Effects of Protein, Calcium, and pH on Gene Transcription, Cell-Envelope Peptidase Activity of Lactococcus lactis Strains, and the Formation of Bitter Peptides

Calcium- and protein-rich fermented milk products, such as concentrated yoghurts and fresh cheeses, may contain undesired bitter peptides, which are generated by the proteolytic cleavage of casein. Up to now, it is not clear whether this process is caused by endogenous milk enzymes, such as plasmin and cathepsin D, or whether proteolytic enzymes from applied starter cultures, such as the lactococcal cell-envelope peptidase PrtP, are involved. A sensory analysis of fresh cheese products made from milk concentrates fermented with prtP-negative and -positive Lactococcus lactis strains revealed bitterness in the products fermented with prtP-positive L. lactis strains. Two prtP-positive strains, LTH 7122 and LTH 7123, were selected to investigate the effect of increased calcium concentrations (additional 5 mM and 50 mM CaCl2) at neutral (pH 6.6) and acidic (pH 5.5) pH-values on the transcription of the prtP gene and its corresponding PrtP peptidase activity in milk citrate broth (MCB). For both strains, it was shown that prtP transcription was upregulated only under slightly elevated calcium conditions (5 mM CaCl2) after 5 h of growth. In concordance with these findings, PrtP peptidase activity also increased. When higher concentrations of calcium were used (50 mM), prtP expression of both strains decreased strongly by more than 50%. Moreover, PrtP peptidase activity of strain LTH 7123 decreased by 15%, but enzymatic activity of strain LTH 7122 increased slightly during growth under elevated calcium concentrations (50 mM CaCl2). Fermentations of reconstituted casein medium with 3.4% (w/v) and 8.5% (w/v) protein and different calcium concentrations using strain LTH 7122 revealed no clear relationship between prtP transcription and calcium or protein concentration. However, an increase in PrtP peptidase activity under elevated protein and calcium conditions was observed. The activity increase was accompanied by increased levels of bitter peptides derived from different casein fractions. These findings could be a possible explanation for the bitterness in fermented milk concentrates that was detected by a trained bitter panel.


Introduction
The production of fermented milk products, such as curd cheese, strained (Greek-style) yogurt, or different types of fresh cheese, generates acid whey. Due to its high content of lactic acid, insoluble whey proteins, minerals such as calcium, enzymes, microorganisms Escherichia coli strains were routinely grown in Luria-Bertani (LB) broth [28]. Overnight cultures were prepared by inoculating LB broth with a single colony and incubating them at 37 • C, 180 rpm, in a rotary incubator.

Experimental Setup for Fermentation
In order to investigate the impact of the selected L. lactis strains on the fermentation characteristics and sensory properties of fresh cheese (CoF) made from concentrated milks obtained by microfiltration, a so-called micro fresh cheese system (µFCS) was used. The µFCS, which was described previously by Schäfer et al. [29], is an experimental setup at laboratory scale for fermentation of (concentrated) milk combined with pH monitoring and logging. For fermentation experiments, micellar casein powder (MCC60), which was prepared according to Schäfer et al. [29], was used. The contents of protein, dry matter, lactose, and calcium were 57.06 ± 0.12 g/100 g, 96.05 ± 0.12 g/100 g, 29.96 ± 0.85 g/100 g, and 19.82 ± 0.33 g/kg, respectively. The MCC60 powder was reconstituted (protein concentration: 8.5% (w/w); calcium content: 2.98 g/kg (74 mM; see Table 2)) with deionized water for 5 min at 50 • C and approx. 8.000 rpm using a rotor-stator mixer (Ultra-Turrax T50, IKA-Werke GmbH & Co. KG, Staufen, Germany). A batch-pasteurization (72 • C for 15 s) of the concentrate was conducted by means of a water bath adjusted to 85 • C. Pasteurization was stopped by cooling in ice water, and the concentrates were stored at 6 • C until use. The concentrated milk was warmed up to the fermentation temperature of 22.5 • C and transferred into sterilized plastic containers (volume = 400 mL) under a laminar flow cabinet. Each concentrate-filled container was inoculated with approx. 4.5 × 10 9 colonyforming units (cfu) of each L. lactis strain. Containers were closed and sterilized pH probes (SE555X/2-NMSN; Knick Elektronische Messgeräte GmbH & Co. KG, Berlin, Germany) were inserted. Table 2. Chemical composition and fermentation characteristics of (reference) fermented-concentrated and concentratedfermented fresh cheeses manufactured by means of fermentation and without fermentation (control, direct acidified). Values represent the means and standard deviations (i = 2, n = 2). The fermentation was conducted in a water bath (Kiss K 25; Peter Huber Kältemaschinenbau GmbH, Offenburg, Germany) at a temperature of 22.5 • C, and acidification was continuously monitored until a pH-value of 4.7 was achieved, or no change in pH-values for ≥4 h was observed. Obtained fermented concentrated milks and fresh cheese samples, respectively, were cooled to ≤6 • C and stored until use. The data were converted (Advantech-ADAM-6217, Sphinx Computer Vertriebs-GmbH, Laudenbach, Germany; MemoRail A1401, Knick Elektronische Messgeräte GmbH & Co. KG, Berlin, Germany), and pH-values as a function of fermentation time were recorded by means of a software (DAQ Factory Express, AzeoTech Inc., Ashland, MA, USA). The initial and final pH-values as well as the fermentation time were evaluated.
Furthermore, a CoF fresh cheese control sample was manufactured without fermentation of the pasteurized concentrated milk, i.e., direct acidification, as described previously by Schäfer et al. [30]. An amount of about 400 mL concentrate contained in a double-walled beaker was cooled to ≤2 • C in a water bath. During continuous stirring, the pH-value was adjusted to pH 4.7 by dropwise addition of 1 M citric acid (Carl Roth GmbH & Co. KG, Karlsruhe, Germany). To ensure a steady final pH, 10 min equilibration time were met and, if necessary, citric acid was added dropwise again. The acidified concentrated milk was filled into a sterilized plastic container (volume = 400 mL), and gel formation was initiated by means of warming using a water bath (15 • C) for a holding time of 2 h. The control CoF fresh cheese sample was cooled to ≤6 • C and stored until use. The manufacture of all CoF fresh cheeses was performed in duplicate.

