference was likely due to the lower protein and INSOL Ca content.Direct acidification of the cheese milk was more effective at reducing the Ca content of cheese.The lower storage temperature was also able to reduce proteolysis in both types of string cheese.There was a large decrease in stringiness during the initial storage period, which coincided with ongoing loss of INSOL Ca and an increase in cheese pH.

NOTES
We thank the commercial manufacturers who supplied the string cheese used in this study and Dairy Management Inc. (Rosemont, IL) for funding this research.This work could not have been accomplished without the help of CDR's cheese group, analytical team, and sensory panelists.We thank Valeria Rizzi (PepsiCo, White Plains, NY), who was involved in developing the visual stringiness scales.Supplemental material for this article is available at http: / / digital .library.wisc.edu/1793/ 85398.No human or animal subjects were used, so this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.The authors have not stated any conflicts of interest.

Figure 1 .
Figure 1.Scales for the stringiness sensory attributes for feathering and fiber quantity (FQ).

Figure 2 .
Figure 2. Changes in the pH (a, b) and insoluble (Insol) Ca content (mg/g of protein in cheese; c, d) of the direct acid (a, c) and cultured (b, d) string cheeses that were high-pressure processed (squares) or not (circles) and held at either 4°C (closed [black] symbols) or 0°C (open [white] symbols) during storage.Error bars represent SD; n = 3.
Figure 3. Changes in the % pH 4.6-soluble N (sol-N) of the total N in the direct acid (a) and cultured (b) string cheeses that that were highpressure processed (squares) or not (circles) and held at either 4°C (closed [black] symbols) or 0°C (open [white] symbols) during storage.Error bars represent SD; n = 3.

Figure 4 .
Figure 4. Changes in levels of intact α S1 -CN as a percentage of the level at 2 d (a, b), of intact β-CN as a percentage of the level at 2 d (c, d), and of the formation of α S1 -CN f(102-199) fraction as a percentage of the intact α S1 -CN level at 2 d (e, f) of direct acid (a, c, e) and cultured (b, d, f) string cheeses that were high-pressure processed (squares) or not (circles) and held at either 4°C (closed [black] symbols) or 0°C (open [white] symbols) during storage.Error bars represent SD; n = 3.

Figure 5 .
Figure 5. Changes in the texture profile analysis (TPA) hardness values in the direct acid (a) and cultured (b) string cheeses that were highpressure processed (squares) or not (circles) and held at either 4°C (closed [black] symbols) or 0°C (open [white] symbols) during storage.Error bars represent SD; n = 3.

Figure 6 .
Figure 6.Changes in the sensory feathering (a, b) and fiber quantity scores (c, d) of the direct acid (a, c) and cultured (b, d) string cheeses that were high-pressure processed (squares) or not (circles) and held at either 4°C (closed [black] symbols) or 0°C (open [white] symbols) during storage.Error bars represent SD; n = 3.

Table 1 .
Definition of the attributes used by trained panelists to evaluate the sensory properties of direct acid and cultured string cheeses at 11°C 1

Table 2 .
Becher et al.: PERFORMANCE SHELF LIFE OF STRING CHEESE Composition of the milks used to make the 2 types of string cheese from different manufacturing facilities Means within a row sharing the same letter are not significantly different (P < 0.05).

Table 3 .
Composition of the directly acidified (DASC) and cultured string cheese (CSC) samples, that were either high-pressure processed (HPP) or not, at 14 d of storage at 4°C Becher et al.: PERFORMANCE SHELF LIFE OF STRING CHEESE

Table 5 .
Summary for the effects 1 of cheesemaking day, high-pressure processing (HPP), storage temperature, and storage time on the casein breakdown in the direct acid (DA) and cultured string cheese samples as determined by urea-PAGE Becher et al.: PERFORMANCE SHELF LIFE OF STRING CHEESE

Table 6 .
Probabilities and R 2 values for descriptive sensory texture and visual stringiness attributes of the direct acid (DASC) and cultured string cheeses (CSC) during 180 d of storage