lable at ScienceDirect

Food Microbiology 38 (2014) 179e191

Contents lists avai

Food Microbiology

journal homepage: www.elsevier .com/locate/ fm

Microbiological, physico-chemical, nutritional and sensorycharacterization of traditional Matsoni: Selection and use ofautochthonous multiple strain cultures to extend its shelf-life

Grazia Marina Quero a, Vincenzina Fusco a,*, Pier Sandro Cocconcelli b,Lubomila Owczarek c, Mehlika Borcakli d, Cecilia Fontana b, Sylwia Skapska c,Urszula T. Jasinska c, Tarik Ozturk d, Maria Morea a

a Institute of Sciences of Food Production, National Research Council of Italy, Via Amendola 122/O, 70126 Bari, Italyb Institute of Microbiology, Faculty of Agriculture, Università Cattolica del Sacro Cuore, Via E. Parmense 84, 29100 Piacenza, Italyc Institute of Agricultural and Food Biotechnology (IAFB), Department of Fruit and Vegetable Product Technology,36 Rakowiecka Street, 02-532 Warsaw, Polandd Tübitak, Marmara Research Center, Food Institute, P.O. Box 21, 41470 Gebze, Kocaeli, Turkey

a r t i c l e i n f o

Article history:Received 13 December 2012Received in revised form19 July 2013Accepted 15 September 2013Available online 25 September 2013

Keywords:Georgian fermented milkAutochthonous lactic acid bacteriaMicrobial ecologyDGGESensory analysis

* Corresponding author. Institute of Sciences ofResearch Council of Italy, Via Amendola 122/O, 705929322; fax: þ39 080 5929374.

E-mail address: vincenzina.fusco@ispa.cnr.it (V. Fu

0740-0020/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.fm.2013.09.004

a b s t r a c t

Matsoni, a traditional Georgian fermented milk, has variable quality and stability besides a short shelf-life (72e120 h at 6 �C) due to inadequate production and storage conditions. To individuate its typicaltraits as well as select and exploit autochthonous starter cultures to standardize its overall qualitywithout altering its typicality, we carried out a thorough physico-chemical, sensorial and microbialcharacterization of traditional Matsoni. A polyphasic approach, including a culture-independent (PCR-DGGE) and two PCR culture-dependent methods, was employed to study the ecology of Matsoni. Overall,the microbial ecosystem of Matsoni resulted largely dominated by Streptococcus (S.) thermophilus andLactobacillus (Lb.) delbrueckii subsp. bulgaricus. High loads of other lactic acid bacteria species, includingLb. helveticus, Lb. paracasei and Leuconostoc (Leuc.) lactis were found to occur as well. A selectedautochthonous multiple strain culture (AMSC) composed of one Lb. bulgaricus, one Lb. paracasei and oneS. thermophilus strain, applied for the pilot-scale production of traditional Matsoni, resulted the best interms of enhanced shelf-life (one month), sensorial and nutritional quality without altering its overalltypical quality. This AMSC is at disposal of SMEs who need to exploit and standardize the overall qualityof this traditional fermented milk, preserving its typicality.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Matsoni (also known as matzoon, matsoon, matsoun, matzoun,madzoon, madzounmacun), is a traditional fermented milk pro-duced in Georgia, a region listed among the cradles of fermentedmilks; this product, popular throughout thewhole Caucasus region,is consumed not only as such but also as ingredient for makingcakes and dough (e.g. for khachapuri) typical of the Caucasiancuisine (http://georgiaabout.com/2012/08/31/about-food-potato-rice-and-herb-soup-with-matsoni). It is so much appreciated by

Food Production, National16 Bari, Italy. Tel.: þ39 080

sco).

All rights reserved.

the Georgians that it was declared complementary food for babiesaged over 6 months by the Georgian Ministry of Labour’s Healthand Social Affairs Department (Nemsadze, 2004).

Matsoni is usually manufactured in farmhouses following atraditional protocol foreseeing the use of cow’s milk (and seldomewe’s, goat’s, buffalo milk or their mixtures) and whose productionprocess, usually carried out in glass bottles, is based on sponta-neous fermentation and back-slopping, i.e. “inoculation of milkwith a small quantity of the previous performed successfulfermentation” (Leroy and De Vuyst, 2004). This ancient but obso-lete practise makes Matsoni quality and stability variable. More-over, inadequate hygienic conditions as well as occasionalinterruptions of the cold-chain distribution may affect the finalquality of this fermented milk resulting in a high sour and bittertaste product with a short shelf-life of max 70e120 h. To overcomethese drawbacks, commercial starter cultures could be used,

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191180

resulting in a loss of the typical traits that make unique eachtraditional fermented milk (Leroy and De Vuyst, 2004).

Use of preservatives and stabilizers is becoming subject topublic concern and ever-tighter legislative control. Thus, manu-facturers need procedures suitable to control the fermentationprocess, stabilizing microbial loads and extending Matsoni shelf-life without altering its traditional taste and flavour.

An in-depth characterization of traditional Matsoni as well asthe development of autochthonous starter cultures to be used bydairy factories seem to be essential pre-requisites to reach thesegoals and favour the production of Matsoni at industrial-scale.Indeed, although Matsoni has been manufactured since morethan 100 years, only few studies have been performed to charac-terize it and mainly from the microbiological point of view(Erzinkjan, 1971; Merabishvili and Chanishvili, 2001; Reddy et al.,1986; Uchida et al., 2007).

The present study aimed at performing a thorough physico-chemical, nutritional, sensory and microbiological characteriza-tion of Matsoni and develop an autochthonous starter culture toimprove and standardize the shelf-life of this product maintainingits typical traits. The bacterial community of this fermented milkwas analysed using culture-dependent and -independent tech-niques. Isolates of the dominant species were characterized forpromising technological features. Selected autochthonous startercultures were successfully applied at lab- and pilot-scale formanufacturing Matsoni, following the traditional productionprotocol.

2. Materials and methods

2.1. Sampling and fermented milks preparation

Sixteen batches of Matsoni, named M1-M16, were collected infarmhouses located in Georgia. These fermented milks were man-ufactured following the traditional protocol (Chanishvili et al.,2001) foreseeing raw cow’s milk (60 L) pasteurized at 90 �C for10 min, cooled up to 42 �C and started with a 3% (vol/vol) back-slopping obtained from the batch production of the previous day.Matsoni is produced in glass bottles, each containing 250 mL ofmilk, which is fermented at 42 �C until the pH value of 4.6 isreached (after ca. 5e7 h). Thereafter, Matsoni bottles are placed in acooling cell and stored at 6 �C for 72e120 h.

Samples (raw milk, pasteurized milk, natural starter, artisanaland commercial Matsoni) were transported under refrigerationcondition to the laboratories and immediately analysed.

The analyses were carried out in triplicate.

2.2. Physico-chemical analyses

Matsoni samples were analysed for pH, total titratable acidity(TTA), fat, sugar, total proteins, dry matter and short chain fatty acid(SCFA) content.

The pH value and TTAwere measured in fresh Matsoni and afterits storage at 6 �C for 72 h.

The pH value was determined by direct insertion of a pH meter(Oakton Benchtop pH 510 Meters, Cole-Parmer, Vernon Hills, Illi-nois, USA).

TTAwas measured as follows: 10 g of each fermented milk wereweighted in a 250 mL Erlenmeyer flask and distilled water wasadded up to 50 mL including few drops of phenolphthalein. Themixture was titrated with 0.1 N NaOH, according to the AOACmethod n. 947.05 (AOAC, 2000) and expressed as percentage oflactic acid in 100 g of sample.

Fats were first methylated and then extracted with hexane.Methylated fatty acids were analysed by gas chromatography using

a Varian Wcot fused silica column, as reported in the EuropeanCommission Regulation no. 2568 (1991) and EuropeanCommission Regulation no. 1429 (1992).

As concerns the calculation of total protein content, the total Noccurring in the fermentedmilks was determinedwith the Kjeldahlmethod, as described for Protein in Beer (AOAC 920.53 method,AOAC, 1998). Briefly, fermented milks (0.7e2.2 mL) were digestedboiling themwith H2SO4 for at least 30min and the developing NH3was titrated with NaOH.

Dry matter (d.m.) content was determined applying the gravi-metric method AOAC 925.23 (AOAC, 1999). In accordance with thismethod, 3 mL of sample were desiccated for 3 h at 130 �C; theremaining amount of the fermented milk was weighted andexpressed as dry matter in percentage of the fresh sample.

The ash content percentage, based on the initial weight of 105 �Cdried materials, was determined in accordance with the followingformula:

Ash% ¼ ðW1=W2Þ � 100

where:W1¼weight of ash, andW2¼ initial weight of 105 �C driedsample. (Method provided by the National Renewable EnergyLaboratory, Midwest Research Institute, Kansas City, Mo 64110,United States).

The available saccharide content was calculated as differencebetween the fermented milk dry matter value and the sum of thecontents of fermented milk ash and two determined nutrients.

The Matsoni samples were evaluated for their content in simplesugars (mono- and disaccharides). A rough assessment of thecontent of these saccharides was obtained by the determination ofglucose and galactose originated after acid hydrolysis of the originalsample. The analyses were performed by HPAE-HPLC (Dionex) withpulsed amperometric detection.

The amount of lactic acid and D(�) and L(þ) lactate weredetermined by the UV-method (Roche, R-Biopharm, Darmstadt,Germany) on spectrophotometer absorption of NADH at 340 nm.

The above mentioned analyses were carried out in triplicate.

2.3. Nutritional value

The method used to estimate the Matsoni energy value wasbased on the average conversion factors for proteins, fat and car-bohydrates (Kunachowicz et al., 1998). These factors are: i) 1 g ofprotein, 17 kJ/4 kcal; ii) 1 g of fat, 37 kJ/9 kcal; iii) 1 g of carbohy-drates, 17 kJ/4 kcal. Energy value of the fresh Matsoni was calcu-lated as the sum of the energy values of its basic nutrientscontained in 100 g of fresh Matsoni (means � standard deviationsof the 16 Matsoni analysed).

