Introducción

Metagenomics is based on the use of molecular biology techniques to analyze the diversity of microbial genomes, also called metagenomes, from environmental samples. The microbial diversity of metagenomes has been analyzed using the 16S rRNA gene, which encodes for the ribosomal RNA that forms the small subunit of the ribosomes. This gene comprises preserved and variable regions in bacteria and archaea. The 16S rRNA gene has been used as a molecular marker, since it allows the classification of bacteria and archaea into taxonomic groups according to families or genera.

The first studies of microbial diversity in environmental samples were carried out using culture-dependent methods, where only those microorganisms that could be isolated in the laboratory were studied. Through the advance of molecular biology techniques, it has been possible to analyze microbial diversity through the use of independent culture methods, obtaining more precise information about bacterial genomes. One of the most widely used methods is PCR amplification of 16S rRNA gene fragments, in some cases followed by denaturing gradient gel electrophoresis (DGGE). These techniques have been used to analyze ruminal bacterial diversity, changes in the microbial community, and gene expression after changes in the ruminants’ diet^1,2. Another advance that has allowed a broader analysis of microbial diversity in the rumen is the targeted sequencing of the variable regions of the 16S gene in order to differentiate microorganisms that are phylogenetically very close, analyze the genes and genomes that degrade the biomass in the rumen, characterize the rumen microbiota, and study the effects of yeasts on bacterial diversity in the rumen^3,4,5.

The recent development of metagenomics has allowed the study of microbial diversity in environmental samples by isolating and analyzing the total genetic material present in an environmental sample^6,7. At the beginning, this strategy was used to search for new enzymes with biotechnological potential, extracting the total DNA contained in an environmental sample, fragmenting it and cloning genes of different size in vectors such as plasmids (15 kb), phages (up to 20 kb), phosmids and cosmids (up to 40 kb), as well as Artificial Bacterial Chromosomes (for larger fragments). These vectors were inserted into different host strains, and fluorogenic substrates were used as expression indicators. However, in the functional search for genes through clones, protein expression and enzymatic activity were of a small magnitude^8,9,10.

A crucial part in the construction of metagenomic libraries is the extraction of the nucleic acids from the sample. There are two main strategies for metagenomic DNA extraction: chemical treatment and direct lysis with mechanical methods. Both methods have advantages and disadvantages. DNA of greater microbiological diversity is recovered with mechanical lysis than with chemical treatment; however, chemical treatment allows obtaining DNA of greater molecular weight. Regarding RNA extraction, the same extraction methods are used for any expression analysis in which RNAsase inhibitors are included, and it is recommended to freeze the samples at -80 ºC immediately after collection to avoid RNA degradation⁹.

To select the ideal extraction method, the type of sample, the nucleic acid to be purified and the type of analysis to be performed must be taken into account. Different strategies have been used for metagenomic analysis. Within the mechanical methods, magnetic beads have been used for oral, dermal or fecal samples, as well as samples of soil and water, from which high-quality sequences have been obtained¹¹. For the analysis of ruminal microbiomes, methods combining magnetic bead extraction (mechanical lysis) with extraction columns (chemical treatment) have been used to purify ruminal microbial DNA^11,12. This combination increased extraction performance over the use of separate magnetic beads and extraction columns. Other identification methods use Stable Isotope Probing (SIP), which identify the microorganisms that incorporate these isotopes through the use of marked substrates. In particular, the nucleic acid stable isotope probe technique (Nucleic acids-SIP) uses substrates with ¹³C and/or ¹⁵N isotopes, which are incorporated into bacterial genomes and can thus be traced¹³. Other substrates with stable isotope probes are 13CH3-OH, 13C-phenol and 5-bromo-2-deoxyuridine. However, limitations of the use of substrates marked with stable isotopes include the crosslinking and recycling of the isotopes within the microbial community, resulting in the loss of specific enrichment of the analyzed microorganisms¹³.

