Impact of NaCl reduction in Danish semi-hard Samsoe cheeses on proliferation and autolysis of DL-starter cultures

International Journal of Food Microbiology 213 (2015) 59–70

Contents lists available at ScienceDirect

International Journal of Food Microbiology journal homepage: www.elsevier.com/locate/ijfoodmicro

Impact of NaCl reduction in Danish semi-hard Samsoe cheeses on proliferation and autolysis of DL-starter cultures Lise Søndergaard a,⁎, Mia Ryssel a, Carina Svendsen a, Erik Høier b, Ulf Andersen c, Marianne Hammershøj d, Jean R. Møller d,1, Nils Arneborg a, Lene Jespersen a a

Department of Food Science, Faculty of Science, University of Copenhagen, 1958 Frederiksberg C, Denmark Chr. Hansen A/S, 2970 Hørsholm, Denmark Arla Strategic Innovation Centre, Rørdrumvej 2, 8220 Brabrand, Denmark d Department of Food Science, Aarhus University, Blichers Allé 20, Postboks 50, 8830 Tjele, Denmark b c

a r t i c l e

i n f o

Article history: Received 27 February 2015 Received in revised form 21 June 2015 Accepted 26 June 2015 Available online 14 July 2015 Keywords: Semi-hard cheese NaCl reduction DL-starter cultures Viability Autolysis Free amino acids

a b s t r a c t Reduction of sodium chloride (NaCl) in cheese manufacturing is a challenge for the dairy industry. NaCl has a profound role on microbial development influencing cheese sensory and technological properties. The purpose of this work was to investigate how proliferation, distribution and autolysis of two commercial DL-starter cultures (C1 and C2) used in the production of Danish semi-hard Samsoe cheeses were affected by reduced NaCl levels. Cheeses containing b0.3% (unsalted), 2.3% (reduced-salt) and 3.4% (normal-salted) (w/v) NaCl in moisture were produced and analyzed during 12 weeks of ripening. Lactic acid bacteria (LAB), distribution of bacteria as single cells or microcolonies, their viability in the cheeses and cell autolysis were monitored during ripening, as well as the impact of NaCl content and autolysis on the formation of free amino acids (FAA). Reduction of NaCl resulted in higher LAB counts at the early stages of ripening, with differences between the two DL-starter cultures. The unsalted cheeses produced with C1 had retained a significantly higher number of the initial LAB counts (cfu/g) after 1 and 2 weeks of ripening (i.e. 58% and 71%), compared to the normal-salted cheeses (i.e. 22% and 21%), whereas no significant difference was found between the reduced-salt (i.e. 31% and 35%) and normal-salted cheeses. At the later stages of ripening (i.e. 7 and 11 weeks) NaCl had no significant influence. For cheeses produced with C2, a significant influence of NaCl was only found in cheeses ripened for 7 weeks, where the unsalted and reduced-salt cheeses had retained a significantly higher number of the initial LAB counts (cfu/g) (i.e. 39% and 38%), compared to the normal-salted cheeses (i.e. 21%). In the Samsoe cheeses, bacteria were organized as single cells, in groups of 2–3 cells or in groups of ≥4 cells. During ripening the decrease in the number of viable bacteria was mainly due to a reduction in the number of viable bacteria organized in groups of ≥4 cells. A negative correlation between NaCl content and PepX activity was observed. At the end of ripening the total FAA content was lower in the unsalted cheeses, compared to the reduced- and normal-salted cheeses. In conclusion, NaCl had a significant influence on proliferation of both DL-starter cultures. However, the influence of NaCl on culture development was more pronounced in cheeses produced with DL-starter culture C1. As both texture and taste are parameters known to be affected by the development of the starter culture, the design of starter cultures for reduced NaCl cheeses is recommended. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Dietary sodium reduction programs have been initiated worldwide, as growing evidence since the 1960s have linked excessive sodium consumption with hypertension, and thereby the risk of cardiovascular diseases (Anon, 2007). The main part of sodium in the diet is consumed as

⁎ Corresponding author at: Department of Food Science, Faculty of Science, University of Copenhagen, Rolighedsvej 26, DK-1958 Frederiksberg C, Denmark. E-mail address: [email protected] (L. Søndergaard). 1 Present address: Lindholmvej 13, 1tv, 8200 Aarhus N, Denmark.

http://dx.doi.org/10.1016/j.ijfoodmicro.2015.06.031 0168-1605/© 2015 Elsevier B.V. All rights reserved.

sodium chloride (NaCl), with average intakes in many countries at 9 to 12 g per day per person (Brown et al., 2009). The World Health Organization (WHO) advocates for a reduction of the NaCl intake, to less than 5 g per person per day (Anon, 2003). Manufactured foods account for up to 75% of the intake of NaCl in the European and Northern American countries (Brown et al., 2009). Cheese has among other foods been targeted for NaCl reductions (Saint-Eve et al., 2009). Consequently, there is a growing interest in the dairy industry to produce cheeses with reduced NaCl content. Reduction of NaCl in cheese manufacturing possesses several technological, sensory and microbiological challenges as NaCl is central for textural properties, microbial growth, autolysis, enzyme activities during ripening and consequently cheese flavor (Guinee

60

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

and Fox, 1999). The sensory defaults that usually have been described in reduced NaCl semi-hard cheeses are increased cohesiveness, adhesiveness, acidity, bitterness and an unpleasant aftertaste together with a decrease in saltiness and hardness (Rulikowska et al., 2013; Schroeder et al., 1988). Additionally, lowering of the NaCl content in cheese reduces the safety of the product, with potential growth of pathogens such as Listeria monocytogenes (Hystead et al., 2013). Samsoe cheese is a semi-hard pressed Danish cheese with few round holes, contains on average 1.8% (w/w) NaCl (g/100 g cheese), maximum 52% moisture in the type with 30% fat in dry matter (30 +) and is ripened in plastic bags. Even though not surface ripened, Samsoe cheese is included in the Danbo cheese group where cheeses mainly differ from one another with respect to size and shape. Danbo are light yellow colored cheeses with a consistency that permits slicing of the cheeses when consumed after most frequently 2–3 months ripening (Werner et al., 1999). The starter cultures used in the production of Danbo cheeses are mesophilic DL-types, which include strains of the species Lactococcus lactis ssp. lactis, Lc. lactis ssp. cremoris, Lc. lactis ssp. biovar. diacetylactis and Leuconostoc spp. (Daly, 1983). The DL-starter culture is an essential part of cheese production and affects both texture and flavor following milk acidification (Cogan and Hill, 1999). During cheese ripening, the starter bacteria will autolyse and in this way release intracellular enzymes of importance for secondary proteolysis. This is an event, which not only leads to an increase in free amino acids (FAA) but also a decrease in cheese bitterness due to degradation of hydrophobic peptides (Bourdat-Deschamps et al., 2004). FAA can either directly contribute to flavor or act as precursors for other flavor compounds (Sousa et al., 2001). Numerous experiments have been conducted, especially in Cheddar cheese, to study the influence of NaCl reduction on viability of the starter culture. The results from Cheddar cheese have led to the conclusion that increasing the NaCl in moisture content from 1.1% up to 6.4% (w/v) resulted in decreased viability of the mesophilic starter culture (Mistry and Kasperson, 1998; Møller et al., 2012; Rulikowska et al., 2013; Upreti et al., 2006). Autolysis of the starter culture has been evaluated by measuring the release of intracellular enzymes, e.g. post-proline dipeptidyl aminopeptidase (PepX). Cheese contains high levels of proline-rich peptides, originating from β-casein, which can be hydrolyzed only by proline-specific peptidases, such as PepX (Wilkinson et al., 1994). These proline-rich peptides can result in bitter flavor development in cheese and PepX is therefore important for the debittering process in cheese (Booth et al., 1990; Singh et al., 2005). Wilkinson et al. (1994) reported that increasing NaCl in moisture levels in Cheddar cheese from 0.5% to 4.9% (w/v), resulted in a 13-fold decrease in activity of PepX in cheeses after 60 days ripening. In contrast, two recently published studies with Cheddar cheese observed a positive correlation between increasing NaCl in moisture levels, in the range of 1.1% to 6.3% (w/v), and PepX activity (Møller et al., 2012; Rulikowska et al., 2013). The influence of starter culture autolysis and NaCl level in cheese on the formation of FAA were evaluated by Møller et al. (2012) and Rulikowska et al.(2013). Møller et al. (2012) found a positive correlation after 270 days ripening between PepX activity and NaCl level in Cheddar cheese on the formation of FAA, whereas no clear correlation was reported by Rulikowska et al.(2013). The above mentioned contradictory results, illustrate the complexity of NaCl reduction in cheese manufacturing. Furthermore, comparisons between studies are complicated due to variations in the cheese type, production method, composition, starter culture and the analytical methodologies used (Kristiansen et al., 1999; Rulikowska et al., 2013). This highlights the importance for studying the influence of NaCl in the specific cheese type, targeted for NaCl reduction. To the best of our knowledge, no studies have evaluated the effect of NaCl reduction on the microbiological proliferation, viability and autolysis in Samsoe cheese. Our aim was therefore to study to which extent NaCl reduction in Samsoe cheese affects proliferation, distribution and autolysis of two DL-starter cultures during ripening. Additionally, the

impact of NaCl level and starter culture autolysis on the formation of free amino acids (FAA) was assessed.