2.4. Sensory analyses

Sensory profiles of four Matsoni samples (M1eM4), randomlychosen among the 16 fresh Matsoni, were evaluated in laboratory,meeting the requirements of ISO 8589:1988 (Anonymous, 1988), bythe trained panel of six assessors (ISO 8586-1:1993; Anonymous,1993) using the Quantitative Descriptive Analysis (QDA) method(Stone et al., 1974) to establish descriptors of sensory notesperceived as the most essential within the attributes of odour, tasteand texture. The note descriptors were proposed, discussed andagreed by the panel members during the training sessions. Noteintensities and the values of the overall Sensory Quality (SQ) factorsfor the samples were delimitated using scales (ISO 4121:1987,Anonymous, 1987): non-structured scales with the border re-straints denominated as “slight/none”e“high” and the 10-unit ratioscales. The evaluation of segments, marked by assessors on

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 181

unstructured scales corresponding in lengths to the magnitudes ofthe note intensities or SQ values, was performed by the appositionof segments to the ratio (numeral ordinal, 10-unit) scale.

The complete assessments of the sensory profiles and overallsensory quality of Matsoni samples consisted of the followingsteps:

1. Development of fingerprint of fermented milk’s sensory profileby setting down the descriptors of the perceived notes;

2. Quantification of sensory notes using scales (unstructured andratio scales);

3. Hedonic assessment of the overall Sensory Quality (SQ) with thescales;

4. Presentation of the sensory profile and SQ on histograms.

The list, including 13 descriptors of Matsoni sensory noteswithin 3 attributes (texture, odour and taste), set-down by theassessors as the most typical at the beginning of the sensory ses-sions, comprised the notes of homogeneity, fluidity, smoothnessand oral cavity coating for texture, the notes of sour, yogurt-like,milky and cheese-like for odour and the notes of sour, yogurt-like, bitter, milky and cheese-like for taste.

Matsoni was served in 40mL portions inwhite polystyrene cupsof 150 mL, labelled randomly with selected codes. These portionswere served at room temperature (ca. 20 �C) to better differentiateodours and flavours and facilitate the characterization and com-parison of each sample. Each assessor received 2 portions of eachMatsoni during one session. The results of sensory evaluation werereported as mean values with standard deviations of two sensorysessions.

The above mentioned sensory analyses were also carried out onfresh and cold-stored (at 6 �C for one month) samples of Matsoni Aand B produced by using the two selected autochthonous multiple

Table 1Primers used in this work.

Target Primer Primer sequence (50e30)

Lb. helveticus PeCf CTGTTTTCAATGTTGCAAGTCPeCr TTTGCCAGCATTAACAAGTCTPeNf CGCTGATTCTAAGTCAAGCTPeNr CGACTAAGAAGTGGAACATTATrif TCTTATTACGCAATGGACCAATrir AATACCGTTCTTGAGGTTAGA

Leuc. lactis Llac-f AGGCGGCTTACTGGACAACLlac-r CTTAGACGGCTCCTTCCAT

16S rRNA P0 GAGAGTTTGATCCTGGCTCAGP6 CTACGGCTACCTTGTTACGA

Lb. paracasei Y2 CCCACTGCTGCCTCCCGTAGGAPara CACCGAGATTCAACATGG

L. delbrueckiisubsp. bulgaricus

SS1 GTGCTGCAGAGAGTTTGATCCTDBI ACCTATCTCTAGGTGTAGCGCA

lacZ LCZF316 CACTATGCTCAGAATACALCZR1283 CGAACAGCATTGATGTTA

PCR-DGGE V6eV8 U968-GC CGC CCG GGG CGC GCC CCG GGGG GCA CGG GGG GAA CGC

L1401 GCG TGT GTA CAA GAC CCRAPD-PCR XD9 GAAGTCGTCCRAPD-PCR Coc1 AGCAGCGTGGprtS prtS-f GTGAGGCTTTGGCAGCTAAC

prtS-r TCGCGATATAGACCGGATTCgdh gdh-f TTGCCAAAGCTTCATGACTG

gdh-r ACATGGGAAAGCCAAGTCAGcsp csp-f TTATTACCTCTGAAGATGG

csp-r ACGTTGACCTACTTCAACATureC ureC-f GGGGATAGCGTACGTCTTGG

ureC-r TCAGCCAGCATCACCCATAACAeps eps-f AGTGATGAAATCGACGTACT

eps-r CCAACCGACTTTTCTACGAC

a Expected size of the amplified PCR fragment.

strain cultures (AMSC1 and AMSC2), described in Section 2.8, andon four fresh artisanal Matsoni (used as control).

2.5. Microbiological and molecular analyses

Twenty-five mL of each Matsoni were dispersed in 225 mL of 2%(w/vol) sodium citrate solution, homogenized and serially dilutedin sterile 0.1% buffered peptone water. The appropriate dilutionswere plated in triplicate on: M17 with 0.5% lactose (LM17, OxoidS.p.A., Garbagnate, Milan, Italy; Terzaghi and Sandine, 1975) andMRS agar (Oxoid; de Man et al., 1960) for cocci- and rod-shapedlactic acid bacteria (LAB), respectively; Violet Red Bile Lactose Agar(VRBLA, Difco Laboratories, Detroit, MI; American Public HealthAssociation, 1992) and Violet Red Bile Glucose agar (VRBGA,Difco; American Public Health Association, 1992) for total coliformsand Enterobacteriaceae, respectively; Potato Dextrose Agar (PDA,Difco; American Public Health Association, 1992) supplementedwith chloramphenicol (0.1 g/L) for both yeasts and moulds. LM17and MRS plates were incubated under anaerobic conditions(AnaeroGene, Oxoid S.p.A) for 48 h, at 42 and 30 �C, respectively.VRBLA and VRBGA plates were incubated aerobically at 37 �C for48 h and PDA plates aerobically at 25 �C for 120 h. The enumerationof Staphylococcus aureus and coagulase positive staphylococci wasperformed using Baird Parker (Oxoid), as described by Fusco et al.(2011a). Moreover, the absence of Salmonella spp. and Listeriamonocytogenes was ascertained by applying the relevant ISO stan-dard (ISO 6785:2001 (Anonymous, 2001), ISO 11290-1:1996 and itsamendment ISO 11290-1:1996/Amd 1:2004 (Anonymous, 1996,2004), ISO 6888-1:1999 and its amendment, respectively). Micro-bial analyses were performed in triplicate.

In order to select LAB strains to be used in autochthonousmultiple strain cultures, a total of 310 colonies were randomlypicked up from the highest plate dilutions of MRS and LM17 agar

Ampliconsize (bp)a

Reference

524 Fortina et al., 2001

726

918

742 Lee et al., 2000

1500 Di Cello et al., 1997

GT 290 Ward and Timmins, 1999

GGCTCAG 1400 Drake et al., 1996

968 Lick et al., 1996

GC GGG GCGGAA GAA CCT TAC

450 bp Zoetendal et al., 1998

Variable Moschetti et al., 1998Variable Cocconcelli et al., 19951501 This study

1353 This study

119 This study

1340 This studyC

1425 Stingele et al., 1996

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191182

and purified by streaking on the relevant media. The resultingisolates were stored at �80 �C in MRS broth with 20% glycerol.

DNA isolation was carried out as previously reported (Fuscoet al., 2011b).

Bacteria strain typing was performed by RAPD-PCR usingprimers XD9 and Coc1 (Table 1), for rod- and cocci-shaped LAB,respectively, as previously reported (Cocconcelli et al., 1995;Moschetti et al., 1998).

Taxonomic strain identification of Streptococcus (S.) thermophi-lus, Lactobacillus (Lb.) delbrueckii subsp. bulgaricus, Lb. helveticus, Lb.paracasei, and Leuconostoc (Leuc.) lactis was performed by species-specific PCR and/or by amplifying and sequencing the 16S rRNAgene (Table 1). The DNA sequences were obtained using the ABIPRISM Big Dye Terminator Cycle Sequencing Kit ver3.1 (PE AppliedBiosystems, Inc., Foster City, CA, USA) with the forward (P0) primer(Di Cello et al., 1997). The reaction products were analysed as pre-viously reported (Fusco et al., 2011b). The resulting sequences werecomparedwith those present in theBasic BLASTSearch, as describedby Altschul et al. (1997) and were finally deposited in GenBank.

PCR-DGGE (Polymerase Chain Reaction-Denaturing GradientGel Electrophoresis) was carried out as previously reported (Ampeet al., 1999; Zoetendal et al., 1998). V6eV8 16S rRNA gene fragmentswere obtained by amplifying the DNA either directly extracted fromraw and pasteurized milk, artisanal and commercial starter, arti-sanal (from Georgian farmhouses) and commercial (from Amaltea-Didube, Tbilisi, Georgia) Matsoni or extracted from the bulk (i.e.,the mass) of colonies harvested from one of the two agar plates(highest dilutions) used to enumerate LAB, while the second agarplate was used to isolate and identify colonies.

Primers used are listed in Table 1.

2.6. Characterization of S. thermophilus strains

The presence of genes encoding for technologically importantproteins was checked in the S. thermophilus strains with theattempt at highlighting some genetic diversity among them (Moraet al., 2002). In particular, the presence of the prtS gene, coding forproteases involved in milk casein breakdown (Courtin et al., 2002),the ureC gene related to urease enzyme (Mora et al., 2005), the gdhgene, coding for glutamate dehydrogenase (Tanous et al., 2002),and the csp gene, involved in the strain heat shock stress resistance(Wouters et al., 1999) and the eps gene cluster, targeting the exo-polysaccharide locus sequence (Stingele et al., 1996) was ascer-tained. Primers used to amplify these genes were designed usingthe sequences obtained from the Genbank database (accessionnumber AF243528, CP000023 region 410792..412162, AF023492,NC006449 region 276683..278401) using the Primer 3 website(http://www-genome.wi.mit.edu). Primer sets and the relevantamplicon size are reported in Table 1.

