Proteomic interrogation of the gut microbiota: potential clinical impact

KEYWORDS:

• Metaproteomics
• microbial communities
• gut microbiota
• clinical implementation

1. Characterization of the gut microbiome

The gut microbiome plays a pivotal role in protecting the host against pathogenic microbes, in modulating immunity and in regulating metabolic processes [1]. Of clinical interest is not only which microbes are present, but more importantly their metabolic function. Microbes can greatly alter their metabolism in response to changing environmental factors, which can considerably impact the host and have implications in disease.

At present, most holistic approaches for analyzing the intestinal microbial community rely on 16S rRNA gene sequencing, a proven method to determine taxonomic composition of communities [2]. However, amplicon sequencing only provides information on who is present and not what they are doing. Metagenome sequencing can provide information on the potential function of the different taxa but still does not describe gene expression or translation of gene products. To overcome the lack of functional information from 16S rRNA gene sequencing without the need of sequencing the metagenome, an artificial metagenome is often constructed by collecting partial or whole genomes from the taxa identified by 16S rRNA gene sequencing from publically available depositories such as NCBInr or UniProt/Swiss-Prot [3,4]. These artificial metagenomes are used for functional analysis of the microbiome. However, the major drawback is the low resolution of 16S rRNA analysis since at species level there is still large diversity in genes present in different strains. For instance, 90,000 different gene families have been identified in all known strains of the model organism Escherichia coli, while a genome from a single strain only contains around 4000 gene families [5]. But even taking this into account, this predictive functional profiling method still suffers from the drawbacks of a conventional metagenome. Whole metatranscriptomics does reveal actual gene expression and therefore gives greater insights into functions in the microbial community, but not all transcripts are translated to proteins since posttranscriptional regulation and activation also affect function [6]. Furthermore, protein abundance is not only a product of protein synthesis but also a result of protein degradation which is of course ignored in metatranscriptomics. Metabolomics provides valuable information on the products/metabolites but is unable to link specific microbial taxa to certain metabolic functions, and consequently, only the metabolome of the community in its entirety can be analyzed [7,8].

Metaproteomics is potentially at the analysis level closest to reveal true function [9,10]. Identified proteins can be assigned to taxa as well as functions and therefore is an optimal method to investigate the different functions in the community, especially if posttranslational modifications are also analyzed. There are a number of challenges to be addressed mainly resulting from the high complexity of the microbial community. One problem is the low coverage of the theoretical expected complete metaproteome. Currently, only a few thousand protein groups can be precisely identified. This is, of course, only a fraction of the >1,000,000 gene-coding sequences expected in the human gut microbiome. The low number of identified peptides also results in low sequence coverage. A further difficulty is the high sequence similarity between many proteins, especially proteins from different taxa with the same function. This makes the identification of unique proteins challenging and is further confounded by the large databases used. To help mitigate this problem to some extent, protein grouping can be implemented. Here, proteins are grouped together if they can be inferred from the same set of identified peptides. The proteins binned to the same group usually exhibit the same function but have different taxonomic origins; therefore, the lowest common ancestor can be used to describe the taxonomic origin of the entire protein group; de novo protein identification tools without a database still perform very poorly but have the potential to be greatly improved. Since database size is an issue in regards to protein identification, more tailored databases will give better results. In future analysis, MS/MS spectra searches could be performed using high-quality metagenome databases derived directly from the measured sample. For this to become standard, sequencing quality has to be improved and costs reduced. Assessing the metaproteome of the microbiome on broad functional classes can give some hints to functional changes, but ideally, changes in function on a pathway level within specific taxa should be investigated.

To alleviate the problem of relative low protein identification, the use of gnotobiotic animal models with simplified microbiotas composed of only a few well-characterized species can greatly increase the protein coverage of functional pathways and thereby improve the information available to elucidate meaningful conclusions [11]. These gnotobiotic animal models provide a similar environment to the natural intestinal tract, including the physical–chemical parameters in the gut, the metabolic exchange between host and microbiota, very similar nutrient availability, and a host immune system interacting with the community [12]. Although many complex interactions between microbes present in conventional gut microbiota are reduced in simplified microbiota. Metaproteomics studies have already revealed functional differences in the microbiota at different locations in the gut [13], associated with obesity [14] and with inflammation of the intestine [15].

