Usability of reference-free transcriptome assemblies for detection of differential expression: a case study on Aethionema arabicum dimorphic seeds.

BACKGROUND: RNA-sequencing analysis is increasingly utilized to study gene expression in non-model organisms without sequenced genomes. Aethionema arabicum (Brassicaceae) exhibits seed dimorphism as a bet-hedging strategy - producing both a less dormant mucilaginous (M+) seed morph and a more dormant non-mucilaginous (NM) seed morph. Here, we compared de novo and reference-genome based transcriptome assemblies to investigate Ae. arabicum seed dimorphism and to evaluate the reference-free versus -dependent approach for identifying differentially expressed genes (DEGs). RESULTS: A de novo transcriptome assembly was generated using sequences from M+ and NM Ae. arabicum dry seed morphs. The transcripts of the de novo assembly contained 63.1% complete Benchmarking Universal Single-Copy Orthologs (BUSCO) compared to 90.9% for the transcripts of the reference genome. DEG detection used the strict consensus of three methods (DESeq2, edgeR and NOISeq). Only 37% of 1533 differentially expressed de novo assembled transcripts paired with 1876 genome-derived DEGs. Gene Ontology (GO) terms distinguished the seed morphs: the terms translation and nucleosome assembly were overrepresented in DEGs higher in abundance in M+ dry seeds, whereas terms related to mRNA processing and transcription were overrepresented in DEGs higher in abundance in NM dry seeds. DEGs amongst these GO terms included ribosomal proteins and histones (higher in M+), RNA polymerase II subunits and related transcription and elongation factors (higher in NM). Expression of the inferred DEGs and other genes associated with seed maturation (e.g. those encoding late embryogenesis abundant proteins and transcription factors regulating seed development and maturation such as ABI3, FUS3, LEC1 and WRI1 homologs) were put in context with Arabidopsis thaliana seed maturation and indicated that M+ seeds may desiccate and mature faster than NM. The 1901 transcriptomic DEG set GO-terms had almost 90% overlap with the 2191 genome-derived DEG GO-terms. CONCLUSIONS: Whilst there was only modest overlap of DEGs identified in reference-free versus -dependent approaches, the resulting GO analysis was concordant in both approaches. The identified differences in dry seed transcriptomes suggest mechanisms underpinning previously identified contrasts between morphology and germination behaviour of M+ and NM seeds.

Draft genome sequence of Leucobacter chromiiresistens, an extremely chromium-tolerant strain.

Here we present the draft genome of Leucobacter chromiiresistens. This is the first genome sequence of an organism belonging to the genus Leucobacter. L. chromiiresistens was sequenced due to its capability to tolerate up to 300 mM Cr(VI) in the medium, which is so far a unique feature for microorganisms.

Prediction and validation of promoters involved in the abscisic acid response in Physcomitrella patens.

Detection of cis-regulatory elements, such as transcription factor binding sites (TFBS), through utilization of ortholog conservation is possible only if genomic data from closely related organisms are available. An alternative approach is the detection of TFBS based on their overrepresentation in promoters of co-regulated genes. However, this approach usually suffers from a high rate of false-positive prediction. Here, we have conducted a case study using promoters of genes known to be strongly induced by the phytohormone abscisic acid (ABA) in the model plant Physcomitrella patens, a moss. Putative TFBS were detected using three de novo motif detection tools in a strict consensus approach. The resulting motifs were validated using data from microarray expression profiling and were able to predict ABA-induced genes with high specificity (90.48%) at mediocre sensitivity (33.33%). In addition, 27 genes predicted to contain ABA-responsive TFBS were validated using real-time PCR. Here, a total of 37% of the genes could be shown to be induced upon ABA treatment, while 70% were found to be regulated by ABA. We conclude that the consensus approach for motif detection using co-regulation information can be used to identify genes that are regulated under a given stimulus. In terms of evolution, we find that the ABA response has apparently been conserved since the first land plants on the level of families involved in transcriptional regulation.