In order to test the S. thermophilus strains exopolysaccharides(EPS) capability, they were streaked on ruthenium red milk (RRM)plates (0.5% yeast extract, 10% skim milk powder, 1% sucrose, 1.5%agar and 0.08 g L�1 ruthenium red per litre), as reported by Stingeleet al. (1996). EPS-positive colonies were white after 48 h of incu-bation at 42 �C under anaerobic conditions.

2.7. Acidification activity of LAB strain cultures

The acidification activity in reconstituted skim milk of theselected LAB strain cultures was evaluated by pH measurementusing a pH meter (Oakton Benchtop pH 510 Meters, Cole-Parmer,Vernon Hills, Illinois, USA). Briefly, 1% (ca. 6 log cfu/mL) of eachstrain culture was inoculated in 9% skim milk pasteurized at 90 �Cfor 20 min. The analysis was carried out in triplicate using non-inoculated skim milk as negative control. The pH values were

registered during fermentation at 42 �C at time 0 and after 4 and 6 hof fermentation at 42 �C as well as during storage at 6 �C after one,two, three and four weeks.

Multiple strain cultures (in skim milk), composed of combina-tions of three strains (1% i.e., ca. 6 log CFU/mL of each strain), werealso evaluated by pH measurement as reported above.

2.8. Experimental pilot-scale Matsoni production

Two Matsoni, named A and B, respectively, were manufacturedwith a pilot-plant, inoculating 150 L of pasteurized cow milk withtwo autochthonous multiple strain cultures (AMSC1 and AMSC2),respectively. The two newly Matsoni A and B were produced, intriplicate, at the AMALTEA DIDUBE Milk dairy plant (Tbilisi, Geor-gia), following the traditional protocol reported in the Section 2.1,using 3% inoculum (ca. 6 log cfu/mL per each strain).

Values of pH were measured at time 0 and during storage at6 �C, at intervals of 24 h up to 1 week and at intervals of one weekup to one month.

Physico-chemical, nutritional, microbiological and molecularanalyses were carried out, as reported in the Sections 2.2, 2.3 and2.5, respectively.

To monitor the strains composing AMSC1 and AMSC2, 180 col-onies were picked up from the LM17 and MRS agar plates (seededwith the highest sample dilutions) at time 0 and after one, two and 4weeks at 6 �C. They were isolated and subjected to DNA extraction,species-specific PCR and RAPD-PCR, as described in the Section 2.5.

Sensory analyses were carried out as reported in the Section 2.4.

2.9. Statistical analysis

Microbiological and physico-chemical analyses were performedin triplicate. The relevant means and standard deviations valueswere calculated. The significance of differences between meanvalues of sensory notes of Matsoni samples was searched with 1-Way Analysis of Variances ANOVA and Duncan test (p < 0,05) us-ing Statistica 7.1 StatSoft.

3. Results

3.1. Physico-chemical analyses of traditional Matsoni

Physico-chemical characterization of 16 artisanal Matsoni wascarried out as reported in the Section 2.2. Values of pH ranging fromca. 4.6 in the fresh Matsoni to ca. 4.0 in the fermented milks after72 h of storage at 6 �C were found, while TTA values ranged from105 �T in fresh samples to 130 �T in cold stored Matsoni.

Basic nutrient contents of Matsoni, its proteins, fat and carbo-hydrates, enabling an estimation of the energy value, were evalu-ated as mean values based on the nutrient content of each of the 16artisanal Matsoni as reported in the Section 2.3. In particular,87.56 � 0.3% water, 0.75 � 0.1% ash content, 4.9 � 0.25% proteins,1.47 � 0.1% fat and 5.32 � 0.05% total carbohydrates (all availablecarbohydrates of which 3.71 � 0.1% lactose, 1.00 � 0.09% glucose,0.61 � 0.15% galactose) were found, on average.

High amounts of lactic acid (8.69 � 0.1 g/L), with a preponder-ance of the isomer L(þ) lactate (5.13 � 0.15 g/L) on the isomer D(�)lactate (3.56 � 0.1 g/L) and 0.009 � 0.03 g/L of acetic acid werefound in the 16 traditional Matsoni analysed.

The mean energy value of the 16 fresh Matsoni samples, calcu-lated based on the mean content of Matsoni basic nutrients as thesumof thenutrient energy values, i.e.: 19.6�0.25kcal/83.3�0.25kJ(proteins) plus 13.2� 0.1 kcal/54.4� 0.1 kJ (fat) plus 21.3�0.05 kcal/90.4 � 0.05 kJ (available carbohydrates), was evaluated in total as54 � 0.25 kcal/228 � 0.20 kJ per 100 g of Matsoni.

Fig. 1. Sensory profiles of four artisanal Matsoni (M1, M2, M3 and M4) on the basis of sensory notes established and evaluated according to QDA and with the use of 10-unit scale.Data are means of three independent experiments. Bars of standard deviations are also represented. On each pillar, different lowercase letters indicate a significant difference(P � 0.05).

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 183

3.2. Sensory profile of traditional Matsoni

The histograms of the texture, odour and taste attributes as wellas the overall SQ values of four Matsoni (M1eM4), randomly cho-sen among the 16 fresh artisanal Matsoni, are shown in Fig. 1. Nosignificant differences were found among most of the sensory pa-rameters of the four Matsoni compared (P � 0.05). The histogramsof texture of Matsoni M1eM4 showed the oral cavity coating as thedominating texture note, followed by smoothness, fluidity andhomogeneity. In the case of odour profile, the notes of the fourMatsoni analysed were generally shown as not significantly

Table 2Distributionwithin each species of the 310 isolates from the 16 traditional Matsoni analyscultures for Matsoni pilot-plant production.

Species Number of isolates withthe same RAPD profile

Lb. delbrueckiisubsp. bulgaricus

9251015106776

1510

Lb. helveticus 37

Lb. paracasei 20Leuc. lactis 6

4S. thermophilus 27

6182017111815126

different (P � 0.05) each other, with the exception of cheese-likeodour, which resulted very low and alike in M2, M3 and M4 andslightly but significantly higher in M1 (P � 0.05). In the case of thetaste histograms, there were no significant differences (P � 0.05)among the notes of the samples, except for the yogurt-like note inM1, which was significantly lower (P� 0.05) than in M4. The highlyassessed notes of the Matsoni taste belonged to yogurt-like, milkyand sour. Cheese-like taste notes of M1eM4 resulted very low,whereas their bitterness was perceived a little higher; the in-tensities of both these latter taste notes did not differ significantly(P � 0.05) among samples.

ed and list of the 26 strains further tested to select the autochthonous multiple strain

Matsoni sample Selected isolate pereach RAPD profile

M1, M14, M15 MY1M5, M8, M9, M10, M11, M13 My60M5, M12, MY15 MY163M6, M11, M15, M16 MY196M2, M14, M15 My197M3, M4, M15 My213M7, M9, M16 UC8059M9, M10, M15 UC8960M9, M12, M15 UC8061M8, M14, M15, M16 UC8096M4, M13, M15 MAT-1AM4, M14 MY203M7, M9, M13 MY208M8, M11, M13 MY165M13 MY177M13 MY178M1, M2, M11, M12 UC8055M1, M7 UC8056M2, M7, M11 UC8057M3, M5, M9 UC8058M3, M5, M9 UC8457M4, M6 UC8461M4, M6, M10 MAT-11M8, M13 UC8052M14, M15 UC8053M16 UC8054

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191184

Overall, within the four artisanal Matsoni analysed, theperception of odour and taste notes such as sour, yogurt-like andmilky was the strongest and not substantially different for the fourfermented milks analysed, except for the significant differences inyogurt-like taste notes of M1 and M4. The overall SQ factors ofMatsoni M4 were significantly lower (P � 0.05) than those ofMatsoni M1 and M3.

3.3. Microbiological and molecular analyses

Sixteen fresh Matsoni (M1eM16) were analysed in order toachieve a deeper understanding of themicrobiota of this traditionalfermented milk and select bacterial strains endowed with physio-logical traits of interest in fermentation milk process. All samplesshowed a high number of LAB, varying between 7 � 0.16 and8 � 0.20 log cfu/mL for cocci-shaped that dominated rod-shapedcommunity, ranging from 5 � 0.027 to 6 � 0.47 log cfu/mL.

A total of 310 colonies of presumptive LAB, randomly picked upfrom MRS and LM17 agar plates seeded with the highest sampledilutions, were clustered by RAPD-PCR and identified by 16S rRNApartial gene sequencing and/or species-specific PCR using theprimers listed in Table 1. Duplicates and a selection of positive(corresponding to the LAB species commonly found in fermentedmilks) and negative controls were used in each species-specific PCRto evaluate the reproducibility and the specificity of the identifi-cation. The GenBank accession numbers for the partial 16S rRNAgene sequences obtained in this study are KF180135eKF180160.

In Table 2, the distribution of the 310 isolates within each LABspecies is summarized and the 26 strains further tested to select theautochthonous multiple strain cultures are listed.

Overall, the microbial ecosystem of Matsoni resulted largelydominated by the thermophilic LAB species S. thermophilus and Lb.delbrueckii subsp. bulgaricus. High density of LAB other than thesetwo species, including Lb. helveticus, Lb. paracasei and Leuconostoc(Leuc.) lactis, were found to occur as well (Table 2).