Sensory Evaluation of the Bitterness Level
The level of bitterness of the fresh cheese samples was evaluated after one week of storage by a trained sensory panel. The panel consisted of 25 members (male/female/diverse; smoker/non-smoker; age: approx. 22 to 58 years), and they were trained to evaluate the level of bitterness using caffeine solutions of 0.05, 0.1, 0.15, 0.175, 0.2, 0.25, and 0.3 g/L, respectively, according to the standard DIN ISO 3972:2013-12 [31] of the German Institute for Standardization (DIN)/International Organization for Standardization (ISO). All panelists could identify a caffeine concentration of at least 0.25 g/L as bitter, which fits the requirements recommended for a bitterness evaluation of food samples [31]. The training and evaluations were carried out at 22-25 • C in the Dr. Rainer-Wild sensory laboratory of the Institute of Food Science and Biotechnology, University of Hohenheim, Stuttgart, Germany, in separated booths.
In order to compare the fresh cheeses, they should have nearly the same appearance and consistency. Thus, prior to the sensorial evaluation, all fresh cheeses were mechanically treated with a hand blender (ESGE-Zauberstab M180, ESGE AG, Mettlen, Switzerland) at level two (n = 17,000 rpm) for 10 to 180 s until visibly smooth and homogenous masses, i.e., liquid to paste-like microgel suspensions, were obtained. The challenge was that some samples were liquid-like and contained gel particles; thus, a very short blending time of approximately 10 s had to be applied, and other samples appeared to be too thin. In order to evaluate the bitterness level of the fresh cheese samples, a polarized sensory positioning was carried out [32,33].
Depending on each trial, the number of panelists varied between 12 and 13. At each session, prior to the sensorial evaluation, two samples were served as reference for the panelists: (1) a commercially available FCo fresh cheese (Cremiger Quark mit frischem Joghurt, Berchtesgadener Land Chiemgau eG (D-BY 110 EG), Piding, Germany) and (2) a CoF fresh cheese which was manufactured from the reconstituted MCC60 powder and was fermented with the strain LTH 7122 at 22.5 • C. As this fresh cheese was ranked as very bitter by the sensorial panel, this fresh cheese was used as a bitter reference.
The FCo fresh cheese product was chosen as a reference based on results of an inhouse sensorial evaluation indicating that the bitterness level was ≤2 (slightly bitter) during 5 weeks of storage at approx. 6 • C, as described by Schäfer et al. [5].
Samples (approx. 20-30 g; temperature = approx. 10 • C) of each fresh cheese were served in plastic containers labeled with randomized 3-digit numbers. To neutralize any carry-over taste during the sensorial evaluation, tap water and matzo bread (Pico Food GmbH, Tamm, Germany) were served. As proposed by Sebald et al. [6], each panelist wore a nose-clip (Vitalograph GmbH, Hamburg, Germany) to prevent cross-modal interactions with other odorants during sample evaluation using the scale as mentioned above and at individual speed.

Isolation of L. lactis Plasmid DNA
Plasmid DNA was isolated according to the protocol of Anderson and McKay [34] with the following modifications: after the addition of lysozyme (10 mg/mL, SERVA Electrophoresis GmbH, Heidelberg, Germany), 20 µL RNase A (100 mg/mL, Qiagen GmbH, Hilden, Germany) was added to the reaction mixes, which were incubated at 37 • C for 60 min. Phenol, which was saturated with TE-buffer and chloroform-isoamyl alcohol (both Carl Roth GmbH & Co. KG, Karlsruhe, Germany), was used for the protocol. Purified plasmid DNA was dissolved in 30 µL of TE-buffer and analyzed by electrophoresis on 0.8% (w/v) agarose gels.