Techniques have also been developed to identify genes that change their expression levels during various biological processes. For example, Suppression Subtractive Hybridization (SSH) has been used to identify variations between complex DNA samples such as those in the ruminal environment^9,13. Differential expression analysis allows to compare the gene expression profile of a microbial community before and after being exposed to a specific condition and/or substrate and, thus, to identify important genes that exhibit changes in gene expression profiles due to the effect of such condition and/or substrate¹³. Another technique that is widely used in gene expression studies is microarrays, which offer the advantage of rapidly identifying and characterizing a large number of clones. Although microarrays can be used to identify a large number of conserved genes, they depend on known sequences previously reported in databases, thus eliminating the possibility of identifying new genes^8,10. More recently, mass sequencing has been used to obtain as much information as possible about the metagenome present in a sample. One of the first works with massive sequencing was the identification of the metagenome of the Sargasso Sea, where 1,045 trillion base pairs of non-redundant sequences were generated, noted and analyzed in order to identify the genetic content, diversity and relative abundance of the microorganisms. It was estimated that the data obtained came from at least 1,800 genomic species that included 148 phylotypes of unknown bacteria and more than 782 genes never before described that code for rhodopsin-like photoreceptors^10,14.

The massive sequencing of the metagenome by "shotgun" has the characteristic of sequencing all the DNA present in the sample so that the microorganisms can be classified taxonomically up to the species level. Furthermore, with the sequences obtained by this type of sequencing, genes with functions never before described can be discovered, and even sequences belonging to the 16S rRNA gene can be selected for taxonomic annotation. These classifications are made with the use of bioinformatic tools that search for homology with the sequences analyzed in different existing databases¹⁵. Specifically in ruminal environments, metagenomic libraries have been analyzed in order to evaluate the effects of diets on ruminal microbiome by means of metagenomic profiles, and the 16S rRNA gene marker has been used to determine and classify the microbial diversity of the sequences^3,5. However, some of the sequences of these samples have not been adequately classified; therefore, using at least one molecular phylogenetic marker other than the 16S rRNA gene may improve taxonomic classification¹⁵.

In the present work it was reviewed the tools used for metagenome analysis, ranging from classical molecular markers to those used with data obtained from massive sequencing, with an emphasis on metagenomes from ruminal environments.

Molecular markers for metagenomic analysis

A molecular marker is a segment of DNA that corresponds to a non-coding gene or regions of the genome, these segments of DNA allow different variants (alleles) to be identified and are located at a particular site on the chromosomes (locus). The differences obtained in these DNA fragments are known as polymorphisms and can be detected by hybridization of nucleic acid sequences, nucleotide sequencing, comparison of the length of the fragments produced by the polymerase chain reaction (PCR) and through sites recognized by restriction enzymes. Molecular markers can be used to classify taxonomic groups, populations, families or individuals in both eukaryotes and prokaryotes^16,17. Various molecular markers have been used in genetic studies in domestic animals, in wildlife, in endangered species, and in forensic and paternity tests. The best known are RFLPs, mini-satellites, AFLPs, RAPDs, microsatellites and SNPs (Table 1).

The most relevant characteristics that molecular markers must have in order to optimize metagenomic studies include (1) that they are single copy genes (genes that have only one or two copies in the entire genome), as they provide less uncertainty than markers for genes with multiple copies (genes with repeated copies in the genome); (2) that the sequence of the marker gene is easily aligned to facilitate phylogenetic analysis; (3) that the proportion of the gene replacement region is sufficient to provide information needed for classification; (4) that primers are selective to amplify the marker gene, but not universal, in order to avoid false positives; (5) that there is no excessive variation in the marker sequence that limits the determination of ancestry. The genes that are used as molecular markers to classify microorganisms are described below.

The use of mass sequencing in metagenomics

Although metagenomic analysis started with the use of different molecular markers such as AFLP, RAPDs, 16S rRNA etc. (Table 1), some of these markers have been observed to improve their efficiency when the technique used to identify them includes their sequencing instead of characterizing them by means of reactions with restriction enzymes and/or amplification by PCR. From its inception, DNA sequencing with Sanger's technology has had a major impact on virtually every branch of the biological sciences, including microbial community studies. Currently, the use of Sanger sequencing can generate up to 96 sequences per run with an average length of 650 bp, which may be sufficient for phylogenetic marker analysis¹⁵. This type of study is known as first generation sequencing and results in high quality sequences of a length between 500 and 1,000 bp. However, its disadvantage is that the proportion of molecular markers that can be sequenced in a run, compared to the total number of microorganisms present in a metagenomic sample, is very low¹¹.