Table 3Microbial species identification (closest match) after sequencing of the variable V6eV8 refinal products of commercial and artisanal Matsoni production as well as of microbial bioidentity with sequences found in the GenBank database are reported in parenthesis.

Unpasteurized milk Pasteurized milk Natural starter Com

Direct PCR-DGGEStreptococcus parauberis

(AB675654; 99%)S. parauberis(AB675654; 99%)

S. thermophilus(HE793101; 99%)

S. t(HE

Enterococcus faecalis(HF558530; 100%)

E. faecalis(HF558530; 100%)

Lactobacillusdelbrueckii subsp.bulgaricus(JQ754466; 100%)

Lb.bul(JQ

Lactococcus raffinolactis(HF562962; 99%)

L. raffinolactis(HF562962; 99%)

L. raffinolactis(HF562962; 99%)

ND

Lactococcus lactis subsp.lactis (HF562968; 99%)

L. lactis subsp. lactis(HF562968; 99%)

ND ND

Macrococcus caseolyticus(GU197532; 99%)

M. caseolyticus(GU197532; 99%)

ND ND

Carnobacterium spp.(KC311553; 97%)

ND ND ND

PCR-DGGE from bulkNDa ND S. thermophilus

(HE793101; 100%)S. t(HE

ND ND Lb. delbrueckiisubsp. bulgaricus(JQ754466; 100%)

Lb.bul(JQ

ND ND L. lactis subsp. lactis(HF562968; 99%)

ND

ND ND ND ND

ND ND ND ND

a ND, not detected.

In order to achieve a thorough characterization of Matsonimicrobiota, PCR-DGGE analyses of V6eV8 16S rDNA fragmentsobtained, amplifying DNA extracted from the samples shown inTable 3, were carried out. Both unpasteurized and pasteurizedmilk revealed the same diversity at species level, with theexception of Carnobacterium spp. not found in the heat treatedmilk (Table 3). Lactococcus (Lc.) raffinolactis, found by PCR-DGGE inmilk also after the heat treatment, was also retrieved in the arti-sanal starter (Table 3). Furthermore, the PCR-DGGE of bulk (i.e.,the mass of colonies) from the artisanal Matsoni, confirmed thepresence of Lb. paracasei (Table 3) at high loads (ca. 6 log cfu/mL;data not shown). This species was also isolated from other Matsonisamples (Table 2), indicating that it could significantly contributein conferring the typical quality to this fermented milk. The arti-sanal starter revealed a higher biodiversity than the commercialstarter, whereas the microbiota of the fresh commercial Matsonireflected the microbial composition of the commercial starter(Table 3).

3.4. Genotypic and phenotypic characterization of S. thermophilusstrains

The presence of genes encoding for technologically importantproteins was also checked in the S. thermophilus strains, as reportedin the Section 2.6, with the attempt at highlighting some geneticdiversity among them (Mora et al., 2002). No PCR products specificfor prtS gene, coding for proteases involved in casein breakdown(Courtin et al., 2002), were obtained for all S. thermophilus analysed,whereas all strains were positive to the DNA amplification of theureC gene, related to urease enzyme, responsible for urea hydro-lysis (Mora et al., 2005). All S. thermophilus strains also resulted toharbour the following genes: gdh, the key gene of the amino acidmetabolism and aroma formation in dairy products, coding forglutamate dehydrogenase (Tanous et al., 2002), csp, involved in thestrain heat shock stress resistance (Wouters et al., 1999) and epsC-

gion of the 16S rRNA gene purified from PCR-DGGE profiles of raw, intermediate andmass (bulk) grown on countable agar plates. Accession numbers and percentage of

mercial starter Commercial Matsoni Artisanal Matsoni

hermophilus793101; 100%)

S. thermophilus(HE793101; 100%)

S. thermophilus(HE793101; 100%)

delbrueckii subsp.garicus754466; 100%)

Lb. delbrueckii subsp.bulgaricus (JQ754466; 100%)

Lb. delbrueckii subsp.bulgaricus(JQ754466; 100%)

ND L. raffinolactis(HF562962; 99%)

ND ND

ND ND

ND ND

hermophilus793101; 100%)

S. thermophilus(HE793101; 100%)

S. thermophilus(HE793101; 100%)

delbrueckii subsp.garicus754466; 100%)

Lb. delbrueckii subsp.bulgaricus (JQ754466; 100%)

Lb. delbrueckii subsp.bulgaricus(JQ754466; 100%)

ND L. lactis subsp. lactis(HF562968; 99%)

ND Staphylococcus pasteuri(JX517276; 100%)

ND Lb. paracasei(KC137275; 100%)

Fig. 2. Acidification activity of lactic acid bacteria monocultures (A) and multiple cultures (B) during milk fermentation at 42 �C for 6 h and storage at 6 �C for four weeks. Data arethe means of three independent experiments. Bars of standard deviations are also represented. The best performing combinations are marked with an asterisk.

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 185

D, related to the exopolysaccharide production (Stingele et al.,1996). In addition, all S. thermophilus strains, as assayed on RRMplates, resulted positive for exopolysaccharide production, wellknown to positively affect viscosity and texture of fermented milks(Shene et al., 2008).

3.5. Acidification activity of LAB strain cultures

Preliminary tests (data not shown) on the acidification activityof monocultures of the 26 biotypes listed in Table 2 allowed usindividuating 12 potential starter strains.

In Fig. 2A, the acidification activity of the 12 selected strainsduring six h of fermentation in skim milk at 42 �C and storage at6 �C for four weeks is shown. A different behaviour was observedfor the cocci and rod-shaped LAB tested. In particular, theS. thermophilus strains showed a very weak acidification duringfermentation with a final pH value of ca. 5.6, on average; a furtheracidification, lowering the pH values between ca. 4.8 and 4.3, wasfound at the end of the cold storage (Fig. 2A).

Conversely, the seven lactobacilli strains revealed a high di-versity in the ability to lower pH values during four weeks of coldstorage. In particular, Lb. paracasei My165 and Lb. helveticus My203proved to be the most acidifying strains since they did drop the pH

Fig. 3. Changes in presumptive lactic acid bacteria loads in both Matsoni samples A and B fresh (T0) and after one week (T7), two weeks (T15) and one month (T30) of storage at6 �C. Data are the means of three independent experiments. Bars of standard deviations are also represented.

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191186

value from ca. 4.5 and 5.0, respectively at 6 h of fermentation to ca.3.8 at the end of the cold storage. A different behaviour at strainlevel was found within the Lb. delbrueckii subsp. bulgaricus species;indeed, their pH values varied from ca. 4.36 to ca. 6.0 at the end ofthe fermentation, and from ca. 3.8 to ca. 5.6 after four weeks ofstorage at 6 �C (Fig. 2A).

Subsequently, the 12 autochthonous strains were tested in allthe possible combinations as multiple strain cultures for the acid-ification activity during Matsoni fermentation and cold storage. InFig. 2B, the best performing combinations are shown. Each com-bination of strains showed a particular behaviour with differentacidification curves. Overall, after the first six h of fermentation at42 �C, pH values varied between ca. 4.8 and 5.7 and at the end of thestorage period, ranged from ca. 4.1 to ca. 4.8 (Fig. 2B). The bestcombinations as possible candidates to be further assayed in apilot-plant Matsoni production were selected based on their fastacidification ability within the first six h of fermentation and theirlow post-acidification activity during the cold storage period; theywere chosen among those marked with an asterisk in the legend ofFig. 2B.

3.6. Pilot-scale Matsoni production with autochthonous multiplestarter cultures

As reported in the Section 2.8, two Matsoni, named A and B,were produced with a pilot-plant, using two autochthonous mul-tiple strain cultures (AMSCs), named AMSC1 and AMSC2. Theywere chosen among the multiple strain cultures marked with anasterisk in the legend of Fig. 2B and include strains named here asS. thermophilus T, Lb. delbrueckii subsp. bulgaricus X, Lb. paracasei Yand Lb. helveticus Z. Since the research activities reported in thepresent paper have been carried out within the Sixth FrameworkProgramme (Horizontal Research Activities Involving SMEs-Co-operative Research Project FERBEV - contract N� 031918), thenames of the selected strains assayed for the pilot-plant Matsoniproduction are not specified here being the exclusive property ofthe SMEs involved in the project. Matsoni A was produced usingAMSC1 composed of S. thermophilus T, Lb. delbrueckii subsp. bul-garicus X and Lb. paracasei Y, whereas Matsoni B was manufactured

with AMSC2, including S. thermophilus T, Lb. paracasei Y and Lb.helveticus Z.

The pH values of both fresh Matsoni A and B were 4.38 and 4.40,respectively; they remained quite stable in the first 3 days of coldstorage, decreasing to ca. 4.32 and 4.22 after 10 days and to ca. 4.26and 4.18 after one month of cold storage.

Protein, fat, dry matter and moisture contents of both Matsoni Aand B resulted similar to those of the artisanal Matsoni included ascontrol (data not shown). As concerns the lactic acid content, alower amount was found in Matsoni A than in Matsoni B (7.21 �0.1and 8.74 � 0.08 g/L, respectively), with a preponderance of theisomer L(þ) (5.36 � 0.1 and 6.09 � 0.09 g/L in Matsoni A and B,respectively) on the D(�) lactate (1.85 � 0.1 and 2.65 � 0.09 g/L, inMatsoni A and B, respectively), whereas acetic acid was not foundin both Matsoni A and B.