PCR Analysis of prtP
The presence of prtP was analyzed by amplifying a conserved region of 163 bp by PCR with the primers prtP_for (5 -CGC TTA GCC GCA TGT CTT TA-3 ) and prtP_rev (5 -ACA GTC ACA TTG GCG AAA GT-3 ). The PCR reaction mix consisted of 2.5 µL ThermoPol ® Buffer, 0.5 µL deoxynucleoside triphosphate (dNTP) mix (10 mM each) and 0.125 µL Taq DNA Polymerase (5,000 U/mL) (all New England Biolabs Inc., Ipswich, MA, USA), 0.5 µL of each primer (10 pmol/µL, Eurofins Genomics GmbH, Ebersberg, Germany), and 50 ng of plasmid DNA as template. Plasmid DNA of strains LTH 1657 and LTH 1418 was used as prtP-positive and prtP-negative controls, respectively. A notemplate control (NTC) was included using sterile ultrapure water (GIBCO TM , Thermo Fisher Scientific Inc., Waltham, MA, USA) instead of template DNA. The volume was adjusted to 25 µL with sterile ultrapure water (GIBCO TM , Thermo Fisher Scientific Inc., Waltham, MA, USA). The PCR reaction was conducted in a T1 thermocycler (Biometra GmbH, Göttingen, Germany) with the following PCR conditions: initial denaturation, 95 • C, 5 min; 30 cycles of denaturation, 95 • C, 0.5 min, annealing, 51 • C, 0.5 min, and elongation, 68 • C, 0.5 min; and a final elongation at 68 • C, 5 min. The primer pair was constructed with the software Primer3web (version 4.1.0; http://primer3.ut.ee, accessed on 1 February 2018) using a conserved region of the prtP gene. This conserved region was identified by aligning known prtP sequences of different strains of L. lactis (Table S1) available at the National Center for Biology Information (NCBI).

Construction of a Plasmid Standard for the Absolute Quantification of prtP Gene Expression in a Quantitative Real-Time RT PCR (qPCR) Transcriptional Assay
For preparation of a plasmid standard for the absolute quantification of prtP transcription, the pGEM ® -T Easy Vector System (Promega Corporation, Madison, WI, USA) was used. Therefore, the prtP PCR product of strain LTH 1657 (see Section 2.4) was purified with a QIAquick ® PCR Purification Kit (Qiagen GmbH, Hilden, Germany) following the corresponding purification protocol. The purified PCR products were cloned into plasmid pGEM ® -T Easy following the pGEM ® -T and pGEM ® -T Easy Vector Systems technical manual (revised 12/18, TM042). Fifty nanogram of plasmid DNA and a three-fold molar excess of the PCR product was chosen for a ligation reaction, which was incubated at 4 • C overnight. Preparation of chemically competent E. coli DH5α cells and transformation procedures were carried out according to standard procedures [28]. Selection of transformants was carried out by blue-white screening on LB agar containing 100 µg/mL ampicillin, 0.5 mM isopropyl β-D-thiogalactopyranoside (IPTG), and 80 µg/mL 5-bromo-4-chloro-3-indoxyl-β-D-galactoside (X-Gal) (Carl Roth GmbH & Co. KG, Karlsruhe, Germany), according to the pGEM ® -T and pGEM ® -T Easy Vector Systems technical manual. Positive transformants were chosen for the isolation of plasmid DNA with a QIAprep Spin Miniprep Kit (Qiagen GmbH, Hilden, Germany), according to the handbook (July 2006). The resulting plasmid was analyzed by gel electrophoresis and designated pCM01.  autoclaved medium was 6.6. This medium is designated in the following as basic MCB (pH 6.6).
In order to adjust the calcium content of the broth into the range of the CoF fresh cheese evaluated as bitter, sterile 0.5 M and 5 M calcium chloride stock solutions were used.
To get close to protein concentrations present in milk, the powders were reconstituted to a protein concentration of 3.4% (w/v), as described below. The respective calcium concentrations were 28 mM for SdCa-mCN (3.4% protein (w/v)) and 15 mM for HdCa-mCN (3.4% protein (w/v)). A high protein fermentation was conducted at a protein concentration of 8.5% (w/v), resulting in calcium concentrations of 70 mM for SdCa-mCN (8.5% protein (w/v)) and 38 mM for HdCa-mCN (8.5% protein (w/v)). Prior to fermentation, the solid casein powders were reconstituted in distilled and autoclaved water at 4 • C for 3-4 h, followed by a sterilization treatment at 95 • C for 30 min in a water bath. The sufficiency of the sterilization process was examined by use of plate count agar (0.5% (w/v) peptone, 0.25% (w/v) yeast extract, 0.1% (w/v) glucose, 1.6% (w/v) agar, pH 7). After sterilization, the reconstituted casein medium was stirred slowly over night at 4 • C and then used for the fermentation experiments on the next day.

Cultivation of L. lactis Strains in Milk Citrate Broth and Reconstituted Casein Medium with Different Calcium and Protein Concentrations
Precultures in M17 broth were grown overnight (see Section 2.1) and the cells were harvested by centrifugation at 4255× g, resuspended in 0.1 M sodium phosphate buffer pH 7.2, and the optical density at 600 nm (OD 600 ) was determined with a DU ® 730 UV/Vis Scanning Spectrophotometer (Beckman Coulter Inc., Brea, CA, USA). With this suspension, basic MCB or reconstituted casein medium (SdCa-/HdCa-mCN, each with 3.4% (w/v) and 8.5% (w/v) protein concentration) were inoculated to an OD 600 of 0.1. In the experimental approaches with basic MCB and high calcium concentrations (in the range of the CoF fresh cheeses), 10 µL of the 0.5 M calcium chloride solution/mL was added to basic MCB (for additional 5 mM calcium chloride) and 10 µL of the 5 M calcium chloride solution/mL (for additional 50 mM calcium chloride) before addition of the cells.
After addition of 50 mM calcium chloride to basic MCB, the pH-value dropped from 6.6 to 5.5. In the approaches with MCB (pH 5.5), 3.4 µL 4 N HCl/mL was added before addition of the cells to adjust the pH-value to 5.5.
The cultures were then incubated under aerobic conditions at 30 • C without shaking for 24 h, and samples were taken hourly for serial dilution and were spread-plated in duplicate on M17 agar. The viable counts were determined after incubation at 30 • C for 24-48 h.
For RNA extraction, samples containing approx. 2 × 10 8 cfu were taken in duplicate at the respective time points and treated with RNAprotect ® Bacteria Reagent (Qiagen GmbH, Hilden, Germany), according to the corresponding Handbook (January 2015) with the following modifications: centrifugation was performed at 4255× g and samples were stored at −70 • C until further processing.

