Microbial Technology for Agro-Ecosystems: Crop Productivity, Sustainability, and Biofortification

Microbial Technology for Agro-Ecosystems: Crop Productivity, Sustainability, and Biofortification describes the application of competent microbes in plant growth promotion, nutrient management and recycling from molecular perspectives. Understanding of molecular mechanism of Microbial diversity in association with plant roots is very imperative for plant health and ecosystem equilibrium.

• Covers fundamental mechanisms, molecular approaches and function aspects of microbial technology
• Describes innovative approaches to the management, development and advancement of agro-ecosystem green technologies
• Highlights improving soil biological health, microbial biomass, soil fertility and plant productivity

• Language: English
• Publisher: Academic Press
• Release date: Mar 26, 2024
• ISBN: 9780443184475

Book preview

Part I

Microbes in crop productivity and sustainability

Outline

Chapter 1. A guide for the analysis of plant microbial communities through high-throughput sequencing methods

Chapter 2. Design and application of microbial biofertilizers

Chapter 3. Bioprospecting of microbial bioactive molecules for the management of biotic and abiotic stress

Chapter 4. Upscaling plant defense system through the application of plant growth-promoting fungi (PGPF)

Chapter 5. Trichoderma as ecofriendly tools for sustainable agriculture management: Improving plant health, crop productivity and soil fertility

Chapter 6. Microbes in plant health, disease, and abiotic stress management

Chapter 7. Review on plant-microbe interactions, applications and future aspects

Chapter 8. Plant growth promoting rhizobacteria (PGPR): A green approach to manage soil-borne fungal pathogens and plant growth promotion

Chapter 9. Cell wall degradation: Microbial enzymes in service of sustainable agriculture

Chapter 1: A guide for the analysis of plant microbial communities through high-throughput sequencing methods

Lihuén Iraí González-Dominicia,b, Ezequiel Peral-Aranegaa,b, Alexandra Díez-Méndezc, Marta Marcos-Garcíad, Esther Menéndeza,b, Paula García-Frailea,b,e, and Zaki Saati-Santamaríaa,b,f     aDepartamento de Microbiología y Genética, Universidad de Salamanca, Salamanca, Spain     bInstituto de Investigación en Agrobiotecnología (CIALE), Universidad de Salamanca, Salamanca, Spain     cFaculty of Sciences and Art, Universidad Católica de Ávila (UCAV), Ávila, Spain     dInstituto de Recursos Naturales y Agrobiotecnología de Salamanca (IRNSA), Consejo Superior de Investigaciones Científicas (CSIC), Salamanca, Spain     eGrupo de Interacción Planta-Microorganismo, USAL, Unidad Asociada al CSIC por el IRNASA, Universidad de Salamanca-IRNASA-CSIC, Salamanca, Spain     fInstitute of Microbiology of the Czech Academy of Sciences, Prague, Czech Republic

Abstract

Plant microbiome research allows an understanding of microbial functions and their ecology. This knowledge can be applied to develop efficient biofertilizers based on the most relevant taxa or functions of the microbiome. Nowadays, the advance of high-throughput sequencing methods have enabled us to get sequences from (meta)genomes, (meta)transcriptomes, and marker genes from complete microbial communities. To reach this goal, several programs allow the decoding of many features of the plant microbiome, such as its taxonomic composition and its functions. However, the analysis of this kind of data is not so easy and requires some expertise in bioinformatics. Here we aim to provide a brief introduction to the analysis of high-throughput sequencing data to help researchers be immersed in these methodologies. We include key steps and common workflows in plant microbiome research, including many examples of bioinformatic tools and databases, and mention their inherent biases. We hope this chapter allows researchers that are not used to utilizing next-generation sequencing to be started with these methods easily.

Keywords

Amplicon sequencing; Bioinformatics; Metagenomics; Metatranscriptomics; Microbial communities; NGS; Plant microbiome; Plant-microbe interactions; Tutorial

1. Introduction

Like any organism, plants harbor millions of microorganisms, collectively known as microbiome, which play an important role in their host's life and fitness (Trivedi et al., 2020; Hartmann and Six, 2022). Outcomes from decades of research have led us to consider microorganisms as a reservoir of additional genes and functions for the plant host, including plant growth promotion, nutrient use efficiency and control of phytopathogens. Based on the study of the most relevant taxa and functions of the microbiome, we will be able to develop more efficient biofertilizers (D'hondt et al., 2021; Hartmann and Six, 2022).