With the emergence of mass sequencing technologies, known as "Next Generation Sequencing technologies (NGS)" millions of DNA molecules can be sequenced simultaneously, which greatly facilitates the study of microbial diversity¹⁵. One of the first high-throughput sequencing technologies was 454 pyrosequencing, which was used for targeted sequencing of ribosomal RNA gene amplicons²⁹. This technique had the advantage of allowing the obtainment of sequences of up to 1,200 bp, albeit with a significantly higher error than other sequencing platforms (1%) and at a higher cost¹⁵. Second generation sequencing, also known as short reading sequencing (50 to 400 bp) uses mainly the Illumina platform¹¹. Among its advantages, it is worth mentioning that it allows a greater number of readings, with an approximate error rate of 0.1% and at a comparatively low cost¹⁵. It is currently the most popular technology, but it requires a more complex bioinformatic analysis phase than other platforms.

Traditionally, when these two platforms (454 pyrosequencing and Illumina) are used for metagenomic analysis with the 16S rRNA marker, a previous amplification step by PCR is performed, limiting the identified species to bacteria and archaea only, since the primers will always be used for amplifying fragments of the 16S rRNA gene. If the population also includes eukaryotic microorganisms such as yeasts and protozoa, they cannot be detected. On the other hand, this step of amplification by PCR entails an enrichment of the DNA which produces a bias towards the species that are found in greater proportion causing that the species that are found in smaller percentage to become hard to detect. Finally, this type of analysis identifies microorganisms down to the gender level²⁹.

An alternative for increasing resolution at the taxonomic level lies in the metagenomic study with the mass sequencing techniques called "Whole-Genome Shotgun sequencing" (WGS) and "Shotgun metagenomics sequencing (SMS)", in which the total metagenomic DNA is sequenced^30,31. The major advantage of these methods is that microorganisms can be classified down to the species level and that not only prokaryotes but also eukaryotes can be identified; also, it does not require the previous amplification step by PCR, and therefore the bias is eliminated. Another advantage of these sequences is that by having sequences of all the DNA present in the sample, those corresponding to the 16S rRNA gene can be selected for use as taxonomic molecular markers; sequences of genes of other constituent polymorphic markers (MLSA) can also be sought in order to achieve a better classification of the microorganisms. The main disadvantages are that it has a higher cost than targeted sequencing of the 16S rRNA gene and requires more complex bioinformatic data analysis³². Several studies have been conducted to identify metagenomes in a wide range of population environments, using both 16S rRNA gene targeted sequencing and full metagenome sequencing with WGS and/or SMS.

Bioinformatic tools for metagenomic analysis

It is important to point out that bioinformatic tools must be used to analyze data obtained from massive sequencing. The greater the amount of data generated, the greater the need for bioinformatics resources¹⁵, both for applications implementing analysis algorithms and for databases with information on microbial genomes (Table 2).

One of the most used applications since its launch is the MG-RAST³³ server, which assigns functional annotations to the analyzed sequences by comparing them with protein and nucleotide homology databases, in addition to allowing phylogenetic analysis. This tool is free and easily accessible, and it is fed with information provided by researchers; therefore, it helps to end the main bottleneck in metagenome sequence analysis, which lies in the availability of information to assign genomic annotations³³. Two other widely used bioinformatic tools in metagenomics are MOTHUR³⁴, which is also freely accessible and which feeds on metagenomic information that users add to a database with monthly updates, and QUIIME³⁵, which is used for the analysis of microbial communities from bacterial and archaeal data.

Another software widely used for metagenome analysis is PhaME³⁶ (Phylogenetic and Molecular Evolutionary), which uses whole genome SNPs to measure interspecific diversity by phylogenetic analysis. PhaME³⁶ can be used to measure inter-species and inter-strain divergence and minimize errors in sequencing and assembly. Comparative genomics, including phylogenetic analysis based on ortho genes and SNPs, requires assembled or finished genomes. PhaME uses the SNP-based approach of complete genomes available in the databases, assembled sequences (contigs) and raw sequences to perform phylogenetic and molecular evolutionary analysis. This software combines algorithms for genome-wide alignment, reading mapping, and phylogenetic construction; it uses internal commands to infer the main genome and SNP, infer trees, and perform other molecular evolution analysis. PhaME is especially useful for the analysis and detection of organisms that are not very abundant in metagenome samples and has been used in data on bacterial samples, viruses, such as Ebola in Zaire, and yeasts, among others³⁶.

