Plant material
Twelve individual plants of Ceropegia sandersonii, commonly known as Giant Ceropegia, were purchased from Paul Shirley Succulents (https://www.paulshirleysucculents.nl/) and cultivated in the Hortus botanicus Leiden, The Netherlands.
Fixation of flowers for micromorphology (micro-CT, SEM)
Fresh mature flowers and buds were harvested in different stages and fixed with standard formalin-aceto-alcohol (FAA: absolute ethanol, 90%; glacial acetic acid, 5%, formalin; 5% acetic acid). The samples were stored at room temperature until further use.
Scanning electron microscopy (SEM)
Floral buds at different developmental stages were dissected in 70% ethanol and subsequently washed twice each with 70% and 96% ethanol. The material was then transferred to 100% acetone which was changed after 30 min. Subsequently, the samples were critical point dried using liquid CO2 with a Leica EM CPD300 critical point dryer (Leica Microsystems, Wetzlar, Germany), and mounted on aluminium stubs using either double-sided carbon tape or Leit-C carbon cement. A Quorum Q150TS sputter coater (Quorum Technologies, Laughton, East Sussex, UK) was used to coat the samples with a 20nm thin layer of Platina-Palladium. Imaging of samples was performed with a JEOL JSM–7600 F Field Emission Scanning Electron Microscope (JEOL Ltd., Tokyo, Japan).
3D X-ray micro-computer tomography (micro-CT)
Fresh mature flowers were stained for 5 days with 1% phosphotungstic acid (PTA) in 70% ethanol as contrast agent whereby PTA was change daily. After staining, flowers were washed twice with 70% ethanol and embedded in 1.5% low melting point agarose in a plastic container. Embedded flowers were scanned using a Zeiss Xradia 510 Versa 3D X-ray equipped with a sealed transmission X-ray source (settings: voltage/power: 40 kV/3 W; source current: 75 μA; exposure time: 2 sec; camera binning 2; optical magnification: 4x; pixel size: 3.5 μm; total exposure time: ~3–2 h). Single 2D images were stacked to build a 3D image which was processed using Avizo 3D software version 8.1 (Thermo Scientific™).
RNA isolation
Early floral buds (<3 mm) and mature flowers (first day of anthesis) were harvested from at least three different plant individuals. Mature flowers were dissected to sepals, petals (tip, tube, base), gynostegium, and corona. Similar tissue types of different mature flowers were pooled to reach the required amount of tissue needed for RNA isolation (~30–90 mg). All samples were snap-frozen in liquid nitrogen and stored at –80°C until RNA isolation. Plant tissue was ground in a 2.2 ml micro centrifuge tube with a 7 mm glass bead using TissueLyser II (QIAGEN). Total RNA was extracted using the RNeasy Plant Mini Kit (QIAGEN) and an adapted protocol which included a step to digest single- and double-stranded DNA (DNase I; Amp Grade, Invitrogen 1U/µl). The amount of RNA for RT-PCR was measured using a NanoDrop (ND–1000 Spectrophotometer, Marshall Scientific). Samples used for RNA sequencing were further quality checked by determining the integrity (RNA Integrity Number; RIN) using the Plant RNA nano protocol on an Agilent 2100 Bioanalyzer (Agilent Technologies). Only samples with a RIN >9.5 were used for sequencing. To obtain a full petal sample, RNA extracted separately from petal tips, tubes and bases was pooled after quality control. RNA samples were sent to Beijing Genomics Institute (BGI) for NGS sequencing on an Illumina HiSeq platform.
Transcriptome analyses and MADS-box gene identification
An in-house designed bioinformatics pipeline was used for quality control, assembly, annotation and differential expression analyses. Raw read pairs obtained from NGS sequencing were quality checked using FastQC v0.10.1 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Low quality reads were trimmed or removed with Trimmomatic v0.32 (45) and remaining reads were again quality checked. Trinity v2.5.1 (46) was used for de novo assembly of cleaned reads, and alignment of reads was performed using Bowtie2 v2–2.3.3.1 (47). CDHIT-EST (48) was used to cluster contigs and to create consensus sequences (unigenes) against which the reads were aligned using RSEM v1.3.0 (49) and Bowtie2 v2–2.3.3.1. Raw counts per read were quantified and a count table was generated with an in house designed bioinformatics script (https://github.com/naturalis/orchid-transcriptome-pipeline/tree/master/Scripts).
All Ceropegia sandersonii gene sequences were blasted against a local database of Gentianales MADS-box gene homologs (Rubiaceae: Coffea arabica, Gardenia jasminoides; Gentianaceae: Gentiana scabra; Apocynaceae: Allamanda cathartica, Catharanthus roseus),, which was created by retrieving DNA sequences from NCBI GenBank. Sequences for Actin were also retrieved to identify the C. sandersonii actin homolog required as control gene for the RT-PCR experiments (see below). For identified C. sandersonii MADS-box gene homologs (see Additional file 2), expression differences between sample types and among sample replicates were visualized in a heatmap (based on the count table for the according sequences; Additional file 6) using an in house designed bioinformatic script (https://github.com/naturalis/orchid-transcriptome-pipeline/tree/master/Scripts). With this script, the number of matches between a specific read in the transcriptomes and a reference gene was scored. In a separately generated ‘Color Key and Histogram’, the number of hits was translated to color codes. Color codes were based on the number of counts per gene and sample divided by the total number of counts, where cold colors correspond with a relative low number and war colors with a relative high number. Additional differential expression analyses were carried out using DESeq in R to calculate the log2 fold change of expression of the genes investigated in the different floral organs and developmental phases. Six pairwise tests between the four sample types ‘early buds’, ‘sepals’, ‘petals’, and ‘gynostegium’ were performed (see Additional file 4). These tests identify those genes with significant differential expression (minimum log2 fold change of 0.25) between a given pair of sample types. All samples had >100 counts so that a cut-off for the analyses was not necessary.
