Microbial Ecology: Where are we now?

Abstract

Conventional microbiological methods have been readily taken over by newer molecular techniques due to the ease of use, reproducibility, sensitivity and speed of working with nucleic acids. These tools allow high throughput analysis of complex and diverse microbial communities, such as those in soil, freshwater, saltwater, or the microbiota living in collaboration with a host organism (plant, mouse, human, etc). For instance, these methods have been robustly used for characterizing the plant (rhizosphere), animal and human microbiome specifically the complex intestinal microbiota. The human body has been referred to as the Superorganism since microbial genes are more numerous than the number of human genes and are essential to the health of the host. In this review we provide an overview of the Next Generation tools currently available to study microbial ecology, along with their limitations and advantages.

Introduction

Molecular techniques for analysis of microbial ecology and characterization have taken over conventional analytical tools, which were laborious and time consuming. These new approaches provide better insight into the complex interactions taking place within indigenous microbiota of various environments, a few of which include: host and host-pathogens, plants and soil microbiota, aquatic microbes with surrounding water. With the advent of Next-Generation sequencing (NGS) technologies, it has become even faster and more economical to comprehensively evaluate the complexity of microbiota in an environment. This review focuses on these molecular approaches for understanding complex microbial communities and their implications.

Microbiology really began in the late 17^th century with Robert Hooke and Antoni van Leeuwenhoek, who were the first to individually publish on the original observations of single celled organisms, or as van Leeuwenhoek referred to them “little animals” (Gest 2004). Traditionally, microbial populations were solely understood through culture based techniques, of which less than 1% of microbes are culturable, and those that are require numerous physiological and biochemical tests for full characterization. These techniques were very time consuming, laborious, as well as requiring prior knowledge of the organisms of interest to selectively and successfully culture from a complex microbial network.

There are several limitations to culture based techniques, most important of which is a biases towards readily cultivatable organisms, completely overlooking non-culturable bacteria (i.e. ‘the great plate count anomaly’), the largest proportion of microbes (Staley and Konopka 1985). To some extent immunological based methods for characterization and identification of bacteria, including enzyme linked immunosorbent assay (ELISA) requiring specific antibodies, can help circumvent these problems but are still limited in assessing functionality and have not been extensively used due to the need for specific antigens/antibodies (Ricke and Pillai 1999). Consequently, focus was turned towards development of molecular approaches for understanding these complex communities, eliminating the need for culturing beforehand.

Dawn of a new era began with the innovation of NGS technologies, which revolutionized the existing gold standard techniques of microbial community analysis. NGS enables the generation of massive sequencing results in parallel utilizing phylogenetic markers, such as 16S rRNA gene, inexpensively and in considerably less time. The generated high throughput results not only enable comprehensive analysis of community microbiota from various environments (for instance, the soil, intestine, fresh/saltwater etc.) but also insights into their interactions with the environmental system. There are endless possibilities for applying NGS technologies. Sequencing also finds its utility in profiling the microbial community of many complex environments, and has been extensively reviewed for performing such projects (Hamady and Knight 2009). Herein we include a brief overview of the molecular techniques readily used in environmental microbiological investigations, along with their various advantages and limitations. We initially begin with molecular approaches on bacterial characterization (16S rRNA, DGGE, T-RFLP) and a section on typing bacterial pathogens. Followed by the various NGS technologies currently used (targeted amplification, metagenomics, RNA-seq, proteomics, Tn-seq). Included in each section are selected examples where the individual research groups aim at identifying key species (or species specific genes) responsible for a dynamic state within a niche. The tools reviewed here are being readily implemented and provide a comprehensive overview of microbial ecology today.

Unravelling Microbial Ecology

Current trends in understanding complex microbial communities are focused on molecular approaches that not only provide unbiased, comprehensive information but are also less labor intensive and can be accomplished in a shorter duration of time. There are several molecular approaches that were developed for classification of complex microbial communities and provide a clearer understanding of the microbe's ecology and interactions when compared to conventional culturing methods.

Several recombinant DNA techniques based on 16S rRNA [originally proposed by Carl Woese (1987)], chaperonin-60, rpoB sequences etc., for unbiased characterization of mixed microbial communities have been developed (Hill et al. 2002; Theron and Cloete 2000) and extensively used. For instance, a study on cohort of nonpregnant, healthy women to assess their vaginal microflora based on chaperonin-60 sequencing revealed that the dominant species found there were Lactobacillus spp (Hill et al. 2005). Also, a study on the intestinal tract of market age broilers based on 16S rRNA gene sequencing revealed a shift towards higher ratio of Firmicutes to Bacteroidetes upon administration of Penicillin (Singh et al. 2013b).

16S rRNA gene sequencing

Microbial community analysis based on 16S rRNA gene sequencing is a rapid and reliable tool that is being readily used for comprehensive analysis of microbial communities within all kinds of niches, be it deep sea sub-surface (Kimura et al. 2007) or an estuary (Bernhard et al. 2005). 16S rRNA is an excellent phylogenetic marker that is conserved across all prokaryotic species and is more resistant to mutations as compared to other conserved genes, however some regions called hot spots are different in all species (Clarridge III 2004). An important note: while it is unlikely for 16S genes to be horizontally transferred (horizontal gene transfer, HGT), such a thing is not impossible (Kitahara and Miyazaki 2013; Tian et al. 2015). Universal microbial primers (for both bacteria and archaea) have been designed to complement the constant while amplifying the variable regions of the ~1500bp 16S rRNA genes, which may then be sequenced for phylogenetic analysis.

Quantitative polymerase chain reaction (qPCR) has also been readily used for complementation and verification of 16S rRNA gene sequencing data, as studies have shown sequencing biases and variability based on the protocol used (Hang et al. 2014). Recent advancements in sequencing platforms, from obsolete 454 pyrosequencing to short read based Illumina, has made use of 16S rRNA gene sequencing. More recently Illumina based sequencing for microbial community analysis has been the robust and economical sequencing platform of choice. The Human Microbiome Project Consortium studied the complex microbial communities from various sites of the human body, including the gut, skin and vagina, making use of the Illumina GAIIx platform (The Human Microbiome Project Consortium 2012). Unfortunately, 16S rRNA gene sequencing is not infallible and sometimes exhibits low resolution due to sequence similarities even when exhibiting different phenotypes. An alternative has been suggested by utilizing 16S rDNA sequencing in combination with DNA-DNA hybridization techniques, like microarray (Janda and Abbott 2007). Several open source software tools have been consequently developed to efficiently analyze the massive amount of data being generated. The most commonly used software tools include Mothur (Schloss et al. 2009) and QIIME (Caporaso et al. 2010) which include steps for quality filtering the data sets to purge ambiguous and sequencing artifacts. Chimera are a major sequencing artifact, and both Mothur and QIIME include a chimera checker (Edgar et al. 2011), to eliminate such issues from the sequencing data before clustering and phylogenetic analyses. Phylogenetic classification is carried out with the help of well-maintained and periodically updated 16S rRNA sequence databases, including the green genes (http://greengenes.lbl.gov/) and the Ribosomal Database Project (Cole et al. 2014). Further downstream statistical analysis tools are also freely available from various sources, which include but are not limited to: LEfSe for identifying differentially abundant microbial features (Segata et al. 2011), Picrust for metagenomic prediction (Langille et al. 2013) and R tools for generating heat maps and graphs.

Denaturing Gradient Gel Electrophoresis (DGGE)