Other tools focus on the analysis of the hypervariable regions of the 16S rRNA gene, such as VITCOMIC1³⁷, which combines the information obtained from the targeted sequencing of the 16S rRNA gene as well as from the massive WGS or SMS sequencing to better visualize the phylogenetic composition of metagenomic samples, in addition to generating a more accurate record of the microbial community. Similarly, the 16SPIP³⁸ application has also been used for rapid detection of pathogenic microorganisms in clinical samples based on 16S rRNA metagenomic sequence data.

As for "predictive metagenomics" approaches, the PICRUSt³⁹ algorithm, which uses evolutionary models to predict metagenomes from 16S rRNA gene data and a reference genome database, should be highlighted. This tool has been used with data from soil microbiome samples, mammalian intestines, microbial mats, and humans³⁹, such as the human oral microbiota study which analyzed 6,431 samples of the 16S rRNA gene from the Human Microbiome Project^39,40.

Examples of metagenomic characterization with high throughput methodologies

Several metagenomic characterization works have been carried out to identify microorganisms living in environments of interest due to their great variability and ecological importance (Table 3). The following are a few examples of these works, without being exhaustive. For example, a massive sequencing of 29 metagenomes from samples from three marine stations that are part of the global Tara expedition was performed²⁹. The taxonomic analysis carried out with the sequence data corresponding to the 16S rRNA gene made it possible to identify all the variable regions of the gene (V1 to V9). Targeted sequencing of the 16S rRNA gene was also performed for comparative purposes. The results obtained indicated that the efficiency in taxonomic classification with the use of ribosomal database RDP (Ribosomal Database Project) is similar for both types of sequencing. However, massive sequencing offers two major advantages: it reduces the error caused in amplicon PCR and it generates a large amount of functional data that can be analyzed along with the taxonomic analysis.

Another metagenomic work in the field of mass sequencing focused on analyzing the diversity and bacterial distribution present in sediments of the tropical mangrove of Sundarban⁴¹. For this identification, it was used the 16S rRNA directed sequencing through 454 pyrosequencing, obtaining a total of 153,926 sequences. The analysis with MG-RAST software made possible the identification of 56,547 species belonging to 44 different phylotypes, being the most dominant the phylotype Proteobacteria. On the other hand, metagenomic analysis of sediments from the Arabian Sea⁴² with Sanger 16S rRNA sequencing classified the sequences obtained into seven different phylotypes where the phylotype Proteobacteria also predominated.

A large number of papers have focused on the characterization of metagenomes from extreme environments. For example, sequencing of 16S rRNA and complete genomes has been used to identify the diversity of thermophilic bacteria present in thermal waters in Malaysia whose temperature varies between 50 and 110 ºC⁴³. An analysis of the 16S rRNA data identified approximately 35 phylotypes, of which Firmicutes and Proteobacteria represented 57 % of the microbiome. As for thermophiles, 70 % of those detected were strictly anaerobic; however, Hydrogenobacter spp. (forced chemolithotrophic thermophilotypes) constituted one of the most frequent taxa, and a large number of thermophilic photosynthetic microorganisms were found as well. Most of the identified phylotypes coincided with the findings of the sequencing of complete genomes. Thanks to this type of analysis, it was possible to identify and classify extreme microorganisms, such as thermophilotypes, anaerobes and chemolithophytes, that would have been difficult to characterize with classic microbiological methods⁴³.

