Introduction

Dairy products are the second most consumed staple food in Algeria, following cereals (Hales and Torry 2018). Due to the local insufficient production of milk caused by the absence of a robust national milk industrial chain (Sraïri et al. 2013), Algeria importations arrive to 250,000 up to 280,000 tons of powdered milk per year (Kardjadj and Dachung Luka 2016). Other limitation to technological advancement of milk self-sustaining is the high level of microbial contamination originated from poor hygienic practices in all stages of the chain, starting by the milking, collection, transporting/distribution, and storage of milk.

Previous studies on Algerian milks showed that they are reservoirs of several LAB, such as Lacticaseibacillus, Lactiplantibacillus, Latilactobacillus, Pediococcus and Lactococcus spp. with antifungal and antibacterial activity (Mechai et al. 2020), which may be exploited to ensure the safety of fermented foods.

The isolation and characterisation of indigenous microbial diversity is a key step to design tailored starter cultures for artisanal/traditional fermented food that increase the safety and quality of such highly appreciated foodstuffs (Capozzi et al. 2020; Saidi et al. 2020). In Algeria arid regions, camel's milk for example is considered as one of the most important sources of dairy products for human diet with potential therapeutic effects.

Specific sensory characteristics are generated from LAB metabolic pathways, resulting in a diversity of aromatic compounds. They convert the sugar into lactic acid, resulting in the rapid acidification of raw material, and produce other metabolites, such as ethanol, diacetyl, acetate, and acetaldehyde that improve the flavor, texture, taste, storage, and safety of the end products (Leroy and Vuyst 2004; Perin et al. 2017) and might be selected to improve or replace currently used starters and adjunct cultures (Brandsma et al. 2008; Alegria et al. 2016).

The LAB also contribute to the proteolysis of cheese, as they can degrade the products derived from the rennet action on the casein (peptides of high and low molecular mass). (Herreros et al. 2003) They are also the object of intensive international researches for their ability to produce several antimicrobial compounds, such as bacteriocin (De Vuyst and Leroy 2007; Reis et al. 2012; Benmechernene et al. 2013), for their essential role in the food fermentation and degradation of protein that lead to the synthesis of a wide range of compounds, such as organic acids, peptides, aromatic compounds and exopolysaccharides (Saidi et al. 2019; Mende et al. 2016).

Obviously, the selection of potential starter cultures must focus not only on their functional properties but also on the absence of production of undesirable factors, such as biogenic amines (BA) and antibiotic resistance (AR) genes. Antibiotic resistant bacteria constitute a serious problem for the health of both humans and animals (Berendonk et al. 2015), and fermented foods could spread AR genes along the food chain to the human gastrointestinal tract (Founou et al. 2016). BAs are low molecular weight nitrogenated compounds which can accumulate in foods though the microbial decarboxylation of certain amino acids (Linares et al. 2011). Certain LAB strains are the main responsible for production and accumulation of BA in dairy products (Linares et al. 2012), thus assessment of the ability to produce BA by potential protective LAB starters is essential to improve food safety and the consumers health (Ladero et al. 2017).

The bioprospecting of recovering new safe protective LAB strains from Algerian milks configures an excellent strategy to save the local biological heritage and to maintain the geographical identity of the obtained fermented product (Saidi et al. 2020; Merabti et al. 2019).

For these reasons, the aim of the present study was to unambiguously identify thirty-eight indigenous bacteria collected from dairy Algerian milks, assess their genetic diversity through molecular fingerprinting, evaluate their technological potential and safety.

Materials and methods

Bacterial strains, sample collection and growth conditions

A total of thirty-eight LAB, previously isolated from different milk sources (camel, cow, sheep and goat milk), from environmental samples (pollen, olive oil, traditional cheese) and from laboratory collection collected from nine areas of south and northwestern of Algeria (Tindouf, Bechar, Adrar, Tiaret, Oran, Tlemcen, Saida, Jijel, Kabilie) (Table 1). Coccus-shaped isolates were cultured in M17 medium (Oxoid, UK) supplemented with 0.5% glucose and incubated at 32 ºC for 24–48 h. Rod-shaped strains were grown in MRS medium (Oxoid) incubated in anaerobic conditions under a 10% H2, 10% CO2 and 80% N2 atmosphere in a MACS MG-500 anaerobic chamber (Don Whitley Scientific, West Yorkshire, UK) at 37 °C for 48 h. The strains were stored as frozen stocks at − 80 °C in the respective culture medium supplemented with 20% (v/v) glycerol.

Table 1 List of LAB isolates with their isolation source and 16S rRNA gene-based identification
Full size table

Molecular identification and typing

DNA extraction

Genomic DNA was extracted from 1.5 mL overnight cultures. In brief, cells were harvested by centrifugation (12,000 rpm for 2 min), washed with distilled water, and then resuspended in a lysozyme solution (30 μg/mL) supplemented with mutanolysin (100 U/mL). After incubation at 37° C for 30 min, proteinase K (20 μg/mL) was added and a second step of incubation at 55 °C for 30 min was carried out. Genomic DNA was purified using the GenElute Bacterial Genomic DNA kit (Sigma-Aldrich, USA), following the manufacturer’s recommendations. DNA yield and purity were quantified using the Nanodrop ONE UV–Vis Spectrophotometer (Thermo Scientific, USA). DNA was stored at − 20° C for downstream analysis.

Molecular identification of the isolates

Total DNA was used as template to amplify the universal region of the 16S rRNA gene by PCR using the primers 27F and 1492R, according to Lane (1991). PCR amplicons were examined by 1.2% (w/v) agarose gel stained with ethidium bromide (0.5 µg/mL) and visualized by GelPrinter plus (TDI, Spain). PCR products were purified by theATP™ Gel PCR Fragment DNA Kit (ATP Biotech Inc., Taipei, Taiwan) and delivered to Macrogen (Amsterdam,The Netherlands) for Sanger sequencing. Lactobacilli were identified by species-specific PCRs according to the protocols described for Lb. plantarum group (Torriani et al. 2001), Lb. casei group (Bottari et al. 2017) and for the Lb. brevis species (Guarneri et al. 2001)⁠⁠.

Sequence identities were analyzed using BLAST. The 16S rRNA gene sequences were deposited in the NCBI database. The unrooted phylogenetic tree was constructed to determine the closest LAB species by the neighbor-joining method (Felsenstein 1985).

Typing of isolates

Genetic fingerprinting of the isolates was assessed by rep-PCR using the primer (GTG)5 (Iacumin et al. 2006) following the previously reported protocol of Versalovic et al. (1994). PCR amplicons were electrophoresed in 1.5% (w/v) agarose gels for 150 min, and then revealed on UVITEC (UK). The O’GenRuler DNA ladder Mix (Thermo Scientific) was used as molecular size marker Digitalized images were analyzed using the Uvitec Fire Reader Acquisition System and the dendrogram was constructed by the UVIB and Map software (Uvitec, UK).

Technological and functional characterisation

Acidifying activity

Acidifying activity was determined in reconstituted, sterile skim milk (Sigma-Aldrich), according to the protocol of Olasupo et al. (2006). Briefly, single colonies of each strain grown on MRS agar plates incubated at 37 °C for 48 h were pickled and inoculated in skim milk. Then, 1 mL of overnight milk cultures was inoculated in 10 mL of skim milk. Incubation was performed at 37 °C for up to 48 h, and pH variations were measured at 24 and 48 h using a pH-meter (Crison Instruments S.A., Spain). Visual inspection of the clotting regarding whey drainage, curd firmness, presence of gas bubbles and curd breaking was also recorded. The assay was performed in triplicate.

