The microbial diversity and community structure of three different kefir grains from different parts of Brazil were examined via the combination of two culture-independent methods: PCR-denaturing gradient gel electrophoresis (PCR-DGGE) and pyrosequencing. PCR-DGGE showed Lactobacillus kefiranofaciens and Lactobacillus kefiri to be the major bacterial populations in all three grains. The yeast community was dominated by Saccharomyces cerevisiae. Pyrosequencing produced a total of 14,314 partial 16S rDNA sequence reads from the three grains. Sequence analysis grouped the reads into three phyla, of which Firmicutes was dominant. Members of the genus
Lactobacillus were the most abundant operational taxonomic units (OTUs) in all samples, accounting for up to 96% of the sequences. OTUs belonging to other lactic and acetic acid bacteria genera, such as Lactococcus, Leuconostoc, Streptococcus and Acetobacter, were also identified at low levels. Two of the grains showed identical DGGE profiles and a similar number of OTUs, while the third sample showed the highest diversity by both techniques. Pyrosequencing allowed the identification of bacteria that were present in small numbers and rarely associated with the microbial community of this complex ecosystem.
► The microbial diversity of kefir grains was assessed by DGGE and pyrosequencing. ► Lactobacillus spp. were found predominant by both methods in all three grains. ► DGGE showed that the yeast community was dominated by Saccharomyces cerevisiae. ► Pyrosequencing allowed the identification of minor bacterial groups.
Kefir is a viscous, acidic, and mildly alcoholic milk beverage produced via the fermentation of milk using kefir grain as a starter culture. Thought to be native to the Caucasus and Middle East, production and consumption of kefir has now spread throughout the world, spurred by its long history of beneficial health effects. Kefir grains are white to yellowish-white, cauliflower-like “florets” composed of an inert polysaccharide/protein matrix containing a stable and specific microbial community of different lactic acid bacteria (LAB), acetic acid bacteria (AAB) and yeast species in a complex symbiotic relationship. Kefir grains are supposed to have developed spontaneously in milk stored in containers made from animal skins, intestines or bladders. Grains may have arisen independently in different locations and may be associated with specific microbial populations that produce beverages with distinct sensory properties.
The microbial diversity of kefir has traditionally been assessed by culturing methods by which different LAB species have been identified. A wide variety of Lactobacillus species have been isolated from both kefir beverages and grains, including Lactobacillus kefiri, Lactobacillus kefiranofaciens, Lactobacillus kefirgranum and Lactobacillus parakefiri, which constitute dominant populations. Lactococcus lactis subsp. lactis and Lactococcus lactis subsp. cremoris are also commonly reported; these bacteria are thought to be loosely associated with the grains and responsible for acidification. Not surprisingly, Lc. lactis has been identified as dominant in the fermented product by both culturing and culture-independent techniques. Leuconostoc and other Lactobacillus species have also been isolated in small numbers AAB have received less attention, although they are presumed to be essential in both the microbial consortium and the organoleptic characteristics of the final product. Among the yeasts, Kluyveromyces marxianus, Torulaspora delbrueckii, Saccharomyces cerevisiae, Candida kefir, Saccharomyces unisporus, Pichia fermentans, Kazachastania aerobia, Lachancea meyersii, Yarrowia lipolytica and Kazachstania unispora have all been detected.
Culturing methods have proven to be unreliable for the complete characterization of microbial ecosystems, including those of food fermentation. Culture-independent microbial techniques, such as denaturing gradient gel electrophoresis (DGGE) and the construction and analysis of libraries of conserved genes such as the 16S rRNA gene are therefore now used, and indeed have already been employed to study the microbiology of kefir grains. These techniques have detected all of the commonly cultured species, as well as some previously undetected microorganisms. However, in spite of this extensive knowledge, the inventory of the microbial species associated with the kefir grains is thought to be far from complete.
Pyrosequencing, an automated high-throughput sequencing technique that involves the synthesis of single-stranded deoxyribonucleic acid and the detection of the light generated by pyrophosphate released in a coupled reaction with luciferase (Margulies et al., 2005), has recently made its debut in the study of food fermentation. This technique allows the rapid and accurate sequencing of nucleotide sequences that can then be used to analyze the population structure, gene content and metabolic potential of the microbial communities in an ecosystem. Pyrosequencing has recently been used to study the diversity and dynamics of the bacterial populations of an Irish kefir grain and its corresponding fermented product (Dobson et al., 2011).