Sample Preparation for prtP Transcriptional Analysis
For RNA extraction, the samples (see Section 2.6.2) were thawed at room temperature and the RNA was isolated with the RNeasy ® Mini Kit (Qiagen GmbH, Hilden, Germany) following protocols 4 and 7 of the RNAprotect ® Bacteria Reagent Handbook with minor modifications. The samples were incubated at 37 • C with 750 rpm for 30 min on a ther- momixer (protocol 4). The RNA was eluted (protocol 7) with 50 µL RNase-free water preheated to 37 • C and kept on ice for further processing or stored at −70 • C. Residual DNA within the RNA preparations was degraded with the Turbo DNA-free TM Kit (Thermo Fisher Scientific Inc., Waltham, MA, USA) following the routine treatment protocol described in the TURBO DNA-free TM Kit User Guide (TURBO TM DNase Treatment and Removal Reagents, Revision G). The RNA integrity was examined by formaldehyde-based denaturing agarose gel electrophoresis. A total of 300 ng of RNA were reverse transcribed into cDNA with the iScript TM Select cDNA Synthesis Kit (Bio-Rad Laboratories Inc., Hercules, CA, USA) following the instructions (10001023 Rev B) for the use of random primers. An additional approach with a no-reverse transcriptase control (-RT) was performed for each sample to detect residual DNA.
2.6.4. Generation of Standard Curves for the Absolute Quantification of prtP Expression in the qPCR Transcriptional Assay Plasmid pCM01 was serially diluted with sterile ultrapure water (GIBCO TM , Thermo Fisher Scientific Inc., Waltham, MA, USA) to obtain 6.0 × 10 2 to 6.0 × 10 7 copies/µL according to the technical manual "Creating Standard Curves with Genomic DNA or Plasmid DNA Templates for Use in Quantitative PCR" (Applied Biosystems, Waltham, MA, USA).
One microliter of each of pCM01 dilutions as well as 50 ng plasmid template DNA of the L. lactis type strain LTH 1657 were used in the respective prtP-qPCR assay (see Section 2.6.5) in order to confirm that the gene copy numbers were covered by the dynamic range of the plasmid standard curves. In the qPCR assays, the standard curves were automatically created with the threshold cycle values (C t values) of the plasmid dilution series and the corresponding decadic logarithms of the starting quantity (either concentration or gene copy number) by the qPCR software (see Section 2.6.5). Three biological replicates with three technical replicates each were performed for proving the reproducibility of the standard plasmid curve, which all met the MIQE criteria [35].

Establishment of a Real-Time PCR Assay for Absolute Quantification of prtP Transcription
For the qPCR assays, an iQ TM 5 Multicolor Real-Time PCR Detection System with the corresponding software (Bio-Rad Laboratories Inc., Hercules, CA, USA) was used. PCR reactions with the prtP primer pair (see Section 2.4) were analyzed in gradient-PCR followed by melting curve analysis (Table 3) in order to prove the specificity of the PCR reaction and to determine the respective annealing temperatures. For these analyses, the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories Inc., Hercules, CA, USA) was used together with the primers (0.8 µL) and template DNA of the strain LTH 1657 (50 ng plasmid DNA) in a 20 µL reaction. The qPCR was set to the following conditions: initial denaturation, 95 • C, 5 min, and 40 cycles of denaturation (95 • C, 0.5 min), annealing (49.8 • C, 0.5 min), and elongation (68 • C, 0.5 min; collection of optical data for gene expression analysis). Table 3.
Cycling programs of the gradient-PCR and melt curve analysis for the prtP-qPCR establishment.

Reaction Step
Temperature (

Determination of PrtP Peptidase Activity
The determination of PrtP activity was conducted as described by Hebert et al. [36]. Samples (5 mL) were taken and harvested by centrifugation (4255× g, 4 • C, 10 min). Cells were washed with 0.9% (w/v) sodium chloride solution containing 10 mM of calcium chloride to avoid detachment of PrtP from the cell surface [37]. Subsequently, the washed cells were suspended in 50 mM sodium phosphate buffer (pH 7.0) and applied in the peptidase activity assay. The PrtP activity was determined by use of the synthetic substrate Suc-Ala-Ala-Pro-Phe-pNA (Bachem Holding AG, Bubendorf, Switzerland), which was dissolved in dimethylsulfoxide (12 mg/mL). The assay was conducted at a temperature of 30 • C in a thermoshaker. A volume of 30 µL of cell suspension was given to 200 µL of sodium phosphate buffer (pH 7), 100 µL of 5 M sodium chloride, and 20 µL of substrate solution. The substrate hydrolysis was stopped by addition of 100 µL of 80% (v/v) acetic acid. The assay's incubation time was chosen with respect to the linear range of product formation as well as the assay's limit of detection (data not shown). Precipitated protein was removed via centrifugation (17,950× g, 4 • C, 10 min), and 200 µL of the supernatant was transferred into a 96-well plate for measuring the absorption at 405 nm. One katal of PrtP peptidase activity was defined as the amount of PrtP that released 1 mol of pNA per second. Additionally, PrtP activity was standardized to its corresponding cfu (see Section 2.6.2).

