The tobacco genomic sequences were downloaded from SGN (Sol Genomics Network)¹. The Hidden Markov Model (HMM) profile of NAC domain (PF02365) retrieved from Pfam² was used to conduct HMM search against the annotated protein database with an E-value cutoff of 1.0 using HMMER (v3.0) (Johnson et al., 2010). All non-redundant hits within expected values were collected and each newly identified hit was subsequently used as a query to perform BLASTP search against the annotated tobacco genome. The protein sequences obtained from the two above-described approaches were combined and redundant entries were removed manually. The resulted non-redundant sequences were manually analyzed using InterPro to ensure presence of the NAC domain (Apweiler et al., 2001). Moreover, TMHMM Server ver.2.0 was used to predict the membrane-bound NtNAC members (Krogh et al., 2001).

Exon-intron structures of tobacco NAC genes were analyzed and illustrated with the Gene Structure Display Server (GSDS)³ by comparison of gene CDS sequences with genomic DNA sequences obtained from the Sol Genomics Network. The program MEME version 4.3.0 was employed for the detection of conserved motifs in tobacco NAC proteins⁴. MEME was run locally with the following parameters: distribution of motif occurrences, zero or one per sequence; maximum number of motifs, 10; optimum motif width, ≥6 and ≤200 (Bailey et al., 2015).

Multiple sequence alignments of full-length tobacco NAC amino acid sequences were performed using the MAFFT program with the default parameters together with 105 Arabidopsis NAC protein sequences and two representative NAC proteins (StNAC010 and StNAC039) from potato (Solanum tuberosum L.) which have been reported to be in the Solanaceae-specific subfamily (TNACs) of the NAC family (Katoh and Standley, 2013). Sequences of Arabidopsis and potato NAC proteins were downloaded from TAIR10⁵ and SGN⁶, respectively. Un-rooted phylogenetic trees were constructed via MEGA 6.06 using the Neighbor-Joining (NJ) method with the following parameters: Poisson model and bootstrap values of 1000 replicates. Pairwise deletion mode was employed to make sure that the divergent C-terminal domains could contribute to the topology of the phylogenetic trees (Tamura et al., 2013).

Common tobacco (Nicotiana tabacum L. Cv. K326) plants were grown in the field. Middle leaves (Leaf No. 9-10 from the base of a tobacco plant) were collected at 15, 45, 65, 75 days after topping (DAT) (defined as YL, young leaf; ES, early senescing leaf; MS, mid-senescing leaf; LS, late senescing leaf, respectively) for analyzing gene expression during leaf senescence. All harvested leaves were wrapped in aluminum foil, immediately placed in liquid nitrogen and stored at −80°C until used. Three biological replicates were analyzed for each time point. In gene functional analysis, The T1 plants of NtNAC080 CRISPR-Cas9 transgenic lines and wild-type plants (K326) were grown in the greenhouse under normal growth conditions.

Arabidopsis Col-0 and transgenic plants were grown in a plant growth chamber (Conviron, Canada) at 22°C with a relative humidity of 55% under long-day conditions (16 h light/8 h dark) with white light illumination (120 μmol photons/m²s).

The data used for expression profiling of tobacco NAC genes were from the tobacco Illumina RNA-seq data generated in the Guo lab (Li et al., 2017) (archived by NCBI under the accession numbers: SRP102153). For RNA-Seq data of tobacco leaf senescence, RPKM values of the 154 NtNAC genes were retrieved and normalized. A heatmap was generated based on the log₂ fold-change values at 25/35/45/55/65/75DAT when compared with 15DAT and visualized with R package (R Development Core Team, 2013). In addition, the data used for expression profiling of Arabidopsis NAC genes (ANAC genes) were retrieved from AtGenExpress leaf developmental data bySchmid et al. (2005). Based on the previous Guo et al. analysis, we performed differential expression analysis of all the ANAC genes between the fully expanded leaves (young leaves) and senescing leaves (Guo and Gan, 2012). Genes that are up-regulated twofold or more were designated as being senescence up-regulated ANACs.