Another study for identifying microbiota from extreme environments was conducted from samples of microorganisms that grow in the fungi that inhabit Yellowstone Park through the directed sequencing of 16S rRNA⁴⁴. Over the years, the study of microorganisms in this habitat has focused on chlorphototrophic bacteria belonging to the Cyanobacteria and Chloroflexi. However, the results of the study revealed that microbial variation is dominated by a single taxon: Roseiflexus spp. which belongs to the group of anoxigenic phototrophic microorganisms⁴⁴. Targeted 16S rRNA sequencing, along with full genome sequencing, has equally been used in glaciers, for which microbial information is also very limited. The first reported metagenomic study of glaciers⁴⁵ identified nine different genomes, including Anaerolinea, Synthrophus and Thiobacillus, and metabolic pathways involved in sulfur oxidation and nitrification were identified.

There are examples of the use of mass sequencing in metagenomic populations within the health and agro-food sectors. As an example within the field of human health, studies of directed sequencing of 16S rRNA to describe the microbiota present in the blood of healthy individuals have shown that this body fluid is not a sterile tissue⁴⁶. At the phylotype level, more than 80% of the microorganisms present in the blood belonged to Proteobacteria, although phylotypes of Actinobacteria, Firmicutes and Bacteriodetes were also found. Ranjan et al.³² used different strategies to characterize the human fecal microbiome. From a single sample they obtained 194.1 x106 readings from different sequencing strategies (16S rRNA directed sequencing, Illumina HiSeq, Illumina MiSeq). When comparing these, especially the 16S rRNA gene directed sequencing with the WGS sequencing, they concluded that the latter has more advantages, as it increases the ability to identify bacterial species and the detection of diversity and gene prediction, and it also improves the accuracy of species detection by increasing the length of the sequences.

In the agro-food field, directed sequencing of 16S rRNA has also been used to identify microorganisms present in Gouda cheese⁴⁷ whether prepared with pasteurized or unpasteurized milk, and to evaluate changes due to the effect of aging. This study identified 120 genera in unpasteurized cheese and 92 in pasteurized cheese. In addition, depending on the aging time, it had a significant influence on the presence of microbiota. The most abundant genera in all samples were Bacillaceae, Lactococcus, Lactobacillus, Streptococcus and Staphylococcus.

In the case of growing broilers, the variation of ileal and cecal microbiota through time has been studied⁴⁸. In order to do this, the hypervariable V3 region of the 16S rRNA gene was amplified and sequenced. The results showed that the cecal microbial communities were more diverse than the ileal ones. In addition, the presence of (potentially pathogenic) Clostridium bacteria was observed to increase as the animals grew and that the population of beneficial microorganisms such as Lactobacillus was low in all intervals⁴⁸.

In the case of ruminal metagenomes, it should be noted that one of the first sequencing studies was conducted to search for cellulolytic enzymes never before described³. In this study, 454 pyrosequencing was performed, obtaining 268 gigabases of metagenomic DNA information. From this information, 27,755 supposed genes of carbohydrate-active enzymes were identified, of which 90 codified for possible proteins, and 57% of them were enzymatically activated by cellulosic substrates. Another study focusing on ruminal metagenome in dairy calves and beef cattle steers²⁰ used 16S rRNA targeted pyrosequencing to assess population variation according to the type of livestock. This study found 8 phylotypes, 11 classes, 15 families and 17 different genera, and differences in the abundance of phylotypes found between dairy and beef cattle. The most abundant phylotypes were Bacteriodetes, Firmicutes, Proteobacteria, Fibrobacteres and Spirochaetes in both types of cattle, but with a lower abundance of Bacteriodetes and Proteobacteria in beef cattle. The use of yeast as a nutritional additive in cattle is known to improve milk production and weight gain. However, there is no knowledge of whether the effect caused by yeasts is a general stimulus to all microbial species or only affects some of the ruminal environment. Due to the above, a study was conducted to evaluate the changes in the rumen microbiota when the animals were fed with a yeast additive compared to when they consumed only the basal diet⁵. In this work, 454 pyrosequencing of the V1 region of the 16S rRNA gene was used to identify the population of ruminal microorganisms. The results showed that a change was observed in the main fibrolytic bacteria (Fibrobacter and Ruminococcus) and in lactate-using bacteria (Megasphaera and Selenomonas) when the yeast additive was added. Targeted sequencing of the 16S rRNA gene in the adult dairy cattle ruminal microorganism population when combining thiamine with high grain diets has been used to evaluate its effect as an additive in animal nutrition⁴⁹. The results confirmed that thiamine supplementation can improve ruminal function, as the number of cellulolytic bacteria increased when this amino acid was administered.

