Four different cultivars of P. davidiana were acquired from the National Institute of Forest Science (NIFoS), Suwon, Korea. The cultivars were divided into two different groups (drought tolerant group and drought susceptible group) based upon their overall growth response to water shortage/drought stress. The tolerant group contained two cultivars named Palgong 2 and Seogwang 9, whereas the susceptible group contained two cultivars named Palgong 1 and Junguk 6–2. The seeds of the four P. davidiana cultivars were surface-sterilized with H₂O₂ and washed five times with sterile distilled water and germinated on a half Murashige and Skoog (MS) medium. After 4 weeks, the plantlets were transferred on full-strength MS salts and vitamins supplemented with 3% Sucrose, 0.27% Gelrite, 0.5 ppm NAA at pH = 5.8 for organogenesis and incubated at 16 h/8 h light–dark cycle at 22–23°C for 4 weeks.

Five plants from each cultivar of each group were selected. For the dehydration stress treatment, whole 8-week-old plants were carefully uprooted from the MS medium and allowed to stand/dry at room temperature and humidity on a dry filter paper for up to 30 min.

Stomatal conductance plays a crucial role in regulating plant responses to dehydration. To determine whether PdNCED3 is involved in stomatal regulation, we investigated the stomatal structure of all the four poplar cultivars after 30 min of dehydration. Furthermore, the same investigations were performed for the wild type (WT), overexpressing (OE), and RNA interference (RNAi) lines under control and dehydration stress. For this purpose, we collected over 10 leaves each from five plantlets of each cultivar or line. A thin layer of transparent nail polish was applied to the bottom of the leaves and allowed to dry for several minutes. Transparent scotch tape was applied to the dried nail polish and carefully pressed starting from one side to remove the air bubbles. The tape was carefully removed from the leaves. This imprints the stomata and the surrounding cells onto the tape. The tape was then mounted on a microscopic slide. The abaxial leaf surface of each sample was observed this way using an Axioplan 2 imaging microscope (Axioplan 2 imaging; Carl Zeiss, Jena, Germany). Pictures were taken and analyzed using ImageJ (Schneider et al., 2012) to measure the stomatal apertures as described by Wu et al. (2017).

The plants were grown as described above and were subjected to dehydration stress at 8-weeks old. The plants were uprooted and kept on a dry filter paper at room temperature. The weight of all the plants was measured 1, 2, 3, and 4 h after uprooting to calculate the percent loss in weight.

As described above, the production of ROS such as H₂O₂ is one of the inevitable consequences of drought/water stress. For the quantification of drought-induced H₂O₂ accumulation, we performed 3,3′-diaminobenzidine (DAB) staining of plant leaves after every 2 min of uprooting for up to 14 min. The DAB staining was performed on three leaves of each cultivar from both the groups as described by Daudi and O'brien (2012). Images of the stained leaves were taken with a Canon EOS700D SLR camera (Canon, Ota City, Tokyo, Japan) and were edited in Adobe Photoshop CS6 (Adobe, San Jose, California, United States) as described by Imran et al. (2016). Briefly, the pictures were imported to Photoshop and the intensity of the DAB stain was measured inside a fixed rectangular shape (drawn through the marquee tool) using the luminosity tool. The measurement was performed at a minimum of three different places per picture.

After the dehydration stress treatments, the whole plants were stored in liquid nitrogen and freeze-dried using a freeze dryer (ISE Bondiro, South Korea). The freeze-dried samples were ground to a fine powder and the endogenous ABA content was measured as described by Kamboj et al. (1999). Briefly, the samples were treated with 10 ml of an extraction solution containing 95% of isopropanol, 5% glacial acetic acid, and 100 ng of [(±)-3,5,5,7,7,7-d6]-ABA. After collecting the aqueous phase, the extracts were evaporated and the residues were dissolved in 4 ml of 1N sodium hydroxide solution. The suspension was again evaporated and the residues were washed three times with 3 ml of methylene chloride to remove the lipophilic material. The aqueous phase was brought to approximately pH 3 with N hydrochloric acid and partitioned three times with ethyl acetate (EtOAc). The EtOAc extracts were combined and evaporated. The dried residue was dissolved in a phosphate buffer (pH8) and mixed with 1 g of polyvinylpolypyrrolidone (PVPP) powder on a shaker for 1 h. The extracts were added into reaction vessels (1 ml) and were methylated by adding diazomethane for the GC/MS-SIM (6890N network GC system, 5,793 networks mass-selective detector; Agilent Technologies, USA) analysis and quantification using the Lab-Base (ThermoQuest, Manchester, UK) software.

