Identification of CrGLK and CrLHCB2.2
AtGLK1 (NP_565476.1) and AtGLK2 (NP_199232.1) protein sequences were used in a Protein Basic Local Alignment Search Tool (BLASTP) search in the C. roseus v.2 translated transcriptome (Franke et al., 2019). The top two hits, CRO_T112335 (41% identity, 99.5% overlap) and CRO_T119410 (72% identity, 18% overlap), were aligned against other GLK, Arabidopsis Response Regulator (ARR), and Arabidopsis Pseudo Response Regulator (APRR) proteins using CLC Main Workbench 21.0.3 (default parameters: gap open cost = 10, gap extension cost = 1). Protein sequences used in the alignment were downloaded from GenBank with accession numbers: AtGLK1 (Arabidopsis thaliana, NP_565476.1); AtGLK2 (Arabidopsis thaliana, NP_199232.1); ZmG2 (Zea mays, AAK50392.1); ZmGLK1 (Zea mays, AAK50391.1); OsGLK1 (Oryza sativa, BAD62070.1); OsGLK2 (Oryza sativa, BAD81484.1); SlGLK1 (Solanum lycopersicum, AFF60404.1); SlGLK2 (Solanum lycopersicum, AFN69447.1); AtAPRR2 (Arabidopsis thaliana, AT4G18020.1); SlAPRR2-like (Solanum lycopersicum, AFX68729.1); CaAPRR2-like (Capsicum annuum, AGF37241.1); AtARR1 (Arabidopsis thaliana, AEE75875.1, outgroup). Domains were annotated on the amino acid alignment according to Fitter et al. 2002 and Makino et al. 2000 (Fitter et al., 2002; Makino et al., 2000).
To identify a homologue of the Light Harvesting Complex Subunit B2.2 (LHCB2.2) in C. roseus, we performed a BLASTP search in the C. roseus v.2 translated transcriptome (Franke et al., 2019) with the AtLHCB2.2 protein sequence (AAM13371.1). This search returned one highly homologous protein, CRO_T101917 (89% identity, 100% overlap), which was named as CrLHCB2.2.
Analysis of tissue-specific RNAseq reads
RNAseq reads (SRP005953) from differing tissue types were downloaded from the European Nucleotide Archive (ENA) and imported into KBase for analysis: Flowers (SRR122239, CRA_AA); Cell Suspension Cultures (MJ 0h) (SRR122250, CRA_AL); Sterile Seedling (SRR122243, CRA_AE); Mature leaf (SRR122251, CRA_AM); Immature leaf (SRR122252, CRA_AN); Stem (SRR122253, CRA_AO); Root (SRR122254, CRA_AP); Hairy root (SRR122257, CRA_AS) (Arkin et al., 2018; Góngora-Castillo et al., 2012). Adapters (TruSeq3-SE) were clipped and sequences were trimmed using Trimmomatic v0.36 default parameters (sliding window size = 4; sliding window minimum quality = 15) (Bolger, Lohse, & Usadel, 2014). Quality of sequences was confirmed using FastQC v0.11.5 (Wingett & Andrews, 2018). Trimmed sequences were aligned to the C. roseus genome v. 2 (Franke et al., 2019) using HISAT2 – v2.1.0 with default parameters (Kim, Langmead, & Salzberg, 2015). Transcripts were assembled and Transcripts per Million (TPM) values calculated using StringTie with default parameters (Pertea et al., 2015). The KBase narrative can be accessed at https://kbase.us/n/95510/53/. Z-scores ((value – mean) / standard deviation) were calculated for the following genes in the varying tissue types: T16H2, CRO_T110598; 16OMT, CRO_T110596; T3O, CRO_T113994; T3R, CRO_T124298; NMT, CRO_T111273; D4H, CRO_T127167; DAT, CRO_T120021; CrGLK, CRO_T112335; CrGATA1, CRO_T134526; G10H, CRO_T133061; TDC, CRO_T125328; STR, CRO_T125329; CS/HL1, CRO_T139139.
