Overnight cultures of the isolates were used to inoculate ultrahigh temperature (UHT) skimmed cow's milk at 1% (V/V), incubated at 30 °C). The change in the pH was recorded every 30 min for 18 h with a pH meter (Orion™ Versa Star™, Thermo Fischer Scientific, Madrid, Spain) only values after 6 and 18 h are shown. Milk clotting was assessed at the end of fermentation.

The proteolytic activity of the isolates was examined by a qualitative method on plate count agar (PCA; Oxoid) supplemented with 2% UHT skimmed cow's milk (Oxoid). Isolated cultures were streaked on these plates and incubated at 30°C for 24h. A clear zone around the colonies indicated proteolytic activity. In addition, the proteolytic activity of the isolates was quantitatively determined using the o-phthaldialdehyde (OPA) method (12). Briefly, after incubation of the strains in the same type of skimmed cow's milk at 30 °C for 24h, the protein fraction was precipitated out of 1-mL samples by adding 2 mL of 0.75 M trichloroacetic acid (Sigma-Aldrich, Merck) and 0.2 mL of water. The mixtures were then vortexed for 2 min and filtered through a 0.20-μm pore diameter membrane (Millipore, Bedford, MA, USA). The OPA reagent (Sigma-Aldrich, Merck) was added to the filtrates and the absorbance was measured at 340 nm using a Benchmark Plus Microplate Spectrophotometer (Bio-Rad, Hercules, CA, USA). Results were expressed as glycine (Gly) using an appropriate calibration curve (concentration range 0.1–10 mM). Positive controls were established by inoculating UHT skimmed cow's milk with Lactococcus lactis NCDO 604^T and Lc. lactis SH4109 (15) strains known for their strong proteolytic activity. Non-inoculated UHT skimmed milk samples were incubated under the same conditions as negative controls.

Kempler and McKay (KMK) culture medium containing iron(III) citrate and potassium hexacyanoferrate(III) (Biochem Chemopharma, Cosne-Cours-sur-Loire, France) was used to assess the capacity of the isolates to utilise citrate (16). Cultures that turned blue after incubation at 30 or 45 °C for 24-48 h were considered able to use citrate.

Acetoin production was tested by inoculating the isolates into 10 mL of Clark and Lubs medium, incubating at 30 and 45 °C for 24 h, and testing via the Vosges-Proskaur (VP) reaction by adding to1 mL of the culture 0.5 mL of the VP I reagent (prepared by adding α-naphtol to absolute alcohol to a volume fraction of 6%) and 0.5 mL of the VP II reagent (a 16% solution of NaOH prepared in distilled water). The reaction tubes were then agitated for 1 min. After a 10-minute rest, the presence of a pink ring on the surface of the culture was deemed to indicate the production of acetoin.

Dextran production was investigated on Mayeux, Sandine and Elliker (MSE) agar medium rich in sucrose. After an incubation at 30 or 45 °C for 24-48 h, viscous colonies were deemed to be those of dextran producers.

The production of volatile compounds was assessed by inoculating UHT skimmed cow's milk with overnight cultures at 1% (V/V). After 24 h of incubation at 30 °C, the volatile compounds were quantified using a head space gas chromatograph (Agilent Technologies, Wilmington DE, USA) connected to a mass spectrophotometer detector (HS/GC/MS) equipped with a DB-WAXetr capillary column (60 mm×0.25 mm×0.25 μm, Agilent). Sample preparation and gas chromatographic separation were performed as previously described (17). Compounds were quantified as the normalized value of their chromatogram peak area using cyclohexanone (3.6 μg/mL) as an internal standard, which was given a value of 100 (12). Non-inoculated UHT skimmed cow's milk samples were incubated under the same conditions as negative controls. The difference between the values obtained for the sample and negative controls were calculated.

The antibacterial activity of the isolates was determined via well diffusion assays as reported by Schillinger and Lücke (18). Lactobacillus sakei CECT 906^T, Lactococcus lactis ssp. cremoris MG1363, Listeria innocua CECT 910^T, Micrococcus luteus NCIMB 8166, Streptococcus thermophilus LMD9 and S. thermophilus CNRZ 1066 were used as microbial indicators. Briefly, 45 mL of 1.2% agar medium at 45 °C were vigorously mixed with 40μL of an overnight culture of each indicator and poured into Petri dishes. Supernatants from overnight cultures, in duplicate, of the tested strains were adjusted to pH=6.57.0 with 0.1 M NaOH, and filtered through a 0.20-μm pore diameter membrane (Millipore). Aliquots of 40 μL of each supernatant were placed in a 4-mm well excavated in the agar plates. Antimicrobial activity was determined by measuring the diameter of the inhibition halo after 24 h of incubation under appropriate conditions. A halo diameter of >8 mm was considered a positive result. The average result of duplicate assays is shown.