RNA extraction and sequencing were performed as described by Hussain et al. (2016). Briefly, the samples were collected after 10 min of the treatment. The finely ground homogenized samples were lysed in a TRIzol reagent (Invitrogen, Carlsbad, CA, USA), and the total RNA was extracted from each sample. For the sequencing, the complementary DNA (cDNA) was synthesized from 2 μg of the total RNA, using a DiaStar^TM RT Kit (Solgent, Daejeon, South Korea). The RNA-seq was performed with the Hiseq2500 sequencer (Illumina, San Diego, California, United States). The raw reads were processed to identify the high-quality reads with a threshold level of Q20 > 40% and the reads with more than 10% ambiguous bases or with Q20 < 40% were removed. The high-quality reads were compared with the reference genome of P. trichocarpa in Ensembl (Flicek et al., 2014) and aligned using TopHat (Trapnell et al., 2009) with default values. The gene expression levels were calculated using Cufflinks (Trapnell et al., 2012). The DEGs were identified at p < 0.05 using Cuffdiff (Trapnell et al., 2012). The data were submitted to the Gene Expression Omnibus (GEO) at the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/) with accession numbers GSE98170 (Junguk 6–2), GES98172 (Palgong 1), GES98173 (Palgong 2), and GES98175 (Seogwang 9).

Large-scale genome sequencing projects such as RNA-seq usually yield data comprising of several thousand genes and transcripts. Handling such a large amount of data is often very challenging and it is often difficult to extract meaningful information at the functional level. To get around this problem, we analyzed the transcriptomic data through MapMan (version 3.6.0RC1) as described in Hussain et al. (2016). MapMan is an omics data analysis software that allows the visualization of the DEGs at functional and pathway levels (Thimm et al., 2004). For this purpose, the list of all the DEGs obtained from each cultivar of each group was uploaded to MapMan and mapped against the poplar database “Ptrichocarpa V3.0_210.” The involvement of DEGs in various pathways and cellular processes was determined by applying the respective BIN and Sub-BIN numbers on custom-made images uploaded to MapMan.

The validation of the RNA-seq data was carried out using the real-time quantitative reverse transcription PCR (qRT-PCR) analysis. For this purpose, the genes related to the ABA-pathway were chosen. The RNA was extracted using a Trizol reagent (Invitrogen, USA) and the cDNA was synthesized from 2 μg of the total RNA using a DiaStar^TM RT Kit (SolGent, Korea). A two-step PCR was performed using the primers given in Supplementary Table 1 in the Eco^TM Real-Time PCR machine (Illumina, USA). The PCR reaction was performed in a total volume of 20 μl using 2 × QuantiSpeed SYBR Green Kit (PhileKorea, Seoul, Korea) with 10 nM of each primer and 100 ng of the template DNA under the following conditions: polymerase activation at 95°C for 5 s, 18 concurrent annealing, and extension steps at 60°C each for 30 s. Poplar actin (Potri.001G309500) was used as a comparative control. A no-template reaction was also used as a control. The qRT-PCR was performed using at least three independent biological replicates and three technical replicates of each biological replicate.

The comparison of the NCED sequence of P. davidiana with that of P. trichocarpa was carried out using the Integrative Genomics Viewer (IGV 2.3.72) program (Robinson et al., 2011). The amino acid sequences of two NCED genes of P. davidiana were submitted to the I-TASSER server for the protein model prediction. The I-TASSER produced one to five models for each of the poplar sequences. Among these, the model with the highest confidence score (C-score) was selected and analyzed using the PyMOL software (LLC, http://www.pymol.org/).

The Populus davidiana overexpression (OE) and RNAi transgenic lines were generated using the Gateway (Thermofisher Scientific, Waltham, Massachusetts, United States) system as described in Figures 8A,B. The full coding DNA sequence of PdNCED3 (Potri.011G112400) for the OE DNA construct and a 200 bp segment (starting from the start codon) for the RNAi DNA construct was amplified using a proof-reading polymerase enzyme for one-step cloning into PCR^®8/GW/TOPO plasmid and transformed into TOP10 competent cells according to the instructions of the manufacturer and plated on Luria Broth (LB)-agar plates containing 50 mg/L of spectinomycin. The successfully transformed colonies were genotyped through PCR to confirm the presence of inserts. Plasmids were extracted and the sequences of inserts were confirmed using M13 universal primers. The sequences of primers used are given in Supplementary Table 1.