2. Materials and methods 2.1. Starter cultures Two different available and commercially produced frozen directvat-set DL-starter cultures (C1 and C2) were used (Chr. Hansen A/S, Hørsholm, Denmark). The starter cultures differed from each other with respect to production method and strain composition. However, both starter cultures comprised multiple strains of Lc. lactis ssp. lactis, Lc. lactis ssp. cremoris, Lc. lactis ssp. biovar. diacetylactis and Leuconostoc spp. C1, a traditional multiple DL-starter culture, was propagated and produced as a mixture of all these strains. C2, a DL-starter culture combined of defined strains belonging to the above-mentioned groups, where strains have been isolated from a traditional multiple DLstarter culture, grouped and grown separately, before combined into the final starter culture. The same batch of the two DL-starter cultures was used for all cheese trials.

2.2. Cheese manufacture and sampling For analyses of microbial cell counts, autolysis and FAA, Samsoe 30+ (30% fat in dry matter) cheeses were produced at Arla Strategic Innovation Centre, Brabrand, Denmark. Cheeses were produced from fresh low-pasteurized (72 °C, 15 s) bovine milk and standardized to a fat/ protein ratio of 0.5 (1.88% fat and 3.75% protein). Milk was heated to 32 °C in cheese vats containing 1000 L. Cheeses were produced on two consecutive days using milk from the same batch, which was stored cold at 4 °C between productions. On each day, three batches of cheese were produced, with the two starter cultures (C1 and C2) applied on separate days. Cheeses with b 0.3% (unsalted), 2.3% (reduced-salt) and 3.4% (normal-salted) (w/v) NaCl in moisture were produced from all three batches. The starter cultures were added according to the manufacture prescriptions at a rate of 0.0108% (w/w) (C1) and 0.0066% (w/w) (C2) together with CaCl2 0.005% (w/w). Coagulation of the milk was obtained by addition of CHY-MAX® Plus (bovine chymosin) at a rate of 0.050% (w/w) (Chr. Hansen A/S), 30 min after the starter culture was added and at a pH-value of ~ 6.5. The coagulum was cut after 30 min. Following this, curd grains were stirred and heated to a target temperature of 36.5 °C. The coagulum was pre-pressed for 20 min (0 to 3.5 bar) resulting in whey drainage. Cheeses were cut into sizes of 28 cm × 28 cm × 15 cm (length × width × thickness) and pressed for 35 min (2.5 to 6.0 bar). If the cheese had the target pHvalue of 6.0 it was soaked in 15 °C water for approximately 18 h. For cheeses with higher pH-values, pH was monitored as acidification continued at room temperature (r.t.) until the target pH of 6.0 was reached, before cheeses were soaked in the 15 °C water. Cheeses were placed in a saturated NaCl solution (23.3% (w/v)) with 0.25% CaCl2 at 11.5 °C for 12 h (2.3% (w/v) NaCl) and 24 h (3.4% (w/v) NaCl), respectively. Cheeses with b0.3% (w/v) NaCl were omitted this step. Cheeses were cut in half (5 kg each), vacuum-wrapped separately, and ripened at 13 °C until analysis. Sampling was performed before brining and after 1, 2, 7, and 11 weeks of ripening. Only samples from 2, 7 and 11 weeks ripened cheeses were used for the FAA analysis. A half cheese was drawn at each sampling point and grated. For evaluation of cell distribution with SEM and CLSM, Samsoe cheeses were produced at Thise dairy, Roslev, Denmark. These cheeses were produced using the same procedure as described above, except that 6000 L vats were used, two batches of cheese were produced and cheeses were sliced before analysis. Sampling was performed after 2, 7 and 12 weeks of ripening. Composition analyses were performed for cheeses from both trials.

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

2.3. Cheese composition Analyses of fat, dry matter, NaCl and protein content were performed according to standard methods (Anon, 2001, 2004, 2006, 2008). pH was measured potentiometrically. All analyses were conducted at Eurofins Steins Laboratory, Holstebro, Denmark.

61

relatively low. Micrographs were analyzed manually and for each micrograph the total numbers of viable and nonviable cells were counted. Additionally, the numbers of viable and nonviable bacteria organized as single cells, groups of 2–3 cells, and groups of ≥ 4 cells were recorded. In total, 15 micrographs of different sections (75 μm × 75 μm) within a sample were analyzed, for each of the two batches. In total 1080 images were evaluated.

2.4. Microbial cell counts 2.7. Autolysis Lactic acid bacteria (LAB) were enumerated on spread-plated M17 agar, with 0.5% (w/v) lactose (Biokar Diagnostics, Allonne, France) incubated aerobically at 30 °C for three days. Non-starter lactic acid bacteria (NS-LAB) were enumerated on spread-plated MRS agar (Oxoid, Hampshire, United Kingdom), pH 5.4, after anaerobic incubation at 37 °C for three days. A 10.0 g sample of grated cheese was diluted with 90 mL of 2% (w/v) trisodium citrate buffer and homogenized in a Stomacher® blender (Seward, Worting, UK) for 2 min at approximately 300 rpm. Serial 10-fold dilutions were prepared in 0.9% (w/v) NaCl solutions. Samples were inoculated on media in duplicates. 2.5. Scanning electron microscopy (SEM) The cheese structure and diversity of bacteria was studied by SEM. A 1 mm3 cube of cheese was fixed in PIPES (2.5% gluteraldehyde in 0.1 M piperazine-N,N′-bis(2-ethanesulfonic acid)) buffer. The cheese sample was dehydrated by washing in ethanol in a series of stepwise increasing 10% concentrations from 10–100% ethanol each for 15 min. For the 20% ethanol step and onwards, the cheese sample was transferred to a critical point drying (CPD) capsule. The sample was washed in 100% dry ethanol for 2 × 15 min and stored at T = 4 °C for N 1 h before CPD. The CPD procedure was performed with liquid CO2 using a Leica CPD300 (Leica Microsystems, Heidelberg, Germany), and the dried sample was stored in a sealed container at r.t. until analysis. Before analysis, the sample was secured onto an aluminium SEM stub with Ag paint and fractured in the horizontal plane. Hereby, a free-break surface facing upwards was obtained, and the surface was covered with a thin layer of Au using an agar high resolution sputter-coater (Agar Scientific, Stansted, UK). The prepared sample was observed at 3 kV with a Zeiss Supra 55VP FEG Scanning Electron Microscope (Carl Zeiss, Oberkochen, Germany), at a working distance of ~ 5 mm at magnifications ranging from 1,000 to 95,000 times. Several pictures were captured for each cheese sample. 2.6. Confocal laser scanning microscopy (CLSM) The distribution of bacteria was studied by CLSM. Freshly cut exterior (depth ~2.0 mm) and interior (7 cm from the surface) cheese samples were added 2 μl LIVE/DEAD viability stain (BacLightTM, Molecular Probes, Invitrogen) and a coverslip was placed on top. The samples were incubated 20–30 min in the dark at r.t. All microscopy work was performed using a Leica TCS SP2 confocal laser scanning microscope (Leica Microsystems, Wetzlar, Germany). For each sample, 15 randomly selected areas were imaged using a 100x magnification objective, corresponding to 1000 × magnification, with a numerical aperture of 1.3. Confocal illumination was provided by an Ar/ArKr and HeNe laser. The excitation wave lengths were 490 and 541 nm and emission were recorded at 520 and 572 nm with acousto optical beam splitter system. Simultaneous dual-channel imaging using pseudocolor was used to display green and red fluorescence. Viable cells with intact membranes and nonviable cells with compromised cell membranes were clearly discriminated due to the green and red color, respectively. Bacteria with various extents of lysis of the cell wall were observed and even after complete lysis of cells, residual cell material were still detectable. However, completely lysed cells were not included in the number of nonviable bacteria. Viable cells could be discriminated from the green background fluorescence originating from milk proteins, as this was