High loads of presumptive thermophilic cocci-shaped LAB(9.14� 0.09 log cfu/mL and 8.65� 0.03 log cfu/mL inMatsoni A andB, respectively) were found after one week of refrigeration storage(Fig. 3). These loads decreased of ca. 1 log cfu/mL in both Matsoni Aand B during the remaining storage period (Fig. 3). This samebehaviour was found for both presumptive mesophilic and ther-mophilic lactobacilli inMatsoni A, whereas in Matsoni B mesophiliclactobacilli increased of ca. 1.5 log cfu/mL after two weeks of stor-age, remaining stable up to the end of the refrigeration period(Fig. 3). Yeasts and moulds as well as Salmonella spp., L. mono-cytogenes, Enterobacteriaceae and coagulase positive staphylococciwere undetectable in all Matsoni analysed, as ascertained by theISO protocols reported in the Section 2.5.

To monitor the strains composing AMSC1 and AMSC2, 180 col-onies were picked up from the LM17 and MRS agar plates seededwith the highest sample dilutions and used to enumerate rod- andcocci-shaped LAB in Matsoni A and B, fresh (T0) and after one week(T7), two weeks (T15) and one month (T30) at 6 �C (30 isolates pereach sample; Fig. 4). They were isolated and subjected to DNAextraction, followed by species-specific PCR and RAPD-PCR, as re-ported in the Sections 2.5 and 2.8. In Matsoni A, the RAPD-PCRprofiles corresponding to S. thermophilus T, Lb. delbrueckii subsp.bulgaricus X and Lb. paracasei Ywere all detected at time 0, one, twoweeks and onemonth of storage at 6 �C. In particular, Lb. paracasei Y

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 187

load increased of ca. 1 log during the first week of cold storage,decreasing to ca. 8 log cfu/mL after two weeks and remaining quitestable until one month. Lb. delbrueckii subsp. bulgaricus X remainedquite stable during thewhole storage period (Fig. 4). S. thermophilusT load increased of 1 log (ca. 9 log cfu/mL) after one week, while itpersisted at a load of ca. 8 log cfu/mL until one month of coldstorage (Fig. 4). Similarly, S. thermophilus T slightly increased duringthe first week of storage in Matsoni B, remaining quite stable up tothe end of the cold storage (Fig. 4). On the other hand, Lb. paracaseiY persisted at ca. 7.5 log cfu/mL for one week, increasing of ca. 2 logafter two weeks and remained stable up to the end of the coldstorage. Starting from a load of ca. 7 log cfu/mL, Lb. helveticus Zslightly decreased after one week and increased again of ca. one logat two weeks and slightly increased up to one month (Fig. 4).

3.7. Sensory evaluation of Matsoni produced at pilot-scale withAMSCs

Sensory profiles of both Matsoni A and B were analysed asdescribed in the Section 2.4. Texture, taste and odour notes, influ-encing sensory profiles of Matsoni A and B, are shown in Fig. 5.

Almost no differences were observed among the most impor-tant texture notes of Matsoni A and B. Decreasing values werefound in both fermented milks for the following notes: smoothness(A and B: 8.88 points), homogeneity (A and B: 8.60 points), fluidity(A: 8.14 points; B: 7.40 points) and oral cavity coating (A: 5.02points; B: 5.58 points). However, the differences among Matsoni Aand B within the same descriptor notes were not statistically sig-nificant (P� 0.1). In the case of odour, only the sour and yogurt-likenotes were perceived as a little higher in Matsoni A than in MatsoniB. However, the differences of the appropriate note values amongthe samples were not statistically significant (P � 0.1). A significantdifference (P � 0.1) among the taste notes of Matsoni A and B wasperceived exclusively in the case of sourness; Matsoni A totalized5.80 points, differently from Matsoni B with 8.04 points. Otherquite comparable taste notes in both Matsoni were: high yogurt-like note (A: 5.88 points; B: 6.42 points), moderate cheese-like

Fig. 4. Changes in autochthonous multiple starter cultures in both fresh (T0) Matsoni A (A(AMSC2, S. thermophilus T, Lb. paracasei Y and Lb. helveticus Z) and after one week (T7), twindependent experiments. Bars of standard deviations are also represented.

note (A: 2.26 points; B: 2.74 points) and remnant milky note (A:0.48 points; B: 0.32 points) (Fig. 5).

The hedonic assessment of the sensory quality of both freshMatsoni A and B and stored for one month at 6 �C was also carriedout as reported in the Section 2.4. Both overall SQ factors for freshand cold-stored Matsoni were evaluated as higher for Matsoni Athan for Matsoni B. Fresh Matsoni A revealed a better quality inhedonic assessment (8.38 points) than freshMatsoni B (7.12 points)and this difference was statistically significant (P � 0.1); the sig-nificant difference (P � 0.1) of SQ factors among Matsoni A and Bwas also found at day 30 of cold storage (Matsoni A: 8.06 points;Matsoni B: 6.86 points).

BothMatsoni A and Bmaintained their high SQ up to onemonthof cold storage, with better quality in hedonic assessment forMatsoni A.

Thereafter, a comparison between Matsoni A and four artisanalMatsoni was carried out in order to ascertain the maintenance oftypicality and overall SQ of the newly Matsoni A. As shown inTable 4, the overall sensory quality score of Matsoni A was notsignificantly different from that of the artisanal Matsoni in spite ofdifferences in the intensities of some notes.

In particular, as concerns texture, important texture notes suchas homogeneity, fluidity and smoothness resulted significantlyhigher (P � 0.05) in Matsoni A than in the artisanal Matsoni. Asconcerns the odour attribute, the sour and the yogurt-like noteswere found not significantly different, whereas the milky andcheese-like notes were perceived slightly divergent in the artisanalthan in the newly fermented milk. For the taste attribute, it isnoteworthy that the bitter note was perceived significantly(P� 0.05) lower inMatsoni A than in the artisanalMatsoni (Table 4).

4. Discussion

Fermented milks are among the most ancient fermented prod-ucts dating back to ca. 6000 years BC in several locations all over theworld (Hui et al., 2012; Kurmann et al., 1992). Fermentation wasoriginally a mean to preserve this highly perishable product, then,

MSC1, S. thermophilus T, Lb. delbrueckii subsp. bulgaricus X and Lb. paracasei Y) and Bo weeks (T15) and one month (T30) of storage at 6 �C. Data are the means of three

Fig. 5. Results of texture, taste and odour notes of fresh Matsoni A and B.

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191188

(a mean) to improve its nutritional value, organoleptic quality, foodsafety and shelf-life (Caplice and Fitzgerald, 1999).

Matsoni, as described in the Encyclopedia of fermented freshmilk products (Kurmann et al., 1992), is a fermented milk with afirm consistency and an acidic flavour.

Although Matsoni is being produced since ancient times, itsquality and stability still relies on back slopping and spontaneousuncontrolled fermentations, with inherent hygienic, nutritionaland organoleptic challenges. Thus, a thorough physico-chemical,sensorial and microbial characterization of traditional Matsoni iscrucial to pinpoint its typical traits as well as to select and exploitautochthonous starter cultures for standardizing and improving theoverall quality without altering the typicality of this fermentedmilk. To the best of our knowledge, a comprehensive character-ization of Matsoni beverage is not available in literature.

Herein, a physico-chemical, nutritional, microbiological andsensory characterization of traditional Matsoni was carried out.

Table 4Analytical sensory evaluation (Quantitative Descriptive Analysis)a of artisanal Mat-soni and Matsoni A (mean � SD).

Sensoryattributes/notes

Samples analysed Significanceof differences

TraditionalMatsonib (n ¼ 4)

MatsoniAc (n ¼ 2)

TextureHomogeneity 2.6 � 0.8 8.6 � 0.6 *Fluidity 2.9 � 1.0 8.1 � 0.9 *Smoothness 3.6 � 1.6 8.9 � 0.3 *Oral cavity coating 4.9 � 0.6 5.0 � 0.9 n.s.OdourSour 6.3 � 1.8 4.6 � 1.5 n.s.Yogurt like 5.8 � 1.8 6.3 � 1.6 n.s.Milky 3.5 � 1.0 1.0 � 0.3 *Cheese like 1.4 � 0.7 2.2 � 1.2 *TasteSour 6.4 � 1.1 5.8 � 1.6 n.s.Yogurt like 6.9 � 0.8 5.9 � 2.3 n.s.Bitter 1.0 � 0.7 0.0 � 0.0 *Milky 6.9 � 1.4 0.5 � 0.2 *Cheese like 0.4 � 0.2 2.2 � 1.3 *Overall sensory

quality8.1 � 0.2 8.4 � 1.0 n.s.

*P � 0.05.n.s. not significant.

a The evaluation of the intensities of sensory notes (descriptors) was done using a10 unit-scale.

b Each value was the mean of 48 results (4 samples, 6 panelists, 2 repetitions).c Each value was a mean of 24 results (2 samples, 6 panelist, 2 repetitions).

Physico-chemical composition and nutritional profile of theartisanal Matsoni analysed mainly reflected the composition of themilk from which they were made but with a lower content oflactose, which being used by LAB as primary carbon source for thefermentation process, allowed the formation of a considerableamount of lactic acid and glucose as well as of acetic acid andgalactose traces. These results are in accordance with those re-ported in literature for other fermented milks (Chandan, 2008;Gambelli et al., 1999; Manzi et al., 2007). Moreover, the culture-dependent and -independent microbial characterization of Mat-soni microbiota confirmed the pivotal role played by LAB in thefermentation process. Indeed, high loads of both cocci- and rod-shaped LAB (on average ca. 7.5 and 5.5 log cfu/mL, respectively)were found in the 16 Matsoni analysed. Overall, streptococciseemed to dominate lactobacilli probably due to their higher sur-vival at low pH and cold storage, as previously reported (Tamimeet al., 1999). Different microbial compositions were found in theartisanal Matsoni analysed, probably due to the back-sloppingpractise usually adopted at house-hold level. As previously sug-gested (Limsowtin et al., 1996; El-Baradei et al., 2008), repeatedsub-culturing or back-slopping from the previous day productioncould result in changes in LAB species and strains proportions andloss of viability and activity of some microbial cultures. Overall, themost common LAB dominating the stable ecosystem of Matsoniresulted to be the thermophilic species of S. thermophilus and Lb.delbrueckii subsp. bulgaricus. This result is in accordance withprevious studies (Erzinkjan, 1971; Merabishvili and Chanishvili,2001; Uchida et al., 2007) and permits to include Matsoni withinthe thermophilic fermented milks group on the basis of the clas-sification of fermented products recently provided by Shiby andMishra (2013). High loads of LAB other than these two typicalyogurt-LAB species (Xanthopoulos et al., 2001), namely Lb. helve-ticus, Lb. paracasei and Leuc. lactis, were found by PCR basedculture-dependent approach.