Afterward, the inserts with the correct sequences were transferred to the destination vectors via LR reaction according to the manufacturer's instruction. The OE DNA construct was transferred to the pK2WG7 destination vector whereas, the RNAi construct was transferred to the pK7GWIWG2(1) destination vector. The recombinant vectors were transformed into DH5α competent cells and plated on LB-agar plates containing 50 mg/L of Kanamycin (pK2WG7) and 50 mg/L of Kanamycin [pK7GWIWG2(1)]. The successfully transformed colonies were genotyped using PCR as described above to confirm the presence of the inserts. The destination vectors containing the OE and RNAi constructs were transformed into Agrobacterium tumefaciens strain GV3101.

The Agrobacterium tumefaciens containing the OE and RNAi constructs were cultured in LB at 28°C in a shaking incubator at 150 rpm. After 48 h, 5 ml of this culture was added to 50 ml of a fresh LB culture medium and incubated again for 24 h. The bacteria were harvested by centrifugation at 3,000 rpm at 4°C for 15 min. The bacterial cells were re-suspended in a 25 ml saline solution on ice. The stems of hybrid poplar (P. alba × P. glandulosa cv. Bonghwa) cultivated for 1 month in a root-induction medium were cut into 5 mm tissue disks. The cut stem tissues were immersed in 5 ml of 0.85% sodium chloride (NaCl). The stem tissues were dipped in a bacterial saline suspension for 15 min after adding 25 ml of acetosyringone (10 mg/ml). The cut stem tissues were then taken out of the bacterial saline and gently rinsed in a sterile saline solution. The process of inoculation and rinsing was repeated two times. The inoculated stem tissues were then cultured in a callus induction medium without antibiotics and incubated for 2 days and then transferred to a fresh callus induction medium supplemented with 50 mg/ml of kanamycin (for OE construct) and 50 mg/ml of kanamycin (for RNAi construct) for 2 weeks. The resulting calli were transferred to a fresh media after every 2 weeks in the same way. The successful calli were formed within 4 weeks of culture. The calli of 2 to 3 mm size were transferred to a stem-induction medium in the presence of kanamycin as described above. Afterward, the plants were transferred to a root induction medium (Supplementary Figure 7).

All the experiments were replicated at least three times. Data were collected at regular intervals. The means and SDs were calculated in Microsoft Excel. The means were separated for significant differences using Student's T-test at a 5% level of confidence.

While screening different P. davidiana cultivars for their response to drought stress, we identified two drought-resistant cultivars (Palgong 2 and Seogwang 9) and two drought susceptible cultivars (Palgong 1 and Junguk 6–2). According to the field and screen house-based drought stress experiments, the detached leaves of the cultivars Palgong 1 and Junguk 6–2 showed severe curling and leaf rolling after 10, 20, and 30 min of detachment which are characteristic symptoms of moisture stress (Figure 1). On the other hand, Palgong 2 and Seogwang 9 exhibited significantly less and delayed curling and leaf rolling. The dehydration resistance shown by Palgong 2 and Seogwang 9 was supported by the stomatal closure after 30 min (Figure 2A). This was also accompanied by a significantly higher linear loss of fresh weight by the susceptible cultivars Palgong 1 and Junguk 6–2, due to a high rate of evapotranspiration from the leaves. The leaves of these cultivars lost 37 and 43% of their fresh weight, respectively within the first 4 h of the treatment (Figure 2B). As the shortage of moisture/water inside plant tissues is characterized by high levels of H₂O₂ in leaf tissues, we quantified the amount of H₂O₂ produced in the leaves via DAB staining, following 30 min of dehydration. The results supported our previous findings as Palgong 1 and Junguk 6–2 showed significantly higher DAB stain intensity as compared with the tolerant cultivars indicating a much higher level of H₂O₂ production in the leaves of these plants (Figure 2C). The ability of plants to resist water loss is also reflected by their ability to regulate stomatal movement under stress. We subjected the four cultivars to 30 min of dehydration following detachment and then observed their stomata through an electron microscope. The stomata of Palgong 1 and Junguk 6–2 (susceptible cultivars) had a significantly higher aperture as compared to the drought-tolerant cultivars Palgong 2 and Seogwang 9 (Supplementary Figure 6A).