Currently, there is a paradigm shift away from single, specific microbes to a more holistic microbiome approach for enhancing crop productivity and the restoration of soil health (Ray et al., 2020). Recent advances provided by next generation sequencing (NGS) have significantly improved our access to genetic information in microbial ecology and have enabled the study of the microbiome as a whole. This technology can help to unveil the microbial communities of a sample and their functions in an accurate and efficient way (Lavelle and Sokol, 2018; Saati-Santamaría et al., 2022). In sum, NGS allows us to elucidate taxonomic composition (Who is there?), as well as functional (What can they do?) and biochemical aspects (How do they do it?) of plants' microbiomes (López-Mondéjar et al., 2017; Levy et al., 2018).

However, NGS frequently results in a high amount of data that must be correctly interpreted, not only considering expected biases and errors but also selecting an appropriate experimental design, sequencing and bioinformatic methodologies. Due to this requirement, the main aim of this chapter is to summarize the procedure and most relevant considerations on data analysis of the most commonly used NGS techniques in plant microbiome research. For this purpose, here we present detailed workflows for massive amplicon, shotgun metagenome, and metatranscriptome sequencing (Fig. 1.1), and we provide a compilation of most commonly used bioinformatics tools for each step of NGS data analysis in plant microbiome studies.

1.1. Massive amplicon sequencing of phylogenetic marker genes

The most typical and cost-effective method to analyze microbial community profiles in plant microbiome studies is high-throughput sequencing of marker gene amplicons (Knief, 2014). The high specificity of this technique allows elucidating the composition and spatial distribution of microbial members in different environments, including unculturable microorganisms (Schloss and Handelsman, 2005). However, this is a technique prone to contamination. Thus, a proper experimental design, with both negative and positive controls, technical and biological replicates, and specific protocols to reduce to a minimum the non-microbiome DNA (blocking primers …) is crucial for a further reliable sequencing data analysis (Lundberg et al., 2013).

The 16S rRNA gene is generally chosen as a bacterial taxonomic marker to study prokaryotic communities (bacteria and archaea). It is frequent to select one or more hypervariable regions (Zhang et al., 2014; Xiong et al., 2017) for short reads NGS equipment (e.g., Illumina Miseq). Full-length 16S rRNA gene sequence consists of nine hypervariable regions that are separated by nine highly conserved regions (Wang and Qian, 2009). Many studies have attempted to establish which region is the most accurate for phylogenetic analysis and taxonomic classification. Some studies point at V4–V5 (Claesson et al., 2010), V4–V6 (Yang et al., 2016), V4 (Barb et al., 2016), V5–V7 (Beckers et al., 2016) or V3–V4 (García-Lopez et al., 2020) regions as the most resolutive hypervariable regions, but the question is still under debate. In rhizosphere and plant-associated microbiome studies, hypervariable regions V3 and V4 have been the most widely used (Klindworth et al., 2013). Although amplicons spanning both the V3 and V4 regions bore a higher taxonomic resolution, nowadays V4 is considered sufficiently resolutive and is recommended by the Earth Microbiome Project (https://earthmicrobiome.org/) to assess microbial communities in soil and plant-rhizosphere (Gilbert et al., 2014).

Fig. 1.1  Overview of the workflow for analyzing data from massive amplicon, shotgun metagenome, and transcriptome sequencing.

For fungi profiling, Internal Transcribed Spacer (ITS) region, between the small (18S) and the large (28S) subunit ribosomal rRNA genes, has been established as formal barcode (Schoch et al., 2012). This region is composed of two hypervariable subregions (ITS1 and ITS2), but as above mentioned for prokaryotes, there is also no consensus about which is the most reliable to identify fungal taxa. The subregion ITS1 is the most commonly and traditionally used for fungi community profiling. Indeed, it is the subregion recommended by the Earth Microbiome Project (Gilbert et al., 2014), but Nilsson et al. (2019) concluded that ITS2 is a better-suited target to extend the coverage of the fungal kingdom. In any case, in the case of profiling fungal communities, it is common to use also the 18S gene as a phylogenetic marker (Panzer et al., 2015).