Cell autolysis was assessed by measuring the release of intracellular post-proline dipeptidyl aminopeptidase (PepX). The enzymatic activity was measured in the cheese serum based on the method by Ardö et al. (1989) with small modifications. A mass of 10.0 g of grated cheese was dissolved in 40 mL of 0.1 M trisodium citrate pH 9.0. The solution was rotated for 30 min at r.t., followed by rotation at 45 °C for 30 min and then rotated at r.t. for 15 min. The liquid phase was centrifuged (10.000 × g, 15 min, 5 °C) and the supernatant was stored at − 80 °C until analysis. The supernatant was filtered (0.45 μm) before analysis and the cheese extract was diluted in 500 mM Tris–HCl pH 7.5 if required. Four aliquots of 30 μL of each cheese extract sample were added to a black microtiter plate together with 60 μL 500 mM Tris– HCl pH 7.5. followed by addition of 10 μL substrate solution (5 mM Gly-Pro-AMC (7-amino-4-methyl coumarin) (Bachem, Bubendorf, Switzerland)). The enzymatic cleavage of Gly-Pro-AMC was measured by monitoring the fluorescence at λ ex/em = 355/460 nm every 5 min for 1 h at 37 °C (Perkin Elmer Wallac 1420 Victor2, Waltham, Massachusetts, USA). The activity of PepX was calculated using the linear regression of an AMC standard curve (100 μM AMC (Bachem, Bubendorf, Switzerland) diluted with 500 mM Tris–HCl pH 7.5). Results were expressed as nmol of AMC released per min per mL of cheese extract at pH 7.5 at 37 °C. 2.8. Free amino acid The compositions of FAA in cheeses were analyzed by GC–MS. Cheeses were analyzed for twenty-one amino acids (nineteen from casein; γ-amino butyric acid from glutamic acid; ornithine from arginine). Pre-column derivitization with methyl chloroformat followed by liquid-liquid extraction were performed prior to the GC–MS step. Extraction of cheeses were performed by rotating 3.0 g of grated cheese in 15 mL of deionized water for 1 h (40 rpm) followed by centrifugation (2500 × g, 30 min, 4 °C). Aliquots of 500 μL supernatant were stored at −20 °C until derivatization. Supernatant volumes of 25 μL were mixed with 125 μL deionized water in 2.0 mL injection vials. A volume of 150 μL 1 mM norvaline mixed with deionized water was added as internal standard followed by 200 μL 1-methanol/pyridine 32/8% and samples were mixed. Methyl chloroformate, 25 μL, was added following addition of 500 μL 1% methyl chloroformate/chloroform. A GC combined with autosampler (HP 5890) and mass selective detector (HP 5972) were used (both Agilent Technologies, Santa Clara, USA). Samples and a standard solution with amino acids were injected (2 μL) onto the column (DB-XLB 15 m × 0.25 mm × 0.25 μm) at a split ratio of 1:15. Helium was used as a carrier gas (25.5 mL/min), and the oven inlet temperature was set at 110 °C (0 min) and raised at 30 °C/min to 350 °C (0 min) with a head pressure at 50 kPa. The MS was operated in scan (60–600 m/z) and selected ion monitoring mode and with the transfer line kept at 350 °C. Concentration of FAA was calculated based on the peak height and standard curves. 2.9. Statistical analysis Data were analyzed using one-way analysis of variance (ANOVA) at 95% level of significance using GraphPadPrism software version 6.03 (http://www.graphpad.com). To further determine the difference between the means, Tukey's post hoc analysis was applied. All error bars

62

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

displayed in figures represents 1 standard deviation above and below the mean. The FAA data was analyzed by Principal Component Analysis (PCA). The analysis was performed using MATLAB (MATLAB R2014a, The MathWorks Inc., USA) and the PLS-Toolbox (PLS-Toolbox 7.5, Eigenvector Research Inc., USA). 3. Results 3.1. Ripening of cheeses and enumeration of lactic acid bacteria Initial chemical composition of cheeses is shown in Table 1. Cheeses with significantly different (P b 0.001) contents of NaCl were produced in this study; i.e. b 0.15%, 1.2% and 1.7 / 1.8% (w/w), corresponding to b 0.27%, 2.3% and 3.2 / 3.4% (w/v) NaCl in moisture, respectively. There was a marginal variation in the NaCl in moisture content of 0.2% (w/v) for normal-salted cheeses produced with DL-starter culture C1 and C2, respectively. However, it was not significant (P = 0.4). The moisture content was higher in cheeses produced with less NaCl content. Protein and fat content were comparable in all cheeses. The normal-salted cheeses produced with C1 had a significantly lower (P = 0.01) pH (5.28), compared to the unsalted (5.34) and reducedsalt cheeses (5.35). There was no significant difference in pH for cheeses produced with C2 (pH = 5.28 to 5.30). At the end of ripening, i.e. after 11 weeks, pH had increased to 5.41 to 5.53 with the highest pH values in cheeses with less NaCl, although not significant (results not shown). Fig. 1 shows the LAB count (cfu/g) (M17 agar with 0.5% (w/v) lactose) for the ripened cheeses, as compared to the LAB counts (cfu/g) in the cheeses before brining, which were found to have LAB counts of 8.9 × 108 cfu/g and 1.7 × 109 cfu/g, for cheeses produced with C1 and C2, respectively. The population decreased progressively in all cheeses during the ripening period in a starter culture dependent manner. For the normal-salted cheeses (i.e. 3.4% (w/v) NaCl), the most pronounced decrease was detected within the first week of ripening. After 1 week of ripening, the LAB counts (cfu/g) were reduced to 22% and 58% of the initial LAB counts (cfu/g) in cheeses produced with C1 and C2, respectively. After 7 weeks of ripening, the LAB counts (cfu/g) were reduced to 5% and 21% of the initial LAB counts (cfu/g) for normal-salted cheeses produced with C1 and C2, respectively. Hereafter, the LAB counts (cfu/g) remained constant up to 11 weeks of ripening (i.e. 3% for C1 and 23% for C2). The NaCl content was found to have a significant influence on the LAB counts (cfu/g). For cheeses produced with C1, significantly higher (P = 0.04) LAB counts (cfu/g) were found in the unsalted cheeses after 1 week of ripening (58%), compared to the normal-salted cheeses (22%), whereas no significant difference was seen for the reduced-salt

cheeses (31%) (Fig. 1A). Similar results were observed after 2 weeks of ripening, where significantly higher (P b 0.001) LAB counts (cfu/g) were found in the unsalted cheeses (71%), compared to the normalsalted cheeses (21%), while no significant difference was seen for the reduced-salt cheeses (35%). After 7 and 11 weeks of ripening, there was no significant influence of NaCl. For cheeses produced with C2, no significant influence of NaCl was found until week 7 of ripening, where the unsalted and reduced-salt cheeses had significantly higher (P b 0.05) LAB counts (cfu/g) (i.e. 39% and 38%) compared to the normal-salted cheeses (21%) (Fig. 1B). After 11 weeks of ripening, no significant influence of NaCl was found, as the LAB counts (cfu/g) had decreased for both the unsalted cheeses (14%) and reduced-salt cheeses (23%) to the same level as for the normal-salted cheeses (23%). When comparing C1 and C2 after 1 week of ripening, a significant difference (P = 0.009) was observed for the normal-salt cheeses, i.e. 22% and 58% of the initial LAB counts (cfu/g) retained, respectively. A significant difference (P = 0.03) between C1 and C2 was also found for the reduced-salt cheeses (i.e. 31% and 56% of the initial LAB counts (cfu/g) retained for C1 and C2, respectively), whereas no significant difference was found for the unsalted cheeses (i.e. 58% and 61% of the initial LAB counts (cfu/g) retained for C1 and C2, respectively). After 2 weeks of ripening, a significant lower percentage of LAB counts (cfu/g) were observed in both the reduced-salt cheeses (P = 0.01) and the normal-salted cheeses (P = 0.01), produced with C1 compared to C2. No significant difference between C1 and C2 was observed for the unsalted cheeses, after 2 weeks of ripening. At later stages of ripening there were a significant difference (P b 0.05) between C1 and C2 for all type of cheeses despite their NaCl concentration, with the lowest percentage of the initial LAB counts (cfu/g) retained in cheeses produced with C1 compared to C2. The NS-LAB counts were not significantly influenced by the NaCl content in cheeses. For cheeses produced with C1, the highest number of NS-LAB was observed after 2 weeks of ripening i.e. 1.5 × 107 cfu/g. After 11 weeks of ripening the NS-LAB had decreased to 3.0 × 106 cfu/g. Cheeses produced with C2 had NS-LAB counts that decreased from 3.1 × 107 cfu/g after 7 weeks of ripening to 1.4 × 107 cfu/g after 11 weeks of ripening (results not shown). 3.2. Microscopic examination of bacteria in cheeses The SEM micrographs of cheeses showed that bacteria were organized as single cells, in groups of 2–3 cells or in groups of ≥ 4 cells (Fig. 2A). Bacteria were attached to the protein matrix and positioned in the interface between the protein structure and the whey pockets/ fat globules (Fig. 2B). The holes in the protein matrix originate from

Table 1 Chemical composition and pH of semi-hard Samsoe cheesesa produced with various levels of NaCl after 1 week of ripening. DL — starter culture

C1

C2

a b c d e

Chemical compositionb

% moisture (w/w) % NaCl (w/w) % NaCl (w/v) % protein (w/w) % fat (w/w) pH % moisture (w/w) % NaCl (w/w) % NaCl (w/v) % protein (w/w) % fat (w/w) pH