Gateway cloning for virus-induced gene silencing
For silencing experiments, two fragments from CrGLK (Fragment 1 = 188 bp, 519–706 from start codon; Fragment 2 = 216 bp, 1040–1255 from start codon) were amplified from either C. roseus cDNA or a previously cloned plasmid using primers listed in Table S1. Fragments were gel extracted and then cloned into pDONR221 and pTRV2-GATEWAY (Yule Liu, Schiff, & Dinesh-Kumar, 2002) using Gateway® Cloning (Invitrogen). As a positive silencing control, a 499 bp fragment (2450–2948 from start codon) of the protoporphyrin IX Magnesium Chelatase Subunit H (CHLH) (David K Liscombe & O’Connor, 2011) was amplified from C. roseus cDNA and cloned into pTRV2-GATEWAY. As a negative control, a 471 bp fragment (436–906 from start codon) of the Green Fluorescent protein (GFP) was amplified from pPD95_77 (Addgene plasmid #1495) and cloned into pTRV2-GATEWAY. This fragment was checked for off-targets by using it as a query sequence in the Sol Genomics Network (SGN) VIGS tool (n-mer = 20, mismatches = 0) (Fernandez-Pozo, Rosli, Martin, & Mueller, 2015). No matches in the C. roseus v2 transcriptome were found, confirming that the fragment is unlikely to have off-target effects in C. roseus. All sequences were confirmed using Sanger Sequencing at Azenta Life Sciences. All plasmids were electroporated into Agrobacterium tumefaciens GV3101 (pMP90). pTRV2-CHLH and pTRV2-CrGLK (fragment 1) were deposited at Addgene (IDs: 203886, 203888).
Golden Gate Modular Cloning of CrGLK for overexpression
The CrGLK overexpression plasmid and vindoline pathway promoter reporter plasmids were constructed using Golden Gate Modular Cloning. Specific parts are from the MoClo toolkit (Addgene Kit #1000000044) (Weber, Engler, Gruetzner, Werner, & Marillonnet, 2011) or MoClo Plant Parts Kit (Addgene Kit #1000000047) (Engler et al., 2014) unless otherwise noted.
The CrGLK coding sequence (CRO_T112335) was amplified from C. roseus var. Little Bright Eye cDNA prepared from seedlings using Phusion High-Fidelity DNA Polymerase (New England BioLabs) and primers listed in Table S1. One silent mutation was introduced into the coding sequence of CrGLK to remove a BpiI recognition site and domesticate the sequence for Golden Gate cloning (Table S2). Amplified fragments were visualized on agarose gels and then purified using the Zymoclean™ Gel DNA Recovery Kit. Fragments were ligated into the pICH41308 level zero (L0) CDS backbone, and the coding sequence was confirmed with Sanger sequencing at Azenta Life Sciences. The CrGLK coding sequence was then amplified without a stop codon from this plasmid and moved into the pAGM1287 level zero (L0) CDS1ns backbone to allow addition of C-terminal tags in the future, if needed.
To facilitate future cloning of coding sequences with and without C-terminal tags, we constructed a pAGM1301 L0 CT plasmid containing a stop codon, as well as a few extra base-pairs that add a glycine and serine to the C-terminus of a CDS to allow fusion to occur. We constructed a transcriptional unit expressing CrGLK in the pICH47732 Level 1 Forward position 1 vector backbone consisting of the cauliflower mosaic virus (CaMV) 2x35S promoter, the tobacco mosaic virus (TMV) omega 5’UTR, the CrGLK CDS1ns, a stop codon (with an additional glycine and serine at the C-terminus), and the Agrobacterium tumefaciens MAS terminator.
This transcriptional unit was moved into the pSB90 backbone (Addgene plasmid #123187), which includes a VirGN54D gene in the plasmid backbone to enhance Agrobacterium virulence (Mortensen et al., 2019). As a negative control, pSB161 (Addgene plasmid #123197) was used, which contains Beta-glucuronidase (GUS) with an intron under control of the same promoter, 5’UTR, and terminator as for CrGLK (Mortensen et al., 2019).
After preliminary experiments suggesting that CrGLK was inactive when overexpressed, out of precaution, the glycine and serine at the C-terminus of CrGLK, artifacts of cloning the stop codon separately, were removed from the final L2 plasmid using the Q5® Site-Directed Mutagenesis Kit (NEB). The transcriptional unit in this plasmid was again sequence-confirmed with Sanger sequencing.
L2 plasmids were electroporated into Agrobacterium tumefaciens GV3101 (pMP90). L0 were deposited at Addgene (IDs: 203903–203905).
Golden Gate Modular Cloning of vindoline pathway promoters for sequence confirmation
The promoters of the vindoline pathway genes with their 5’UTRs (approximately 1kb upstream of start codon) were amplified in parts using Phusion High-Fidelity DNA Polymerase (New England BioLabs) and sequenced from C. roseus var. Little Bright Eye gDNA using primers listed in Table S1. Mutations were introduced to mutate BpiI and BsaI recognition sites and domesticate the sequences for Golden Gate Modular Cloning (Table S2). Some of these sequences (pT16H2, p16OMT, pT3O, and pD4H) differed from the sequences predicted from the C. roseus genome v.2. The genome was sequenced from the Sunstorm Apricot cultivar, so the discrepancies likely arose from cultivar differences. Amplified promoter + 5’UTR sequences were deposited in Genbank (Accessions: OR052132-OR052138). Sequences were cloned into the pICH41295 L0 promoter + 5’UTR vector and deposited at Addgene (IDs: 203889–203895).