Biogenic amine production was measured in supernatants obtained after 48 h of incubation of the strains in 10 mL of MRS broth supplemented with either 1 mM tyrosine, 1 mM histidine, 1 mM ornithine, 1 mM agmatine (ornithine and agmatine are precursors of putrescine via different pathways) or 1mM lysine (all from Sigma-Aldrich, Merck). Tyramine, histamine, putrescine and cadaverine were detected as previously described (7). Briefly, 100 μL of supernatant, obtained after centrifugation of the cultures at 3000×g for 10 min (5424 benchtop centrifuge; Eppendorf, Hamburg, Germany), were derivatised with diethyl ethoxymethylenemalonate (Sigma-Aldrich, Merck) following a previously described protocol (19). Derivatised samples were filtered through a 0.2-μm pore diameter membrane and analysed by ultrahigh performance liquid chromatography (UHPLC) using a Waters H-Class ACQUITY UPLC apparatus with a UV detector (Waters, Milford, MA, USA) controlled by Empower v. 2.0 software (20) (Waters) under the conditions described by Redruello et al. (19).

The ability of the strains to produce acid in commercial skimmed cow's milk was assessed by measuring the pH of the medium over 18 h (Table 2). All the isolates were slow acidifiers; none reduced the pH by more than 1 unit (Lactobacillus between 0.18 and 0.35, Leuconostoc 0.17 and 0.69, and Enterococcus isolates 0.24-0.78) in the first 6 h. After 18 h, the final pH varied from 6.14 by Ln. mesenteroides isolate LEY3 to 4.71 by E. hirae isolate LMA18. The Lb. rhamnosus isolates showed medium acidifying capacity and could be used as part of a mixed starter culture (30). Only four isolates (Lb. rhamnosus LEY15, E. hirae LMA16 and LMA18 and E. faecium LMA9) partially clotted the milk at the bottom of the tube after 18 h (Table 2). This low acidification activity of all the isolates indicates that they must be used in combination with strong acidifying starter cultures, for example of the species Lc. lactis, for the elaboration of dairy products.

Proteolytic activity against casein is a property of interest for any strain to be used as an adjunct culture (31). This ability of LAB strains to grow and develop in milk is essential (32). In the present work, proteolytic activity was examined qualitatively via growth on PCA supplemented with skimmed cow's milk, and quantitatively via the OPA assay. All the tested isolates showed a clarification halo on the supplemented PCA plates and were therefore considered proteolytic (data not shown). The OPA assay was more discriminating, revealing inter- and intra-species differences in proteolytic activity between the isolates (Fig. 1). All the lactobacilli had a proteolytic activity expressed in equivalents of glycinic acid (Gly) to about 1.00 mM), with isolate LEY14 showing the weakest activity (0.70±0.05) mM and isolate LEY15 the strongest (1.3±0.1) mM. E. hirae isolates LMA16 and LMA18 showed similar activity (0.9±0.3) and (0.9±0.2) mM, respectively. Among the E. faecium isolates, isolate LMA17 had the maximum activity, and the isolate LMA5 minimum (1.2±0.3) and (0.67±0.08) mM, respectively (denoting intraspecies variation). Intraspecies variation was also found among the Leuconostoc isolates. Isolate LEY10 showed notable activity with an equivalent of (1.2±0.4) mM; in contrast, none of the other Leuconostoc isolates exceeded 0.79 mM. As other authors report, milk proteolytic capacity appears to vary both within and among species isolated from natural sources (33, 34). This ability is therefore strain-dependent (28, 35, 36). Interestingly, some isolates, such as Lb. rhamnosus LEY15 and Ln. mesenteroides LEY10, showed good proteolytic activity, suggesting they might have practical applications and be responsible for releasing peptides and amino acids responsible for the product texture and aroma (37, 38). Some amino acids are involved in the production of aroma compounds, serving either directly or indirectly as precursors of aldehydes, acids, alcohols and esters (35), and thus contributing to the sensory profile of the end product (5).

The capacity of LAB to utilise the citrate present in milk is a desirable technological trait; its metabolism results in an excess of pyruvate that can be converted via α-acetolactate to diacetyl, acetoin and 2,3-butanediol - important flavour and aroma components of certain fermented dairy products (39). Table 3 shows that all the isolates of lactobacilli were positive for citrate utilisation, and in the case of Leuconostoc, only isolate LEY10 was unable to use it. The two E. hirae isolates were also able to utilise citrate, as were all but three E. faecium isolates (LMA1, LMA5 and LMA8) (Table 3).

Acetoin production, reflected by the presence of a pink ring in Clark and Lubs broth, was positive for all the tested enterococci, both the E. hirae and E. faecium isolates, all the Lb. rhamnosus isolates, and one Leuconostoc isolate (LEY10) (Table 3). Acetoin, which is produced via the catabolism of pyruvate, is responsible for the development of flavour and aroma. Enterococci and lactobacilli are the predominant genera able to produce it (34). Although Ln. mesenteroides LEY10, and E. faecium LMA1, LMA5 and LMA8 were unable to utilise citrate (Table 3), they seem to catabolise some of the pyruvate, produced from sugar, into acetoin. All the isolates of lactobacilli and enterococci produced acetoin in this work, and may play an important role in the development of the distinctive organoleptic properties of fermented camel dairy products.