NaCl levelc Unsalted

Reduced

Normal

P-valued

54.7 ± 1.0 b0.15e b0.27 25.4 ± 0.7 15.6 ± 0.2 5.34 ± 0.02 54.1 ± 1.0 b0.15 b0.27 25.5 ± 0.5 15.7 ± 0.4 5.29 ± 0.05

53.0 ± 0.8 1.2 ± 0.1 2.3 ± 0.1 25.8 ± 0.4 15.3 ± 0.2 5.35 ± 0.03 52.9 ± 0.3 1.2 ± 0.1 2.3 ± 0.1 25.9 ± 0.2 15.4 ± 0.3 5.30 ± 0.01

52.4 ± 0.7 1.8 ± 0.1 3.4 ± 0.2 25.7 ± 0.5 15.3 ± 0.2 5.28 ± 0.02 51.2 ± 0.1 1.7 ± 0.1 3.2 ± 0.2 26.6 ± 0.3 15.6 ± 0.3 5.28 ± 0.03

0.04 b0.001 b0.001 0.6 0.2 0.01 0.003 b0.001 b0.001 0.02 0.5 0.6

Chemical composition of cheeses produced at Arla Strategic Innovation Centre, comparable to cheeses produced at Thise dairy. Means of three replicated batches ± standard deviation. Unsalted = b0.27% (w/v) NaCl; Reduced = 2.3% (w/v) NaCl; Normal = 3.4% or 3.2% (w/v) NaCl. One-way analysis of variance at 95% level of significance. Limit of quantification was 0.15 g/100 g cheese.

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

63

Fig. 1. Lactic acid bacteria (LAB) counts (cfu/g) (M17 agar with 0.5% (w/v) lactose) for the ripened semi-hard Samsoe cheeses compared to the initial number of LAB in the cheeses i.e. before brining (8.9 × 108 cfu/g for C1 and 1.7 × 109 cfu/g for C2). Cheeses with b0.3% w/v ( ), 2.3% w/v ( ), and 3.4% w/v ( ) NaCl and two DL-starter cultures i.e. A) C1, and B) C2 are shown. The mean value was calculated from duplicated samples of triplicated batches. Asterisks denote the values significantly different (P b 0.05) from the normal-salted cheeses (i.e. 3.4% (w/v) NaCl).

fat and whey, which were removed during sample preparation. The CLSM micrographs of cheeses (Fig. 3) showed that bacteria were organized as described above for the SEM micrographs. The fat was dispersed within the protein matrix and bacteria cells were preferentially positioned at the interface between the two (Fig. 3B). Fig. 4A and B shows the total number and distribution of viable bacteria on exterior cheese samples, from the CLSM micrographs, produced with starter cultures C1 and C2, respectively. The number of viable bacteria was reduced during the ripening period in all cheeses. Cheeses produced with C1 had, in general, lower numbers of total viable bacteria compared to cheeses produced with C2. For the normal-salted cheeses produced with C1, the most marked decrease (60% reduction) in the number of viable bacteria was seen between week 2 and week 7 (Fig. 4A). After 12 weeks of ripening, the number of viable bacteria was reduced with 90%, compared to the number of viable bacteria after 2 weeks of ripening. For the normal-salted cheeses produced with C2, the most marked decrease (44% reduction) in the number of viable bacteria was, on the other hand, observed between week 7 and week 12 (Fig. 4B). The reduction in the number of viable bacteria were 19% after 7 weeks of ripening and 63% after 12 weeks of ripening, compared to the number of viable bacteria after 2 weeks of ripening, respectively. In general, an inverse relationship between the total number of viable bacteria and NaCl content was observed, but significant differences were only found in cheeses ripened for 2 weeks. Cheeses produced with C1 had significantly higher (P b 0.04) numbers of viable

bacteria in the unsalted cheeses ripened for 2 weeks, compared to the reduced- and normal-salted cheeses, whereas the difference between the reduced- and normal-salted cheeses were not significant. For cheeses produced with C2 significantly higher (P = 0.04) numbers of viable bacteria were found in the reduced-salt cheeses ripened for 2 weeks, compared to the normal-salted cheeses, whereas the difference between the unsalted cheeses and the normal-salted cheeses was not significant. Bacteria organized in groups of ≥ 4 cells accounted for the largest proportion of the total bacteria after 2 weeks of ripening. Furthermore, it was observed that the decrease in the total number of viable bacteria during ripening, was mainly due to a reduction in the number of viable bacteria organized in groups of ≥4 cells. For the normal-salted cheeses produced with C1, bacteria positioned in groups of ≥4 cells constituted 77% and 45% of the total number of viable bacteria after 2 and 12 weeks of ripening, respectively, whereas bacteria in groups of 2–3 cells accounted for 13% and 33% after 2 and 12 weeks and single cells for 10% and 22% after 2 and 12 weeks, respectively. For the normalsalted cheeses produced with C2 these numbers were 74% and 63% for bacteria organized in groups of ≥ 4 cells, 14% and 16% for bacteria in groups of 2–3 cells, and 12% and 21% for single cells after 2 and 12 weeks, respectively. Results from the exterior cheese samples were basically confirmed by the results from the interior cheese samples presented in Fig. 4C and D. An even larger reduction in the number of viable bacteria organized in groups of ≥ 4 cells was observed during ripening for the interior

Fig. 2. Scanning electron microscopy micrographs (10,000× magnification) of 12 week ripened semi-hard Samsoe cheeses with b0.3% w/v NaCl. From the micrographs are observed A) bacteria organized as single cells, in groups of 2–3 cells and in groups of ≥4 cells, and B) bacteria attached to the protein matrix and positioned in the interface between the protein structure and the whey pockets/fat globules.

64

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

A

1 µm

B

C

1 µm

1 µm

Fig. 3. Confocal laser scanning microscopy micrographs of 2 week ripened semi-hard Samsoe cheeses with 3.4% w/v NaCl stained with LIVE/DEAD BacLight viability stain. Viable and nonviable bacteria are colored green and red, respectively. The background consists of protein matrix with green fluorescence and dark spaces containing fat. Background color differs due to variation in the intensity of fluorescence. From the micrographs are observed A) three single viable bacteria, B) viable bacteria in a microcolony forming a circle around fat, and C) viable and nonviable bacteria in a microcolony.

samples. For the normal-salted cheeses produced with C1, bacteria organized in groups of ≥ 4 cells constituted 72% and 21% of the total number of viable bacteria after 2 and 12 weeks of ripening, respectively,

while bacteria in groups of 2–3 cells accounted for 13% and 41% after 2 and 12 weeks, respectively, and single cells for 15% and 38% after 2 and 12 weeks, respectively. For the normal-salted cheeses produced

Fig. 4. Number and distribution of viable bacteria on (A, B) exterior and (C, D) interior semi-hard Samsoe cheese samples during ripening, measured with confocal scanning laser microscopy combined with LIVE/DEAD staining. Cheeses with b0.3%, 2.3%, and 3.4% (w/v) NaCl and two DL-starter cultures i.e. (A, C) C1 and (B, D) C2 were produced. The number of viable cells organized as single bacteria ( ), 2–3 bacteria ( ), and in groups of ≥4 cells ( ) are shown. The mean value was calculated from duplicated batches. For each batch 15 micrographs from different sections within a sample were analyzed (6 to 13 micrographs in week 2 and week 7 for C2). Asterisks denote the values significantly different (P b 0.05) from the normal-salted cheeses (3.4% (w/v) NaCl).

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

65

Table 2 Percentage of nonviable bacteria from the exterior and the interior of semi-hard Samsoe cheeses produced with various levels of NaCl, determined with confocal scanning laser microscopy using LIVE/DEAD viability stain. Culture

Position

Week

NaCl level† Unsalted

C1⁎

Exterior

Interior

C2⁎

Exterior

Interior

2 7 12 2 7 12 2‡ 7‡ 12 2 7 12

b

Reduced ┼

25 ± 3 (n = 843) 44a ± 14 (n = 581) 81a ± 4 (n = 361) 37b ± 1 (n = 995) 55a ± 9 (n = 784) 62b ± 10 (n = 792) 19a ± 1 (n = 729) 36a ± 2 (n = 890) 57a ± 9 (n = 1121) 13a ± 7 (n = 1037) 34a ± 13 (n = 1461) 49a ± 24 (n = 926)

a

55 ± 7 (n = 413) 44a ± 11 (n = 189) 57a ± 11 (n = 378) 49a ± 2 (n = 864) 65a ± 9 (n = 316) 92a ± 3 (n = 333) 30a ± 14 (n = 1310) 23b ± 2 (n = 502) 60a ± 17 (n = 1127) 25a ± 7 (n = 1291) 19a ± 14 (n = 992) 60a ± 14 (n = 862)