The analysis of the transcriptomic data showed highly significant changes in expression at a cutoff Q value of 0.05 (Figure 3A). An average of 45.34 million reads was obtained from the stressed plants compared with 44.89 million reads (Figure 3B; Supplementary Table 2) from the control samples. Up to 55.36% of these reads were successfully mapped onto the P. trichocarpa reference genome. An average of 27,501 genes and 98,321 transcripts in the control samples and an average of 27,623 genes and 98,575 transcripts in the treated samples were identified (Supplementary Table 2). Further, a total of 16,318 DEGs were obtained following the stress treatment (Figure 3C). Among these, the drought-tolerant cultivars, Palgong 2 and Seogwang 9, showed an increase in the expression of 2,076 and 1,891 genes and a reduction in the expression of 1,573 and 2,719 genes, respectively. On the other hand, the drought susceptible clones, Palgong 1 and Junguk 6–2, yielded 1,822 and 2,244 upregulated genes along with 2,119 and 1,874 downregulated genes, respectively (Figure 3C). Interestingly, two genes that encode the NCED protein, PdNCED1 (Potri.001G393800) and PdNCED3 (Potri.011G112400), were upregulated in the drought-tolerant cultivars Palgong 2 and Seogwang 9 by more than 5 and seven folds, respectively, but significantly downregulated in the drought-susceptible cultivars (Figure 7).

All the DEGs were analyzed to identify specific drought-related genes with statistically significant and at least two-fold change in their expression, among the drought-tolerant and drought-susceptible plants. The results showed that both the drought-tolerant cultivars expressed at least six drought-specific genes with the same expression patterns that were not present in the plants of the drought susceptible group (Supplementary Figure 1). Furthermore, there were at least three drought-specific genes that had contrasting expression patterns between the tolerant and susceptible groups, i.e., these were downregulated in the drought-tolerant plants but upregulated in the drought susceptible plants (Supplementary Figure 1) indicating their negative regulatory role in drought tolerance. These DEGs included Potri.001g131400.3, Potri.005g226000.4, and Potri.011g009900.1. All three DEGs encode early-responsive to dehydration stress (ERD3) proteins.

For a further detailed analysis of the transcriptome profiles of drought-tolerant and susceptible cultivars, we analyzed the transcriptomic data through MapMan (Thimm et al., 2004). For this purpose, we chose genes with statistically significant and at least a two-fold change in their expression. The results showed that the genes belonging to a plethora of important physiological processes involved in abiotic stress responses were differentially expressed in the drought-tolerant and susceptible clones (Figure 4). These include more than 30 genes responsible for heat stress regulation, with differential expression patterns between the resistant and susceptible cultivars. Other groups with differential expression patterns among resistant and susceptible poplar plants included genes involved in the signaling of developmental hormones such as ABA, IAA, GA, and Ethylene. Drought and other abiotic stresses are often accompanied by a concomitant change in the cellular redox signature. We found several groups of DEGs involved in redox signaling such as multiple genes encoding thioredoxin, peroxiredoxin, glutaredoxin, catalase, and dismutase genes (Figure 4). The MapMan analysis also showed multiple DEGs encoding important transcription factors such as WRKY, ERF, bZIP, MYB, and bHLH that have a well-known role in abiotic stress responses.

Genes encoding the NCED proteins (that catalyze the oxidative cleavage of carotenoids to synthesize ABA) have been shown to play a key role in the ABA-dependent drought tolerance in different plants (Schwartz et al., 1997; Tan et al., 1997; Burbidge et al., 1999; Qin and Zeevaart, 1999; Chernys and Zeevaart, 2000; Iuchi et al., 2000, 2001; Thompson et al., 2000a,b; Rook et al., 2001). We wondered if the response of drought-tolerant and susceptible plants used in this study was related to the ABA content in these plants. As expected, the ABA measurement results showed a significantly higher increase in the ABA content in Palgong 2 and Seogwang 9 poplar after 30 min of the stress treatment as compared with the plants of the drought susceptible group (Figure 5). These results were parallel to the transcriptomic expression patterns of NCED in these cultivars.

To further elucidate the functional role of NCED in regulating drought stress, we evaluated Arabidopsis nced3 (AT3G14440) for the loss of function mutant plants for their response to drought conditions. Results showed that 75% of the nced3 plants died after 9 days of continuous drought stress whereas, 75% of the WT plants recovered when the irrigation was re-started after 9 days (Figures 6A,B). This was also accompanied by a significantly higher weight loss of the nced3 plants as compared to WT plants (Figure 6C).

