Nam V. Hoang, Goh Choe, Yi Zheng, Ana Cecilia Aliaga Fandino, Inyoung Sung, Jaeryung Hur, Muhammad Kamran, Chulmin Park, Hyoujin Kim, Hongryul Ahn, Sun Kim, Zhangjun Fei, Ji-Young Lee
Laser Capture Microdissection (LCM) based RNA sequencing
] was used with minor modifications. Briefly, tissue block (1.2×1.2×1.2 cm3) containing cambium was obtained from the thickest part of a root. Dissected tissue block was immediately fixed in ice-chilled Famer’s fixative (75% ethanol, 25% glacial acetic acid), and stored overnight at 4°C. The fixed tissue samples were gradually dehydrated with ethanol and embedded with Steedman’s wax (9:1 (w/w) of polyethylene glycol distearate and 1-hexadecanol). To obtain tissue sample slides for LCM, about 2 mm of tissue sections were trimmed off (100 steps 20 μm-thickness using a rotary microtome) to reach to the inner part of the tissue block, and then 20 μm-thick tissue sections were prepared on an Arcturus PEN membrane glass slide.
Before LCM, Steedman’s wax was removed by incubating the slide in ethanol for 10 min twice, dried for 5 min, and immediately processed for LCM using an Arcturus Veritas LCM microdissection system. Areas of interest within the tissue were marked using the drawing tool in the Arcturus software to cut-out tissues using a UV laser (355 nm), and the dissected tissues were collected onto the Arcturus Capsure Macro LCM cap. Total RNA was extracted with the Arcturus PicoPure RNA isolation kit following the manufacturer’s instructions. RNA quality was evaluated using RNA Pico chip on an Agilent 2100 Bioanalyzer system (NICEM, Seoul National University, the Republic of Korea).
RNA-seq libraries were prepared using the Nugen Ovation RNA-seq System 1-16 for model organisms (Arabidopsis). An aliquot of 10 ng of each total RNA sample was used as a starting material, and final sequencing libraries were prepared by amplifying the constructed libraries using PCR with empirically determined amplification cycles. Paired-end sequencing data were obtained with Illumina HiSeq 4000 (CLC Genomics and Epigenomics Core Facility, Weill Cornell Medical College, USA).
RNA-seq data analysis to identify novel genes and lincRNAs
]. After trimming, reads shorter than 80 bp were discarded. The resulting reads were aligned to ribosomal RNA (rRNA) database [
] using Bowtie [
] allowing up to three mismatches. The aligned reads were discarded and the remaining high-quality cleaned reads were assembled de novo using Trinity [
] with “min_kmer_cov” set to 5. The resulting assembled contigs were then compared to the GenBank Nucleotide (nt) and non-redundant proteins (nr) databases using the BLAST+ program, and those having hits only to sequences from viruses, bacteria, and archaea were discarded. To remove redundancies in the Trinity-assembled contigs, the contigs were further processed using CD-HIT [
] with minimum sequence percent identify set to 95%.
] with “npaths” set to 0. Contigs overlapping with known gene regions in the reference genome were discarded. The coding potential of the remaining contigs was calculated using Coding Potential Calculator (CPC) [
], and contigs with CPC score > 0 and containing an open reading frame (ORF) > 300 bp were identified as novel radish genes. To annotate the newly identified genes, their protein sequences were compared to GenBank nr, the Arabidopsis protein and UniProt (Swiss-Prot and TrEMBL; https://www.uniprot.org/) databases using DIAMOND [
] with ‘-evalue’ set to 1e-4, as well as the InterPro database using InterProScan (v5.19-58.0) [
]. Arabidopsis genes hit with the lowest e-value were considered the putative ortholog of radish transcripts (Data S1D). GO annotations were obtained using Blast2GO (version 2.5.0) [
] based on the BLAST results against the GenBank nr database and results from the InterProScan analysis.