The presence of spoilage bacteria such as Streptococcus para-uberis, Enterococcus faecalis,Macrococcus caseolyticus, L. raffinolactisand Carnobacterium spp. in the raw milk was also revealed byculture-independent approach. Such alterative microorganismshave been previously found in other traditional fermented milksproduced elsewhere (Abdelgadir et al., 2001; Blandino et al., 2003;Mutukumira, 1996). Very probably spoilage bacteria tend to hidebeneficial LAB that might anyway prevail during milk fermentationand probably contribute to the fast spoilage of the fresh fermentedmilks. Both unpasteurized and pasteurized milk revealed the samediversity at species level, with the exception of Carnobacterium spp.not found in the heat treated milk. L. raffinolactis, found by culture-

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 189

independent PCR-DGGE in milk also after the heat treatment, wasalso retrieved in the artisanal starter. Since no colonies grew on therelevant LM17 and MRS agar plates used to count LAB, this micro-bial population has probably died or was in a viable but unculti-vable state. Indeed, DNA from dead, injured and viable butuncultivable cells may be still amplified by PCR (Fusco et al., 2012;Fusco and Quero, 2012). As expected, the natural starter revealed ahigher biodiversity than the commercial starter, whereas themicrobiota of the fresh product (commercial Matsoni) reflected themicrobial composition of the commercial starter. Lb. helveticus andLeuc. lactis were not found by V6eV8 PCR-DGGE analysis, eventhough they were isolated by the conventional culture-dependentapproach from the same Matsoni samples. The unsuccessfulattempt to find these species by PCR-DGGE could be due to knownbiases of this method such as the DNA yield and quality, possibledifferential or preferential amplification of rDNA genes, formationof chimeric or heteroduplex molecules, etc. (Ercolini, 2004). Vice-versa, we could not find, among the randomly isolated colonies,any Lactococcus lactis strain, even though its presence was revealedby the PCR-DGGE analysis of both DNA directly isolated fromsamples and DNA from the bulk cells. However, L. lactis resulted tobe the dominant species isolated from Matsoni on LM17 agarincubated at 15 �C for 48 h (data not shown). From the methodo-logical point of view, these results highlight the importance of us-ing both culture-dependent and independent methods in the studyof the microbial ecology of fermented foods.

The polyphasic approach used for investigating the microbiotainvolved in Matsoni production allowed us revealing a discretenumber of species and biotypes probably related to differentmetabolic traits and which could lead to the appreciated flavourand taste of this traditional fermented milk.

The Quantitative Descriptive Analysis (QDA), performed in thisstudy to set up the sensory descriptors of traditional Matsoni andused for its sensory profiling, resulted in the list of the odour notesof this beverage as milky, yogurt, sour and cheese as well as thetaste notes as cheese, bitter, sour and yogurt. These sensory de-scriptors for notes have been used by other authors for the char-acterization of fermented milks and yogurt. Indeed, the descriptorsused by Ott et al. (2000) for odour, i.e. milky, yogurt, sour, cottagecheese and for taste, i.e. cottage cheese, bitter, sour and yogurt,were the same proposed and agreed for Matsoni sensory profile byour sensory panel. As concerns the texture descriptors, we usedhomogenous, mouthcoating (oral cavity coating) and smoothness,which are the same reported by Majchrzak et al. (2010), whocombined a comprehensive list of descriptors for the sensory ana-lyses of conventional and probiotic yogurts.

The four Matsoni analysed, as evaluated by the sensory panel,with the exception of M4, whose overall SQ was found significantly(P � 0.05) lower than that of M1 and M3, were close each other forthe overall sensory quality and mainly characterized by the domi-nant notes of sour and yogurt-like of taste and odour. Nevertheless,the intensity of such notes was found significantly different in someMatsoni analysed, highlighting once more the need for standard-izing the fermentation process by applying selected autochthonousstarter cultures.

The initial fast acidification is essential for the production offermented milks, while the post-acidification, occurring during thestorage of these products, leads to several undesired effects such asstrong acidic taste, increase inwhey separation and decrease of LABcount (Abu-Jdayil and Mohameed, 2002; Donkor et al., 2006). Theselection of LAB strains with a weaker post-acidification during thecold storage could contribute to solve this problem (Leroy and DeVuyst, 2004). Thereafter, in order to improve both the sensoryprofile and the shelf-life of traditional Matsoni, we screened,among the dominant LAB isolated from this traditional fermented

milk, some biotypes for a fast acidification activity during the first6 h of fermentation and a low-acidification activity during cold-storage.

As concerns the acidification activity in milk driven by LABmonocultures, a different behaviour was found at species and strainlevel among the autochthonous strains assayed. As expected, allS. thermophilus strains showed a weak acidifying activity, whereasLb. paracasei and Lb. helveticus strains proved to be able to drop pHvalue at 4.5 and 5.0, respectively within six h of fermentation at42 �C. These latter strains showed a weak post-acidification up tofour weeks at 6 �C. The acidification activity variedmarkedly withinthe Lb. delbrueckii subsp. bulgaricus species with pH values rangingfrom 3.8 to 5.6 after four weeks of cold storage. Such differentbehaviour at species and strain level could be mainly due to thespecific attitude of strains in assimilating the nutrient compoundsavailable in the medium (Badis et al., 2004). As compared tomonocultures, the multiple strain cultures resulted in most cases inhigher acidification activity probably due to the symbiotic rela-tionship among the strains. Such mutual benefit is well knownbetween S. thermophilus and Lb. bulgaricus and is referred to asproto-cooperation (Tamime and Robinson, 1999; Herve-Jimenezet al., 2009).

In this study, some strains belonging to the dominant speciesoccurring in the artisanal Matsoni microbiologically characterized,i.e. S. thermophilus, Lb. delbrueckii subsp. bulgaricus, Lb. paracaseiand Lb. helveticus, revealed the best properties, mainly in terms offast acidification during fermentation and low post-acidificationduring cold storage, when used in multiple combinations asautochthonous starter cultures to standardize and improve theoverall quality and shelf-life of traditional Matsoni.

Based on the results obtained at lab-scale, the two best per-forming AMSCs were scaled up in pilot-plant Matsoni productions.A lower post-acidification was found in both Matsoni producedwith the two AMSCs in comparison with the artisanal Matsoni,resulting in a shelf-life of one month instead of 72e120 h f theartisanal Matsoni, confirming that the weaker post-acidificationactivity is a pivotal selection criterion to be considered for devel-oping an autochthonous starter culture. Lactose level in fermentedmilks is lower than in the milk from which they derive due to theLAB fermentation (Shah, 2008). The amount of lactose retrieved inthe artisanal and newly Matsoni A and B was similar to that re-ported in commercial yogurts found able to alleviate or preventlactose a poor digestion (European Food Safety Authority (EFSA)Panel on Dietetic Products, Nutrition and Allergies, 2010), a meta-bolic disorder affecting approximately 75% of the world’s adultpopulation (Swaggerty et al., 2002). According to the EFSA Panel onDietetic Products, Nutrition and Allergies (2010), in order to bearthe claim (improved lactose digestion), the yogurt should contain atleast 108 cfu live starter microorganisms (Lb. delbrueckii subsp.bulgaricus and S. thermophilus) per gram. Considering that MatsoniA and B contained at least 8 log cfu/g of LAB also during one monthof shelf-life, it can be hypothesized that they may improve thelactose digestibility.

Lactic acid (2-hydroxy propionic acid) is themain product of LABfermentation. Depending onwhich stereo-specific NAD-dependentlactate dehydrogenases (LDHs) LAB hold, they may give rise, duringthe fermentation process, to one or two enantiomers (i.e. opticallyactive isomeric forms) of lactate, D(�)- and L(þ)-lactate (Chandan,2008). The exploitation of the two AMSCs for Matsoni productionresulted in a higher percentage of L(þ) lactic acid produced inMatsoni A and B (74.3 and 69.7% of the total lactic acid produced inthe first and the latter, respectively) than in the artisanal Matsonianalysed (59% of L(þ) lactate). This finding is probably due to thedifferent microbial loads and composition of the AMSCs involved inthe fermentation process of the two fermented milks. In particular,

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191190

Matsoni A was produced with AMSC1, composed of strainsbelonging to one D(�) lactic acid producing species (Lb. delbrueckiisubsp. bulgaricus) and two L(þ) lactic acid producing species(S. thermophilus and Lb. paracasei) (Connolly and Lönnerdal, 2004),whereas Matsoni B was produced with AMSC2, including twostrains belonging the two L(þ) lactic acid producing species(S. thermophilus and Lb. paracasei) and one strain belonging to aspecies known to produce the racemate DL-lactic acid (Lb. helveti-cus) (Connolly and Lönnerdal, 2004). In Matsoni A the loads ofAMSC1 strains remained quite stable at ca. 8 log cfu/mL up to theend of the cold storage. By contrast, different microbial populationcounts were found in Matsoni B, where the strain belonging to theL(þ) lactic acid producing species (Lb. paracasei), starting from amicrobial load lower of ca. one log than in Matsoni A, reached ca.the same load retrieved in Matsoni A only after two weeks of coldstorage. Moreover, the racemate DL-lactic acid producing Lb. helve-ticus, starting from ca. 7.5 log cfu/mL, decreased of ca. one log afterone week of cold storage to reach ca. 8 log cfu/mL only at the end ofthe storage period.