Phylogeny of MADS-box gene lineages
To assess phylogenetic relationships between Gentianales MADS-box gene lineages, all publicly available DNA sequences (see Additional file 3) plus those from the newly identified Ceropegia sandersonii homologs (see Additional file 2) were translated to amino acids in the correct translation frame by using translate-protein tools (http://reverse-complement.com/translate-protein/ROOT/), loaded into Geneious Prime® 2019 v.2.3 (www.geneious.com) and aligned with the best matching open reading frame (from start to stop codon) using the ‘Geneious alignment’ function. The created alignment was trimmed down to the most conserved regions (protein domains and amino acid motifs) to ensure all sequences had similar length; regions that did not align were removed prior to further analysis. Separate alignments were made for each MADS-box gene subfamily, and combined in a Maximum Likelihood phylogenetic analysis using the PhyML plugin (50) with the following settings: substitution model ‘Blosum62’; Bootstrap ’100’; proportion of invariable sites ’0, fixed’; number of substitution rate categories ’4’; gamma distribution parameter ’0, estimated’; optimize ‘Topology/length/rate’; topology search ‘NNI (default, fast)‘. As outgroup, sequences which did not fall into the major subfamily clades were chosen, i.e. SEEDSTICK sequences of Gentiana scabra (GsSTK1) and the according homolog of Ceropegia sandersonii (CsanSTK1)..
Primer design and cDNA Synthesis
Six MADS-box gene homologs (CsanFUL2, CsanTM6, CsanGLO, CsanAG2, CsanAGL6, CsanSEP1) were chosen for further analyses as they showed most divergent expression patterns in the transcriptome analyses. Detailed expression patterns were investigated in early floral buds and mature sepal, petal, gynostegium, and corona tissue. Primers were designed using the online software Primer3Plus (https://primer3plus.com/cgi-bin/dev/primer3plus.cgi) with the following settings: max Poly-X = 3; CG-Clamp = 2; max. End GC = 2). All primer pairs (see Additional file 1) were screened for specificity in a gradient PCR reaction with a reaction mixture (25 μl) containing 10x CoralLoad Buffer (Qiagen), 25mM MgCl2 (Qiagen), 100 mM Bovine Serum Albumin, Acetylated-BSA (Promega), 1.25x DMSO (Qiagen), 5x Q-Solution (Qiagen), 0.2 μM of each primer (IDT), 2.5 mM dNTPs (Qiagen), and 1.25 units/50 µl DNA Taq Polymerase (Qiagen), plus 100 ng cDNA template. MQ water (Ultrapure) was used to reach the final volume of 25 µl. The amplification started with an initial denaturation step of 6 min at 94°C, continued with 38 cycles of three steps consisting of 1 min at 94°C, 1 min at 52°C–55°C, and 2 min at 72°C, and was finalized with one amplification step of 12 min at 72°C.
From RNA extracted of early floral buds, and sepals, petals, gynostegia, and corona tissue of mature flowers, cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen). As a first step the reaction mix containing the RNA template (10 pg–5 µg), 1 µl 10 mM dNTP Mix, 50 µM Oligo (dT)20, and sterile distilled water (final volume: 14 µl) was heated at 65C for 5 minutes. After incubation on ice for at least one minute, the mix was briefly centrifuged, then 5X first-stand buffer (4 µl), 0.1 MDTT (1 µl), and 200 units/µL Superscript III (1 µl) were added. This mixture was then incubated at 55C for 50 minutes to dissolve the RNA template while avoiding formation of secondary structures; heating at 70C for 15 minutes inactivated the reaction. A reaction mix without RNA template (non-template control, NTC) was used as negative control. Quantity and quality of cDNA were measured via nanodrop (ND–1000 Spectrophotometer, Marshall Scientific). A total of 90 ng per cDNA sample was then used for PCR-amplification with sequence specific primers (see below). The Actin gene homolog (CsanACT) was used as positive control and a non-template reaction DNA (NTC) was used as negative control.
Reverse Transcription PCR (RT-PCR)
Semi-quantitative reverse transcription-PCR (RT-PCR) was performed for six selected MADS-box gene homologs, i.e. CsanFUL2, CsanTM6, CsanGLO, CsanAG2, CsanAGL6, and CsanSEP1. The thermal cycling regime used in the RT-PCR reaction was similar as for the gradient PCR (see above); however, the annealing temperature was set to 52°C as this temperature yielded the best results and most specific products in the gradient PCR. Actin was amplified as a positive control; the negative control was a non-template control (NTC). All reactions were carried out in a CFX384 Touch Real-Time PCR system thermocycler (Biorad). The PCR products were run on a 1% agarose gel with 1x TAE and a 1 kb plus GeneRulerTM (Thermo Scientific) as ladder. The gel was stained with ethidium bromide and digitally photographed using a gel doc (Ultima 10si, Isogen Life Science).