Normal 52a ± 8 (n = 534) 56a ± 7 (n = 215) 65a ± 26 (n = 141) 42b ± 2 (n = 725) 52a ± 34 (n = 410) 43b ± 11 (n = 486) 21a ± 19 (n = 484) 21b ± 4 (n = 522) 61a ± 20 (n = 683) 21a ± 1 (n = 1359) 12a ± 8 (n = 1039) 34a ± 7 (n = 690)

a–b

Significantly different (P b 0.05) percentages of nonviable bacteria within the same row are indicated with superscript letters. ⁎ Average of duplicated batches ± standard deviation. † Unsalted = b0.27% (w/v) NaCl; reduced = 2.3% (w/v) NaCl; normal = 3.4% or 3.2% (w/v) NaCl. ┼ n = total number of bacteria. ‡ 12–26 micrographs were analyzed for the exterior samples from week 2 and week 7 produced with C2. For all other samples 30 micrographs were analyzed.

with C2 these numbers were 63% and 5% for bacteria organized in groups of ≥4 cells, 15% and 26% for bacteria in groups of 2–3 cells and 22% and 69% for single cells after 2 and 12 weeks, respectively. Total numbers of viable bacteria correlated well between the exterior (A and B) and the interior (C and D) cheese samples. The only significant difference (P = 0.04) was observed after 2 weeks of ripening, where the reduced-salt cheeses produced with C1, had higher numbers of viable bacteria in the interior samples, compared to the surface cheese samples. In addition, no significant influence of NaCl on the organization of viable bacteria on exterior and interior cheese samples was found (Fig. 4). Table 2 shows the percentage of nonviable bacteria during ripening from the cheese exterior and interior produced with C1 and C2, respectively. The proportion of nonviable bacteria increased in all samples between week 2 and week 12, although statistical significant difference between week 2 and week 12 were not found for all samples (results not shown). Cheeses produced with C1 had, in general, a higher proportion of nonviable bacteria in cheeses ripened for 2 and 7 weeks, compared to C2, whereas differences between the two cultures were minimal in cheeses ripened for 12 weeks. For exterior of the normalsalted cheese produced with C1, we found 52%, 56% and 65% nonviable bacteria after 2, 7 and 12 weeks ripening, respectively. These numbers were 21%, 21% and 61% for the similar C2 samples after 2, 7 and 12 weeks of ripening, respectively. The influence of NaCl on the proportion of nonviable bacteria was significant at the early stages of ripening,

i.e. 2 and 7 weeks, but not after 12 weeks. However, statistical significance was not found for all samples and no correlation between NaCl level and the proportion of nonviable bacteria could be established. Still, the total number of bacteria (viable plus nonviable) was higher during ripening for the unsalted cheeses, in general, compared to the normal-salted cheeses (Table 2). Differences between the exterior and the interior samples for each culture were minimal, due to large variation within the samples. Due to NaCl equalization within the cheeses, the differences between the exterior and the interior samples diminished, at the end of the ripening period. 3.3. Autolysis Autolysis was determined by measuring the activity of post-proline dipeptidyl aminopeptidase (PepX) in cheese extract during ripening (Fig. 5). In general, the activity of PepX increased during the ripening period. Furthermore, a negative correlation between NaCl content and PepX activity was found. The activity of PepX before brining was 0.15 nmol AMC/min/mL and 0.11 nmol AMC/min/mL for cheeses produced with C1 and C2, respectively. For the normal-salted cheeses produced with C1 there was a significant increase (P b 0.01) in the activity of PepX up to and including week 7 (Fig. 5A). Between week 7 and week 11 the activity of PepX decreased significantly (P = 0.03). For cheeses produced with C1, NaCl had a significant influence, from week 2 and until the end of ripening, on the activity of PepX. From week 2, the

Fig. 5. Post-proline dipeptidyl aminopeptidase (nmol/min/mL cheese extract) activity in semi-hard Samsoe cheeses during ripening. Cheeses with b0.3% w/v ( ), 2.3% w/v ( ), and 3.4% w/v ( ) NaCl and two DL-starter cultures i.e. A) C1, and B) C2 are shown. Values presented are the means of four measurements of single samples from triplicated batches. Asterisks denote the values significantly different (P b 0.05) from the normal-salted cheeses (3.4% (w/v) NaCl).

66

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

activity of PepX was significantly higher (P b 0.05) in the unsalted and reduced-salt cheeses, compared to the normal-salted cheeses. Additionally, the activity of PepX was significantly higher (P b 0.05) in the unsalted cheeses after 2 and 11 weeks of ripening, compared to the reduced-salt cheeses. For the normal-salted cheeses produced with C2 a significant increase (P b 0.01) in the activity of PepX was seen during the whole ripening period (Fig. 5B). For the cheeses produced with C2, NaCl had a significant influence on PepX activity from week 2, where the unsalted cheeses had higher activity of PepX (P = 0.02), compared to the normal-salted cheeses. No significant difference was seen between reduced- and normal-salted cheeses, after 2 weeks of ripening. At week 7, the activity of PepX did not differ significantly between the unsalted and the normal-salted cheeses, however, the reduced-salt cheeses had significantly higher (P = 0.01) activity of PepX, compared to the normal-salted cheeses. No significant influence of NaCl on PepX activity was observed in cheeses ripened for 11 weeks. Within the first week of ripening, a markedly different increase in the activity of PepX for C1 and C2 was seen. For the normal-salted cheeses produced with C1, the activity of PepX increased with a factor 8, compared to only a factor 4 for cheeses produced with C2. Similar observations were seen in the unsalted cheeses (i.e. a factor 12 increase for C1 and a factor 3 increase for C2) and reduced-salt cheeses (i.e. a factor 9 increase for C1 and a factor 4 increase for C2). 3.4. Free amino acid composition Fig. 6 shows the total amount of FAA as a function of ripening time. The total amount of FAA increased significantly (P b 0.05) during ripening. The NaCl content had a marked influence on the concentration of FAA. For cheeses produced with C1 a significantly higher (P b 0.001) concentration of total FAA was observed after 2 and 7 weeks of ripening in the unsalted cheeses, compared to the normal-salted cheeses, whereas no significant difference between the normal-salted and the reduced-salt cheeses was observed (Fig. 6A). At the end of ripening, the concentration of FAA was, on the contrary, significantly lower (P = 0.02) in the unsalted cheeses, compared to the normal-salted cheeses, whereas a significantly higher (P b 0.001) concentration was found in the reduced-salt cheeses. For cheeses produced with C2, the concentration and development of FAA during ripening were almost comparable to those reported for C1. However, during ripening the unsalted cheeses produced with C1 had higher concentrations of FAA, compared to the unsalted cheeses produced with C2. At the end of ripening, the concentration of FAA in the unsalted cheeses produced with C1 was 76 mmol/kg, compared to 60 mmol/kg for the unsalted cheeses produced with C2. A PCA model with four principal components is shown in Fig. 7. The FAA profiles could efficiently discriminate cheeses according to ripening

time, NaCl content, and DL-starter culture. The PC1 mainly expanded the variation in FAA caused by NaCl content, PC2 the variation caused by ripening time, and PC3 the variation caused by DL-starter culture, explaining 48%, 26%, and 12% of the total variance, respectively. As shown by the loading plot in Fig. 7A2, markers of increasing ripening time were methionine, ornithine and threonine, whereas phenylalanine and proline described cheeses in the beginning of the ripening period. The amino acids tyrosine, tryptophan, isoleucine and alanine characterized the unsalted cheeses, whereas leucine, glycine, asparagine and glutamic acid were associated with increasing NaCl level, i.e. reducedand normal-salted cheeses. The loading plot in Fig. 7B2 shows an inverse correlation between glutamic acid and γ-amino butyric acid. Cheeses produced with culture C1 were correlated with glutamic acid, whereas cheeses produced with culture C2 were correlated with γamino butyric acid. Sensory evaluation of the reduced- and normal-salted cheeses by a trained panel at the dairy showed that taste, mouth feel, and consistency were influenced by the NaCl concentration and the starter culture used. For C1 the normal-salted cheese was found to be significantly more hard, salty, acidic, harmonious and less bitter compared to the reduced version. For the reduced- and normal-salted cheeses produced with C2 there was not found any significant difference by the sensory panel. Furthermore, the reduced-salt cheese produced with C2 was preferred by the sensory panel, compared to the reduced-salt cheese produced with C1 (results not shown). 4. Discussion The LAB counts were 2.0 × 108 cfu/g and 9.7 × 108 cfu/g in normalsalted Samsoe cheeses ripened for 1 week produced with the two DLstarter cultures C1 and C2, respectively. These levels are in agreement with Ardö et al. (2002), who reported LAB counts of 7.9 × 108 cfu/g in Danbo cheese ripened for 1 week. After 7 weeks of ripening, the normal-salted cheeses had LAB counts of 4.5 × 107 cfu/g for C1 and 3.5 × 108 cfu/g for C2. Previously, we have found LAB counts from the interior of 6 weeks ripened Danbo, cheeses with 3.0% to 3.4% (w/v) NaCl, varying between 3.1 × 105 cfu/g and 5.7 × 105 cfu/g (Gori et al., 2013). The difference between the two studies could be caused by variation in the sampling position, i.e. that the present study sampled from a blend of the interior and the exterior of cheese versus sampling from the interior of the cheese as in our previous study. Furthermore, variations in the DL-starter culture composition and production method at the dairies will also influence the LAB counts. Throughout the entire ripening period, the unsalted cheeses had the highest LAB counts (cfu/g), compared to both the reduced- and normalsalted cheeses, with the exception after 11 weeks of ripening for the unsalted cheeses produced with C2. The observed inverse relationship

Fig. 6. Total free amino acid content during ripening of semi-hard Samsoe cheeses. Cheeses with b0.3% w/v ( ), 2.3% w/v ( ), and 3.4% w/v ( ) NaCl and two DL-starter cultures i.e. A) C1, and B) C2 are shown. Values presented are the means of two measurements of single samples from triplicated batches. Asterisks denote the values significantly different (P b 0.05) from the normal-salted cheeses (3.4% (w/v) NaCl).