Table 3 shows that only the Leuconostoc isolates produced dextran, with the exception of isolate LEY1. This would suggest the latter to be Ln. mesenteroides ssp. cremoris, a subspecies characterised by its non-production of this compound (24, 40). The dextran-producing ability of the Leuconostoc isolates makes them candidates for use in secondary starters; dextran improves the appearance, stability and rheological properties of dairy products (34) and has potential health benefits (immunogenic properties, protection against gastric ulcers, improvement of digestive transit, and hypocholesterolaemic, antiviral and antitumoral activity, etc.) (41).

Flavour development in dairy products is essentially an enzymatic process performed (mainly) by microorganisms (35). Lactose fermentation leads to the formation of pyruvate that can be further metabolised to ethanol, diacetyl, acetoin and acetaldehyde (42). By producing volatile compounds, such as ethanol (important in kefir and koumiss), diacetyl (important in butter, buttermilk and cheese) and acetaldehyde (important in yoghurt and buttermilk) (5), LAB contribute to the typical flavours of different dairy products.

The production of volatile compounds was assessed by HS/GC/MS (17). Eighteen major compounds were identified (Table 4). Wide variation was observable in the diversity and quantity of volatile compounds produced by the different isolates. The most common compounds produced were ethanol, acetaldehyde, methyl acetate, acetoin and acetic acid. Acetaldehyde was produced by all the isolates except Ln. mesenteroides LEY9 and LEY10. The production of acetoin in milk agreed with the results of the phenotypic test performed in Clark and Lubs medium. All the isolates of enterococci and lactobacilli gave positive results for both methods (HS/GC/MS, and Clark and Lubs medium) (Table 3 and Table 4). However, while Ln. mesenteroides LEY10 was positive in the phenotypic test, no acetoin was detected when it was grown in milk, indicating the need for confirmatory testing. The production of diacetyl was detected in all Lb. rhamnosus isolates and E. hirae LMA16, but not in Ln. mesenteroides nor E. faecium isolates. The LAB producing the least volatile compounds were the Leuconostoc isolates. This might be explained by the fact that they grow poorly in milk and that they must be combined with acid-producing lactococci in order to act as flavour producers in mixed starters (12, 35, 43). The ability of the Lb. rhamnosus isolates to produce flavour compounds such as diacetyl and acetoin suggests they could be used as adjunct cultures for developing cheese flavour (44).

The potential production of antimicrobial substances by the isolates was tested against a variety of indicator strains, including several LAB species, since bacteriocins produced by Gram-positive bacteria are most commonly active against closely related bacteria. Table 5 shows that none of the isolates inhibited the growth of S. thermophilus CNRZ 1066. The indicator Lb. sakei CECT 906^T was inhibited by all but one of the Lb. rhamnosus isolates, and by 42% of the E. faecium isolates, while M. luteus NCIMB 8166 was inhibited by 52% of the Leuconostoc isolates. Lc. lactis ssp. cremoris MG1363 was inhibited by 16% of the E. faecium isolates, L. innocua CECT 910^T was inhibited by 42% of the E. faecium isolates, and S. thermophilus LMD9 by 26% of the E. faecium isolates. Wider inhibition activity was observed for the E. faecium isolates, which inhibited the growth of four of the indicator strains (all except S. thermophilus CNRZ1066 and M. luteus NCIMB 8166). E. faecium LMA5 showed the strongest activity. LAB have a wide range of antimicrobial activities (45); E. faecium produces bacteriocin-like substances such as enterocins A and P, while Ln. mesenteroides produces leucocin (46). The present results agree with those reported in the literature, but further studies should test the capacity of these strains to inhibit the growth of pathogenic bacteria such as Listeria monocytogenes. They should also evaluate the potential of the isolates as adjunct cultures for improving food safety and shelf life.

Dairy products accumulate the greatest diversity and quantity of biogenic amines (47). The consumption of food with elevated biogenic amine concentrations can lead to symptoms of intoxication. It should not, therefore, be allowed to accumulate to dangerous levels. One of the preventive measures that might be applied is the selection of strains for use in starter or co-starter cultures that are confirmed non-biogenic amine-producers (6). In the present work, the capacity to produce tyramine, histamine, putrescine and cadaverine was examined by UHPLC (19). Table 6 shows that all the E. faecium isolates produced tyramine. This is not surprising since tyramine production is a known species-dependent feature of E. faecium (48). The presence of biogenic amine-producing bacteria, such as enterococci, in the raw milk highlights the potential risk of allowing spontaneous fermentation, since as shown in this work, some harmful microorganisms could grow and accumulate toxic compounds (49). If E. faecium is to be used as an adjuvant culture, the balance between benefits and dangers must take into account the many factors that influence the accumulation of tyramine, such as the availability of tyrosine (due to casein proteolysis) and the presence of an acidic pH (6). Moreover, if a biogenic amine-producing strain also produces an antimicrobial substance, as is the case of E. faecium LMA5, it could dominate the microbiota and the risk of tyramine accumulation would be increased.