DGGE is an important tool for studying complex microbial communities as well as their shift over time (Muyzer and Smalla 1998). Community profiling by DGGE is recommended as an important clinical tool that depends on the differences between the genetic sequences of various bacterial communities, thereby generating different banding patterns based on denaturation and electrophoretic mobility of the denatured DNA (Ariefdjohan et al. 2010). DNA sequences amplified by PCR are chemically denatured as the amplified product runs on polyacrylamide gel with increasing concentrations of denaturants. In addition to being less laborious, DGGE is also considered superior to sequencing with cloning techniques, and has been demonstrated by denaturation analysis of 16S rRNA genes as a useful diagnostic tool for microorganisms including bacteria, fungi and yeast (Muyzer et al. 1993). Even though DGGE has been shown incapable of detecting two or three base variations, the addition of a GC clamp increased the methods ability to detect most single base substitutions for DNA fragments ranging from 25-500bp (Myers et al. 1985). Soil microbial community DGGE analysis using 16S rRNA gene amplification products was found to be efficient in resolving and characterizing a bacterial community (Nakatsu et al. 2000). In another study, DGGE was implemented on a large scale to elucidate marine microbial ecology and its spatial and temporal variations. Significant difference in bacterial community compositions were identified within a relatively small geographic distance (Singh and Ramaiah 2011).

Terminal Restriction Length Polymorphism (T-RFLP)

T-RFLP allows for reproducible profiling of a microbial community, as well as its changing dynamics in response to environmental factors. Terminal end fluorescent labeling of 16S rRNA PCR products are digested by a 4bp cutter enzyme and run on an automated sequencer, leading to differentiating peaks that can then be analyzed and generate a phylogenetic pattern. Extensive examination of gut microbial composition revealed a heterogeneity between bacterial populations of ceca and ileum in broiler chickens by T-RFLP analysis of 16S rRNA genes (Gong et al. 2002). Similar analyses on human intestinal microbiota exhibited approximately 10 phylogenetic bacterial groups (Nagashima et al. 2006). Limitations of this technique include generation of ‘pseudo-T-RFs’ (terminal restriction fragments), generated from undesired enzymatic digestion of secondary single stranded molecular structures. Formation of ‘pseudo-T-RFs’ result in additional peaks that may lead to over estimation of microbial diversity and the need to generate clone libraries due to inaccessibility of sequences (Egert and Friedrich 2003).

Typing of Bacterial Pathogens

The need to characterize and differentiate bacterial pathogens has resulted in various identification schemes to understand their ecology and epidemiology. Typing systems are evaluated in terms of reliability, reproducibility, discriminatory power, as well as data interpretation. The method of choice also depends upon ease of use, time taken and cost involved (Foxman et al. 2005). Perhaps the most important application of identification is during an infectious outbreak by providing insight on the spread of infection (Pitt 1994). Conventional typing methods are phenotypic based and include serotyping, biotyping, bacteriocin typing, phage typing. Even though these gold standard techniques are laborious and time consuming, they have their benefits and can be used depending on the level of discrimination desired, as a standalone technique or in combination with molecular techniques. Moreover, the advent of DNA based techniques were based on the need to overcome shortcomings of conventional phenotypic typing methods.

Genotyping techniques based on total DNA sequences can be divided into two categories, one that is based on enzyme digestion patterns [ex. restriction enzyme analysis (REA) or Pulse Field Gel Electrophoresis (PFGE)] and a second category based on southern hybridization technique (ex. Ribotyping). Ribotyping was shown to have a higher discriminatory power over a combination of conventional methods, including biochemical tests, serotyping, phage-typing and LPS profiling in Enterobacter cloacae [a nosocomial pathogen, isolated from patients (Garaizar et al. 1991)].

PCR based techniques include amplified fragment length polymorphisms, random amplified polymorphic DNA (RAPD), PCR-RFLP and sequencing based on multi locus sequence typing (MLST). Menon et al. (2003) identified RAPD as a quick typing technique with high reproducibility and ease of use, based on an electrophoretic pattern generated for Pseudomonas aeruginosa in contrast to a set of conventional methods. Given that the discriminatory power of RAPD could not be compared due to a lack of correlation, Menon et al. (2003) suggest an advantage to combine more than one technique for typing purposes. This was supported by a study conducted to evaluate PCR based molecular typing techniques like RAPD, PCR fingerprinting and REA to assess the genetic diversity in Saccharomyces cerevisiae, and indicated the need to combine more than one molecular typing strategy to achieve a greater degree of discrimination (Couto et al. 1996). Salmonella species were shown to exhibit higher discrimination with more than one technique, like that of a PFGE pattern generated by XbaI and antibiotic susceptibility tests (performed for typing and source tracking) (Nayak and Stewart-King 2008). As well as a combination of enterobacterial repetitive intergenic consensus (ERIC) fingerprinting and RAPD (Lim et al. 2005).