Production of volatile compounds in milk

Volatile compound analysis was performed after growth of the LAB in screw-cap tubes at 37 °C for 48 h in UHT milk supplemented with cyclohexanone (3.6 µg/mL) as internal standard. Separation and quantification of the volatiles compound were performed by headspace-gas chromatography-mass spectrometry (HS-GC–MS), using Agilent apparatus (Agilent Technologies, USA) equipped with a capillary column DB-WAXetr (60 m × 0.25 mm × 0.25 µm). Sample preparation and gas chromatographic separation were performed as described by Salazar et al. (2008). Compounds were quantified as the normalized value of their chromatogram peak areas; the internal standard was given a value of 100. The experiment was performed in duplicate.

Antibacterial activity

The capacity of LAB isolates to produce antimicrobial substances was determined by the agar well-diffusion assay as described in Saidi et al. (2020). Lb. parabuchneri St2A, Listeria innocua CECT 906 T, Micrococcus luteus NCBI 8166, and Lactococcus lactis subsp. cremoris MG1363 were used as target strains. Supernatants from overnight cultures, in duplicate, of the tested strains were adjusted to pH 7.0 with 0.1 M NaOH and filtered through a 0.20 μm pore diameter membrane (Millipore). Twenty μL aliquots of each supernatant were placed in wells excavated in the agar plates and were incubated at 37 °C for 24–48 h. The clear inhibition zone around the well was measured in mm. A halo above 8 mm was considered a positive result (Saidi et al. 2020).

Proteolytic activity

Proteolytic activity of whole cells in milk was determined using the O-phthaldialdehyde (OPA) test as previously described (Church et al. 1985). In summary, the increase in optical density at 340 nm (OD340), relative to the control, was determined using the Cary 60 UV–Vis Spectrometer (Agilent Technologies, Santa Clara, CA, USA). The OPA solution contained: 2.5 mL of 20% (w/v) SDS, 25 mL of 100 mM sodium tetraborate (RIEDEL Germany), 40 mg of OPA (Fluka, Biochemika) (previously dissolved in 1.0 mL methanol), 100 µL of 2-mercaptoethanol (BIO RAD), and distilled water up to a 50 mL final volume. The samples were incubated with 0.75 M trichloroacetic acid (TCA) (Fisher Bio Reagents) in a proportion of TCA/sample of 1:3 at 4 °C for 30 min and centrifuged at 5000 rpm for 10 min. A 50 µL supernatant aliquot of this mixture was added to 1.0 mL of OPA reagent and incubated at room temperature for 20 min, then read at the spectrophotometer. Proteolytic activity was arbitrarily expressed as µg of leucine (Leu) released/mL using a standard curve of L-leucine (Sigma Chemical Co).

Exopolysaccharide production

Overnight cultures of isolates were spotted (5.0 µL) on the surface of MRS plates supplemented with 0.08 g/L of ruthenium red (Sigma-Aldrich). After incubation at 37 °C for 48 h, exopolysaccharide (EPS) producing strains gave white colonies, while non-producers appeared as red colonies (Kersani et al. 2017).

Safety assessment

Antibiotic resistance assay

Antibiotic susceptibility was assessed by the disk diffusion method, as described by Anisimova et al. (2017). In brief, all overnight cultures of strains were diluted in 0.85% saline solution to obtain a standardized turbidity equivalent to McFarland scale 0.5. Aliquots of these suspensions were pour-plated in Muller-Hinton agar medium plates. Antibiotic discs for tetracycline (30 µg), vancomycin (30 µg), gentamycin (10 µg) and erythromycin (15 µg) (Bio-Rad, Marnes-la- Coquette, France) were dispensed onto the inoculated plates. After 48 h incubation at 37 °C in anaerobic conditions, inhibition halos were measured in mm (means ± SD of 3 trials) and interpreted as susceptible (S), moderately susceptible (MS), or resistant (R), according to Melo et al. (2017).