The aim of the present work was to characterize the microbial diversity of three different kefir grains collected in different regions of Brazil. Several studies of Brazilian kefir have already been undertaken, but most of these focused on the microbial composition of the kefir beverage during fermentation. The present work catalogues the microbial species identified in three kefir grains using two culture-independent microbial methods – PCR-DGGE and barcode pyrosequencing – and compares the results obtained with those reported in the literature.
The three kefir grains used in this study were collected from different cities with similar climates and an average distance of 400 km in southeastern Brazil. Grains AV and AD were kindly provided by researchers from Brazilian universities, meanwhile the third grain (AR) belonged to a family that traditionally cultivated the kefir grain in private household for self consumption. At the laboratory, grains were activated in sterile reconstituted skim milk (10% w/v) at 25 °C for 24 h, filtered through a sieve to remove the clotted milk, and rinsed with sterile water. This activation step was repeated three times.
For microbial genomic DNA extraction, activated kefir grains were first homogenized in 2% sodium citrate, and 2 ml of each homogenate was centrifuged for 10 min at 10,000 g. Total DNA from the pellets was extracted and purified using the FastDNA Spin kit (QBIOgene, Carlsbad, CA, USA) according to the manufacturer’s instructions. The DNA obtained was quantified using a Qubit flourometer apparatus (Invitrogen Detection Technologies, Eugene, OR, USA).
Raw sequences were processed through the Ribosomal Database Project (RDP) pyrosequencing pipeline (http://wildpigeon.cme.msu.edu/pyro/index.jsp). Sequences were excluded from the analysis if they were of low quality, if the read length was less than 300 bp, or if one of the primer sequences was missing. The high-quality partial 16S rDNA sequences were submitted to the RDP-II classifier using an 80% confidence threshold to obtain the taxonomic assignment and the relative abundance of the different bacterial groups (Wang et al., 2007). Multiple sequence alignments for each sample were made using the Aligner tool (default settings) on the RDP website. These alignments served as inputs for constructing the distance matrix and for clustering the sequences into operational taxonomic units (OTUs) using MOTHUR v. 1.14.0 software The clusters were constructed at a 3% dissimilarity cut-off and served as OTUs for generating predictive rarefaction models, and for determining the ACE and Chao1 richness and Shannon diversity indices (Shannon and Weaver, 1949). The MOTHUR program was also used to perform the Fast UniFrac test, which was employed to compare the phylogenetic structure of the libraries and to generate the Venn diagrams. A neighbor-joining tree was constructed with representative sequences of each OTU selected by MOTHUR. These sequences were compared against the RDP database using the Seqmatch option to select for the nearest neighbours. All sequences were then aligned using MEGA 5.0 software and the Jukes–Cantor model. The equivalent sequence of the archaea Halococcus saccharolyticus (AB004876) was used as an outgroup to root the tree.
DGGE fingerprints of the microbial communities in the three Brazilian kefir grains were rather simple, containing from one to five different bands (Fig. 1, panels A through D). Most bands were shared by all three samples. The species profile of the total bacteria amplified with the universal primers involved up to five bands reflecting three different species (Fig. 1, panel A). Bands corresponding to Lb. kefiranofaciens (bands 1, 2 and 5) and to Lb. kefiri (band 4) were found in all samples. An additional band present in sample AV (band 3) was identified as Lc. lactis. The same three species were also found using the group-specific primers for lactobacilli and lactococci ( Fig. 1, panels C and D, respectively). The DGGE fingerprints of the yeast community were also simple and similar for the three grains. A high-intensity band was present in all samples, and was identified as S. cerevisiae (band 6, Fig. 1 panel B). A low-intensity band corresponding to K. unispora was revealed in kefir grain AD (band 7, Fig. 1 panel B).
