Developments and insights into the analysis of the human microbiome

Introduction

Studies of the human microbiome using extracted DNA can be divided in two main categories; the use of the entire DNA for sequencing is often referred to as shotgun metagenomics and offers a random representation of genomic sequences. Alternatively, one can focus on one or few marker genes and sequence only these amplified genomic regions. For this so-called targeted amplicon approach, the ribosomal small subunit RNA (16S rRNA) gene has emerged as the most used marker. This marker has the advantage of having highly variable regions interrupted by more or less conserved regions (Figure 1A), which allow the design of conserved primers. The present article is aimed at giving a short overview on recent studies on metagenomics and new developments in next-generation sequencing (NGS) technology.

Figure 1

Schematic view of the 16S rRNA gene together with an example of long amplicon sequence data achieved using the GS-FLX+ technology.

(A) The schematic view of the 16S rRNA displays the variable regions (V1–9) in orange, and the bars indicate a 100-bp span. The length and approximate position of the long amplicons are indicated with bars. (B) The distribution of the achieved read lengths for each amplicon is plotted against the relative number of reads. The read length achieved using GS-FLX/Junior (indicated in light blue) is compared with the read length achieved when using the GS-FLX+ technology when sequencing two different long amplicons. The exact length of the amplicons is indicated in the graph.

Developments in NGS methods

Before the use of NGS methods, amplicons were usually cloned and individually sequenced using the Sanger method, which can still be regarded as the gold standard. However, the development of new sequencing technologies had a big impact leading to the extensive usage of the Roche/454 and the Illumina technologies. Excellent reviews on this technological leap forward are already available [1, 2].

Pyrosequencing was the first NGS method widely adopted for metagenomics studies [3]. This method, which was developed by 454 Technologies (now part of the Roche group), offers a throughput of about 1 million reads with read lengths of 400–500 bp. A higher throughput of hundreds of millions of reads can be achieved when applying the sequencing-by-synthesis technology developed by Illumina/Solexa. This technology has often been used mainly for the advantage of high throughput in shotgun sequencing, and the limitations resulting from short read lengths of 150 bp are widely accepted [4]. Recently, Illumina has enabled paired end reads of 250 bp to be sequenced on the smaller MiSeq platform (Illumina, San Diego, USA), which now also allows the sequencing of 400- to 500-bp amplicons. Another recent development is the upgrade of the Roche/454 FLX-Titanium technology to the FLX+ technology (Branford, USA), which enables reads of more than 1 kb in length for shotgun sequencing. We have assessed the usability of this method also for long amplicon sequencing showing that long reads of over 900 bp can be achieved (Figure 1B, data not published). Using this technology, one can combine the advantages, which long reads bring about regarding bioinformatic analysis with the high throughput produced by NGS. Another sequencing technology, the Ion Torrent semiconductor technology from Life Technologies (Carlsbad, USA) has also been used for metagenomic analyses [5], but only recently, its read length increased to 400 bp. The single molecule sequencing technology developed by Pacific Biosciences (Menlo Park, USA) has been used for the identification of bacterial isolates [6] enabling sequencing reads of more than 3 kb, but its low accuracy of 85% and lower throughput are hindering its usability for metagenomic samples.

Despite its high adoption, there are some drawbacks preventing the 16S amplicon sequencing from becoming the method of choice for all metagenomic analyses. The design of primers, which are able to amplify all known species remains challenging [7], therefore, optimal primers have to be carefully chosen and tested. Mosaicism through horizontal gene transfer may not be reported by the 16S region [8], and whole-genome sequencing is still necessary in such cases. Additionally, similarity between 16S sequences of two species can be nearly 100%, resulting in misinterpretation/missing information. Shotgun sequencing can overcome these problems, and functional gene grouping can even reveal additional information [9]. Nevertheless, sequencing highly diverse samples like those of the gut microbiome, besides still being expensive, results in large amounts of data and, therefore, still cannot be the method of choice for fast routine analyses. The analysis of the huge amount of data emerging from shotgun or amplicon metagenomic studies remains challenging. More information on recent solutions for this issue can be found in the reviews of Thomas et al. and Kuczinski et al. [1, 10], which describe most of the common bioinformatic methods used to gain insight into this vast and complex amount of data.