Detection of BA-producing genes

The presence of the amino acid decarboxylase genes involved in the production of tyramine [tdcA, encoding the tyrosine decarboxylase from the tyrosine decarboxylase cluster (TDC)], histamine [hdcA, encoding the histidine decarboxylase from the histidine decarboxylase cluster (HDC)] and putrescine [aguD-aguA genes from the agmatine deiminase (AGDI) cluster] was checked by PCR. For that we used the primer pairs tdc1 and tdc2 (Fernández et al. 2004), hdcDG-F and hdcDG-R (Diaz et al. 2016), and AgmSq1 and AgmSq2 (Linares et al. 2011), respectively. Positive controls were performed using total genomic DNA obtained from different BA-producing strains: Enterococcus faecalis V583 for tyramine and putrescine (via AGDI) (Ladero et al. 2015), and Lentilactobacillus parabuchneri IPLA11150 for histamine (Diaz et al. 2016). PCR products were visualized after gel electrophoresis as stated above.

Biogenic amine production

The ability to produce the BA tyramine, histamine and putrescine was evaluated in the isolates showing PCR positive results for the presence of BA-producing genes following the protocol described by Ladero et al. (2015). Briefly, lactococci were grown at 32 °C in GM17 (Oxoid) while lactobacilli were grown in anaerobiosis at 37 °C for 24 h in MRS broth, both supplemented with either 1 mM histidine, 1 mM tyramine or 1 mM agmatine. BA production in culture supernatants was analyzed by Ultrahigh Performance Liquid Chromatography (UHPLC) in a Waters H-Class Acquity UPLC apparatus with a UV detector (Waters, USA) controlled by Empower 2.0 software (Waters), following the protocol described by Redruello et al. (2013).

Results

Molecular identification and typing of LAB isolates

Thirty-Eight LAB isolates from different Algerian dairy and farm sources were identified through 16S rRNA gene sequencing. Based on BLAST analysis, the isolates were preliminarily identified as Enterococcus faecium (n = 15), Enterococcus durans (n = 2), Enterococcus hirae (n = 2), Enterococcus lactis (n = 1), Lactiplantibacillus plantarum group (n = 6), Lactococcus lactis (n = 4), Lacticaseibacillus casei group (n = 4), Levilactobacillus brevis (n = 3), and Pediococcus acidilactici (n = 1) with a similarity value over 99% (Table 1). Since safety aspects related to enterococci have raised questions regarding their use in foods or as probiotics (Berendonk et al. 2015), they were excluded from further analysis.

To identify closely related lactobacilli isolates at species level, species-specific PCRs were performed. Results, showed in the Online Resource 1, allowed the unequivocal identification of Lpb. plantarum (isolates LGMY9, LGMY16, LGMY23, LGMY27, LGMY29, LGMY30), Lvl. brevis (LGMY21, LGMY26 and LGMY33) and within the Lcb. paracasei group, PCR differentiated the isolates of Lcb. paracasei (LGMY31, LGMY32 and LGMY35) from Lcb. rhamnosus LGMY34 (Table 1).

To highlight genotypic differences among isolates, Rep-PCR and cluster analysis were performed. The dendrogram depicted in Fig. 1 differentiated six groups that coincided with the identification at species level. Further, different band patterns were found at the intraspecific level, revealing the presence of sixteen unique profiles associated to single strains (Fig. 1). However, two isolates of Lpb. plantarum (LMGY16 and LMGY29) and two of L. lactis (LMGY4 and LMGY36) showed the same profile, even though they were isolated from different sources.

Fig. 1
figure 1

Dendrogram showing the genetic similarities between LAB isolates based on (GTG)5-PCR fingerprinting. The cluster analysis of genetic distances was performed with the unweighted pair-group method using arithmetic averages (UPGMA)

Full size image

Query coverage of the LAB identification results obtained using BLAST was 99–100%. The partial sequences of isolates were deposited in the GenBank and the accession numbers were reported in Table 1. A phylogenetic tree was constructed using the neighbor-joining method (Fig. 2).