DGGE profiles of the microbial community from three Brazilian kefir grains (samples AR, AD and AV). Panel A: DGGE profile of the eubacterial 16S rRNA gene obtained with universal primers (1) Lactobacillus kefiranofaciens; (2) Lactobacillus kefiranofaciens; (3) Lactococcus lactis; (4) Lactobacillus kefiri; (5) Lactobacillus kefiranofaciens. Panel B: DGGE profile of the eukaryotic domain D1 of the 26S rRNA gene (6) Saccharomyces cerevisiae; (7) Kazachstania unispora. Panel C: DGGE profile of the 16S rRNA gene obtained with specific primers for the lactobacilli group (8) Lactobacillus kefiranofaciens; (9) Lactobacillus kefiri. Panel D: DGGE profile of the 16S rRNA gene obtained with specific primers for the lactococcus group (10) Lactococcus lactis.
A total of 14,314 high-quality partial 16S rDNA sequences, longer than 300 bp, were obtained by pyrosequencing analysis, including 2641 from sample AD, 2690 from sample AR and 8983 from sample AV. Diversity richness, coverage, and evenness estimations calculated for each data set are presented in Table 1. Rarefaction curves showed similar patterns for all samples (Fig. 2), and suggested that the bacterial community was well represented since they became flatter while the number of sequences analyzed increased. Additionally, when re-sampling analyses were performed, normalizing by sample size, the rarefaction curves proved to be saturated (Fig. 2 panel B). Moreover, the coverage at the 97% similarity level was above 0.99 for each of the kefir grains. According to Fig. 2, and the OTU richness estimated by ACE and Chao1 indices (Table 1), sample AV possessed greater species richness than the other grains at 97% similarity level. A similar finding was observed when considering the microbial diversity estimated by the Shannon index at 97% similarity level. Indeed, 14, 18, and 46 OTUs were associated with kefir samples AR, AD, and AV respectively (Table 1).
Estimated OTU richness, sample coverage and diversity index of 16S rDNA libraries of kefir grain samples.
Estimated OTU richness
42.24 (28.17; 75.78)
54.00 (28.27; 144.19)
0.49 (0.45; 0.53)
38.57 (23.67; 76.40)
24.50 (16.03; 68.19)
0.33 (0.29; 0.37)
148.02 (109.96; 208.74)
82.14 (58.65; 149.23)
0.70 (0.67; 0.72)
Abbreviations: ESC, estimated sample coverage; NS, number of sequences for each library; OTU, operational taxonomic unit.
Values in brackets are 95% confidence intervals as calculated by MOTHUR.
a. Calculated by MOTHUR at the 3% distance level.
b. Shannon diversity index calculated using MOTHUR (3% distance).
c. ESC: Cx = 1 − (Nx/n), where Nx is the number of unique sequences and n is the total number of sequences.
Rarefaction curves of partial sequences of the bacterial 16S rRNA gene from Brazilian kefir grains (AD, AR and AV) at a 97% similarity level (A) and rarefaction curves normalized with respect to sample size (B).
The Unifrac test further revealed that sample AV was significantly different from AD and AR (p < 0.01) when the relative proportion of sequences from each community was taken into account (Weighted Unifrac algorithm).
To evaluate the distribution of OTUs among the different kefir grains, a Venn diagram was constructed (Fig. 3). This showed that 11 OTUs, embracing 95.8% of the sequences, were common to all three grains. Further, despite the larger number of specific OTUs in the AV sample (24 OTUs), the incidence of such grain-specific sequences (3.86%) was much lower than those shared by all samples (95.8%). Similarly, specific OTUs of the other two samples were represented by a low percentage of sequences.
Venn diagram showing specific and common OTUs in the Brazilian kefir grains AD, AR and AV, and the percentage of occurrence of the total sequences (in parentheses).