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

67

A1

A2

B1

B2

Fig. 7. Score (A1, B1) and loading (A2, B2) plots generated from a PCA model of free amino acid data from eighteen semi-hard cheese samples. Cheeses with b0.3%, 2.3%, and 3.4% (w/v) NaCl and with two DL-starter cultures (C1 and C2) were produced. The mean value from duplicated measurements of single samples from triplicated batches was used. Numbers in brackets, at each axis label, is the percent variance explained by each PC of the total variance in the data. Score plot A1 is colored according to NaCl content; b 0.3% (red), 2.3% (green), and 3.4% w/v (blue). Score plot B1 is colored according to culture; C1 (blue), and C2 (red). Cysteine was below the limit of quantification (2.5 ppm/kg).

between LAB counts and NaCl content in cheese is in accordance with previous findings (Mistry and Kasperson, 1998; Møller et al., 2012; Rulikowska et al., 2013; Upreti et al., 2006). The unfavorable hyperosmotic environment for bacterial growth at increasing NaCl levels, is the most likely reason for this observation. Beyond week 7, no significant influence of NaCl content on the LAB counts (cfu/g) was found. A similar observation was done by Rulikowska et al. (2013), who reported that the negative correlation between NaCl level in cheddar cheese and LAB counts on M17 medium, was only evident up to 8 weeks of ripening. The same authors also found lower non-starter LAB counts in cheeses ripened for 8 weeks with 1.1% and 2.6% (w/v) NaCl compared to cheeses with 3.6% to 5.0% (w/v) NaCl. As Wilkinson et al. (1994) reported, M17 medium was not selective for Lactococcus spp. and after 9 weeks of ripening the population enumerated on this media, will be dominated, not only by the Lactococcus spp. included in the starter acidification culture, but also include non-starter LAB. Growth of less NaCl sensitive non-starter LAB on M17 medium, at the late stage of ripening, could explain why no significant influence of NaCl level on the LAB counts was evident after 11 weeks of ripening, in the present study. Recently, Ruggirello et al. (2014) evaluated the viability of Lc. lactis throughout cheese ripening by RT-qPCR and plating on M17 medium. The results indicated that the Lc. lactis starter population was mainly present in a viable but not culturable state during ripening. The authors therefore conclude that culture-dependent methods have to be supplemented with direct analyses of cheese samples. Direct observations of the bacteria in the complex matrix of Samsoe cheese were possible by the use of SEM and CLSM in combination with

LIVE/DEAD viability stain. From the SEM micrographs of Samsoe cheeses we observed cocci cells attached to the protein matrix. The cocci cells were located in holes in the protein matrix where whey and fat have been positioned before sample preparation. Most likely, the hydrophilic LAB (Boonaert and Rouxhet, 2000) are dispersed in whey, as the hydrophobic nature of fat makes positioning of LAB, within the fat globules, unlikely. Further investigation of cheeses using CLSM showed that the bacteria were positioned mostly at the interface between fat and protein network, which supported our observations from the SEM micrographs concerning the location of bacteria. Additionally, according to our observations Lopez et al. (2006) using CLSM on Emmental cheese reported that bacterial colonies primarily were localized at the fat/protein interface, dispersed in whey pockets surrounding fat inclusions. In this study, the bacteria in Samsoe cheeses were shown to be organized as single cells, groups of 2–3 and in groups of ≥4 cells and with no measurable influence of NaCl level on the organization of the bacteria. For Emmental cheese, Lopez et al. (2006) observed that bacteria were only organized in colonies during ripening. Differences in the organization of bacteria between our study and Lopez et al. (2006), could be due to variations in the composition of the starter culture, as this may influence the organization of bacteria (Champagne et al., 1995). Furthermore, variations in the production methods applied for Emmental and Samsoe cheeses could affect the organization of bacteria. Differences in the starter culture populations at the time of rennet addition, has previously been reported to influence the distribution of bacteria in model cheeses analyzed by CLSM (Hannon et al., 2006). Hannon et al. (2006) observed that bacterial cells were organized into colonies when the

68

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

starter culture population was low at the time of rennet addition. On the other hand, bacterial cells were dispersed as single cells or small colonies of cells when the starter culture population was high at the time rennet was added. As the starter culture population at the time of rennet addition was not determined in our study or reported in the study by Lopez et al. (2006), it can only be hypothesized whether the population size at the time of rennet addition differed. The bacteria positioned in groups of ≥ 4 cells accounted for the largest proportion of the total viable bacteria after 2 weeks of ripening. After both 7 and 12 weeks of ripening, the decrease in the number of viable bacteria was mainly due to a reduction in the number of viable bacteria organized in groups of ≥4 cells. This could possibly be due to increased cellular stresses caused by lack of nutrients and a more acidic environment, especially in the center of the colonies, compared to the stress experienced at the single cell level. This has previously been observed by Hannon et al. (2006) who found that lysis of the starter culture was delayed in cheeses with an even distribution of single cells, compared to lysis in cheeses with starter cells organized in colonies. It should be noted that we only evaluated the organization of cells in this study. However, the distance between cells in the cheese matrix could potential be of importance for the flavor formation, as diffusion of substrates in the cheese matrix to the enzymes liberated from the cells, is needed for both the lipolytic and proteolytic event to occur (Hannon et al., 2006; McSweeney and Sousa, 2000). In future work it would be interesting to evaluate whether a reduction of NaCl in Samsoe cheese influences the spatial distribution of the cells. In cheeses ripened for 2 weeks, an increase in the number of viable bacteria was observed by CLSM at reduced NaCl levels. At later stages of ripening, there was no correlation between the NaCl content and the number of viable bacteria. This confirms the observation from the LAB counts where no significant influence of the NaCl content was found after 7 weeks of ripening. As the environment within the cheeses changes during ripening, i.e. depletion of nutrients and accumulation of organic acids (Beresford et al., 2001), these factors could possibly become predominant in determining the viability of the starter culture at the end of ripening. This could explain why no significant influence of NaCl was found, at the late stage of ripening. Activity of PepX is not only influenced by the level of autolyzed cells but also the proportion of permeabilized cells, as both groups releases intracellular enzymes into the cheese matrix (Lortal and ChapotChartier, 2005). Additionally, the environment in the cheese will affect the activity of the enzyme (Wilkinson et al., 1994). In our study, an inverse relationship between the activity of PepX and NaCl concentration was observed, in Samsoe cheeses during ripening. This is in agreement with Wilkinson et al. (1994), who observed a 13-fold decrease in the activity of PepX after 60 days ripening, when the NaCl concentration increased from 0.5% to 4.9% (w/v) in Cheddar cheese. On the contrary, in two recently published studies, a positive relationship between NaCl concentration and the activity of PepX in Cheddar cheese was reported (Møller et al., 2012; Rulikowska et al., 2013). The influence of NaCl on autolysis differs between species (Koz et al., 2010). Furthermore, strain related differences in autolysis have been observed to influence proteolytic enzyme activity in Cheddar cheese during ripening (Sheehan et al., 2006). The different findings could therefore be due to variations in the composition of the starter cultures at both species and strain level. Additionally, the different salting procedures for Cheddar cheese (mixing NaCl into the milled curd before pressing) and Samsoe cheese (immersion into a saturated brine after pressing), may influence autolysis of the starter culture. The use of PepX activity, as a marker enzyme of cell autolysis, has previously been discussed, due to instability of the enzyme in the cheese environment (Chapot-Chartier et al., 1994; Wilkinson et al., 1994). The stability of PepX in cheese curd and in a cell free extract was evaluated by Wilkinson et al. (1994). Only 15% of the initial activity of PepX remained after 24 h in the cheese curd and the activity of PepX had decreased to 80% of the initial activity after 4 h in a cell free extract.