To validate the expression patterns of the DEGs observed in the transcriptome, we performed real-time PCR analysis for six different genes following the dehydration stress. As described above, the expression patterns of the two drought-responsive NCED genes that are known to regulate ABA biosynthesis and signaling were very unique i.e., both the PdNCED1 and PdNCED3 were upregulated in the drought-tolerant cultivars Palgong 2 and Seogwang 9 by more than five- and seven-fold, respectively, but downregulated in the drought-sensitive plants by two- and 2.9-fold, respectively. The same expression patterns were produced by the qRT-PCR analysis (Figures 7A,B). As described earlier, members of the NCED gene family have been shown to positively regulate drought stress in plants in an ABA-dependent manner and further investigation showed four different ABA-responsive elements (AREB) genes (Potri.002G125400, Potri.004G140600, Potri.010G248100, and Potri.014G028200) to be upregulated in the drought-tolerant clones but downregulated in the drought-sensitive clones (Figures 7C–F). This indicates strict regulation of drought stress responses in these plants at the transcriptional level.

To further elucidate the role of NCED in the drought tolerance of P. davidiana, we generated transgenic NCED3 OE and RNAi lines using DNA constructs (Figures 8A,B). Generating OE and RNAi lines usually results in multiple lines. However, most often, different transgenic lines show different levels of expression of the transgene. Some OE lines show significantly high expression whereas others show only a small increase in expression of the target gene as compared with that in the WT lines. Similarly, some RNAi lines exhibit only a small reduction in the expression whereas, others may show a significant reduction in expression of the target gene. As far as we have observed in our laboratory, the transgenic lines with the expression profile of the transgene at the extremes (i.e., either the lowest or highest changes in expression compared with WT lines) do not often yield promising and reproducible results for all the parameters under study. In such situations, one may have to choose the lines with optimum changes in the expression of the target gene or lines that give promising and reproducible results in multiple experiments or for all or most of the parameters under study. We got several successful transgenic lines for NCED3 in Poplar. We obtained eight (8) independent OE lines and seven (7) independent RNAi lines (Supplementary Figure 2A). We tested all of these lines for NCED3 expression under normal as well as after drought or water shortage conditions. Under normal conditions, all of the OE lines had significantly higher NCED3 expression compared with WT. On the other hand, all of the RNAi lines had lower NCED3 expression as compared with those in the WT plants (Supplementary Figure 2B). However, the expression profiling after 10, 20, and 30 min of the drought stress treatment showed the highest increase in NCED3 expression in the leaves of OE line #16, whereas the least change in NCED3 expression was recorded in RNAi line #26 (Supplementary Figure 2C). Therefore, we chose these two lines (OE line #16 and RNAi line #26) for further experiments.

According to the field and screen house-based drought stress experiments, the detached leaves of the RNAi lines exhibited severe curling and leaf rolling, which are characteristic of dehydration stress, after 10 and 20 min of detachment as compared with the leaves of the OE lines which showed no rolling and curling (Figure 8C). The PdNCED3 OE line was found to have a significantly tolerant phenotype under dehydration stress. These results were also supported by the significantly smaller stomatal apertures in the leaves of the NCED3 OE plants as compared with those of the WT and RNAi plants (Supplementary Figure 6B). Furthermore, water loss calculations indicated that the RNAi plants lost a significantly higher amount of moisture, whereas the OE plants lost a significantly lower amount of moisture when compared with the WT plants (Supplementary Figure 6C).

Furthermore, the RT-PCR analysis showed a significantly higher PdNCED3 (Potri.011G112400) mRNA accumulation in the WT plants following the stress treatment with the highest expression after 30 min followed by 20 and 10 min of the treatment (Figure 8D). In the case of the OE line, the transcript accumulation of PdNCED3 was higher and faster than in the WT, whereas the RNAi line showed significantly lower and delayed expression of NCED3 (Figure 8D). Furthermore, the PdNCED1 (Potri.001G393800) transcript accumulation was upregulated in both of the OE and RNAi lines after dehydration stress (Figure 8E), but there was no difference in its expression level when compared with the wild type. Among the ABA-signaling genes, the bZIP transcription factors Potri.014g028200, Potri.002g125400, and Porti.004g140600 were ABA-responsive element-binding factors that enhance abiotic stress signaling (Zou et al., 2008; Hossain et al., 2010; Tang et al., 2012). Our results indicated that the expression of these three genes gradually increased in the order of OE, WT, and RNAi under dehydration stress (Figures 8F–H). Furthermore, NCED3 OE resulted in a decrease in the expression of Potri.010g248100, abscisic acid-insensitive 5-like protein 2 (Pt-ABI2), as compared with the RNAi and the WT line (Figure 8I).