Total RNA was extracted according to the method previously described (Li and Guo, 2018). Genomic DNA was eliminated from the RNA samples via DNase I (Takara bio inc., Otsu, Japan) treatment for 30 min at 37°C. The quality and concentrations of RNA samples were determined using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, United States). First-strand cDNAs were synthesized using the RevertAid H Minus First Strand cDNA Synthesis Kit (Thermo Scientific) with Oligo(dT) primers. qRT-PCR was performed on an ABI7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, United States) with the reaction mixtures comprising 1 μL template cDNA, 1 μL each of forward and reverse primers (0.3 μM each) and 10 μL FastStart Universal SYBR Green Master (Rox, Roche Applied Science). For tobacco, NtActin was used as an internal control. NtCP1 (SAG12 homolog in tobacco) and NtRBCS were used as senescence markers. For Arabidopsis, AtActin and AtSAG12 were used as the internal control and senescence marker, respectively. The relevant primers are given in Supplementary Table S1. All reactions were run in triplicate. The relative gene expression values were analyzed using the 2^{−Δ Δ Ct} method.

The coding sequence of NtNAC080 was PCR amplified from K326 cDNA using the PrimeSTAR HS DNA polymerase (Takara Bio, Japan) and cloned into the binary vector pCHF3 (a modified pPZP212 vector) at the BamHI/SacI sites after the CaMV 35S promoter (Hajdukiewicz et al., 1994). Primers used in vector construction are listed in Supplementary Table S2. The construct was then used to transform Arabidopsis using the Agrobacterium floral dip method (Clough and Bent, 1998). Transgenic Arabidopsis plants were selected on kanamycin (50 mg/L) plates. Presence of the transgene was PCR confirmed using genomic DNA from the leaves of putative transformants. Abundance of the NtNAC080 transcript was estimated via qRT-PCR using cDNA from senescing leaves of the transgenic plants.

The DNA sequences of NtNAC080, NtNAC028, NtNAC083, and NtNAC110 are highly similar. To produce a gRNA specific to NtNAC080, an online software⁷ was used for synthesizing two DNA oligos, NtNAC080-gRNA1 and NtNAC080-gRNA2, The NtNAC080-sgRNA forward and reverse primers were denatured by heating at 95°C for 3 min and annealed to form a double-stranded DNA. Thereafter, the double-stranded DNA was inserted into the pORE-Cas9/gRNA vector at the BsaI site (Gao et al., 2015). Tobacco cultivar K326 was transformed with the resulting construct via Agrobacterium-mediated transformation as described earlier (Horsch et al., 1989). The shoots of putative transformants were selected in Murashige and Skoog (MS) medium containing 0.1 mg/L 1-naphthylacetic acid (NAA), 1 mg/L 6-dimethylaminopurine (6-BA), 50 mg/L kanamycin (Kan), and 500 mg/L cefotaxime sodium (Cef). Putative Kan-resistant transformants were transferred to soil pots and grown under greenhouse conditions. The genomic DNA of transgenic tobacco plants from Kan selection were extracted using the DNeasy Plant Mini Kit (Qiagen, CA, United States). The DNA fragments containing the Cas9/gRNA target sequences were amplified by PCR using PrimeSTAR HS DNA polymerase (Takara Bio, Japan). After purification, the PCR product was cloned to the pEASY-Blunt Zero vector (Transgene, China) and the DNA from the single colonies were sequenced to detect the types of mutation. For the target gene edited tobacco plants, selfing was performed and the homozygous mutants were acquired from T1 transgenic plants. The relevant primers are given in Supplementary Table S2.

Chlorophyll was extracted and quantified as described previously (He and Gan, 2002). Briefly, 10 mg of freeze-dried leaf tissue was extracted with 1 mL of 95% ethanol in the dark for 24 h with agitation. The supernatant was quantified via spectrophotometric measurement at 649 and 665 nm. Three biological replicates were used for each sample.