Fig. 2
figure 2

Phylogenetic tree based on 16S rRNA gene sequences depicting the diversity of Lpb. plantarum, Lcb. paracasei, Lvl. brevis, Lcb.rhamnosus, L. lactis and P. acidilactici isolates by the neighbor-joining method

Full size image

Technological and functional characterization

Acidifying activity

The UHT milk acidification assay (Table 2) showed that all strains were able to grow in milk, and provoked milk clotting after 24 h at 30 °C. After 48 h fermentation, pH values between 2.65 and 2.70 were reached in the vats inoculated with Lcb. paracasei LGMY31 and Lpb. plantarum LGMY30, Lvl. brevis LGMY21 and LGMY33, and Lcb. paracasei LGMY31, LGMY32 and LGMY35, resulting in the production of a stable clot. Furthermore, in the vats inoculated with Lvl. brevis LGMY33, the gas production was considerable and jeopardized the clot structure resulting in a strong whey drainage. The strains L. lactis LGMY17 and LGMY36 acidified the UHT milk more rapidly than the other two lactococcal strains, and reached lower pH values after 48 h.

Table 2 Technological and functional properties of the isolates
Full size table

Proteolytic activity

The results of the proteolytic activity of the strains, determined by the OPA test, are showed in Table 2. Lcb. paracasei LGMY31 showed the highest release of amino acids from milk proteins (1.699 mM of leucine equivalents), followed by Lpb. plantarum LGMY29 and Lvl. brevis LGMY21, which released 0.178 and 0.146 mM of leucine equivalents, respectively. Six other LAB showed a low proteolytic activity, ranging from 0.006 for Lpb. plantarum LGMY30 up to 0.092 mM of leucine equivalents for L. lactis LGMY8. The remaining strains showed no detectable proteolytic activity.

Exopolysaccharide production

The production of EPS was assessed by evaluating the color of colonies grown on MRS agar containing ruthenium red. All the six strains of Lpb. plantarum, Lcb. paracasei LGMY31 and LGMY35, and Lcb. rhamnosus LGMY34 grew as white colonies, indicating their ability to produce EPS.

Antagonistic activity

The inhibitory potential of the LAB was investigated against four target strains: L. innocua CECT 906T, a surrogate of the human pathogen L. monocytogenes; M. luteus NCBI 8166 and Llb. parabuchneri St2A, two sensitive strains frequently used as target strains: and L. lactis subsp. cremoris MG1363, a closely related species. Most strains exhibited antagonistic activity against one or more targets in the well-diffusion assay, as shown in Table 2. Nine strains (L. lactis LGMY4, LGMY 9 and LGMY36, Lvl. brevis LGMY21 and LGMY26, Lpb. plantarum LGMY27, Lcb. paracasei LGMY31 and LGMY32, Lcb. rhamnosus LGMY34) inhibited Llb. parabuchneri St2A. Remarkably, only P. acidilactici LGMY3 showed specific inhibitory effect against L. innocua CECT 906T. Regarding M. luteus NCBI 8166, it was inhibited by Lpb. plantarum LGMY29 and LGMY30, Lvl. brevis LGMY26, Lcb. paracasei LGMY31 and Lcb. rhamnosus. Moreover, L. lactis subsp. cremoris was inhibited by six strains (L. lactis LGMY8 and LGMY36, Lpb. plantarum LGMY27 and LGMY30, Lvl. brevis LGMY33 and Lcb. paracasei LGMY31). Overall, Lcb. paracasei LGMY31 exhibited the widest range of inhibition against the selected targets.