The importance of combing typing methods has been stressed by many researchers to enhance organismal discrimination, as demonstrated for 1) Salmonella Newport where a combination of MLST, PFGE and antimicrobial susceptibility (Harbottle et al. 2006) and 2) Salmonella Typhimurium where a combination of MLST, PFGE and Rep-PCR was evident (Foley et al. 2006) as well as 3) Candida albicans (Chen et al. 2005).

Furthermore, Whole Genome Sequencing (WGS) provides more characterization resolution wherein genome wide information of a bacterial pathogen is made available. WGS has been implemented in conjunction with epidemiological data to unravel transmission source. For instance, a study conducted on carbapenem-resistant Klebsiella pneumonia linked dissemination of this pathogen to a single patient providing insight into nosocomial infections based on NGS technologies (Snitkin et al. 2012).

Efficient strategies for NGS technologies

Development of NGS technologies has enabled a great number of genomic facilities with varied applications for large scale sequencing by drastically increasing the capacity of high throughput sequencing in a time and cost effective manner. To completely utilize the potential of NGS technologies, enrichment techniques capable of large scale multiplexing are required. Even though as compared to conventional sequencing technologies, the cost of large scale sequencing has drastically reduced (Stein 2010), it is still very expensive for WGS of complex organisms, such as humans. For this reason, emphasis is being placed on NGS that allows for selective amplification of targeted genes relevant to individual sequencing projects.

Targeted Amplicon Sequencing

A primer based method by which a gene or region of interest is sequenced. Target enrichment can be done in two ways, hybridization based and amplification based methods. Hybridization based methods rely on sequence-specific nucleic acid hybridization in both a solid phase built on microarray based enrichment (Summerer et al. 2009) and a solution phase including molecular inversion probes; both methods have been elaborately reviewed (Fan et al. 2006; Mamanova et al. 2010; Summerer 2009). Tewhey et. al. has developed a hybridization based sequence enrichment technique in solution phase, that can selectively capture and enrich for a 3.9Mb sequence of the human genome and has been reported 99% accurate for SNP (single nucleotide polymorphism) calling (Tewhey et al. 2009). Amplification based methods like nested patch PCR for multiplexing (Varley and Mitra 2008), circular selectors (Dahl et al. 2005) and megaplex PCR (Meuzelaar et al. 2007), have been developed to efficiently and selectively capture target DNA, using either primers or probes each of which have been demonstrated to have unique NGS application. Target enrichment techniques are evaluated in terms of multiplexity, specificity, uniformity, drop-off rate and reproducibility. The above-mentioned techniques were all evaluated using these key parameters (Ting et al. 2009). Senapathy et al. developed an enrichment technique based on fixed-randomization that binds multiple genomic sites with similar sequence complementarity, called Functional Genomic Fingerprinting (FGF) and has been used to enrich functional genomic regions in humans (Senapathy et al. 2010). One important aspect of multiplexing by target enrichment is normalization of the amplicon pool, where many samples are assessed and barcoded for increased performance and efficiency, as well as generation of a complete microbial profile (Harris et al. 2010). Similarly, in another study, enhanced bacterial characterization technique (MLST-seq) for foodborne pathogen utilizing Salmonella as a model system, was demonstrated. Novel target enrichment strategies that selectively amplified housekeeping and virulence genes of Salmonella for NGS sequencing in a high-throughput manner were efficiently developed (Singh et al. 2011; Singh et al. 2012, 2013a. Comprehensive reviews of several other applications in the fields of functional genomics (Olena and Marra 2008), clinical diagnostics (Bosch and Grody 2008), ancient DNA (Knapp and Hofreiter 2010), and transcriptomic characterization (Wall et al. 2009) have been assessed. Numerous other applications could be performed, like studying host-pathogen interactions or carrying out genomic studies for drug discoveries.