Virus-induced gene silencing (VIGS)
C. roseus var. Little Bright Eye seeds (NESeeds, 0.4 g) were sterilized by submersion in 70% ethanol for 45 seconds, 30% bleach and 1X Triton for 6 minutes, triple-rinsed in sterile water, and incubated in 3% Plant Preservative Mixture (PPM) for 18 hours in the dark. The PPM was decanted, and the seeds were spread on full-strength Gamborg’s media (3.1 g/L Gamborg’s basal salts, 1 ml/L Gamborg’s 1000X vitamins, and 6% micropropagation agar type 1, Phytotechnology Laboratory) inside a sterile Magenta™ Plant Culture Box (Sigma) for germination. Seeds were germinated in the dark at 25–27˚C until seedlings were about 2 cm tall (about 7 days). Seedlings were then transferred to 16-hour light / 8-hour dark photoperiods (red and blue LED lights, ~ 80 µmol m− 2 s− 1) for at least two days. Once seedlings had undergone photomorphogenesis, they were planted in soil (Miracle-Gro) in 2.25” square cells and grown under 16-hour light / 8-hour dark photoperiod (red and blue LED lights, ~ 90 µmol m− 2 s− 1) until two true leaves appeared (about 4–6 weeks).
Once seedlings had two true leaves, they were infected with Agrobacterium tumefaciens according to the pinch-wounding method (David K Liscombe & O’Connor, 2011). A single colony of A. tumefaciens GV3101 (pMP90) harboring pTRV1 or pTRV2 was used to inoculate a 10 mL culture of LB with Gentamycin (10 mg/L, selects for pMP90) and Kanamycin (50 mg/L, selects for pTRV1 and pTRV2-GATEWAY) in a 50 mL conical centrifuge tube. This culture was grown at 26˚C and 250 RPM for two days. It was then pelleted, resuspended in 10 mL of induction media (10.46 g/L Agrobacterium minimal medium (PlantMedia), 100 µM acetosyringone) with antibiotics, and grown for another 3 hours. It was then pelleted again and resuspended in 1 mL of VIGS infiltration media (10 mM MgSO4, 10 mM MES pH 5.8, 200 µM acetosyringone). A. tumefaciens strains containing pTRV1 and pTRV2 plasmids were combined in a 1:1 ratio (OD600 of each strain = 2–4). Modified tweezers were dipped into the A. tumefaciens solution and the plant was pinched three times in the highest internode beneath the shoot apical meristem (dipping into the solution between each pinch).
After infection, plants were kept in the dark for two days before being placed back into a 16-hour light / 8-hour dark photoperiod, either under red and blue LED lights (~ 90 µmol m− 2 s− 1) or white, fluorescent lights (~ 15 µmol m− 2 s− 1). Light measurements are an average of five measurements taken with the Apogee SQ-520 Full Spectrum Smart Quantum Sensor. Plants were grown until two pairs of leaves emerged after silencing and the CHLH-silenced plant exhibited yellow leaves (about 2–3 weeks). At this point, a single leaf from the two youngest leaf pairs were individually harvested (unless otherwise noted) for RNA extraction.
Chlorophyll quantification
Using a 1-hole punch, a 6 mm diameter leaf disc (~ 10 mg) was collected from the 1st and 2nd leaf pairs after VIGS infection. Each leaf disc was placed in a 2 mL screwcap microcentrifuge tube containing ten 3 mm glass beads and was flash frozen in liquid nitrogen. Frozen leaf discs were pulverized in a Mini-BeadBeater-16 (Biospec) for 20 seconds and placed back in liquid nitrogen to keep cold. After 1 mL of 80% acetone was added to the crushed tissue, samples were vortexed briefly, incubated on ice in the dark for 15 minutes, and then centrifuged at 19000 RPM for 2 minutes. Supernatant was transferred to a new 2 mL tube and the extraction was repeated. Supernatants from each extraction were combined. Undiluted supernatant was transferred to a clear-bottom 96-well plate (Greiner Bio-One 65509) to measure absorbance at 645 nm and 663 nm with a Biotek Synergy HT microplate reader. Chlorophyll content was calculated using the formulas (Ördög & Zoltán, 1949): [Chl-a] = 12.7(A663) – 2.69(A645) and [Chl-b] = 22.9(A645) – 4.68(A663) where [Chl-a] and [Chl-b] are in mg/L. Values were converted to µg per cm2 using the extraction volume (2 mL) and leaf disc area (0.28 cm2). Linear range and extraction efficiency of the assay were validated before analysis.