] with high-quality cleaned RNA-seq read pairs that were aligned to the radish reference genome. Expression of the assembled transcripts was measured and normalized to the number of fragments per kilobase of exon per million mapped fragments (FPKM) using the Cuffnorm program in the Cufflinks package. The coding potential of the assembled transcripts was calculated using CPC [
]. LincRNAs were defined as transcripts of a minimum length of 200 nt, having CPC score of 100 aa and to be expressed at an FPKM > 0.1. Tissue specificity of lincRNAs was calculated for all tissues and developmental stages by employing the entropy-based measure [
], which defines the expression patterns using Jensen-Shannon (JS) divergence score. The analysis and validation of lincRNAs are available in the Methods S1.
RT-PCR and gene cloning from radish root RNA for validation experiments
RT-PCR validation of lincRNAs and cloning for protein coding genes were proceeded using the cDNA library prepared from 7-week-old radish root RNA. Total RNA was isolated from roots harvested after 7 weeks post seed planting using the RNeasy Plant Mini kit (QIAGEN). Reverse transcription reaction was carried out by mixing 2 ng of the total RNA, 1 μl of Oligo(dT)16 (10 pM/μl), 0.5 μl of dNTP (10mM), and DEPC-treated water to a total volume of 13 μl. The mixture was incubated at 65°C for 5 min and then 4 μl of 5 × first strand buffer was added, together with 1 μl of each of DTT, DEPC-treated water, and SuperScript III reverse transcriptase (Invitrogen). This 20 μl reaction mixture was incubated at 50°C for 1 hour and then 70°C for 15 min.
] were used as an experimental control, while RNA sample without reverse transcription step was used as negative control. Primers were designed by Primer3 (v.0.4.0) [
] using Tm (58-63°C), primer length (18-27 nt) and product size range (150-400 nt).
RNA in situ hybridization
To prepare RNA probes for in situ hybridizations, a PCR product with T7 binding sequence was purified and concentrated to 100 ng/μl. The in vitro transcription reaction using ∼900 ng of this PCR product as template was carried out employing T7 RNA polymerase DIG labeling Kit (Sigma), following the manufacturer’s instruction. The reaction was incubated for 37°C for 3 hours and then DNA was removed from the mixture by adding 1 μl of RNase inhibitor, 5 μl of 10 × DNase I buffer, 22 μl of RNase-free water and 2 μl of RNase-free DNase I (Sigma). The mixture was incubated 15 min at 37°C. Finally, the probe was let to precipitate overnight at −80°C after adding 6 μl of 4M LiCl, 2 μl of 0.5M EDTA (pH 8.0) and 180 μl of absolute ethanol. The dried RNA pellet after centrifugation and washing with 70% of ethanol was resuspended in RNase free water.
The radish root tissue around cambium was cut into a block and immediately submerged in 4% of paraformaldehyde prepared in 1 × Phosphate-Buffered Saline (PBS). After fixation overnight at 4°C, the fixative was discarded, and the tissue block was rinsed with 1 × PBS. Tissue was then dehydrated using the following ethanol series for 1 hour each time: 25%, 50%, 75%, 100%, 100%, and 100%, 75% ethanol/25% Histoclear, 50% ethanol/50% Histoclear, 25%ethanol/75% Histoclear, 100% Histoclear, 100% Histoclear and 100% Histoclear. Finally, a half of the tube was filled with Paraplast chips and incubated overnight at 58°C for infiltration. For the following 4 days, the Paraplast was replaced twice a day and incubated at 58°C. On the fifth day, the radish blocks were placed in a mold with melt Paraplast and solidified at room temperature. Finally, the tissue blocks were cut using a RM 2255 microtome (Leica) creating 15 μm thick slices, which were mounted on a TRUBOND slide and left to dry overnight.