Also, a few recent studies show the importance of micro-eukaryotic (beyond fungi) communities in the rhizosphere and phyllosphere environments (Zhu et al., 2018; Taerum et al., 2022). For assessing this portion of microbiome, 18S rRNA gene is most usually sequenced marker (Amaral-Zettler et al., 2009; Stoeck et al., 2010; Hugerth et al., 2014). However, this kind of eukaryotic diversity analysis still lags behind that of bacteria and fungi in the plant microbiome studies.

Long-reads NGS platforms, such as the PacBio Single Molecule, Real-Time (SMRT) DNA sequencing platform, allow the sequencing of the complete marker genes, but their costs are usually higher. In any case, it is important to bear in mind that the use of 16S rRNA gene, ITS or 18S rRNA gene as phylogenetic markers could be an important source of bias (Schirmer et al., 2015) leading to an over- or underestimation of certain taxa’ richness (Kennedy et al., 2014) due to overrepresentation of some research-relevant species with strain representatives in databases and due to the need to use amplification primers that may exclude some taxa. For that reason, other taxonomic markers, known as housekeeping genes, have been tested as more reliable alternatives to assess microbial diversity; these genes include DNA gyrase subunit B (gyrB) (Barret et al., 2015; Poirier et al., 2018), RNA polymerase subunit B (rpoB) (Ogier et al., 2019), the TU elongation factor (tuf) (Ghebremedhin et al., 2008) and the gene encoding the 60 kDa chaperonin protein (cpn60) (Links et al., 2012), amongst others. Despite their discrimination power, housekeeping genes are used for specific taxonomic groups, and it is recommended to use them in combination with the 16S rRNA gene for global bacterial and archeal community analysis, mainly due to the availability of extensive and accurate databases.

Regardless of the chosen gene marker for the analysis of the microbial communities associated with plants, the output data from sequencing platforms are thousands or millions of sequences presented in FASTQ format that allow the generation of a taxonomic profile in a step-by-step procedure. These raw data always include sequences of primers, adaptors, low quality reads, singletons, chimeras and other artifacts that can disturb our analysis. Thus, the first step in amplicon sequencing data analysis is pre-processing our sequences by filtering bases and reads based on quality scores and length, and to remove barcode sequences with specific tools such as Cutadapt (Martin, 2011), Trimmomatic (Bolger et al., 2014) or FastP (Chen et al., 2018), amongst others. It is recommended to check the sequences quality before and after trimming with FastQC (Andrews, 2010) or MultiQC (Ewels et al., 2016). Furthermore, pair-end sequencing data should be merged into a unique file using FLASH (Magoč and Salzberg, 2011), PANDAseq (Masella et al., 2012), MeFiT (Parikh et al., 2016), or others, to increment the confidence level in uncalled or miscalled bases. For example, PANDAseq is an assembly and error correction tool for paired-end reads of 16S rRNA gene amplicons that determines the proper amount of overlap and reconstructs the entire sequence by correcting errors in the overlapping region.

Once these sequencing data are pre-processed, we aim to elucidate the microbial composition of the samples (Who is there?). The basic unit for taxonomic profiling is species, but in this kind of analysis, it is common to use Operational Taxonomic Units (OTUs) or Amplicon Sequence Variants (ASVs) terms. In the first, sequences are clustered into clusters of sequences based upon a threshold of similarity in a pairwise sequence alignment, usually 97% (Konstantinidis and Tiedje, 2005; Nguyen et al., 2016). In each cluster, a single sequence is selected as representative, which is annotated, and all sequences within the OTU (cluster) inherit that same annotation, reducing the computational costs in subsequent analyses. However, the differences in discriminating power of different hypervariable regions in taxonomy and the existence of taxa which share a higher identity percentage make it difficult to establish a truly representative threshold (Yarza et al., 2014), resulting in a loss of taxonomic resolution. Due to the increase in sequencing accuracy, a new form of clustering, ASVs, has recently emerged. ASV clustering can detect small sequence variants, even of a single nucleotide, forming clusters based on sequence probability rather than sequence identity throughout well-established statistical models (Callahan et al., 2016). This results in a more acceptable precision clustering that is being increasingly used in microbiome research (Nearing et al., 2018; Prodan et al., 2020). Some of the most popular algorithms to OTU picking are Uparse (Edgar, 2013) or Uclust (Prasad et al., 2015), and UNOISE3 (Edgar, 2016), while Deblur (Amir et al., 2017) and DADA2 (Callahan et al., 2016) are some of the most frequently used for ASVs. We recommend the revision of the comparison of these algorithms accomplished by Prodan et al. (2020), where authors show different features of each of them.