Alkaloid extraction
At the time of harvest, individual leaves were weighed to determine their fresh weight (range: 2–180 mg), placed in a 2 mL screw cap tube containing ten 3 mm glass beads and flash-frozen in liquid nitrogen. While still frozen, tissue was crushed by shaking in a Mini-BeadBeater-16 (Biospec) for 15 seconds. Crushing was repeated four times, placing the samples back on liquid nitrogen between each cycle. Alkaloids were extracted by adding 1 mL of methanol, vortexing, and sonicating samples in an ice bath for 30 minutes (vortexing every 10 minutes). Samples were then centrifuged at 10,000 rcf for 10 minutes. Methanol was removed and the extraction with 1 mL methanol was repeated two more times. Methanol extract was combined for each sample into a 15 mL falcon tube and evaporated using a SpeedVac Concentrator (Savant SC210A, RVT4104, VN100, VLP200). Immediately prior to analysis, samples were resuspended in 20 mL methanol per gram of fresh weight. Samples were vortexed and sonicated to ensure full resuspension. To ensure that there were no particulates in the samples, they were placed at 4˚C for at least 4 hours before being centrifuged at 10,000 rcf for 10 minutes. Supernatant was transferred to a new tube, and centrifugation was repeated. Finally, extract was diluted 1:50 in 50% methanol/50% water.
Alkaloid quantification with HPLC-MS/MS
Quantification of the alkaloids was performed at the Mass Spectrometry Facility at Northeastern University. The Thermo Scientific™ Vanquish HPLC system with a Phenomenex Luna Omega LC column (1.6 µm C18 100 A°, 2.1 × 50 mm) was used for alkaloid separation. The mobile phase consisted of 0.1% formic acid in water (solvent A) and 0.1% in acetonitrile (solvent B). The protocol consisted of 15% solvent B for 0.5 min, a gradient of 15–31% solvent B for 15.5 mins to elute the alkaloids, and finally 98% solvent B for 3 mins. The column was re-equilibrated with 15% solvent B for 4 mins prior to the next injection. The flow rate was 0.3 mL/min, the column temperature was maintained at 35°C, and the injection volume was 1 µL.
The compounds were detected on a Tandem HRMS orbitrap mass analyzer (Thermo Scientific Exploris 240) coupled to an electrospray ionization (H-ESI) source in the positive mode with typical settings (supplementary materials). Quantitative analysis was performed in the full scan MS with data-dependent tandem mass spectrometry (full MS/dd MS2 mode with inclusion list, supplementary materials). The parent ion and the confirmatory fragment ion at the optimal collision energies (CE) of each alkaloid was catharanthine 337.19 → 144.08 (CE 31.5), ajmalicine 353.19 → 144.08 (CE 30), serpentine 349.16 → 263.08 (CE 46), and vindoline 457.23 → 188.11 (CE 34). The data processing and area under the curve of the extracted ion chromatogram (XIC) was performed on Thermo Scientific Xcalibur Version 4.5.474.0.
Extraction and quantification of alkaloids were validated as follows; three extractions were sufficient to extract more than 95% of total alkaloids from 50 mg dry weight leaf tissue (~ 500 mg fresh weight) with the percent recovery measured at 100 +/- 10%. The linear range of the quantified alkaloids was validated with a calibration curve of standards prepared in solvent.