Microbial diversity of healthy humans

The human body is the habitat of numerous microbes, which occupy different niches. Beginning with our birth, the colonization of our body starts and is subject to continuous change during a lifetime. Various factors are shaping the microbial composition, but numerous analyses have revealed the four bacterial phyla Actinobacteria, Firmicutes, Proteobacteria, and Bacteriodetes as being the most abundant [11, 12]. The variation in the microbiota compared between individuals and also within one individual is impressively large as it has been determined by large sequencing projects like the Human Microbiome Project [12] and Metagenomics of the Human Intestinal Tract (MetaHIT) [13]. Given the exponential growth of data and knowledge acquirement, which was recently achieved, only some general aspects of the healthy human microbiome can be presented in the following sections.

The gut microbiome consists of about 10¹⁴ bacterial cells, thus, being the biggest community of the human microbiome. This amount of cells, which is 10 times the number of human cells in the body, also harbors a combined gene pool, which is two orders of magnitude greater than the human gene set [12]. It is, therefore, legitimate to regard it as an additional organ with vast metabolic and biochemical abilities. The implication of these bacteria in several aspects of host biology, with effects, which are sometimes essential, are being increasingly studied and reviewed [14].

Although the initial inoculum transmitted from the mother during and after birth has a big impact on the initial microbiota of the child, recent studies have shown that its gut microbiome is no more similar to that of its mother than to that of its biological father. Moreover, the microbiomes of genetically unrelated but cohabiting mothers and fathers were significantly more similar to one another than to members of different families, which indicates that the constant environmental exposure is considerably shaping the gut community [15].

It is not surprising that most of the gut microbiota are anaerobic bacteria, most of which are belonging to the phyla Firmicutes and Bacteroidetes. The study of samples from different locations all over the world lead to the discovery of three enterotypes, but the factors driving the emergence of these different clusters are still unknown [16]. The approximately 500–1000 species inhabit niches shaped by different gradients found in the gastrointestinal tract, starting with the challenging stomach habitat exhibiting the lowest and ending with the colon having the highest abundance and diversity. Metabolically, some species facilitate the uptake of otherwise indigestible polysaccharides and produce essential vitamins. Additionally, they are required for the development and differentiation of the host’s intestinal epithelium [14]. Unlike the microbial taxa divergence, microbiome metabolic pathways are ubiquitous among individuals and body habitats. The most abundant among them are part of the essential pathways for host-associated microbial life, like ATP synthesis and the ribosome machinery. Only a few pathways are highly variable among subjects indicating a high degree of host-microbe and microbe-microbe interactions. Nevertheless, between 78% and 86% of genes annotated from assemblies could not be assigned a metabolic function and probably encode much of the uncharacterized metabolism of the metagenome [12]. Thus, many of these newly described genes and pathways remain to be explored in future metatranscriptomic studies.

The skin also harbors a very diverse microbiome, given the fact that as an ecosystem, it provides very variable niches within short distances. The characterization of 20 distinct sites revealed 19 bacterial phyla being present, with different species dominating the distinct sites [17]. For example, propionibacteria species and staphylococci species predominated in sebaceous sites, while corynebacteria species predominated in most sites, although staphylococci species were also represented. Recent investigations of the skin have also revealed the wide dispersion of fungal communities, which have not been described yet and, therefore, showed that microbial diversity is not limited to bacteria. Interestingly, most of the sites were dominated by fungi of the genus Malassezia, while three foot sites showed high fungal diversity [18]. The skin protects humans from invasion by pathogenic microorganisms, and therefore, such characterizations are of high interest and might provide insight into the delicate balance between skin health and disease.

The human microbiome consists of, and is also influenced by, viruses, most of them being bacteriophages but comprising also a substantial number of eukaryotic viruses. While lacking conserved genes, which could be used for the classification of all viruses, insights in the human virome were facilitated by the development of newer high-throughput sequencing technologies. Among other important discoveries, such studies could reveal that bacteriophages encode antibiotic resistance genes and other genes associated with bacterial metabolic pathways, showing that viruses serve as reservoirs for mobile genetic elements in bacteria [19].