Quantification of Key Bitter Peptides
Certain key bitter peptides [6] were quantified within the fermentations of reconstituted casein medium. The cultivations in SdCa-mCN and HdCa-mCN medium (3.4% and 8.5% (w/v) protein content) were done in triplicates, and samples were taken after 5 and 24 h from every cultivation. The peptides were extracted as described by Sebald et al. [6]. Therefore, 1 ± 0.01 g of sample was extracted twice with 5 mL of 0.1% (v/v) formic acid by use of 3 ± 0.1 g of glass beads (∅ 3 mm). The internal standard angiotensin I (final concentration: 1 µg/mL) was given to the sample. After 3 × 20 s of mixing on a vortex, the sample was centrifuged (12,857× g, 4 • C, 10 min). The supernatant was adjusted to 10 mL in a measuring cylinder, sterile-filtered, and stored at −20 • C until further use. The concentration of bitter peptides was analyzed according to Sebald et al. [6] via mass-spectrometry by the Core Facility of the University of Hohenheim. The same peptides (peptides&elephants, Hennigsdorf, Germany) as published by Sebald et al. [6] were used for the calibration and quantification of the peptides within the fresh cheese samples. The determined peptide concentrations were used to calculate the dose-over-threshold-factor (DoT-factor) for every peptide. For that, the bitter taste thresholds published by Sebald et al. [6] of the peptides were used. On the basis of the DoT-factor, the influence of a peptide on the bitter taste reception can be estimated. At a DoT-factor of >1.0, a peptide can be assumed as being involved in the reception of bitterness in a product. An additive contribution can be performed by subthreshold peptides with DoT-factors of 0.1 ≤ DoT ≤ 1.0 [6].

Statistical Analysis
Of each data set, the means of three biological replicates were used for statistical analysis (of the data set HdCa-mCN 3.4% (w/v) for prtP transcriptional analysis, only two biological replicates could be used for analysis). Data sets of the different cultivation experiments were analyzed on normal distribution with the Anderson-Darling test. MCB data sets were compared to the reference (basic MCB) using an unpaired t-test. Fermentations in reconstituted casein medium were compared using analysis of variance (ANOVA). The variance homogeneity was identified using Brown-Forsythe test. If requirements were given, one-way ANOVA was performed using a post hoc Fisher's least significant difference (LSD) test (α = 0.05). Statistical analysis was performed with the software OriginPro (version 2020, OriginLab Corporation, Northampton, MA, USA). Significant differences concerning the level of bitterness of the fresh cheeses were identified by conducting an ANOVA. If significant, a subsequent Fisher's LSD test (α = 0.05) was applied. The software XLSTAT 2015 (version 2015.4.01.20575, Addinsoft, New York, NY, USA) was used.

Fermentation of Concentrated Milks for the Production of CoF Fresh Cheese Samples and Sensory Evaluation
In order to produce fresh cheese (CoF) samples with prtP-positive and prtP-negative L. lactis strains for sensory evaluation, reconstituted casein medium with initial pH-values in a range of 6.6 to 6.8 were fermented until pH-values of 4.7 to 5.5 were reached ( Table 2). The use of the prtP-negative strains LTH 7117 and LTH 7112 resulted in prolonged fermentation times of approx. 50 h and 66 h, respectively, as compared to fermentation with the prtP-positive strains LTH 7122 and LTH 7123 of approximately 27 h and 42 h, respectively, to achieve the desired pH-values under the same fermentation/growth conditions. Furthermore, prtP-negative strains did not acidify the concentrated milks to pH-values ≤ 5.1. Consequently, no homogenous gel formation was observed by visual inspection ( Figure S1). Whereas the CoF fresh cheese obtained with LTH 7123 with a pH of 4.7 showed a homogenous gel, the CoF fresh cheese with LTH 7122 with same final pH showed a liquid-like (microgel) suspension with gel-particles contained therein.
Sensory evaluation indicated that CoF fresh cheese samples prepared with the prtPnegative strains LTH 7117 and LTH 7112 were perceived with low bitterness levels of 1.7 and 2.8, respectively, whereas the CoF fresh cheeses produced with prtP-positive strains LTH 7123 and LTH 7122 showed significantly higher bitterness levels of 3.9 and 4.1, respectively ( Figure 1). CoF fresh cheeses obtained with the prtP-negative strains LTH 7117 and 7112 were perceived with higher bitterness levels than the commercial reference FCo fresh cheese with a bitterness level of 0.5.
In conclusion, sensory evaluation of fermented CoF fresh cheese samples using prtPnegative and prtP-positive strains revealed significantly higher bitterness levels than the non-fermented, directly acidified, CoF control sample (Figure 1).
The CoF-fresh-cheese sample fermented with strain LTH 7117 was also judged as slightly bitter. However, a significant difference in bitterness between the control CoF fresh cheese and the reference FCo fresh cheese was not noticed. Since the L. lactis strains LTH 7122 and LTH 7123 obviously were involved in the development of bitterness in the CoF fresh cheese, these strains were further investigated with molecular methods in terms of prtP transcription and PrtP activity.