The bacterial sequence reads were grouped into three different phyla: Firmicutes, Actinobacteria, and Proteobacteria. Of these, Firmicutes was the most abundant phylum, and was dominated by members of the class Bacilli belonging to the order Lactobacillales. Three families were found among the sequences belonging to this order: Leuconostocaceae, Streptococcaceae and Lactobacillaceae. The family Lactobacillaceae was predominant in all three grains, and was represented by only one genus, Lactobacillus, which accounted for 99.7, 93.9, and 99.6% of the reads for grains AR, AV and AD respectively. In the family Streptococcaceae, the genus Streptococcus comprised only 0.01% and 0.04% of all sequences identified in grains AV and AD respectively, whereas the genus Lactococcus was detected only in kefir grain AV (4.87% of the reads). The genus Leuconostoc also occurred at a low level in samples AV (0.12%) and AD (0.23%). Few sequences were assigned to the phylum Proteobacteria, which accounted for 0.3% of the total assigned sequences for grain AR, 1% for AV and 0.04% for AD. The sequences of this phylum belonged to the genus Acetobacter in sample AR (0.26%) and AD (0.04%), and to the genus Pseudomonas (0.99%) in sample AV. Phylum Actinobacteria was represented by reads belonging to the genus Solirubrobacter in grain AR (0.04%) and the genus Bifidobacterium in grain AV (0.02%).
Relative abundance at family level, based on the classification of partial 16S rDNA sequences of bacteria from the Brazilian kefir grains AD, AR and AV using RDP-Classifier.
Because of the low diversity found, unique representative sequences from each OTU were selected and used to construct a phylogenetic tree. The different sequences were manually compared against the RDP database and further aligned with up to three of their nearest sequences in the database. The majority of the OTUs represented close phylogenetic lineages of Lactobacillus spp. commonly reported in kefir grains. These alignments and manual investigations further allowed the classification of the reads in a number of Lactobacillus species and subspecies, including among others Lb. kefiranofaciens subsp. kefirgranum, Lb. kefiranofaciens subsp. kefiranofaciens, Lb. kefiri, Lb. parakefiri, Lactobacillus parabuchneri, Lactobacillus amilovorus, Lactobacillus crispatus and Lactobacillus buchneri. Sequences identified as Lc. lactis subsp. cremoris were revealed in kefir sample AV.
The microbial diversity of kefir grains of different origins has been analyzed using both culturing (Simova et al., 2002, Witthuhn et al., 2005, Mainville et al., 2006, Chen et al., 2008, Wang et al., 2008 and Miguel et al., 2010) and culture-independent techniques (Garbers et al., 2004, Wang et al., 2006, Wang et al., 2008, Ninane et al., 2007, Chen et al., 2008, Miguel et al., 2010 and Dobson et al., 2011). In the present study, two culture-independent techniques were used to evaluate the microbial diversity and community structure of three different kefir grains from different locations in Brazil. The dominant populations were identified using PCR-DGGE, while the next-generation sequencing technology of pyrosequencing allowed a more complete view of the grain communities’ overall composition.
As in previous studies (Garbers et al., 2004, Chen et al., 2008, Jianzhong et al., 2009, Miguel et al., 2010 and Magalhães et al., 2010), the bacterial PCR-DGGE profiles were shown to be composed of a small number of bands. These corresponded to several Lactobacillus species that have always been reported as prevalent. Lb. kefiranofaciens ( Chen et al., 2008, Jianzhong et al., 2009 and Magalhães et al., 2010) and Lb. kefiri ( Miguel et al., 2010) have been described as accounting for the more intense DGGE bands. These two bacterial species have also been reported as dominant by culturing in different kefir grains ( Mainville et al., 2006, Chen et al., 2008 and Miguel et al., 2010). The small number of DGGE bands seen for the yeast profile has also been reported for many other kefir grains ( Garbers et al., 2004, Wang et al., 2008, Jianzhong et al., 2009 and Magalhães et al., 2010). The dominant yeasts found in these earlier investigations were Saccharomyces spp., Kluyveromyces lactis, Kazachtania spp. and Candida spp. ( Garbers et al., 2004, Wang et al., 2008 and Jianzhong et al., 2009). S. cerevisiae was the main yeast species detected in the present work. This and other related species have also been identified as majority via culturing ( Simova et al., 2002 and Latorre-García et al., 2007).