Applications in medical microbiology

Today’s medical microbiology faces important tasks in clinical diagnostics, such as characterization and treatment of pathogens, as well as in epidemiology and public health, like the detection and tracking of outbreaks and the identification of new potential sources of infection. Many of the common standard techniques have been developed decades ago and are very specialized and species specific. They are often associated with a substantial delay and require a high degree of expertise of the microbiologist. Furthermore, they strongly rely on the cultivability of microorganisms. As the majority of bacterial species on earth is difficult to grow or does not grow at all in in vitro cultures, reliable, culture-independent alternatives for pathogen identification could facilitate and fasten the diagnostic process especially for culture-negative patient samples [20]. NGS-based metagenomic methods, which are already widely used in ecological microbiology for the community analysis of various environmental samples, e.g., from marine, soil, plant, or animal-associated origin [21–23] are entering the field of medical microbiology. Owing to limited read lengths, which have been obtained in the first years of NGS analyses, poor gene annotation and overestimates of bacterial richness and abundances were some of the technical problems. But over the last years, read lengths were extensively elongated, and the impression from the data became less distorted. Thus, these analyses can serve as highly valuable tools for diagnostic community assessment and are about to revolutionize the field of medical microbiology and its different application areas [24].

Future perspectives and conclusions

Regarding further future perspectives, metagenomic studies looking even beyond bacterial populations might become highly beneficial instruments for the analysis of other possible pathogens or health-influencing populations of the human body. In this context, viral metagenomics is a strongly evolving field. NGS can serve as a diagnostic technique for the identification of viruses and the characterization of the viral community in and on the human body [39]. Comparing the viromes of children with unexplained fever to afebriles by deep sequencing revealed an association of the amount of viral reads as well as viral diversity with the occurrence of fever. With such knowledge, adapted treatment recommendations beyond the default antibiotic therapy, which is often applied if a clear diagnosis is absent, can be made. In the mentioned study, advantages of both long read lengths and a high-sequencing depth were elucidated [40]. As viruses not only directly affect human health, but prokaryotic viruses also influence the bacterial community of the human body, the investigation of the phage community is beginning to gain researchers’ interest, too [12, 41]. However, the application of viral metagenomics in the medical field is still in its infancy [42].

Usage of metagenomic studies for the identification of fungi in humans, especially on the human skin, is also conceivable, although the current interest is only at a fraction of that for the bacterial microbiome. For fungal identification, internal transcribed spacer (especially ITS 2) regions are preferred targets, as they are composed of conserved and highly variable regions, similarly to 16S rRNA in bacteria [43]. Recently, fungal communities from different skin sites were analyzed by ITS 1 sequencing, as discussed above. The experience of this study demonstrated that especially physiologic attributes and topography have influence on bacterial and fungal communities on the human body surface [18]. But at the moment, fungal metagenomics is still at its infancy in the context of diagnosis and medical microbiology and is still very rarely used in human samples in comparison to bacteria-associated metagenomic studies.

Many of the metagenomic NGS-based studies in medical microbiology still only contribute to retrospective analyses. To include metagenomics in an effective way in a routine lab, some obstacles still need to be overcome. Reasonable limitations come from the costs. Although costs decreased for most NGS platforms in the last years, as competition on the market is increasing, still a high amount of money has to be spent per sequencing run [44, 45]. Also, the degree of standardization is still far from the needs of a clinical routine lab. Many steps between sample collection, DNA extraction [46], amplification [47], subsequent library preparation and even data analysis can lead to a high variability of results. To avoid this and to complement existing initiatives in data standards, there are first efforts to propose standards for sequencing library preparation, e.g., with regard to replicate experimental design to achieve data comparability [48], or by the use of standard materials [49] as well as by standardization of bioinformatic data analysis [50].

Furthermore, the time consumption and complexity of laboratory and bioinformatic techniques used for obtaining and analyzing data, which are still high at the moment, need to be reduced to make NGS-based metagenomics approaches accessible for routine labs. But as techniques are evolving fast, and in parallel, knowledge of the human microbiome is extensively growing, metagenomic studies will soon become a part of clinical routine, and thereby, the very conservative field of medical microbiology will undergo considerable evolutionary changes.