To confirm that PdNCED reduces damage to plants by mitigating ROS-mediated oxidative damage, the amount of H₂O₂ accumulation in the leaves of the transgenic plants was assessed after 30 min of dehydration stress. As expected, leaves of the OE line showed significantly lower DAB stain intensity as compared with the WT and RNAi transgenic plants, indicating a much lower level of H₂O₂ production in the leaves of these plants following dehydration stress (Figure 9A).

It was worth knowing if the response of the PdNCED3 OE and RNAi plants was related to the ABA contents in these plants. As expected, the ABA measurement results showed a significantly higher increase in the ABA contents of the OE line as compared with the RNAi plants (Figure 9B). These results were parallel with the expression patterns of PdNCED3 in transgenic plants.

As described above, PdNCED1 (Potri.001G393800) and PdNCED3 (Potri.011G112400) were upregulated in the drought-tolerant cultivars but downregulated in the drought-sensitive cultivars. We aligned the sequence of these genes with their orthologous genes in Populus trichocarpa and found out that the PdNCED genes are unique in their sequence though both the genes showed more than 95% similarity to their respective P. trichocarpa orthologs. Single amino acid differences were found in the amino acid sequence of NCED3 P. davidiana and P. trichocarpa at seven different positions (Supplementary Figure 3). The first difference was a single amino acid difference at position 26 expressing isoleucine (I) in all P. davidiana cultivars compared with threonine (T) in P. trichocarpa. This was followed by another major difference between the NCED3 sequences of the two Populus species with the starting position number 55, where three consecutive proline threonine pairs (PTPTPT) were found in all the cultivars of P. davidiana as compared with only one pair in P. trichocarpa which may have a significant effect on the structure of this protein in the two species. Similarly, other P. davidiana—P. trichocarpa single amino acid differences were also found such as leucine (L66)—arginine (R66), arginine (R117)—glycine (G117), serine (S289)—threonine (T289), aspartic acid (D331)—asparagine (N), and alanine (A431)—proline (P431). Furthermore, the PdNCED3 peptide was found to be significantly smaller than PtNCED3. This was attributed to the multiple amino acid residues absent in the PdNCED3 sequence from the positions 298 to 330. This feature was especially prominent in the drought susceptible cultivars, Palgong 1 and Junguk 6–2. At position number 518, the NCED3 of both the drought-resistant P. davidiana cultivars and P. trichocarpa expressed a tyrosine (Y) residue, but it was absent in the NCED3 sequence of both the susceptible cultivars (Supplementary Figure 3).

Similarly, single and multiple amino acid differences were also found between the sequences of PdNCED1 and PtNCED1 (Supplementary Figure 4). Single-amino acid differences between P. davidiana and P. trichocarpa NCED1 were found at five different places. For example, at position number 48, proline (P) was found in all P. davidiana sequences but serine (S) in the P. trichocarpa sequence. Similarly, at position number 51 we found tyrosine (Y) in all the P. davidiana sequences as compared with cysteine (C) in P. trichocarpa. Another important difference that we observed was at position number 155 where we found valine (V) in all the P. davidiana sequences instead of leucine (L) in P. trichocarpa. We also found threonine (T) in all the P. davidiana sequences at position number 568 instead of asparagine (N) in P. trichocarpa.

With these differences in the sequence of NCED peptides from drought-resistant and susceptible cultivars, we wondered if these have any effect on the protein 3D structure. For this purpose, we submitted the deduced amino acid sequences to the I-Tasser server to predict the 3D structures. The results showed some interesting differences in the protein structure as well as the Co-factor/ligand binding (Supplementary Figure 5). PdNCED1 (Potri.001G393800) peptide was predicted to bind with an iron (Fe/Fe²⁺) co-factor only from the Palgong 2 and Seogwang 9 cultivars which are from the drought-tolerant group. Whereas, the PdNCED1 from the drought susceptible cultivars showed a tetraethylene glycol monooctyle ether (C8E) co-factor as a potential ligand. These structural differences may lead to the differential functioning of NCED in these plants resulting in significantly variable responses to drought stress.