Both enantiomers have been found to improve the casein di-gestibility and aid in retention of calcium in the intestine (Shah,2008), however the L(þ) isomer is physiologically more digestiblethan the D(�) isomer, so that the use of the latter in formula forinfants and young children is controversial (WHO, 1974; Shah,2008; Connolly and Lönnerdal, 2004). Thus, the use of AMSCsmay result in an improved digestibility of Matsoni and may haveconsiderable implications in the use of this fermented milk forinfant feeding.

As concerns the sensory profile, the exploitation of the AMSCsallowed us obtaining a fermented milk with an overall sensoryquality not significantly different from that of the artisanal Matsoni,but with a bitter note of taste significantly lower than in the arti-sanal Matsoni. Moreover, Matsoni fermented with the two AMSCsresulted in an improved texture with homogeneity, fluidity andsmoothness notes significantly higher than those of the artisanalMatsoni. This result is most likely due to the EPS production by LAB,well known to positively affect the viscosity and texture of fer-mented milks (Shene et al., 2008). The in situ production of suchnatural texture-improving polymers could be most likely due toautochthonous S. thermophilus T strain, which was herein found toharbour and express the eps gene, as assayed by the protocolsdescribed by Stingele et al. (1996). Thus, both Matsoni A and Bmaintained high overall SQ factors during one month of storage,probably due to the low post-acidification activity of the AMSCsused to produce the fermented milks. However, Matsoni A revealeda significantly better quality in hedonic assessment than Matsoni Bthat was also found in the fermented milk stored for one month at6 �C.

Thereby, among the two AMSCs assayed for the pilot-plantproduction of traditional Matsoni, AMSC1, composed ofS. thermophilus T, Lb. delbrueckii subsp. bulgaricus X and Lb. para-casei Y, resulted the best in terms of low post-acidification and highsensory quality, given the enhanced shelf-life from the 72e120 h(5e6 days) of the artisanal Matsoni to one month.

For this reason, we propose the use of this AMSC to satisfy in-dustrial requirements of Matsoni manufacturers who need toexploit and standardize the overall quality of this traditional fer-mented milk, preserving its typicality.

In conclusion, the thorough microbial, physico-chemical,nutritional and sensory characterization of the traditional Mat-soni herein performed allowed individuating the typical traits ofthis fermented milk and selecting an AMSC, which was successfullyused to improve and standardize its overall typical quality. More-over, the addition of such AMSC allowed extending the shelf-life toone month instead of 72e120 h of the artisanal Matsoni. This could

make it feasible the distribution of this fermented milk at longerdistances from the production farmhouses, allowing the growth ofSMEs through the expansion of their markets and competitiveness.

Acknowledgements

This workwas supported by the Sixth Framework Programme ofthe European Commission (Horizontal Research ActivitiesInvolving SMEs Cooperative Research Project FERBEV contract N�

031918, www.ferbev.net). We thank AMALTEA DIDUBE Milk (Tbi-lisi, Georgia) and AYGIN SÜT (Konya, Turkey) SMEs’ entrepreneursfor their cooperation in manufacturing and delivering beverages.

References

Abdelgadir,W.S., Hamad, S.H., Moller, P.L., Jakobsen, M., 2001. Characterization of thedominant microbiota of Sudanese fermented milk Rob. Int. Dairy J. 11, 63e70.

Abu-Jdayil, B., Mohameed, H., 2002. Experimental and modelling studies of the flowproperties of concentrated yogurt as affected by the storage time. J. Food Eng.52, 359e365.

Altschul, S.F., Madden, T.L., Schäffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J.,1997. Gapped BLAST and PSI-BLAST: a new generation of protein databasesearch programs. Nucleic Acids Res. 25, 3389e3402.

American Public Health Association, 1992. Standard Methods for the Examination ofDairy Products, 16th ed. APHA Inc., Washington DC.

Ampe, F., Ben Omar, N., Moizan, C., Wacher, C., Guyot, J.P., 1999. Polyphasic study ofthe spatial distribution of microorganisms in Mexican pozol, a fermented maizedough, demonstrates the need for cultivation-independent methods to inves-tigate traditional fermentations. Appl. Environ. Microbiol. 65, 5464e5473.

Anonymous, 1987. Sensory Analysis e Methodology e Evaluation of Food Productsby Methods Using Scales (ISO 4121: 1987). International Organization forStandardization, Geneva.

Anonymous,1988. Sensory AnalysiseGeneral Guidance for the Design of Test Rooms(ISO 8589: 1988). International Organization for Standardization, Geneva.

Anonymous, 1993. Sensory Analysis e General Guidance for the Selection, Trainingand Monitoring of Assessors e Part 1: Selected Assessors (ISO 8586-1:1993).International Organization for Standardization, Geneva.

Anonymous, 1996. Microbiology of Food and Animal Feeding Stuffs e HorizontalMethod for the detection and enumeration of Listeria monocytogenes e Part 1:Detection method. ISO 11290-1:1996. International Organization for Stan-dardization, Geneva.

Anonymous, 2001. Milk and Milk Products e Detection of Salmonella spp. (ISO6785:2001). International Organization for Standardization, Geneva.

Anonymous, 2004a. Microbiology of Food and Animal Feeding StuffsdHorizontalMethod for the Detection and Enumeration of Listeriamonocytogenes. ISO 11290-1: 1996/Amd.1:2004(E). International Organization for Standardization, Geneva.

Anonymous, 2004b. Modification of the Isolation Media and the Haemolysis Test,and Inclusion of Precision Data. ISO 11290-1:1996/Amd 1:2004. InternationalOrganization for Standardization, Geneva.

Association of Official Analytical Chemists,1998. OfficialMethods of Analysis. Proteinin Beer. AOAC official method 920.53. First Action 1920. Final Action 1998.

Association of Official Analytical Chemists, 1999. Official Methods of Analysis. Solids(Total) in Milk. AOAC official method 925.23. First Action 1925. Repealed Action1999.

Association of Official Analytical Chemists, 2000. Acidity of milk. Tritimetricmethod. AOAC official method 947.05. In: Horwitz, W. (Ed.), Official Methods ofAnalysis of AOAC International, p. 7. Gaithesburg, MD, USA (Chapter 33).

Badis, A., Guetarni, D., Mouss Boudjemaa, B., Henni, D.E., Kihal, M., 2004. Identifi-cation and technological properties of lactic acid bacteria isolated from raw goatmilk of four Algerian races. Food Microbiol. 21, 579e588.

Blandino, A., Al-Aseeri, M.E., Pandiella, S.S., Cantero, D., Webb, C., 2003. Cereal basedfermented foods and beverages. Food Res. Int. 36, 527e534.

Caplice, E., Fitzgerald, G.F., 1999. Food fermentations: role of microorganisms infood production and preservation. Int. J. Food Microbiol. 50, 131e149.

Chandan, R.C., 2008. In: Chandan, R.C. (Ed.), Manufacturing Yogurt and FermentedMilks. Blackwell Publishing.

Chanishvili, N., Merabishvili, M., Porchkhidze, K., Natroshvili, G., 2001. The Cauca-sian product Matsoni: its bacterial composition and properties. Bull. GeorgianAcad. Sci. 163, 302e305.

Cocconcelli, P.S., Porro, D., Galandini, S., Senini, L., 1995. Development of RAPDprotocol for typing of strains of lactic proteolysis, and functional properties.Lett. Appl. Microbiol. 21, 376e760.

Connolly, E., Lönnerdal, B., 2004. D(�) e lactic acid producing bacteria: safe to usein infant formulas. NUTRAfoods 3, 37e49.

Courtin, P., Monnet, V., Rul, F., 2002. Cell-wall proteinases PrtS and PrtB have adifferent role in Streptococcus thermophilus/Lactobacillus bulgaricus mixed cul-tures in milk. Microbiology 148, 3413e3421.

de Man, J.C., Rogosa, M., Sharpe, M.E., 1960. A medium for the cultivation of lac-tobacilli. J. Appl. Microbiol. 23 (1), 130e135.

G.M. Quero et al. / Food Microbiology 38 (2014) 179e191 191

Di Cello, F.P., Bevivino, A., Chiarini, L., Fani, R., Paffetti, D., Tabacchioni, S.,Dalmastri, C., 1997. Biodiversity of a Burkholderia cepacia population isolatedfrom maize rhizosphere at different plant growth stages. Appl. Environ.Microbiol. 63, 4485e4493.

Donkor, O.N., Henriksson, A., Vasiljevic, T., Shah, N.P., 2006. Effect of acidification onthe activity of probiotics in yoghurt during cold storage. Int. Dairy J. 16, 1181e1189.

Drake, M., Small, C.L., Spence, K.D., Swanson, B.G., 1996. Rapid detection andidentification of Lactobacillus spp. in dairy products by using the polymerasechain reaction. J. Food Prot. 59, 1031e1036.

El-Baradei, G., Delacroix-Buchet, A., Ogier, J.C., 2008. Bacterial biodiversity oftraditional Zabaday fermented milk. Int. J. Food Microbiol., 295e301.

Ercolini, D., 2004. PCR-DGGE fingerprinting: novel strategies for detection of mi-crobes in food. J. Microbiol. Methods 56, 297e314.

Erzinkjan, L.A., 1971. Biological Peculiarities of Some Races of Lactic Acid Bacteria.Armenian Academy of Sciences, Yerevan.

European Communities, 1991. Regulation No. 2568, L 248, 1991. European Com-munity’s Official Gazzette, Brussels, Belgium.