Pyrosequencing is becoming the state-of-the-art technique for the analysis of microbial populations from different ecosystems. It has been used to study several types of food fermentation (Humblot and Guyot, 2009, Roh et al., 2010 and Jung et al., 2011). Indeed, one report exists in which a kefir grain and its fermented milk were analyzed by this technique (Dobson et al., 2011). The present pyrosequencing analysis revealed the phylum Firmicutes to be strongly dominant in the examined grains, accounting for more than 99% of the sequences. This phylum is composed of a group of low-GC-content Gram-positive bacteria, which includes LAB. Firmicutes was also found dominant in the study of the Irish kefir, in which both the interior and exterior of the grain were analyzed ( Dobson et al., 2011). These authors also showed that all other phyla detected (Actinobacteria, Proteobacteria and Bacteriodetes) were minor components of the kefir community in the interior of the grain. Within the phylum Proteobacteria, Pseudomonas spp. was identified in the grain AV, which has been suggested to be an environmental contamination ( Dobson et al., 2011). The genus Acetobacter (Proteobacteria subgroup) was found in only two of the Brazilian grains (AR and AD). Although AAB have often been mentioned as one of the main components of the bacterial population of kefir grains ( Rea et al., 1996, Garrote et al., 2001 and Miguel et al., 2010), they have in fact only occasionally been detected. ( Garbers et al., 2004, Chen et al., 2008, Jianzhong et al., 2009, Miguel et al., 2010 and Dobson et al., 2011).
In general, the two techniques used in this study were consistent with respect to the detection of the predominant bacteria. However, some microorganisms identified by pyrosequencing were not detected by DGGE analysis, probably because they were part of minority populations in the grains. This limitation of the PCR-DGGE method was previously noted by Ercolini (2004), who reported that minor bacterial groups in complex communities may not be represented in the DGGE profiles. The present results show that pyrosequencing allows the detection of microorganisms that are not part of the dominant community such as bifidobacteria. These minor microorganisms may contribute to the particular sensory characteristics of the kefir beverage fermented by each grain (fizziness, acidic taste, and refreshing flavour), via the production of metabolites such as organic acids, ethanol and aromatic compounds.
Traditional culturing and molecular techniques have indicated that a few specific microbial genera and species may be constantly present in kefir grains, whereas others may or may not occur (Simova et al., 2002, Witthuhn et al., 2005, Mainville et al., 2006, Wang et al., 2006, Wang et al., 2008, Ninane et al., 2007, Chen et al., 2008, Miguel et al., 2010 and Dobson et al., 2011). Further, as Farnworth and Mainville (2008) have noted, the list of bacteria and yeasts in kefir grains should not vary significantly from one part of the world to another if good care, similar growth conditions, and proper sanitary conditions are maintained. However, over time and under different growing conditions, kefir grains may change their microbial make up and fermentation properties. In the present work, even considering that the three kefir grains may have been grown in different locations over the years, the dominant microbiota was similar; only the minority communities varied. These small microbial differences may be associated with distinctive grain-specific sensory profiles (Pintado et al., 1996, Rea et al., 1996 and Simova et al., 2002).
Two culture-independent methods were used to evaluate the microbial diversity of three Brazilian kefir grains: PCR-DGGE and pyrosequencing. Both techniques showed Lb. kefiranofaciens to be dominant, while DGGE showed S. cerevisiae to be the main eukaryotic microorganism. Pyrosequencing analysis also allowed the identification of minor bacterial components. For the complete description of the microbial communities of the kefir grains, pyrosequencing analysis using specific primers for eukaryotic and archaea organisms should also be performed.
The study was supported by the CAPES Foundation (process number PDEE 5019109) and a project from the Spanish Ministry of Science and Innovation (MICINN) (reference AGL2007-61869-ALI). S. Delgado was supported by a research contract from MICINN under the Juan de la Cierva Program (reference JCI-2008-02391).
See complete paper on site for complete list of references.
Copyright © 2012 Elsevier Ltd.
Got Culture Questions? Call the Culture Hotline...
Please Call 1+ 231.269.3261 Between 4PM - 8PM, GMT -5/EST. Copyright 2010-2016 Organic Cultures - Water Kefir Grains, Dairy Kefir Cultures, Kombucha/JUN strains, Japanese culture
spores for home food culturing.
Copyright protected under state, national, and international laws. All rights reserved. Images may be used/copied if www.organic-cultures.com
is quoted or linked as the source.
Contact webmaster at: webmaster (at) organic-cultures.com