Complementarity of medium-throughput in situ RNA hybridization and tissue-specific transcriptomics: case study of Arabidopsis seed development kinetics

Abstract

The rationale of this study is to compare and integrate two heterologous datasets intended to unravel the spatiotemporal specificities of gene expression in a rapidly growing and complex organ. We implemented medium-throughput RNA in situ hybridization (ISH) for 39 genes mainly corresponding to cell wall proteins for which we have particular interest, selected (i) on their sequence identity (24 class III peroxidase multigenic family members and 15 additional genes used as positive controls) and (ii) on their expression levels in a publicly available Arabidopsis thaliana seed tissue-specific transcriptomics study. The specificity of the hybridization signals was carefully studied, and ISH results obtained for the 39 selected genes were systematically compared with tissue-specific transcriptomics for 5 seed developmental stages. Integration of results illustrates the complementarity of both datasets. The tissue-specific transcriptomics provides high-throughput possibilities whereas ISH provides high spatial resolution. Moreover, depending on the tissues and the developmental stages considered, one or the other technique appears more sensitive than the other. For each tissue/developmental stage, we finally determined tissue-specific transcriptomic threshold values compatible with the spatiotemporally-specific detection limits of ISH for lists of hundreds to tens-of-thousands of genes.

Figure 1. Flowchart of the method used for medium throughput in situ hybridization based on/compared to published tissue-specific transcriptomic data.

This flowchart is labelled according to the paragraph numbers appearing in the Material and Methods and Supplementary Methods sections.

Figure 2. Illustration of the in situ hybridization results obtained for a selection of genes and developmental stages.

Results for 10 AtPRXs (a–j) and for 10 non peroxidase genes (k–t) are displayed in the decreasing order of their maximal individual tissue-specific transcriptomic expression values. For each gene and each developmental stage, a screen copy of the seed eFP browser tissue-specific transcriptomic map including the individual absolute heatmap scale (red-to-yellow colour codes correspond to high-to-low expression values) that is different for each gene is compared with the corresponding ISH results (for both the antisense and sense probes). The final corresponding new ISH map (re-coloured after original drawings from ref. available at Seedgenenetwork http://estdb.biology.ucla.edu/seed/) gave increased cellular resolution with a unique colour code for all genes (red corresponding to visually detected strong signals, orange corresponding to moderate signals and white corresponding to the absence of detected ISH signal). Note that the naturally brown colour of seed coat does not correspond to ISH signal. The full ISH results corresponding to a selection of 5 developmental stages for all 39 genes can be found in Supplementary Figs. S4–S42. Scale bars: 100 μm. Abbreviations: AS, anti-sense; CZEN, chalazal endosperm; CZSC, chalazal seed coat; EP, embryo proper; MEN, micropylar endosperm; PEN, peripheral endosperm; S, sense; SC, seed coat; SUS, suspensor.

Figure 3. Thresholds of tissue-specific transcriptomic values compatible with ISH are specific to each developmental stage/tissue.

For each developmental stage/tissue tissue-specific transcriptomic sample ((a,b) globular developmental stage; (c,d) mature green developmental stage; (a,c) embryo proper tissue; (b,d) seed coat tissue), the 23,933 genes present on the microarray were classified according to their tissue-specific transcriptomic expression values. The range 0–44 corresponded to the expression values below the 45 detection limit of the transcriptomic study (55–65% of the genes); the range 45–299 corresponded to the expression values between the 45 detection limit of the transcriptomics and the 300 arbitrary cut-off that we initially defined for our ISH study; the other ranges were arbitrarily set to allow the distribution of the genes in various expression value groups. The resulting number of genes within each range was plotted on individual histograms. All the genes analysed by ISH in this study were positioned above the histograms according to their individual tissue-specific transcriptomic expression value. The ISH results were colour-coded in red/orange/white as in Fig. 2, according to Supplementary Figs. S4–S42. The deducted total number of genes compatible, to various extend, with ISH was posted on the top of each graph within double arrows using shades of the same colour coding. This clearly illustrates that the sensitivity of ISH is dependent on spatiotemporal parameters and the complementarity of transcriptomics and ISH. The corresponding histograms for all developmental stages and tissues are available in Supplementary Figs. S46–S51.