To identify NAC genes in tobacco, Hidden Markov Model (HMM) search was performed against the Sol Genomics Network database⁸ using the Pfam NAC domain (PF02365) as query. Newly identified hits were used as queries to carry out BLASTP search until no further hit can be obtained. After manual removal of redundant hits, all the putative NAC proteins were further screened using the Pfam⁹ and the Interpro¹⁰ programs to ensure that each protein has a N-terminal NAC domain. Following this procedure, a total of 154 non-redundant NAC family proteins were identified in tobacco. The tobacco NAC genes were named NtNACs followed by Arabic numbers. The length of the NtNAC proteins ranges from 130 to 645 amino acids (aa) with an average of 330 aa. The detailed information of NAC family genes in tobacco, including accession numbers, protein lengths, protein sequences, conserved domains and similarities to their Arabidopsis orthologs etc., is listed in Supplementary Tables S3, S4.

To study the evolutionary relationship between the tobacco NAC proteins and known NACs from other plant species, an unrooted phylogenetic tree was constructed from alignments of 154 NtNACs, 105 Arabidopsis ANACs and 2 representative potato NAC protein sequences (Figure 1). All the NAC proteins were divided into 15 distinct subgroups: NAC-a to NAC-o. The tobacco NAC members demonstrated an interspersed distribution in all the subgroups except the subgroup NAC-l, in which all members are from Arabidopsis. The subgroup NAC-o, on the other hand, contains no NAC from Arabidopsis. Further alignment analysis suggested that proteins in subgroup NAC-o shared sequence characteristics of TNACs, which have been reported to be the Solanaceae-specific NAC subfamily (Rushton et al., 2008). The two representative TNAC potato proteins, StNAC010 and StNAC039 (Singh et al., 2013), were also grouped in this subgroup, suggesting NAC-o being a Solanaceae-specific NAC subgroup.

To gain insights into the structural diversity of the NtNAC genes, the exon/intron structure of the coding sequences of individual NtNACs in tobacco was analyzed (Figure 2B). In general, members of the same subgroup share similar exon/intron structure and gene length. For example, the NAC genes in subgroups NAC-a, NAC-b, NAC-d, NAC-e, and NAC-h contain one or two introns. The NAC-f members have two introns except NtNAC110, which harbors three introns. In addition, 7 of the NAC genes have no intron, all of which belong to subgroup NAC-o. The members in subgroups NAC-i and NAC-m on the other hand, have more variable gene structures. Among the 154 NtNACs, the shortest NtNAC gene is 474 bp (NtNAC104) long, whereas NtNAC068 is the longest NtNAC gene with a 10 kb genomic sequence.

Using the MEME program, 10 conserved motifs, named Motif 1-10, were identified in the NtNAC proteins, all located within the N-terminal region which are highly conserved for DNA-binding, as described previously (Singh et al., 2013; Wei et al., 2016; Figure 2C and Supplementary Figure S1). Motif 2/8, Motif 1/7, Motif 3, and Motif 4 comprise the NAC DNA-binding domain while other motifs are dispersed around the NAC domain. Similar motif organization was observed within the same subgroup or in closely related subgroups (Figures 1, 2A,C). For instance, only NAC members in the Solanaceae-specific subgroup NAC-o contain both Motifs 8 and 9. Subgroup NAC-f mainly contains Motifs 1, 2, 3, and 4. Most members of the subgroup NAC-m comprise Motif 3, 5, 8, and 10. A conserved nuclear localization signal (NLS) was identified within Motif 3 (Supplementary Figure S1). As expected, the C-terminal transcriptional regulation region (TRR) is variable.

A trans-membrane (TM) region with a predicted-helix was identified in the far C-terminal region of 13 NtNACs proteins, which belonging to subgroups NAC-i, NAC-k and NAC-o. Remarkably, subgroup NAC-i contains most of membrane-bound NAC members of tobacco and Arabidopsis, including 8 NtNACs and 7 ANACs (Supplementary Table S3). There are 18 membrane-associated NAC TFs in Arabidopsis, 5 in rice, 11 in soybean and 14 in potato (Kim et al., 2007, 2010; Singh et al., 2013). A number of Arabidopsis membrane-bound NAC proteins have been shown to be involved in biotic and abiotic stress responses (Kim et al., 2007, 2010; Lee et al., 2012).