Production of volatile compounds

The production of volatile compounds differed among strains (Fig. 3A). Ethanol was the main volatile compound produced by all strains, except LGMY8, which probably lack the enzyme alcohol dehydrogenase involved in ethanol production from acetaldehyde (Dan et al. 2019). Acetic acid was also frequently produced (89% of the strains), except for L. lactis LGMY4 and P. acidilactici LGMY3. Small amounts of 2-propanone were produced by Lpb. plantarum LGMY27 and Lvl. brevis LGMY21. Acetoine was identified as a minor component in the profile of P. acidilactici LGMY3. Almost all the strains produced 3-methyl 1-butanol, except Lpb. plantarum LGMY23, L. lactis LGMY8 and Lcb. paracasei LGMY35, while acetate-3-methyl-1-butanol was produced by L. lactis LGMY36.

Fig. 3
figure 3

Relative abundance of volatile compounds produced in milk by the LAB strains (A) and hierarchical cluster analysis based on their maximum concentration (B)

Full size image

Furthermore, hierarchical cluster analysis based on the volatile compounds produced by each strain is shown in Fig. 3B. Three main clusters and one outlying group were identified in the dendrogram. The first group include the high producer L. lactis LGMY36, the second group includes Lcb. paracasei LGMY31 and LMGY35. The third cluster grouped seven medium producers belonging to different species (Lpb. plantarum LGMY9, LGMY27, LGMY29, LGMY30, Lvl. brevis LGMY21, LGMY26 and L. lactis LGMY8). In the fourth cluster, the most numerous with 10 strains, includes the lower producers (Lcb. paracasei LGMY32, Lcb. rhamnosus LGMY34, Lpb. plantarum LGMY16 and LGMY23, Lvl. brevis LGMY33, L. lactis LGMY4 and LGMY17, and P. acidilactici LGMY3). This analysis showed that volatile compounds production is strain-specific and not related to the species. Indeed, different species are equally distributed in these two last groups.

Safety assessment

Antibiotic resistance

All the strains were analyzed for antibiotic susceptibility by disc diffusion method and were classified either as resistant (R), moderately susceptible (MS), or sensitive (S) based on zones of growth inhibition (Table 3). Vancomycin resistance was assayed for the L. lactis strains, that resulted all susceptible, since this characteristic is intrinsic for the other LAB species considered, and its evaluation is not required (Aquilina et al. 2012). Most isolates were susceptible to erythromycin (88%), tetracycline (78%), and gentamicin (50%).

Table 3 Safety assessment of isolates based on AR phenotypes and BA producing potential
Full size table

Presence of BA-producing genes and capability to produce BA

The genetic potential to synthesize BA was investigated by PCR assays targeting genes responsible for histamine, tyramine and putrescine production, through the HDC, TDC and AGDI pathways, respectively. None of the isolates were PCR positive for the genes hdc and tdc (Table 3), which indicated the absence of the HDC and the TDC clusters and therefore their inability to produce histamine and tyramine, respectively. However, some isolates were PCR positive for the genes aguD-aguA of the AGDI cluster (Table 3). The 700 bp expected amplicon was obtained for the strains P. acidilactici LGMY3, L. lactis LGMY4, LGMY8, LGMY17 and LGMY36, Lvl. brevis LGMY21, LGMY26 and LGMY33. Among those, only P. acidilactici LGMY3, and Lvl. brevis LGMY21 and LGMY26 were capable to produce putrescine under the examined conditions (Table 3).

Discussion

The genetic characterization of the LAB strains isolated from several Algerian dairy milks and farm sources allowed their accurate taxonomic assignment at species and strain levels. Particularly, the analysis of (GTG)5 REP-PCR fingerprints highlighted a high genetic biodiversity of the strains. In this research, the lactobacilli showed a remarkable ability to cause rapid milk acidification, which is a desired activity for biotechnological applications (Bintsis 2018). Among the investigated strains, LGM32, LGM34, LGM27 and LGMY33 produced an unstable clot, thus they are not suitable for application in dairy processes. In contrast, Lpb. plantarum LGMY23, LGMY29 and LGMY30, Lvl. brevis LGMY21 and LGMY33, and Lcb. paracasei LGM31, LGM32, LGM35 showed a faster and stronger acidification ability, higher than that reported for others Algerian milk isolates (Bousmaha-marroki and Marroki 2015). This capability is crucial in cheese production, since a rapid acidification favors the coagulation and limits the growth of adventitious undesired microorganisms (Hassaïne et al. 2007).