Growth of L. lactis Strains LTH 7122 and LTH 7123 in Basic MCB and Quantification of prtP Transcription
In order to set up a system for quantitative analysis of prtP transcription under model conditions, basic MCB was used, which was previously described to initiate proteolytic activity in Lacticaseibacillus casei [27]. Initially, growth curves were generated with strains LTH 7122 and LTH 7123 in basic MCB (Figure 2, white squares) to evaluate the optimal conditions for qPCR. The cultures were started with a high inoculum of approx. 10 8 cfu/mL to obtain sufficient RNA concentrations for qPCR analysis. Both strains showed similar growth characteristics in basic MCB and multiplied by approximately 1 log (Figure 2A,B, white squares). Only slight differences in viable counts were found. The prtP transcription levels ( Figure 3A,B) were highest after 3 h of growth with 2.2 × 10 5 copies/ng cDNA for LTH 7122 and 1.2 × 10 5 copies/ng cDNA for LTH 7123, demonstrating slight differences between the strains. Afterwards, a continuous decrease in transcription levels (copies/ng cDNA) was observed with increasing cultivation time. Thus, prtP gene expression levels of both strains were highest in the early growth phase; therefore, 3 h and 5 h appeared as the most promising time points for subsequent analysis of the effects of different calcium concentrations and a low pH-value on prtP gene transcription.
FCo fresh cheese with a bitterness level of 0.5.
In conclusion, sensory evaluation of fermented CoF fresh cheese samples using p negative and prtP-positive strains revealed significantly higher bitterness levels than non-fermented, directly acidified, CoF control sample (Figure 1). The CoF-fresh-cheese sample fermented with strain LTH 7117 was also judged slightly bitter. However, a significant difference in bitterness between the control fresh cheese and the reference FCo fresh cheese was not noticed. Since the L. lactis str

Quantitative Analysis of prtP Transcription during Growth of Strains LTH 7122 and LTH
7123 in Basic MCB (pH 6.6), MCB (pH 5.5), and with Additional 5 mM or 50 mM Calcium Chloride As a prerequisite for quantitative expression analysis under certain growth conditions, growth curves of strains LTH 7122 (Figure 2A) and LTH 7123 ( Figure 2B) were generated in basic MCB (pH 6.6), MCB (pH 5.5), and basic MCB with additional 5 mM or 50 mM calcium chloride. Since the supplementation of 50 mM calcium chloride to basic MCB reduced the initial pH from 6.6 to 5.5, control experiments were carried out with MCB (pH 5.5) in order to evaluate the effect of a lowered pH-value on growth and on prtP transcription levels (Figures 2 and 4). The data obtained show a high transcription level of prtP after 3 h of growth in basic MCB (pH 6.6) for both strains and a lower transcription after 5 h of growth. For both strains, the prtP expression in MCB (pH 5.5) was similar to that in basic MCB (pH 6.6) after In the first 5 h after inoculation, both strains grew well with minor differences in generation time and viable counts ( Figure 2). Whereas the decrease of the pH value to 5.5 had only a minor growth-delaying effect on strain LTH 7122 after 5 h of growth, the addition of 50 mM calcium chloride to basic MCB delayed the growth in terms of generation time and viable counts clearly in a strain-dependent manner (Figure 2A,B). However, the addition of 5 mM calcium chloride demonstrated a slightly growth-promoting effect, which was visible throughout the growth curve with strain LTH 7122 and after 5 h also with LTH 7123. These data supported the idea to measure prtP transcription at time points 3 h and 5 h after inoculation.
The prtP transcription levels of strains LTH 7122 and LTH 7123 during growth in basic MCB (pH 6.6), MCB (pH 5.5), and basic MCB (pH 6.6) with addition of 5 mM or 50 mM calcium chloride are summarized in Figure 4A.
The data obtained show a high transcription level of prtP after 3 h of growth in basic MCB (pH 6.6) for both strains and a lower transcription after 5 h of growth. For both strains, the prtP expression in MCB (pH 5.5) was similar to that in basic MCB (pH 6.6) after 3 h but was increased by 212% (LTH 7122) and 140% (LTH 7123) after 5 h of growth. A decrease of prtP expression levels during growth in basic MCB (pH 6.6) with additional 50 mM calcium chloride was observed with both strains ( Figure 4A). After 3 h and 5 h of growth, the prtP transcription of strains LTH 7122 and LTH 7123 was reduced.
To investigate the effect of a lower calcium chloride concentration on prtP transcription, 5 mM calcium chloride was added to basic MCB (pH 6.6) ( Figure 4A). In contrast to the results with 50 mM calcium chloride, prtP expression of strains LTH 7122 and LTH 7123 was affected in different ways. No effect was observed after 3 h, but transcription was increased by 107% in strain LTH 7122 and 217% in strain LTH 7123 after 5 h of growth. These data demonstrated that a high calcium concentration of 50 mM inhibits transcription, whereas the ten-times lower concentration promotes growth as well as prtP transcription in both tested strains. The PrtP peptidase activity after 3 h of growth in basic MCB (pH 6.6) was approximately 200 pkat/10 10 cfu and similar for both strains ( Figure 4B). Accordingly, PrtP peptidase activity increased for both strains during the late exponential growth phase. The PrtP peptidase activity increased also during growth in MCB (pH 5.5) after 3 h of cultivation compared to the PrtP peptidase activity of cells grown in basic MCB (pH 6.6). However, after 5 h of growth, the PrtP peptidase activity of both strains was lower than at time point 3 h.
The addition of 5 mM CaCl 2 to basic MCB (pH 6.6) led also to an increase in PrtP peptidase activity. For strain LTH 7122 a longer adaption to the conditions of elevated calcium was required, as the increased PrtP peptidase activity was determined just after 5 h of growth. In the beginning of the exponential growth at 3 h, this strain showed a PrtP peptidase activity similar to the PrtP peptidase activity of cells grown in basic MCB (pH 6.6) after 3 h of growth ( Figure 4B). Strain LTH 7123 showed an increased PrtP peptidase activity from the beginning of exponential growth.
The PrtP peptidase activity increased for both strains during the exponential growth phase in basic MCB (pH 6.6) supplemented with 50 mM CaCl 2 . The PrtP peptidase activity of strain LTH 7123 was comparable to cells grown in basic MCB (pH 6.6), whereas PrtP peptidase activity of strain LTH 7122 increased to a low extent. At the end of exponential growth, PrtP peptidase activities were similar for both strains and comparable to the PrtP peptidase activity of cells grown in MCB medium.