Additionally, Wilkinson et al. (1994) observed a marked decrease in the activity of PepX after 60 days of ripening in Cheddar cheeses with NaCl contents of 0.5% to 4.9% (w/v), and with no detectable PepX activity after 120 days of ripening. In our study, we also observed a decrease in the activity of PepX between week 7 and week 11 for the reducedand normal-salted cheeses produced with C1. In addition, the large decrease in the LAB counts (cfu/g) for the reduced- and normal-salted cheeses during the early stages of ripening, was not accompanied by a proportional increase in the activity of PepX, as expected. The results therefore highlight the need for further evaluation of PepX, as a marker of cell autolysis during cheese ripening. As expected, a time-related increase in the free amino acid (FAA) content was seen, indicating progression of proteolysis during ripening. At the end of ripening, the concentration of FAA correlated positively with the NaCl concentration in the cheeses, as observed in a previous study in Cheddar cheese (Møller et al., 2012). After 11 weeks of ripening, the concentration of FAA was higher in both the reduced- and normal-salted cheeses, i.e. 89–107 mmol/kg, compared to the concentration previously reported, i.e. ~70 mmol/kg in a Samsoe 30+ cheeses ripened for 11 weeks with 1.9% to 2.2% (w/w) NaCl (Ardö et al., 2002). The different findings can be explained by variations in pH, ripening temperature and composition of the starter culture, which are factors that can influence the peptidase activity in cheese (Gatti et al., 2008). Previously, it has been reported that the NaCl content, ripening time and starter culture influenced the composition of FAA in Samsoe and Danbo cheeses (Ardö et al., 2002; Kristiansen et al., 1999), which is in accordance with our results. Kristiansen et al. (1999) found that the amount of tyrosine, phenylalanine and tryptophan increased, with decreasing NaCl content in Danbo cheese. In this study, similar observations for tyrosine and tryptophan was obtained, which indicate that peptidase activity is not only affected by NaCl but also by moisture content since the moisture content was 2% to 3% higher in the unsalted cheeses, compared to the normal-salted cheeses. The inverse correlation between glutamic acid and γ-amino butyric acid, has previously been described in cheese (Møller et al., 2013). LAB are responsible for the formation of γ-amino butyric acid, which is generated from glutamic acid by the use of glutamic acid decarboxylase (Christensen et al., 1999). Different activity of glutamic acid decarboxylase possessed by the starter bacteria (Nomura et al., 1998), variations in pH and substrate availability (Christensen et al., 1999) could explain the difference in the concentration of glutamic acid and γ-amino butyric acid in cheeses produced with C1 and C2, respectively. No correlation between the activity of PepX and the concentration of FAA in the cheeses was found. This was unexpected, as it is well established that the formation of FAA in cheese primarily is determined by the activity of the intracellular peptidases released into the cheese matrix when bacterial cells lyses (Chapot-Chartier et al., 1994; Fox, 1989; Hannon et al., 2003). More than 16 intracellular peptidases with different specificity have been identified in LAB (Christensen et al., 1999). These could be influenced differently by the NaCl content in cheese (Wilkinson et al., 1994), which in turn could explain the lack of correlation between PepX activity and the concentration of FAA. In addition, the lack of correlation could also be explained by instability of PepX during cheese ripening. Furthermore, results from both the LAB counts (cfu/g) and the CLSM micrographs of cheeses revealed that starter culture viability decreased faster during ripening in the reduced- and normal-salted cheeses, compared to the unsalted cheeses. Leakage of peptidases into the cheese matrix following cell dead (Bourdat-Deschamps et al., 2004), could possibly explain the higher concentration of FAA in the reduced- and normal-salted cheeses, at the end of ripening. The viability of the two DL-starter cultures differed during ripening of the Samsoe cheeses. It was a general observation that the viability of the starter culture in cheeses produced with C1 decreased faster, in comparison to cheeses produced with C2. This was shown from the LAB counts (cfu/g), e.g. for normal-salted cheeses where 22% and 58%

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

of the initial LAB (cfu/g) were found after 1 week of ripening, for C1 and C2, respectively. These results were confirmed, after 2 weeks of ripening, by the CLSM micrographs of LIVE/DEAD stained cheeses where 52% and 21% of the bacteria were nonviable on the exterior of normalsalted cheeses produced with C1 and C2, respectively. The results of PepX activity in the cheeses indicated that cell autolysis, within the first week of ripening, was more pronounced in the normal-salted cheeses produced with C1, as the activity of PepX increased with a factor 8, compared to only a factor 4 for normal-salted cheeses produced with C2. The difference observed could be caused by a more rapid autolysis of starter culture C1 but also reflect differences in strain composition between the two cultures, as the activity of PepX is known to be strain dependent for the species included in a DL-starter culture (Meijer et al., 1996; Pedersen et al., 2013). In Lc. Lactis spp. the regulation of expression of PepX was observed by Meijer et al. (1996) to be both strain dependent and affected by the growth medium. Recently Pedersen et al. (2013) evaluated the aminopeptidase activity of five Leuconostoc spp. strains isolated from traditional DL-starter cultures. The authors concluded that all strains had a low aminopeptidase activity and that the activity of PepX differed between spp. In general, a less significant influence of NaCl concentration on viability of DL-starter culture C2, compared to C1 was shown. It is known that the ability to survive at high NaCl concentrations is strain dependent. Generally, strains of Lc. lactis ssp. lactis have been found to be more NaCl tolerant compared to strains of Lc. lactis ssp. cremoris (Chapot-Chartier et al., 1994; Crow et al., 1993; Guinee and Fox, 1999). Our results could indicate that Lc. lactis ssp. cremoris is more predominant in C1 and that Lc. lactis ssp. lactis is more predominant in C2. Based on the results obtained C1 had a faster reduction in cell viability, a higher rate of cell autolysis, but a lower concentration of FAA at the end of ripening as compared to C2. Additionally, a larger difference between the NaCl concentrations was observed for C1 as compared to C2, which was also reflected in the sensory analysis. In conclusion, a reduction of the NaCl content in semi-hard Samsoe cheeses from 3.4% to b 0.3% (w/v) had a significant influence on the proliferation, viability, and autolysis of the two DL-starter cultures during ripening. Reducing the NaCl level resulted in an increased number of LAB counts (cfu/g), especially at the early stages of ripening. Additionally, the impact of NaCl on the viability of the two DL-starter cultures differed during ripening. From the confocal laser scanning microscopy micrographs of cheeses, it was observed that bacteria were organized as single cells, in groups of 2–3 cells and in groups of ≥4 cells. During ripening, the decrease in the number of viable bacteria was mainly due to a reduction in the number of viable bacteria placed in groups of ≥4 cells. The activity of post-proline dipeptidyl aminopeptidase, used as a marker for the extent of autolysis of the starter culture, increased as the NaCl content was reduced in cheeses. However, our results indicated that this enzyme was unstable in the cheese environment, and its use as a marker of cell autolysis should be evaluated further. At the end of ripening, cheeses with b0.3% (w/v) NaCl had a lower content of free amino acids, compared to the cheeses with higher NaCl content. As flavor and texture are known to be affected by the viability of the starter culture, this highlights that careful selection of specific DL-starter cultures for NaCl reduced cheeses is needed. Acknowledgments This research was supported by the Danish Dairy Board (project no. 1030511001), Arla Foods amba (Denmark) (project no. 1030501001) and Department of Food Science, Faculty of Science, University of Copenhagen (project no. 5001405002). We thank both Arla Foods Pilot Dairy and Thise Dairy for assistance and providing facilities for the production of cheeses used in this study. Louise Salt and Alan Mackie at Institute of Food Research, Norwich Research Park, Colney, Norwich, UK are thanked for providing access to and assistance in the SEM analysis. Pernille Johansen, Department of Food Science, Faculty of Science,

69

University of Copenhagen is kindly thanked for proof-reading the article.