Slides were placed two times in Histoclear for 10 min, then passed through ethanol series (100%, 100%, 95%, 85%, 70%, 50%, 30% in NaCl 0.85%) for 1 min each time, NaCl 0.85% for 2 min, in 0.2M of HCl for 20 min, in DEPC-treated water for 5 min, and in 1 × PBS for 2 min. Slides were then incubated with Pronase (0.135 mg/ml) for 28 min at 37°C and then the reaction was stopped by submerging the slides in glycine 0.2% in 1 × PBS, followed by a 2 min submersion in 1 × PBS and a 10-min submersion in 4% paraformaldehyde fixative. Once, the fixation was finished, slides were rinsed for 2 min in 1 × PBS, then submerged in NaCl 0.85% for 2 min, and finally the ethanol series was repeated backward for dehydration of the tissue.
For hybridization, a chamber was prepared by putting paper towels soaked in 50% formamide. Slides were placed in the chamber covered with 250 μl of prehybridization solution (50% formamide, 1 × salt solution, 1 × Denhardt’s, 200 μg/ml of tRNA, 10 U/ml of RNase inhibitor). Slides were incubated for 1 hour at room temperature and then another 1 hour at 45°C. During this prehybridization, DIG-labeled RNA probes were denatured for 1 min at 80°C. Once the prehybridization incubation was finished, 250 μl of the hybridization solution (50% formamide, 1.25 × salts, 12.55 dextran sulfate, 250 μg/ml tRNA, 1.25 × Denhardt’s, 12.5 U/ml of RNase inhibitor, and DEPC water) with 0.4 μg/ml/kb of RNA probe was added to the slide glass. The slide was covered with a coverslip and incubated in the formamide chamber for 24 hours at 45°C.
After the hybridization, the coverslips were removed and slides were the placed in a rack which stands in a jar with 0.2 × SSC for 1 hour at 55°C. Solution was later replaced and incubated for another hour. After finishing the washes, slides were rinsed in NTE solution (0.5M NaCl, 10mM Tris pH 8.0, and 5mM EDTA) and incubated for 30 min at 37°C in a solution of 10 μg/ml RNase A in NTE solution. Slides were rinsed for 5 min in NTE for stopping the reaction and incubated one more hour in 0.2 × SSC at 55°C. Finally, slides were rinsed in 1 × PBS.
Sliced were placed in blocking solution (100mM Tris, 100mM NaCl, 1% blocking reagent of DIG Nucleic Acid detection kit, Roche) with gentle agitation for 45 min. Then, the blocking buffer was replaced by buffer A (100mM Tris, 100mM NaCl, 1% BSA, and 0.3% Triton X-100) and slides were incubated for another 45 min. Once, the slides were ready, antibody conjugate from the DIG Nucleic Acid detection kit was spread on the slides in a 1:1000 ratio in buffer A. Slides were incubated for 1 hour at room temperature in a chamber with high humidity. Slides were then washed in buffer A with gentle agitation for 20 min, three times and in detection buffer (100mM Tris pH 9.5, 100mM NaCl, and 50mM MgCl2) twice for 5 min. Finally, 500 μl of color substrate (200 μl of NBT/BCIP solution DIG Nucleic Acid detection kit in 10 mL of detection buffer and 100 μl of levamisole) was added to the slide, covered with a coverslip, and incubated at room temperature for 36 hours in the chamber protected from the light. Images were taken in a Nikon eclipse Ni light microscope.
Growth experiments in soils with varying water contents
Radish seeds from two inbred lines 216 and 218 were surface sterilized in 1% bleach for 1 min, rinsed 6 × in water and germinated on moist filter paper in the dark on the bench. Three-day-old seedlings were transferred into pots (20x20x20 cm3) filled with an equal amount of soil. The pots were placed on a bench in a glasshouse (randomly arranged) for 10 weeks. Controlling soil water contents started after three weeks of growth by controlling irrigation to maintain volumetric water content percentage (VWC, %) to ∼45% (T1), 33%–36% (T2), and 22%–26% (T3). VWC was measured three times a week using a portable EC-5 soil moisture sensor (Meter Environment, USA). The stomatal conductance (gs) and assimilation rate (A), were measured using a portable infrared data analysis system (LI-COR 6400, LI-COR Biosciences). PAR was set at 1000 μmol m-2 s-1, and the chamber CO2 concentration was kept at an ambient CO2 concentration of 400 μmol mol−1. A section of a youngest fully expanded leaf was placed in the chamber while still attached to the plant. Stable measurements for each replicate were done between 11:00 am and 2:00 pm. At 10 weeks, shoots and roots were harvested, and fresh mass was measured.