Quantitative Analysis of prtP Transcription during Growth of Strain LTH 7122 in Reconstituted Casein Medium with Different Protein and Calcium Concentrations
In order to confirm the results obtained in the experiments with MCB and to include the influence of elevated protein concentrations, a series of experiments was conducted using reconstituted casein medium with different protein and calcium concentrations (see Section 2.6.1). With this set of experiments, we wanted to shed light on the question of why the fermentation of milk concentrates, in contrast to the fermentation of milk, leads to strongly bitter-tasting fermentation products [5,6].
The results of the transcriptional analyses of the prtP gene in reconstituted casein medium, differing in calcium and protein concentration, are shown in Figure 5A. When comparing the approaches of the protein-rich reconstituted casein media (protein: 8.5% (w/v)) after 5 h of cultivation, it was found that calcium reduction had little effect on the transcription of prtP. In contrast, when comparing the approaches with a native protein concentration of 3.4% (w/v), it was found that transcription was decreased by approximately 85% when using HdCa-mCN medium. Interestingly, the comparison of SdCa-mCN medium with a native (3.4% (w/v)) and an elevated (8.5% (w/v)) protein concentration revealed that transcription was decreased significantly by approximately 47% under an elevated protein concentration. The comparison of HdCa-mCN medium with a native (3.4% (w/v)) and an elevated (8.5% (w/v)) protein concentration showed the opposite effect: an increase of transcription by approximately 210% under an elevated protein concentration.
After 24 h of cultivation, transcription was found to be decreased by more than 90% in all approaches compared with the respective 5 h approaches.
Thus, these results for strain LTH 7122 clearly showed that when reconstituted proteinrich casein concentrates (protein: 8.5% (w/v)) were used as the starting material for fermentation, there were no increased transcription rates of the prtP gene and also that a calcium reduction had little effect on transcription. Interestingly, even after 24 h of cultivation, there were no increased but significantly reduced transcription levels in the approaches with protein-rich concentrates compared to the respective approaches with a native protein concentration of 3.4% (w/v).

Quantitative Analysis of PrtP Peptidase Activity during Growth of Strain LTH 7122 in Reconstituted Casein Medium with Different Protein and Calcium Concentrations
L. lactis strain LTH 7122 was grown in media of reconstituted casein powders at two protein and calcium concentrations, respectively. The activity of PrtP was analyzed during exponential growth (5 h) and at the end of fermentation (24 h). Strain LTH 7122 was grown in reconstituted casein medium at a native protein concentration of 3.4% (w/v), which was compared to fermentation at an elevated protein concentration at 8.5% (w/v). At both protein concentrations, the calcium concentration was varied ( Figure 5). protein concentration.