References Anon, 2003. Diet, nutrition and the prevention of chronic diseases. Report of a joint WHO/ FAO expert consultation. WHO Tech. Rep. Ser. No. 916. Anon, 2006. ISO5943:IDF88. Cheese and Processed Cheese Products – Determination of Chloride Content – Potentiometric Titration Method. Int. Dairy Fed, Brussel, Belgium. Anon, 2007. Reducing salt intake in populations. Report of a WHO Forum and Technical Meeting. 5–7 October 2006, Paris, France. WHO, Geneva, Switzerland. Anon, 2008. ISO3433:IDF222. Cheese – Determination of Fat Content – Van Gulik Method. Int. Dairy Fed, Brussel, Belgium. Anon, 2001. ISO8968-1:IDF20-1. Milk – Determination of Nitrogen Content – Part 1: Kjeldahl Method. Int. Dairy Fed, Brussel, Belgium. Anon, 2004. ISO5534:IDF4. Cheese and Processed Cheese — Determination of the Total Solids Content (Reference Method). Int. Dairy Fed, Brussel, Belgium. Ardö, Y., Larsson, P.O., Månsson, H.L., Hedenberg, A., 1989. Studies of peptidolysis during early maturation and its influence on low-fat cheese quality. Milchwissenschaft 44, 485–495. Ardö, Y., Thage, B.V., Madsen, J.S., 2002. Dynamics of free amino acid composition in cheese ripening. Aust. J. Dairy Technol. 109–115. Beresford, T.P., Fitszimons, N.A., Brennan, N.L., Cogan, T.M., 2001. Recent advance in cheese microbiology. Int. Dairy J. 11, 259–274. Boonaert, C.J.P., Rouxhet, P.G., 2000. Surface of lactic acid bacteria: relationships between chemical composition and physicochemical properties. Appl. Environ. Microbiol. 66, 2548–2554. Booth, M., Donnelly, W.J., Fhaoláin, I.N., Jennings, P.V., O'Cuinn, G., 1990. Proline-specific peptidases of Streptococcus cremoris AM2. J. Dairy Res. 57, 79–88. Bourdat-Deschamps, M., Le Bars, D., Yvon, M., Chapot-Chartier, M.-P., 2004. Autolysis of Lactococcus lactis AM2 stimulates the formation of certain aroma compounds from amino acids in a cheese model. Int. Dairy J. 14, 791–800. Brown, I.J., Tzoulaki, I., Candeias, V., Elliott, P., 2009. Salt intakes around the world: implications for public health. Int. J. Epidemiol. 38, 791–813. Champagne, C.P., Piette, M., Saint-Gelais, D., 1995. Characteristics of lactococci cultures produced in commercial media. J. Ind. Microbiol. 15, 472–479. Chapot-Chartier, M.-P., Deniel, C., Rousseau, M., Vassal, L., Gripon, J.-C., 1994. Autolysis of two strains of Lactococcus lactis during cheese ripening. Int. Dairy J. 4, 251–269. Christensen, J.E., Dudley, E.G., Pederson, J.A., Steele, J.L., 1999. Peptidases and amino acid catabolism in lactic acid bacteria. Antonie Van Leeuwenhoek 76, 217–246. Cogan, T.M., Hill, C., 1999. Cheese starter cultures. Cheese: Chemistry, Physics and Microbiology, 2nd Ed General Aspects vol. 1. Aspen Publishers, Inc., Gaithersburg, Maryland, pp. 193–255. Crow, V.L., Coolbear, T., Holland, R., Pritchard, G.G., Martley, F.G., 1993. Starters as finishers: starter properties relevant to cheese ripening. Int. Dairy J. 3, 423–460. Daly, C., 1983. The use of mesophilic cultures in the dairy industry. Antonie Van Leeuwenhoek 49, 297–312. Fox, P.F., 1989. Proteolysis during cheese manufacture and ripening. J. Dairy Sci. 72, 1379–1400. Gatti, M., De Dea Lindner, J., Gardini, F., Mucchetti, G., Bevacqua, D., Fornasari, M.E., Neviani, E., 2008. A model to assess lactic acid bacteria aminopeptidase activities in Parmigiano Reggiano cheese during ripening. J. Dairy Sci. 91, 4129–4137. Gori, K., Ryssel, M., Arneborg, N., Jespersen, L., 2013. Isolation and identification of the microbiota of Danish farmhouse and industrially produced surface-ripened cheeses. Microb. Ecol. 65, 602–615. Guinee, T.P., Fox, P.F., 1999. Salt in cheese: physical, chemical and biological aspects. Cheese: Chemistry, Physics and Microbiology, 2nd Ed General Aspects vol. 1. Aspen Publishers, Inc., Gaithersburg, Maryland, pp. 257–302. Hannon, J.A., Wilkinson, M.G., Delahunty, C.M., Wallace, J.M., Morrissey, P.A., Beresford, T.P., 2003. Use of autolytic starter systems to accelerate the ripening of Cheddar cheese. Int. Dairy J. 13, 313–323. Hannon, J.A., Lopez, C., Madec, M.-N., Lortal, S., 2006. Altering renneting pH changes microstructure, cell distribution, and lysis of Lactococcus lactis AM2 in cheese made from ultrafiltered milk. J. Dairy Sci. 89, 812–823. Hystead, E., Diez-Gonzalez, F., Schoenfuss, T.C., 2013. The effect of sodium reduction with and without potassium chloride on the survival of Listeria monocytogenes in Cheddar cheese. J. Dairy Sci. 96, 6172–6185. Koz, R.A., Solichov, I.N.A., Ondr, I.V.A., Plockov, M., 2010. Effect of some environmental factors on autolysis of lactococci used for cheese production. J. Food Nutr. Res. 49, 1–9. Kristiansen, K.R., Dening, A.S., Jensen, D.F., Ardö, Y., Qvist, K.B., 1999. Influence of salt content on ripening of semi-hard round-eyed cheese of Danbo-type. Milchwissenschaft 54, 19–23. Lopez, C., Maillard, M.-B., Briard-Bion, V., Camier, B., Hannon, J.A., 2006. Lipolysis during ripening of Emmental cheese considering organization of fat and preferential localization of bacteria. J. Agric. Food Chem. 54, 5855–5867. Lortal, S., Chapot-Chartier, M.-P., 2005. Role, mechanisms and control of lactic acid bacteria lysis in cheese. Int. Dairy J. 15, 857–871. McSweeney, P.L.H., Sousa, M.J., 2000. Biochemical pathways for the production of flavour compounds in cheeses during ripening: a review. Lait 80, 293–324. Meijer, W., Marugg, J.D., Hugenholtz, J., 1996. Regulation of proteolytic enzyme activity in Lactococcus lactis. Appl. Environ. Microbiol. 62, 156–161. Mistry, V.V., Kasperson, K.M., 1998. Influence of salt on the quality of reduced fat cheddar cheese. J. Dairy Sci. 81, 1214–1221.

70

L. Søndergaard et al. / International Journal of Food Microbiology 213 (2015) 59–70

Møller, K.K., Rattray, F.P., Høier, E., Ardö, Y., 2012. Manufacture and biochemical characteristics during ripening of Cheddar cheese with variable NaCl and equal moisture content. Dairy Sci. Technol. 92, 515–540. Møller, K.K., Rattray, F.P., Bredie, W.L.P., Høier, E., Ardö, Y., 2013. Physicochemical and sensory characterization of Cheddar cheese with variable NaCl levels and equal moisture content. J. Dairy Sci. 96, 1953–1971. Nomura, M., Kimoto, H., Someya, Y., Furukawa, S., Suzuki, I., 1998. Production of gammaaminobutyric acid by cheese starters during cheese ripening. J. Dairy Sci. 81, 1486–1491. Pedersen, T.B., Ristagno, D., McSweeney, P.L.H., Vogensen, F.K., Ardö, Y., 2013. Potential impact on cheese flavour of heterofermentative bacteria from starter cultures. Int. Dairy J. 33, 112–119. Ruggirello, M., Dolci, P., Cocolin, L., 2014. Detection and viability of Lactococcus lactis throughout cheese ripening. PLoS ONE 9, 1–14. http://dx.doi.org/10.1371/journal. pone.0114280. Rulikowska, A., Kilcawley, K.N., Doolan, I.A., Alonso-Gomez, M., Nongonierma, A.B., Hannon, J.A., Wilkinson, M.G., 2013. The impact of reduced sodium chloride content on Cheddar cheese quality. Int. Dairy J. 28, 45–55. Saint-Eve, A., Lauverjat, C., Magnan, C., Déléris, I., Souchon, I., 2009. Reducing salt and fat content: Impact of composition, texture and cognitive interactions on the perception of flavoured model cheeses. Food Chem. 116, 167–175.

Schroeder, C.L., Bodyfelt, F.W., Wyatt, C.J., McDaniel, M.R., 1988. Reduction of sodium chloride in cheddar cheese: effect on sensory, microbiological, and chemical properties. J. Dairy Sci. 71, 2010–2020. Sheehan, A., Cuinn, G.O., Fitzgerald, R.J., Wilkinson, M.G., 2006. Proteolytic enzyme activities in Cheddar cheese juice made using lactococcal starters of differing autolytic properties. J. Appl. Microbiol. 100, 893–901. Singh, T.K., Young, N.D., Drake, M., Cadwallader, K.R., 2005. Production and sensory characterization of a bitter peptide from β-casein. J. Agric. Food Chem. 53, 1185–1189. Sousa, M., Ardö, Y., McSweeney, P.L., 2001. Advances in the study of proteolysis during cheese ripening. Int. Dairy J. 11, 327–345. Upreti, P., Metzger, L.E., Hayes, K.D., 2006. Influence of calcium and phosphorus, lactose, and salt-to-moisture ratio on Cheddar cheese quality: proteolysis during ripening. J. Dairy Sci. 89, 444–453. Werner, H., Nielsen, W., Ardö, Y., Rage, A., Antila, V., 1999. North European varieties of cheese. Cheese: Chemistry, Physics and Microbiology, 2nd Ed Major Cheese Groups vol. 2. Aspen Publishers, Inc., Gaithersburg, Maryland, pp. 245–262. Wilkinson, M.G., Guinee, T.P., Fox, P.F., 1994. Factors which may influence the determination of autolysis of starter bacteria during cheddar cheese ripening. Int. Dairy J. 4, 141–160.