NAC TFs play an important regulatory role in leaf senescence (Li and Guo, 2014). Transcriptomic studies have shown that approximately one third of the NAC genes were up-regulated during leaf senescence in Arabidopsis, highlighting their importance in senescence regulation (Breeze et al., 2011). Using the publicly available gene-chip based data by Schmid et al. (2005), 36 ANAC genes which were up-regulated twofold or more during leaf senescence were identified (Supplementary Table S5). Interestingly, 13 of the senescence-up-regulated ANACs belong to the phylogenetic subgroups NAC-b and NAC-f, which totally contain 14 ANAC genes from Arabidopsis. Moreover, several ANAC TFs that have been previously identified as positive regulators of leaf senescence, including ANAC029/AtNAP, ANAC059/ORS1, ANAC092/ORE1/AtNAC2, ANAC002/ATAF1 (Guo and Gan, 2006; Balazadeh et al., 2011; Rauf et al., 2013; Garapati et al., 2015) were clustered together in the NAC-b and NAC-f subgroups. NAC genes with same functions showed a tendency to fall into the same subgroup (Fang et al., 2008; Hu et al., 2015; Wei et al., 2016). We thus predict that the NAC genes of NAC-b and NAC-f subgroups might play an important role in regulating leaf senescence.

To explore the patterns of tobacco NAC expression during leaf senescence, we performed a comprehensive analysis of NtNAC genes expression profiles at different stages of leaf senescence based on data retrieved from our earlier RNA-Seq study of tobacco leaf senescence (Li et al., 2017). We were able to obtain transcripts data from most of the NtNACs (147 out of 154) from the dataset (Supplementary Table S6). As previously reported in Arabidopsis and other species (Breeze et al., 2011; Christiansen and Gregersen, 2014), the analysis of expression changes indicated significant transcriptional responses of the NtNAC genes during tobacco leaf senescence. Twenty four of the NtNACs were up-regulated fourfold or more at least at one time point from 25DAT to 75DAT compared with 15DAT (Figure 3 and Supplementary Table S6). Interestingly, 11 of these 24 genes were clustered in subgroups NAC-b and NAC-f, and are closely related to the Arabidopsis senescence-regulating NAC genes, including AtNAP/ANAC029, ANAC002, ANAC059, and ANAC092 (Figure 1). In addition, four other NtNAC genes in subgroups NAC-b and NAC-f, including NtNAC008, NtNAC018, NtNAC064, and NtNAC098, showed an up-regulation of 2–4-fold and most of the remaining genes in the two subgroups showed somewhat increased expression during senescence. NtNACs in subgroups NAC-b and NAC-f shared similar up-regulation patterns during senescence as their homologs in Arabidopsis, suggesting that these two subgroups may play significant roles in regulating leaf senescence and these regulatory roles might be conserved between species.

To validate the result of RNA-Seq analysis, we analyzed the expression patterns of 15 selected NtNAC genes from the subgroups NAC-b and NAC-f during leaf senescence using quantitative RT-PCR (qRT-PCR). Data from 4 leaf developmental stages, including fully expanded, non-senescent leaf (YL), early senescent leaf (ES), middle senescent leaf (MS), and late senescent leaf (LS) were used (Supplementary Figure S2) and all of the 15 NtNAC genes tested were found to be up-regulated during senescence, with a fold change of two or more (Figure 4). Among them, NtNACs including NtNAC002, NtNAC030, NtNAC046, NtNAC079, NtNAC098, and NtNAC149 were up-regulated at the MS stage with high level expression maintained until the LS stage. Expression of some other NtNACs genes (NtNAC008, NtNAC073, NtNAC099, and NtNAC148) were induced only at the LS stage. Expression of NtNAC028, NtNAC080, NtNAC083, and NtNAC117 increased rapidly at the ES stage, suggesting potential roles of these genes in regulating the onset of leaf senescence. Phylogenetic analysis showed that NtNAC028, NtNAC080, and NtNAC083 were closely related to ANAC029/AtNAP, which acts as a key positive regulator of leaf senescence (Guo and Gan, 2006). Similarly, NtNAC117 was in the same clade that contained the Arabidopsis senescence-associated genes ANAC059/ORS1 and ANAC092/ORE1, which also play significant roles in regulating senescence (Balazadeh et al., 2010, 2011; Rauf et al., 2013). We therefore predicted that these four genes (NtNAC028, NtNAC080, NtNAC083, and NtNAC117) might function as key regulators of leaf senescence in tobacco.