Regarding antagonistic activity, most of the isolated LAB strains were able to inhibit at least one of the target strains, and, remarkably, P. acidilactici LMGY3 inhibited L. innocua CECT 906T. The antimicrobial activity was strain-dependent and could be linked to the production of one or more active compounds during their growth, such as bacteriocins (Gao et al. 2019).

The strains Lcb. paracasei LGMY31, Lpb. plantarum LGMY29 and Lvl. brevis LGMY21 showed interesting results regarding proteolysis. Proteolytic activity is another important characteristic to obtain desirable organoleptic and flavor traits in dairy products (Medjoudj et al. 2020). It was also of interest the capacity of the strains Lpb. plantarum LGMY23 and LGMY30 and Lcb. paracasei LGM31 and LGM35 to produce EPS and acidify rapidly milk, indicating a good technological potential for application in milk fermentation processes to maximize texture and viscosity (Bachtarzi et al. 2019).

The volatilome was mainly represented by high amounts of ethanol and acetic acid. Homofermentative LAB produce mainly lactic acid, from carbohydrates, whereas heterofermentative LAB produce a mixture of lactic acid, acetic acid, ethanol and CO2 (Widyastuti and Febrisiantosa, 2014). Differently from the other strains, L. lactis LGMY8 produced acetaldehyde, which is an important secondary metabolite related to the typical aroma and flavor in yogurts, recognized as “ethereal”, “pungent”, “fresh” and “green” (Dan et al. 2019). On the other hand, 2-propanone was produced only by the strains Lpb. plantarum LGMY27 and Lvl. brevis LGMY21. This compound was identified as an odor-active compounds related to wood pulp or hay odor notes (Picon et al. 2019). The related ester 3-methyl-1-butanol acetate (isoamyl acetate) was exclusively produced by L. lactis LGMY36. Esters give an important contribute to the aroma of cheeses and are associated to fruity flavors in Italian and Swiss-type cheeses (Liu et al. 2004).

Regarding the safety assessment, most isolates were resistant to at least one of the antimicrobial compound tested. The major prevalence was to gentamicin (50%); this resistance was previously observed for other LAB isolated from Algerian dairy products (Naceur and Boudjemâa, 2016). The high number of isolates resistant to gentamicin is related to the intrinsic resistance of lactobacilli against this antimicrobial (Campedelli et al. 2019), while the resistance to tetracycline and erythromycin were related to horizontal gene transfer (HGT) events (Anisimova and Yarullina 2018). BA are present in several food products due to the decarboxylase activity of certain strains of LAB (Ladero et al. 2017). The genes related to histamine and tyramine were not detected in any of the strains, while several strains presented genes related to putrescine synthesis via the AGDI pathway. However, putrescine production was confirmed by UHPLC only for P. acidilactici LGMY3, and Lvl. brevis LGMY21 and LGMY26.

Conclusion

Raw milks and artisanal fermented products represent excellent sources of native LAB strains that can be used as starter or protective cultures. Indeed, the technological and safety aspects evaluated in this research allowed the individuation of some strains that could be good candidates for applications in dairy sector. Particularly, the strains L. lactis LGMY8 from cow milk, Lpb. plantarum LGMY30 from artisanal cheese and Lcb. paracasei LGMY31 from camel milk were safe and unveiled at least one of the desirable technological properties tested, such as high acidification and proteolytic activities in milk, EPS production and antagonistic activity. Further studies will be necessary to evaluate the performances of these strains, singularly or in combination, in artisanal dairy production processes.