Quantitative Analysis of Key Bitter Peptides Formed during Growth of Strain LTH 7122 in Reconstituted Casein Medium with Different Protein and Calcium Concentrations
As seen in the sensorial evaluation of CoF fresh cheese described before, strain LTH 7122 showed a capability to produce bitterness during fermentation. Therefore, samples from fermentation experiments described before were examined for their concentration of bitter peptides (Table 4). Based on the study of Sebald et al. [6], the seventeen published key bitter peptides derived from different casein fractions were used for that investigation.
Sixteen of the seventeen key peptides were found in the fermentation samples although at low concentrations, in most cases. In SdCa-mCN medium (protein: 3.4% (w/v); calcium: 28 mM), the concentration of peptides increased in the progress of fermentation and as well as at elevated protein and calcium conditions (protein: 8.5% (w/v); calcium: 70 mM). Especially the peptides VLPVPQ, MAPKHKEMPFPKYPVEPF (both derived from β-casein), and ARHPHPHLSFM (derived from κ-casein) were determined in higher concentrations. In comparison to that, in HdCa-mCN (protein: 3.4% (w/v); calcium: 15 mM) the peptide concentrations were found in comparable or lower concentrations except for the peptides MAPKHKEMPFPKYPVEPF and KVLPVPQKAVPYPQ, which derived from β-casein. The increase in protein and calcium content in casein concentrate HdCa-mCN (protein: 8.5% (w/v); calcium: 38 mM) led to higher peptide concentrations than in HdCa-mCN medium (protein: 3.4% (w/v); calcium: 15 mM). Compared to SdCa-mCN medium (protein: 3.4% (w/v); calcium: 28 mM), most peptide concentrations were increased although the calcium concentration was at a comparable level. The highest peptide concentrations were determined for ARHPHPHLSFM in the fermentation samples. This peptide derived from κ-casein and is also stated to be a primary contribution to bitterness.
Nevertheless, the DoT-factors were < 0.1 for the majority of the samples except for ARHPHPHLSFM, which additionally showed the highest DoT-factor of 0.7 in SdCa-mCN (protein: 8.5% (w/v); calcium: 70 mM).

Discussion
In the current study, prtP transcription and PrtP peptidase activity of two prtP-positive single starter strains that were isolated from a commercial starter culture were analyzed in MCB and reconstituted casein media with various calcium and protein concentrations. The two strains were chosen since the sensory analyses of fermented concentrated milk (containing approximately 74 mM calcium) revealed that both strains induce a strong sensory bitterness in the fermented dairy products.
Our findings that the CoF-fresh cheese sample fermented with the prtP-negative strain LTH 7112 was significantly more bitter than the CoF-control indicates that even prtPnegative strains may be able to induce bitterness in CoF fresh cheese to some extent.
Even the control CoF fresh cheese was perceived as very slightly bitter (bitterness level of 1.0), which was not completely in line with results of Poulsen and Mogensen [38], who reported that an ultrafiltration retentate, which was acidified with glucono-δ-lactone, was perceived as not bitter. The slight differences between both control and reference samples show that it was difficult for the panelists to differentiate between those low bitter-tasting samples.
The growth curve analyses in different compositions of MCB medium revealed that the addition of 5 mM calcium chloride to basic MCB caused a growth promoting and the addition of 50 mM calcium chloride a growth-impairing effect on both strains LTH 7122 and LTH 7123. Since growth rates of LAB in milk are strongly dependent on casein proteolysis [12,14,39,40], similar effects regarding the transcription levels and the enzymatic activity of the cell-envelope peptidase PrtP were expected. Since proteolysis by the PrtP peptidase activity leads to various casein-derived oligopeptides, among them also bitter tasting peptides [12,13,[24][25][26], an upregulation of prtP could also be considered as a bitterness-increasing process.
The results of the transcriptional analyses in basic MCB coincide with the previously described findings that PrtP peptidase activity is crucial for growth in milk [39] and that the expression level also depends on the growth phase and growth rate [41][42][43].
In the experiments using different compositions of MCB, elevated concentrations of CaCl 2 (50 mM) were shown to decrease prtP transcription, which is in contrast to the initial hypothesis that high levels of calcium lead to an upregulation of prtP. However, upon the addition of 5 mM calcium chloride, higher transcription levels of prtP were observed.
In contrast to the transcriptional analyses, it could be shown that the addition of calcium to basic MCB resulted generally in a higher PrtP activity during bacterial growth. A similar behavior was already published by Exterkate [44], who stated that the PrtP peptidase activity of L. lactis ssp. cremoris strain AM 1 depends on the calcium concentration of the surrounding medium. In their study, an addition of 5 mM calcium to the growth medium increased the PrtP peptidase activity compared to cells grown on the same medium without the added calcium. It was found that the addition of 50 mM also increased the proteolytic activity of the PrtP peptidase although to a lesser extent than 5 mM.
The cultivation experiments in reconstituted casein medium with 3.4% (w/v) and 8.5% (w/v) protein content revealed that prtP transcription was not clearly increased when strain LTH 7122 was grown in protein-rich casein concentrates. Additionally, high calcium concentrations also did not clearly increase prtP transcription. These results suggest that the previously observed increased bitterness in calcium-rich, fermented dairy products, which was also described by Schäfer et al. and Sebald et al. [5,6], cannot be explained by sheer alterations in prtP transcription due to high calcium or protein concentrations.
In contrast to the transcription levels, PrtP peptidase activity was mainly increased at elevated calcium and protein concentrations. These results are in line with the study of Exterkate [44], who observed increased activity at elevated calcium conditions. The quantification of key bitter peptides within those samples revealed higher peptide concentrations at elevated protein-and calcium concentrations. Two of those peptides (MAPKHKEMPFP-KYPVEPF and ARHPHPHLSFM) are known to exhibit a low taste threshold and thus can be assumed to have a strong impact on bitterness perception [6]. Especially for the κ-casein derived peptide ARHPHPHLSFM, concentrations up to 20 µmol/kg were determined after