A major advantage of NGS is the elimination of vector based sequencing that is likely to be biased, laborious, time consuming and would not generate a complete profile of the microbial community. One such study conducted on cecal microbiome of chickens, demonstrated a role of mobile DNA elements in microbial horizontal gene transfer and evolution of a functional microbiome conferring resistance to zoonotic organisms in the host (Qu et al. 2008). Phylogenetic markers like cpn60 are also being explored and utilized for metagenomic analysis (Schellenberg et al. 2009).

Other areas of research and interest include transcriptome analysis, cancer genetics, SNP analysis, mutation discoveries etc. A few limitations to this technology exist since it cannot be applied to research requiring minimal sequencing. Relatively speaking, NGS is an inexpensive technology when comparing its data output with Sanger's sequencing. Another limitation of Illumina technology is the generation of shorter reads with an average of ~300bp, this technology is now being replaced by new improved kits, however, still requires bidirectional sequencing to cover amplicons over 600bp in length.

Metagenomics

A method by which all the DNA extracted from an environmental sample is sequenced, allowing for a comprehensive analysis of microbial communities. Metagenomics, or shotgun sequencing, is dependent upon the method of DNA obtainment. A poor method of DNA removal will lead to false interpretations of the microbiota in, and the functional profile of, an environment. DNA may be extracted through either indirect or direct approaches. Indirect DNA extractions consist of removing cells from an environmental sample, followed by extracting DNA from the separated cells; and direct DNA extractions, involve in situ cell lysis and subsequent removal of DNA from within an environmental sample (Williamson et al. 2011). There are many approaches to choose from for direct DNA extraction, it is important to consider sample characteristics before determining the ideal method. For example, there are many substances in soil that bind nucleic acids or inhibit downstream applications; such as humic acids (inhibitors of PCR) and clay (binds nucleic acids) (Nair et al. 2014; Yankson and Steck 2009). Regardless, extracted DNA is fragmented to about 150bp in size, these fragments are then end labeled with barcodes for downstream sample identification, enabling multi-sample sequencing. Fully sequencing the DNA within a sample allows researchers to examine the taxonomic diversity of all microbes, not just the culturable. However, the degree to which the diversity can be determined, depends on the depth of sequencing, i.e. how much sequencing data is obtained from a sample (Sims et al. 2014). The more abundant an organism in the environment, the more DNA extracted and the more sequencing data obtained. Greater sequencing depth is necessary to obtain information on the least abundant microorganisms. An important note is that the DNA obtained from a sample will depend on the method of extraction, as well as the sample characteristics from which the DNA is extracted (Sagova-Mareckova et al. 2008).

Upon extracting the DNA and sequencing the samples, one must then analyze the data. The obtained sequencing reads are sorted according to the barcodes, poor quality reads are removed, then the remaining quality reads are assembled, annotated, and analyzed for taxonomic and/or functional diversity (i.e. who's there and/or what are they doing?) (Albertsen et al. 2013; Sharpton 2014; Thomas et al. 2012). Each stage is computationally laborious and there are many programs available to assist in each of the steps (Howe et al. 2014; Miller et al. 2010; Wang et al. 2015). Fierer et al. (2012) investigated the impact of nitrogen (N) fertilization on terrestrial ecosystems utilizing various approaches, including shotgun sequencing, and indicate that anthropogenic inputs of N are favoring copiotrophic (fast growing) microbes.

Transcriptomics

In addition to metagenomics, there are other similar ‘omic's to assist in broadly examining the taxonomic and/or functional diversity of environments. Transcriptomics, or RNA sequencing (RNA-seq), allows for the examination of RNA present at the time of extraction (i.e. which genes were considered necessary at that time?) (Wang et al. 2009). As noted above for DNA extraction, the method of nucleic acid removal will inform the results. Additional consideration is required when working with RNA since it is a short-lived molecule, and not to mention that RNases are everywhere and are difficult to avoid (Dineen et al. 2010; Saleh-Lakha et al. 2005). Also, during the extraction process, all RNA is obtained but usually, only mRNA is of interest, therefore before beginning with sequence analysis of prokaryotic RNA, mRNA (no poly-A tail) needs to be ‘selected’ above other RNA molecules (tRNA, rRNA) (Giannoukos et al. 2012).