Arabidopsis lines, WT (Col-0), erf-1, and 35S::ERF-1: GR were grown in trays with 32 pots under 16-hour light/8-hour dark and 125 μmol m-2 s-1. Trays were randomized on a daily basis to avoid any bias caused by subtle differences in growth conditions. The controlled irrigation started after 9 days of seed planting, ∼two times a week, depending on the need to keep a specific VWC with three different levels of water treatment: T1 (∼45%), T2 (∼32%), and T3 (∼20%). For dex treatment, plants were treated with dex (10 μM) once a week. VWC was measured every day using a portable EC-5 soil moisture sensor.
Perturbation experiments and quantitative (q)RT-PCR
For the in vivo cambium GRN inference, loss-of-function mutant lines were obtained for nine genes, ASL9, WOX4, PXY, LHW, ERF-1, ERF2, MYB15, WRKY33 and STZ, by CRISPR-CAS9 and T-DNA insertion. All T-DNA insertion lines were confirmed by PCR and Sanger sequencing. Mutant lines and WT (Col-0) used were grown on MS medium for 10 DAT from 4°C to the growth chamber. For further gene expression and phenotypic analyses in the ERF-1 perturbation lines, the WT, erf-1, and 35S::ERF-1:GR line was grown on MS medium for 14 DAT. Here, a half of 35S::ERF-1:GR seedlings grown for 9 DAT were treated with 10 μM of dex for another 5 days.
]. List of primer sequences is provided in Data S5E. The expression level of each gene was normalized against that of the reference gene GAPDH, and compared against that of the control WT. Three technical replicates were run for each sample. Data were analyzed using Student’s t tests in Bio-Rad CFX Maestro software 1.0. Expression values were quoted as Mean ± SEM, and p
Analysis of ERF-1 function through network dynamics
]. The seed set in propagation was made by selecting the highly expressed genes for each ERF-1-altered condition (WT, erf-1 mutant, ERF-1:GR control, and ERF-1:GR treated). The qRT-PCR values were normalized by calculating z-scores for each gene across the conditions. Then, the top five highly expressed genes were selected as seeds for each condition as follows, in WT: ERF105, STZ, WRKY18, WRKY33, and WRKY46; in erf-1 mutant: ERF2, LHW, MYB15, SCL7, and WOX14; in ERF-1:GR control: ERF-1, STZ, WRKY18, WRKY33, and WRKY46; and in ERF-1:GR treated: ERF-1, ERF105, LHW, SCL7, and WOX14. Using the seed genes and expression values for each condition, the network propagation was performed on the radish root-specific GRN as described in Methods S1. Then, genes in the GRN were ranked by order of the network propagation scores, resulting in four condition-specific rank profiles (RWT, Rerf-1, RERF-1GRcontrol, and RERF-1GRtreated). To elucidate the biological phenotypes of these network dynamics results, we performed an association analysis of four condition-specific rank profiles with data from the two radish lines (216 and 218). For each radish line, three time-point RNA-seq values were normalized by calculating z-scores, and the mean absolute deviation (MAD) value of gene was calculated. Then, the values were ranked in descending order of MAD value, two rank profiles (R216 and R218). To quantify the association between the ERF-1-altered networks and the radish root-specific networks, Spearman’s rank correlation coefficient was calculated for the network propagation rank profiles of each ERF-1 altered condition and the MAD rank profiles of each radish genotype.