European Communities, 1992. Regulation No 1429/92, L 150, May 26, 1992. Off. J.Comm. Eur. Comm..

European Food Safety Authority (EFSA) Panel on Dietetic Products, Nutrition andAllergies, 2010. Scientific opinion on the substantiation of health claims relatedto live yoghurt cultures and improved lactose digestion (ID 1143, 2976) pur-suant to article 13(1) of regulation (EC) no 924/2006. EFSA J. 8, 1763.

Fortina, M.G., Ricci, G., Mora, D., Parini, C., Manachini, P.L., 2001. Specific identifi-cation of Lactobacillus helveticus by PCR with pepC, pepN and htrA targetedprimers. FEMS Microbiol. Lett. 198, 85e89.

Fusco, V., Quero, G.M., Morea, M., Blaiotta, G., Visconti, A., 2011a. Rapid and reliableidentification of Staphylococcus aureus harbouring the enterotoxin gene cluster(egc) and quantitative detection in raw milk by real time PCR. Int. J. FoodMicrobiol. 144, 528e537.

Fusco, V., Quero, G.M., 2012. Nucleic acid-based methods to identify, detect andtype pathogenic bacteria occurring in milk and dairy products. In: AmerEissa, Ayman (Ed.), Structure and Function of Food Engineering. InTech,pp. 371e404.

Fusco, V., Quero, G.M., Morea, M., Stea, G., Visconti, A., 2011b. Novel PCR-basedidentification of Weissella confusa using an AFLP-derived marker. Int. J. FoodMicrobiol. 145, 437e443.

Fusco, V., Riccardi, M., Quero, G.M., 2012. Thin agar layer- versus most probablenumber-PCR to enumerate viable and stressed Escherichia coli O157:H7 andapplication in a traditional raw milk pasta filata cheese. Int. J. Food Microbiol.159, 1e8.

Gambelli, L., Manzi, P., Panfili, G., Vivanti, V., Pizzoferrato, L., 1999. Constituents ofnutritional relevance in fermented milk products commercialised in Italy. FoodChem. 66, 353e358.

Herve-Jimenez, L., Guillouard, I., Guedon, E., Boudebbouze, S., Hols, P., Monnet, V.,Maguin, E., Rul, F., 2009. Postgenomic analysis of Streptococcus thermophiluscocultivated in milk with Lactobacillus delbrueckii subsp. bulgaricus: involve-ment of nitrogen, purine, and iron metabolism. Appl. Environ. Microbiol. 75,2062e2073.

Hui, Y.H., Evranuz, Ö., Chandau, R.C., 2012. Handbook of Animal-based FermentedFood and Beverage Technology, second ed. CRC Press, p. 798.

Kunachowicz, H., Nadolna, I., Przygoda, B., Iwanow, K., 1998. Tabele warto�sciod _zywczej produktów spo _zywczych (Food Composition Tables). National Foodand Nutrition Institute, Warsaw.

Kurmann, J.A., Lj Ra�si�ceremija, J., Manfred, K., 1992. Encyclopedia of FermentedFresh Milk Products: an International Inventory of Fermented Milk, Cream,Buttermilk, Whey, and Related Products. Springer, p. 212.

Lee, H., Park, S., Kim, J., 2000. Multiplex PCR-based detection and identification ofLeuconostoc species. FEMS Microbiol. Lett. 193, 243e247.

Leroy, F., De Vuyst, L., 2004. Lactic acid bacteria as functional starter cultures for thefood fermentation industry. Tr. Food Sci. Technol. 15, 67e78.

Lick, S., Keller, M., Bockelmann, W., Heller, K.J., 1996. Rapid identification of Strep-tococcus thermophilus by primer-specific PCR amplification based on its lacZgene. Syst. Appl. Microbiol. 19, 74e77.

Limsowtin, G.K.Y., Powell, I.B., Porente, E., 1996. Type of starters. In: Cogan, T.,Accolas, J. (Eds.), Dairy Starters Cultures. VCH Publishers, Inc., USA.

Majchrzak, D., Lahm, B., Dürrschmid, K., 2010. Conventional and probiotic yogurtsdiffer in sensory properties but not in consumers’ preferences. J. Sens. Stud. 25,431e446.

Manzi, P., Marconi, S., Pizzoferrato, L., 2007. New functional milk-based products inthe Italian market. Food Chem. 104, 808e813.

Merabishvili, M., Chanishvili, N., 2001. Identification of the lactic acid bacterialcultures isolated from the samples of the Caucasian Matsoni. Proc. GeorgianAcad. Sci. 27, 91e95.

Mora, D., Fortina, M.G., Parini, C., Ricci, G., Gatti, M., Giraffa, G., Manachini, P.L., 2002.Genetic diversity and technological properties of Streptococcus thermophilusstrains isolated from dairy products. J. Appl. Microbiol. 93, 278e287.

Mora, D., Monnet, C., Parini, C., Guglielmetti, S., Mariani, A., Pintus, P., Molinari, F.,Daffonchioa, D., Manachini, P.L., 2005. Urease biogenesis in Streptococcus ther-mophilus. Res. Microbiol. 156, 897e903.

Moschetti, G., Blaiotta, G., Aponte, M., Catzeddu, P., Villani, F., Deiana, P., Coppola, S.,1998. Random amplified polymorphic DNA and amplified ribosomal DNAspacer polymorphism: powerful methods to differentiate Streptococcus ther-mophilus strains. J. Appl. Microbiol. 85, 25e36.

Mutukumira, A.N., 1996. Investigation of Some Prospects for the Development ofStarter Cultures for Industrial Production of Traditional Fermented Milk inZimbabwe (Doctor Scientiarum thesis). Agricultural University of Norway.

Nemsadze, K., 2004. Report from the country of Georgia: protecting and promotingbreast feeding through regulation of artificial-feeding marketing practices.J. Perinat. Educ. 13, 23e28.

Ott, A., Hugi, A., Baumgartner, M., Chaintreau, A., 2000. Sensory investigation ofyogurt flavour perception: mutual influence of volatiles and acidity. J. Agric.Food Chem. 48, 441e450.

Reddy, N.S., Miyamoto, T., Kataoka, K., Nakae, T., 1986. Isolation and identification oflactic acid bacteria from milk products in Eastern Europe. Japan. J. Dairy FoodSci. 35, 5e12.

Shah, N.P., 2008. Health benefits of yogurt and fermented milks. In: Chandan, R.C.(Ed.), Manufacturing Yogurt and Fermented Milks. Blackwell Publishing,pp. 327e340 (Chapter 21).

Shene, C., Canquil, N., Bravo, S., Rubilar, M., 2008. Production of the exopoly-saccharides by Streptococcus thermophilus: effect of growth conditions onfermentation kinetics and intrinsic viscosity. Int. J. FoodMicrobiol.124, 279e284.

Shiby, V.K., Mishra, H.N., 2013. Fermented milks and milk products as functionalfoods-a review. Crit. Rev. Food Sci. Nutr. 53, 482e496.

Stingele, F., Neeser, J.-R., Mollet, B., 1996. Identification and characterization of theeps (exopolysaccharide) gene cluster from Streptococcus thermophilus Sfi6.J. Bacteriol. 178, 1680e1690.

Stone, H., Sidel, J., Oliver, S., Woolsey, A., Singleton, R.C., 1974. Sensory evaluation byquantitative descriptive analysis. Food Technol. 28, 24e33.

Swaggerty, D.L., Walling, A.D., Klein, R.M., 2002. Lactose intolerance. Am. Fam. Phys.65, 1845e1850, 1855e1856.

Tamime, A.Y., Robinson, R.K., 1999. Yoghurt Science and Technology. PergamonPress, Oxford.

Tamime, A.Y., Muir, D.D., Wszolek, M., 1999. Kefir, koumiss and kishk. Dairy Ind. Int.64, 32e33.

Tanous, C., Kieronczyk, A., Helinck, S., Chambellon, E., Yvon, M., 2002. Glutamatedehydrogenase activity: a major criterion for the selection of flavour-producinglactic acid bacteria strains. Antonie van Leeuwenhoek 82, 271e278.

Terzaghi, B.E., Sandine, W.E., 1975. Improved medium for lactic streptococci andtheir bacteriophages. Appl. Microbiol. 29, 807e813.

Uchida, K., Urashima, T., Chanishvili, N., Arai, I., Motoshima, H., 2007. Majormicrobiota of lactic acid bacteria from Matsoni, a traditional Georgian fer-mented milk. Anim. Sci. J. 78, 85e91.

Ward, L.J.H., Timmins, M.J., 1999. Differentiation of Lactobacillus casei, Lactobacillusparacasei and Lactobacillus rhamnosus by polymerase chain reaction. Lett. Appl.Microbiol. 29, 90e92.

WHO, World Health Organization, 1974. Toxicological Evaluation of Some FoodAdditives Including Anti- Caking Agents, Antimicrobials, Antioxidants, Emulsi-fiers and Thickening Agents. In: Food Add. Ser. No 5. WHO, Geneva.

Wouters, J.A., Rombouts, F.M., de Vos, W.M., Kuipers, O.P., Abee, T., 1999. Cold shockproteins and low-temperature response of Streptococcus thermophilus CNRZ302.Appl. Environ. Microbiol. 65, 4436e4442.

Xanthopoulos, V., Petridis, D., Tzanetakis, N., 2001. Characterization and classifica-tion of Streptococcus thermophilus and Lactobacillus delbrueckii subsp. bulgaricusstrains isolated from traditional Greek yogurts. J. Food Sci. 66, 747e752.

Zoetendal, E.G., Akkermans, A.D.L., De Vos, W.L., 1998. Temperature gradient gelelectrophoresis analysis of 16S rRNA from human fecal samples reveals stableand host-specific communities of active bacteria. Appl. Environ. Microbiol. 64,3854e3859.