Once the additional considerations when working with RNA are accounted for, the downstream analysis are quite similar to those discussed for metagenomics. For example, RNA-seq was recently used to examine the differential expression of genes from Sphingomonas wittichii RW1 [debatably the only organism in isolation that is able to utilize the toxin dibenzo-p-dioxin (DD) as its sole carbon and energy source] in the presence of either DD, dibenzofuran (DF; a similar molecule with lower toxicity), or succinate (SUC; negative control carbon source) with and without clay particles (a prominent soil component) (Chai et al. 2016). Such investigations can assist in understanding why dibenzo-p-dioxin is so resistant to biodegradation, whereas dibenzofuran biodegradation is relatively common (toxicity, molecular structure, or something else entirely?).

Proteomics

Similar to transcriptomics, there is another method for examining the functional profile of a microbial community or organism of interest. However, instead of looking at which genes were transcribed, proteomics examines the proteins that have actually been produced (Simon and Daniel 2011). An important distinction, since not all amplified mRNA leads to synthesized protein (Ingolia 2014). Similar to metagenomics and transcriptomics, the method of extracting proteins from a sample requires due consideration (Keiblinger et al. 2012; Wang et al. 2006). One of the main issues is to successfully recover a representative profile from samples with various characteristics, the challenges of soil proteomics is reviewed by Renella et al. (2014). The purified protein samples are then run on a mass spectrophotometer to determine the sequences of proteins present (Armengaud 2013). An alternative is to run the purified protein samples on two-dimensional polyacrylamide gel electrophoresis (2D-PAGE), but this method will only provide a protein profile for comparison between samples, not the specific proteins present. Toueille et al. (2012) used proteomics to comparatively analyze the nucleoid associated proteins (NAPs) from Deinococcus spp, with the goal of identifying whether novel NAPs play a role in conferring radiation stability to Deinococcus organisms.

Transposon Sequencing

Transposon sequencing (Tn-seq), is a method that can be utilized to examine gene functions and/or genes of essentiality within an organism under varying growth conditions. Before beginning a Tn-seq experiment, there are numerous considerations to be made, much like with any method (Kwon et al. 2016). Briefly, Tn-seq involves a mutating transposable element that is introduced into a dense culture of an organism, ideally resulting in a collection of mutants where under such growth conditions, all non-essential genes are interrupted (i.e. transposon library for that organism) (Opijnen et al. 2009). Subsequently, the transposon library can then be grown up under various conditions of interest, as well as the starting/negative control condition (which provides the background essential genes). The DNA is then extracted and sequenced (targeting the region where the transposable elements were inserted), genes with a transposable element are not essential for that organism's growth under that condition (i.e. the cell grew without that gene). Through Tn-seq analyses, Pechter et al. (2016) examined the essential aerobic heterotrophic genes of Rhodopseudomonas palustris, an organism that is able to fluctuate between anaerobic photoautotrophic growth and aerobic heterotrophy. Obtaining a better understanding of the metabolic capabilities of organisms such as R. palustris is a most worthwhile endeavor. A tool such as Tn-seq is useful when working to understand the metabolic diversity of microbes: which genes enable an organism to metabolize a compound, but not others? What are the functional differences between two similar organisms with differing metabolic capabilities? Is the ability conferred by metabolic pathway enzymes, or through reduced stress/toxicity affects, or something else entirely? Tn-seq is an increasingly utilized method to address these and many other questions.

Conclusion

Current approaches to microbial ecology have come a long way in the past few centuries, enabling a broader and more thorough investigation into the small, single celled organisms on which we all depend. NGS technologies have been utilized for high throughput screening and have not only become more efficient and accessible tools for studying complex microbial communities, but also provide better insight into the interactions taking place within complex microbial communities. NGS technologies have aided in determining the core factors responsible for the dynamic nature of these complex communities and their interaction with host and native environments. One can only wonder what new advances, and wondrous discoveries will come tomorrow.