RNA extraction and quantitative PCR
Expression levels of CrGLK and vindoline pathway genes were monitored using quantitative real-time PCR (qRT-PCR) with primers listed in Table S3. mRNA was extracted from liquid nitrogen flash-frozen leaf tissue or seedlings (5 whole seedlings pooled as one biological replicate) placed in a 2 mL screw cap tube containing ten 3 mm glass beads and stored at -80˚C until needed. While still frozen, tissue was crushed by shaking in a Mini-BeadBeater-16 (Biospec) for 15 seconds. Crushing was repeated twice, placing the samples back on liquid nitrogen between each cycle. Afterwards, RNA was extracted with RNAzol-RT (Molecular Research Center) and the Direct-zol RNA Miniprep Plus Kit (Zymo Research) with on-column DNAse treatment to remove genomic DNA. RNA integrity was assessed using agarose gel electrophoresis, and concentration and purity were quantified with a NanoDrop (ND-1000 Spectrophotometer; ThermoScientific). cDNA was synthesized using either the SuperScript II First-Strand Synthesis System (Invitrogen) or the LunaScript RT SuperMix Kit (New England Biolabs) with up to 2.5 µg of RNA, according to manufacturer’s instructions. cDNA was diluted 1:4, and 1 µL was used in a 10 µL reaction with SYBR Green ROX qPCR Master Mix (Qiagen or ABClonal) and 300 nM primers on the MX3000P (Agilent) or CFX96 (Bio-Rad) qPCR instrument using the following thermocycler protocol: 10 min at 95˚C (Taq activation), 30 sec at 95˚C (denaturing), 45 sec at 60˚C (annealing), 30 sec at 72˚C (extension), steps 2–4 repeated for 40 cycles, followed by a melt curve. For G10H, TDC, and STR primers, concentrations were used according to previous optimizations (G10h-forward = 100 nM; G10h-reverse = 600 nM; Tdc forward = 300 nM; Tdc reverse = 600 nM; Str_F = 100 nM; Str_R = 300 nM) (Goklany, Loring, Glick, & Lee-Parsons, 2009; Rizvi, Weaver, Cram, & Lee-Parsons, 2016). Ct values for each biological replicate were calculated as the average of two technical replicates. Transcript levels were normalized to the housekeeping gene, SAND (Pollier, Vanden Bossche, Rischer, & Goossens, 2014), and fold changes relative to the negative control condition were calculated according to the 2−∆∆Ct method (Livak & Schmittgen, 2001). Amplification efficiency for each primer set was confirmed using Ct values over a range of cDNA dilutions and was 100% ± 10% for each gene monitored. Specificity of the primers were confirmed by gel electrophoresis and sequencing. SAND Ct values in no reverse-transcriptase controls were confirmed to be at least 5 Ct values above the respective experimental sample, indicating minimal genomic DNA contamination (Svec, Tichopad, Novosadova, Pfaffl, & Kubista, 2015).
Efficient Agro-mediated seedling infiltration (EASI)
To overexpress CrGLK, C. roseus seedlings were transformed according to the efficient Agro-mediated seedling infiltration (EASI) method (Mortensen et al., 2019) with an A. tumefaciens strain containing a CaMV2x35S driven CrGLK or a CaMV2x35S driven GUS (negative control) at an OD600 of 0.2. After infiltration, seedlings were kept in the dark for 2 days and then moved to continuous red and blue LED lights for 24 hours prior to harvest. Five seedlings were pooled for each biological replicate.
Chloroplast retrograde signaling treatments
C. roseus var. Little Bright Eye (NESeeds, 0.8 g) were sterilized by incubation in 4% Plant Preservative Mixture (PPM) for 18 hours in the dark; the PPM was decanted, and the seeds were spread on full-strength Gamborg’s media with appropriate treatments in Magenta™ Plant Culture Boxes. Lincomycin hydrochloride (Thermo Scientific) was dissolved in water at a concentration of 0.5 M, filter-sterilized, and added to media after autoclaving at a final concentration of 0.5 mM. Norflurazon was dissolved in methanol (Crescent Chemical Co Inc.) at a concentration of 1000 µg/mL and added to media after autoclaving at a final concentration of 5 µM. An equal volume of methanol was added to media for the mock treatment. Norflurazon and mock-treated media were left open in a laminar flow hood for about 1 hour to allow residual methanol to evaporate.
After germination in the dark at 25–27˚C for about 9 days, seedlings were transferred to 16-hour light / 8-hour dark photoperiods (red and blue LED lights, Benchmark MyTemp 65HC Digital Cooling Incubator) at 25˚C for two full days. Seedlings were harvested at the beginning of the third day. Five biological replicates were harvested, each containing a pool of five seedlings.
Statistical analysis
The effect of CrGLK-silencing, light intensity, and developmental state on chlorophyll levels, alkaloid levels, and gene expression was analyzed using a univariate general linear model with IBM SPSS Statistics v. 28.0.0.0. “Leaf developmental state” was chosen as a within-subject repeated measure while CrGLK-silencing and light were between-subject fixed factors. For each dependent variable, a full factorial linear model was fitted and p-values were determined from a type III sum of squares F-test.
Fold change in gene expression of lincomycin-treated, norflurazon-treated, mock-treated, or untreated seedlings (Fig. 6, Fig. S4) were compared using a one-way ANOVA performed in JMP Pro 15. The resulting p-values were adjusted for FDR among the 13 genes measured using a two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli (Q = 5%) in GraphPad Prism v. 9.5.1. The Dunnett’s test was performed on genes with significant ANOVAs in JMP Pro 15 with mock-treated as the control condition (Fig. 6) or untreated as the control condition (Fig. S4).