After clustering our reads, the next step to know who is there? is the taxonomic assignation of these clusters using a sequence reference database. The most widely used database for 16S sRNA gene and ITS region classification are Greengenes (DeSantis et al., 2006), NCBI (Federhen, 2012), Silva (Quast et al., 2013), RDP (Cole et al., 2014), EzBioCloud (Yoon et al., 2017), and UNITE (Abarenkov et al., 2010). Inevitably, identification and interpretation of the community composition are affected by the chosen database. Park and Won (2018) evaluated the accuracy of Greengenes, Silva, and EzBioCloud for bacterial community analyses using mock communities, concluding that EzBioCloud performs better than the other two databases, likely due to the higher curation level of taxonomic information for each sequence, which that might facilitate a correct identification at species level. Furthermore, the updating of databases is crucial for a more thorough taxonomic assignment. For example, Greengenes was not updated since 2017, respectively, which means that they lack all newly described taxonomic taxa, whereas RDP, Silva and EzBioCloud were updated in 2020, and UNITE in 2022. Recently, Liao et al. (2022) published a new database, MetaSquare, which aims to merge many of the most widely used databases and a collection of novel sequences systemically into an integrated resource without limitations on either the sequence redundancy or the delay on new sequence recruitment. In any case, bioinformatic tools will add taxonomic to each representative OTU or ASV sequence. In the end, the chosen software will generate a taxonomic profile of each sample, providing information of its composition and taxa abundance at different taxonomic levels.

In plant microbiome research, we frequently want to compare taxonomic profiles from different samples or treatments in order to establish hypotheses. For a correct comparison between samples, the sequencing library size (total reads per sample) has to be normalized. We can opt to calculate the relative abundances of taxa, perform random down-sampling, reduce overdispersion using available packages such as DESeq (Anders and Huber, 2010), and edgeR (Robinson et al., 2010) or metagenomeSeq (Paulson et al., 2013), amongst others. Any choice of normalization will affect how data are interpreted downstream.

Data normalization allows us to perform a comparative analysis of the microbial diversity of each sample, which is the second main purpose of amplicon sequencing analyses. In ecology, there is a distinction between alpha-diversity (diversity within a single sample or set of replicates) and beta-diversity (degree to which different group of samples are different) (Whittaker, 1960). Alpha-diversity may be estimated throughout species richness (with indices such as Chao1, ACE or Jackknife, that try to estimate the number of species in the sample), throughout evenness (with Shannon and Simpson indexes, that estimate how even species distribution is), and throughout phylogenetic diversity (i.e., Faith Phylogenetic Diversity index), that considers the phylogenetic distances between the community members. Beta-diversity is a rather different issue than within-sample richness and evenness, and can be measured with distance metrics such as Euclidean, Jensen-Shannon and Kullback-Leibler, with correlation coefficients, such as Pearson's product moment, Spearman's rank correlation and Bray–Curtis dissimilarity, or also using phylogenetic distances (i.e., with weighted UniFrac distances). Finally, composition and diversity measures will enable comparisons and assumptions for our study, which require a statistical demonstration or better known as hypothesis testing. Due to non-normality of most microbial data, it is necessary to use a nonparametric test, such us Kruskal–Wallis’ H-test, also known as ANOVA on ranks when there are only two sample groups, or PERMANOVA (Anderson, 2014) and ANOSIM for multiple comparisons. Also, it is possible to estimate which are the main predictors of the microbial communities. For that, the use of PERMANOVA has been implemented within the function adonis (Oksanen et al., 2013) that allows the use of distance matrices from beta-diversity analyses (i.e., Bray Curtis) as input.

Apart from the knowledge of the microbial community diversity or composition directly obtained from a taxonomic profile, computational progress and integration of public scientific data provide a powerful tool to infer more information about microbiomes. In fact, the integration of results and multi-omics approaches have enabled the development of bioinformatic tools capable of predicting functions and molecular interactions from taxonomic profiles data, widening our knowledge about microbial roles and interactions. For instance, it is possible to predict functional portrayals from taxonomic profiles with tools such as PICRUSt (Langille et al., 2013) or Tax4fun (Aβhauer et al., 2015), that infer functions likely associated with an OTU or ASV based on a reference genome for the corresponding taxon. It is important to note that the accuracy predicting cannot be guaranteed, and shotgun sequencing is more recommended for this purpose. In the same way, we can also perform a network analysis, which allows us to identify microbial taxa that are functionally linked to others revealing biochemistry aspects of interacting microbes in a studied ecosystem. Currently, there are many approaches to carry out a network analysis (Jiang et al., 2019) and some of the most used tools are GeneMANIA algorithm (network integration method) (Mostafavi et al., 2008), SCNIC (co-occurrence network analysis method) (Shaffer et al., 2023) or CONEXIC (Bayesian network method) (Akavia et al., 2010).

In summary, to avoid biases introduced along the workflow or results misinterpretations, analysis of the metabarcoding data requires an exhaustive previous knowledge about the whole data processing procedure and the critical steps of the process. Fortunately, in the last few years several pipelines to carry out an end-to-end workflow for a correct bioinformatic analysis have been developed. The most popular current bioinformatic pipelines/tools for 16S rRNA gene amplicon sequencing analyses are MOTHUR (Schloss et al., 2009), phyloseq (McMurdie and Holmes, 2013) and QIIME2 (Bolyen et al., 2018). Although one of the most important purposes of those software is to simplify the sequences analyses, all of them require knowledge on the basic concepts of programming languages. More user-friendly tools are available online, such as Galaxy Project (Afgan et al., 2018), an open source platform for data and metadata analysis, SEED2 (Větrovský et al., 2018), which is a downloadable user-friendly program, or the 16S rRNA gene-based Microbiome Taxonomic Profiling (16S-based MTP) tool from EzBioCloud (Yoon et al., 2017), which was specifically developed for 16S amplicon sequence analysis. However, these webservers have many limitations in parameters modification or error correction.

2. Shotgun metagenomics

Shotgun sequencing targets all the DNA of a certain sample, which is useful to obtain genomes and metagenomes sequences without any previous targeted-PCR and its inherent amplification bias. Also, it captures any type of DNA in the sample, including that of fungi, bacteria, viruses, oomycetes, etc. The main advantages of this technique is that it allows to study, not only the taxonomy of the microbial communities (Who is there?), but also their functions (What can they do?). For instance, it allows the knowledge of which genes or functions are enriched in a certain ecosystem compared to another one (i.e., plant rhizosphere or endosphere vs. bulk soil). Thus, through shotgun metagenomic analyses it is possible to study the microbial pathways that are associated with plant-microbe interactions, including mechanisms for plant-growth promotion, resistance to host defenses, entrance to the plant, alleviation of plant stresses, etc.

For that, metagenomic sequences produced by the chosen sequencing platform should first be cleaned, as in any DNA sequence-based method. To do so, several bioinformatic tools that are capable of simultaneously processing millions of sequences might be used (Table 1.1). Then, there are several programs (Table 1.1) to assemble the reads into contigs; these programs differ not only in the sensitivity, but also in the RAM usage and processing speed, something that, depending on the available computing facilities, may be a crucial step (Vollmers et al., 2017).

The gene calling on the metagenome contigs is usually biased toward prokaryotic or to eukaryotic organisms since annotation programs do not use different genetic codes simultaneously. A solution to this may be an initial classification of the contigs based on their taxonomy (Pronk and Medema, 2022; Saraiva et al., 2022) and to perform a specific and targeted structural annotation for each specific kingdom afterward (Table 1.1). The taxonomic assignment of the contigs can be done through different methods, which based on different principles: (I) comparison of the sequence similarity of the contigs with those available in databases; (II) DNA characteristics (codon usage, tetranucleotide content, G + C%, etc.) as biomarkers; (III) the combination of those two approaches; and (IV) search for marker genes in the contigs and classify the sequences based on the matches of those genes with those present in specific taxonomic databases. When scoring the closeness of a contig to a certain taxa, there are also two possible methods: one takes into account just the best hit of the comparison and returns it as the taxon of the contig, the other looks for the Last Common Ancestor (LCA), looking for the immediate higher shared taxa of the different hits