In the field of animal health, the sequencing of complete metagenomes has also been used. For example, skin metagenome with active and receding bovine digital dermatitis has been compared with the skin of healthy cattle to see if pathogens involved in the pathogenesis of the disease were detected³⁰. The sequences obtained were analyzed with MG-RAST and six main phylotypes were identified, among which Firmicutes and Actinobacteria predominated in the microbiome of healthy patients, while Spirochetes, Bacteroidetes and Proteobacteria were the most abundant in active and recession patients; this confirms that the presence of the disease changes the population of the metagenome.

Rumen metagenomic profiles have been obtained by sequencing complete metagenomes from samples of ruminal fluid from three different cattle and between different locations in the rumen³¹. In addition to comparing with the metagenome from feces of the same animals, the results indicated that the variation in metagenomic profiles was less among samples taken from the same animal, even if they were taken from different regions of the rumen. Contrary to expectations, no relationship was found with the metagenomic profile of faeces and ruminal fluid from the same animal.

Conclusions

Traditionally, metagenomic analysis used laborious methodologies, such as denaturing gradient gel electrophoresis, the digestion of genomes with restriction enzymes, and their visualization by means of agarose and/or acrylamide gels. The development of nucleic acid sequencing methodologies, especially new mass sequencing technologies, has helped to reduce this problem.

The 16S rRNA gene has traditionally been considered the gold standard for classifying prokaryotic microorganisms (bacteria and archaea), as it meets all the characteristics required to be a molecular marker. However, despite the large number of works that have used the sequencing of the hypervariable regions of this marker, it has the disadvantage of not being able to determine taxa at an infra-generic level. A strategy used to improve taxonomic classification has been the combination of the 16S rRNA marker with some other constitutive expression genes such as the genes sodA, hps65, gyrB, among others, and even genes encoding for subunits of the cytochrome c enzyme complex have been used to classify microorganisms into species.

In the last decade, mass sequencing technologies have made it possible for microbial populations to be analyzed in greater depth, either by sequencing the entire 16S rRNA gene, thus increasing the resolution of that marker, or by combining the information of that gene with the sequencing of complete metagenomes. In this last type of analysis, sequences of all the genomic material present in the sample are obtained, which offers the great advantage that in addition to making the taxonomic classification it is also possible to obtain functional information of the detected genes. Thus, despite the limitations of the required bioinformatic analysis, the use of these methodologies allows for more complete analyses.

However, despite the development of high-performance sequencing techniques, the targeted sequencing of 16S rRNA on the Sanger platform is not entirely obsolete, and the selection of the analysis strategy will depend on the objectives of the study, the degree of precision desired, the sample size and the financial resources that can be allocated by the research team. For example, if you are looking for the presence and/or absence of a single bacterial genus, Sanger sequencing would be ideal because it has the ability to sequence relatively large fragments with greater precision than any mass sequencing platform. If what is wanted is to discriminate between species of a single bacterial genus, two strategies can be utilized: the sequencing of some hypervariable region of the 16S rRNA together with some other constitutive gene (MLSA), or the sequencing of the whole gene in order to obtain the information of all the hypervariable regions.

Today, metagenomics faces numerous challenges arising from the large amount of information generated, its storage and the way in which it must be treated. Although many tools and applications have been designed for bioinformatic analysis of metagenomes, there is no single "protocol" of analysis; therefore, each study must be adapted to the nature of the samples and the objectives of the experiment.

In conclusion, microbial diversity studies will always use the 16S rRNA molecular marker to make taxonomic classifications, either through the sequencing of one or two of its hypervariable regions or through that of the whole gene, and it can even be combined with the use of another constitutive gene as a molecular marker to achieve a better taxonomic classification. On the other hand, mass sequencing technologies have greatly improved the study capacity and speed of metagenome analysis. This has occurred particularly in environmental samples with ecological importance, in both human and animal health, in studies on symbiosis of plants with endophytic fungi, and in the evaluation of ruminal metagenomes, to mention a few.