2. Clinical implementation of gut microbiome screenings

New personalized medicine approaches aim to monitor, diagnose, and treat the patients according to their specific traits (genetically/phenotypically). For example, in 2012, the first integrative personal OMICs profile (iPOP) was published where analyses of genomic, transcriptomic, proteomic, metabolomic, and autoantibody profiles from a single individual over a 14-month period were combined to reveal various medical risks [16]. Related to this, the integration of gut microbiome samples for personalized proteome profiles should become realistic in the near future. The advanced proteome technologies introduced during the last decade have led to novel opportunities in identifying biomarker candidates based on human microbiome studies and accelerating the implementation of new clinical diagnostic tests based on these methods. However, missing worldwide standardization in sample collection, bio-banking, processing, and clinical validation hinders the application of whole proteome profile assays for routine clinical analysis. The high variabilities between samples (intrasubject and intersubject variability) further complicate the analysis which substantially influences the results of human gut proteome analysis.

3. Challenge of the high variability in the intestinal microbiome

The microbial protein concentrations can be precisely assessed by state-of-the-art quantitative mass spectrometry techniques to generate comprehensive and condition-dependent protein-abundance maps (e.g. for E. coli) [17]. However, the extreme dynamic range of microbial species abundance in the gut often spanning more than 10 orders of magnitude is particularly challenging. The more complex and diverse the microbial community, the fewer proteins are identified for each taxa. To obtain biologically meaningful proteome data, about 1–5% proteome coverage of a single species should be met for reliable functional categorization.

Relative label-free quantification of proteins is the most common quantification technique in metaproteomics. But unlike in pure cultures, the high sample variability exhibited in metaproteomics leads to only a minority of the identified proteins being detected across samples and therefore quantified.

Counting of peptide spectral matches (PSMs) is often used to relatively quantify the protein abundance [18]. However, the protein groups are usually identified only by very few PSMs, so a small change in MS/MS-spectra counts can significantly alter the calculated protein abundance. Alternatively, the area under the curve from the precursor ion which is determined on the full MS level can be applied [18]. Though very few protein groups are identified by three or more peptides. Therefore, when relative quantifying between samples, ideally only the peak areas from same peptide sequences should be compared because the peak area is related to not only the protein abundance in the sample, but also the ease of extraction, i.e. hydrophobicity or ionization efficiency.

Possible solutions to increase the protein identification rate/coverage are comprehensive sample preparation processes (i.e. isoelectric focusing and pH gradients) before LC–MS/MS analysis, although this is almost always impractical in clinical practice. Although improvements in LC–MS instrumentation has occurred, LC gradients greater than 1 h are inherently limited to lower sample throughput compared to microarrays or ELISA.

In contrast to the technical challenges posed by measuring bacterial proteins, the analysis of host-secreted proteins in feces can serve as valuable reporter on the status of host–microbiota interaction [19]. Here, the change in abundance of host proteins can be sensitively assayed to elucidate the host response to microbial dynamics from the intestinal tract.

4. Commercial stool profile analysis

Currently, PCR-based assays such as the GI Effects® Comprehensive Stool Profile (Genova Diagnostics, Asheville, North Carolina, USA) offer a commercial available test that provides immediate clinical information of gut health. They routinely measure the commensal bacteria in stool, which reveal the composition, diversity, and relative abundance of a key set of 24 clinically relevant genera/species that map to seven major phyla. However, such comparable large-scale proteomics analyses can at present only be carried out in dedicated laboratories with access to high-end equipment and expertise.

5. Conclusions

Comprehensive microbial community surveys such as NIH’s Human Microbiome project and the European MetaHIT project started to describe the potential molecular functional profile and the microbial composition of the human gut microbiota. Such projects will definitely promote future clinical implementations of standard gut microbiota analyses [20]. However, we expect in the future that gut metaproteome analysis of feces will definitely become more widespread and routinely offered by service companies. Personalized proteomics approaches have started to be applicable in routine analysis, but considerable improvements in regards to high costs, time-consuming data analyses, and reliable LC–MS-based approaches are necessary.

Declaration of interest

This work was supported by DFG SPP 1656. The authors have no other relevant affiliations or financial involvements with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript.