The FibroChip, a functional DNA microarray to monitor cellulolysis and hemicellulolysis activities of rumen microbiota

Ruminants are the most efficient herbivorous animals to transform plant biomass into edible products, principally thanks to the rumen microbiota that produces a large array of enzymes responsible for the hydrolysis of plant cell wall polysaccharides. Several biotic and abiotic factors influence the efficiency of fiber degradation, which can ultimately impact the animal productivity and health. To provide more insight on mechanisms involved in the modulation of fibrolytic activity, a functional DNA microarray targeting genes coding for key enzymes involved in cellulose and hemicellulose degradation by rumen microbiota was designed. The DNA microarray, designated FibroChip, was validated using targets of increasing complexity. This new tool demonstrated sensitivity and specificity as well as explorative and semi-quantitative potential. The FibroChip was designed with the objective to propose a high throughput tool that enables to get a rapid picture of the capacity of rumen microorganisms to degrade cellulose and hemicellulose based on a targeted metatranscriptomics approach. We chose to focus on a few number of genes by targeting main ruminal fibrolytic microorganisms and selected CAZyme families that may have a pivotal role in cellulose and hemicellulose degradation. The microarray targets the coding sequence of catalytic domains from 8 CAZyme families involved in cellulose and hemicellulose degradation (i.e. glycoside hydrolases GH5, GH9, GH10, GH11, GH43 and GH48, and carbohydrate esterases CE1 and CE6). Taken together, these families present complementary activities needed for the complete degradation of cellulose and hemicellulose. In total, 392 nucleic sequences encoding 363 GH and 29 CE were kept for the microarray design. These sequences originated from 41 bacterial species, 4 protozoal species and 10 fungal species. The microarray is composed of 1631 25-mer probes. GoArrays strategy consisting by associating 2 25-mer probes targeting the same gene (Rimour et al., 2005) was also emplyed to determine 2618 composite probes of 54-mers. Triplicate of probes of 25 and 54-mers were synthetized in situ on an Agilent 8x15K DNA microarray. The microarray contained also 382 Agilent internal control probes including positive controls, negative controls and quality control probes. Probes were randomly placed on the array to avoid position bias. The objective of this work was to validate the FibroChip design and evaluate its sensitivity and specificity. For this, targets of increasing complexity were tested: PCR products, then bacterial genomic DNA, RNA from one bacteria, and finally RNA from a cow rumen content (complex microbial community). (1) PCR products were amplified from 7 genes selected to represent the diversity of targeted CAZyme families and microorganisms. Genes were cloned beforehand to ensure that a unique sequence was hybridized. (2) Genomic DNA from the strain Escherichia coli K12 was used to assess the specificity of probes as no gene from this bacterium was targeted on the FibroChip whearas the gDNA from 3 targeted bacteria strains (Fibrobacter succinogenes S85, Ruminococcus albus 20 and Bacteroides xylanisolvens XB1A) were used to assess the sensitivity of the DNA microarray. (3) To test the ability of the FibroChip to detect the differential expression of genes, the microarray was used to analyze the transcriptome of the rumen bacterium Fibrobacter succinogenes S85, cultivated in two media differing by their carbon source, i.e. the simple sugar cellobiose (C) or the complex substrate wheat straw (WS). (4) The FibroChip was then used to analyze the metatranscriptome of a cow rumen content. The rumen sample was from a dairy cow fed a mixed diet and was previously analyzed for fibrolytic activity using a RNAseq metatranscriptomics approach (Comtet-Marre et al., 2017).