To further assess the function of senescence-associated NtNAC genes in planta, we obtained transgenic Arabidopsis plants overexpressing NtNAC080, expression of which increased significantly at early stages of leaf senescence in tobacco, with a 48-folds change at the ES stage. Two lines of NtNAC080-overexpressing transgenic Arabidopsis, with expression of NtNAC080 confirmed by qRT-PCR (Figure 5D), exhibited a premature senescence at 45 days after sowing (DAS) (Figure 5A). The precocious leaf yellowing phenotype was also supported by changes in total chlorophyll content and the maximal photochemical efficiency of PSII (Fv/Fm), which reflects the photochemical quantum efficiency of PSII and photosynthetic activity (Figures 5B,C). Expression of the senescence-specific cysteine protease SAG12, which is a widely used molecular marker of leaf senescence, was strongly induced in leaves of the two NtNAC080-overexpressing lines (Figure 5E). These data indicated that, like its close homolog ANAC029/AtNAP in the NAC-f subgroup, NtNAC080 also acts as a positive regulator of leaf senescence.

To confirm the function of NtNAC080 in regulating tobacco leaf senescence, we mutated NtNAC080 using the CRISPR/Cas9 system in tobacco. Two gRNAs (sgRNA1 and sgRNA2) targeting the NAC domain of NtNAC080 were created and used in Agrobacterium-mediated transformation of tobacco plants (Figure 6A). Twelve T0 plants for NtNAC080 sgRNA1 and 14 T0 plants for NtNAC080 sgRNA2 were obtained after kanamycin screening. PCR-amplified target regions of these plants were sequenced to identify mutations. Eleven of the 12 (91.7%) sgRNA1 and 9 of the 14 (64.3%) sgRNA2 T0 plants were found to contain mutations with deletion or insertion of nucleotides (Supplementary Table S7). Homozygous mutant plants were obtained from the T1 generation and two representative lines ntnac080-1 and ntnac080-2, were further used for phenotype analysis (Figure 6C). The ntnac080-1 line had a 1 bp insertion at the 3′ end of sgRNA1 sequence which introduced a stop codon after position P121, and the ntnac080-2 plants harbor a 12 bp deletion causing a deletion of four amino acids and transition of one amino acid at position 117 (Figure 6B). In addition, we have also sequenced NtNAC028, the closest homolog of NtNAC080, in the two ntnac080 mutants. The sequencing results confirmed that these CRISPR mutants contain no mutation on NtNAC028 gene. Thus these mutant lines might express abnormal NtNAC080 proteins with a truncated/mutated NAC domain (Figures 6B,C). Both ntnac080-1 and ntnac080-2 plants showed a similar delayed senescence phenotype compared to WT (K326) tobacco (Figure 6D). Quantitative RT-PCR indicated no difference in transcript levels of NtNAC080 between these two mutants and WT plants (Figure 6E). We also assessed the progression of leaf senescence between mutants and WT by measuring changes in chlorophyll content and the maximal photochemical efficiency of PSII (Fv/Fm). Under glasshouse conditions, chlorophyll levels in individual leaves (leaf 1-6, numbered from the top to the bottom of a plant) of the two mutants were generally higher than that in counterpart leaves of the age-matched WT plants (Figure 6F). The Fv/Fm ratios in leaf 5 and 6 of the mutants plants were also higher than that in counterpart leaves of WT plants (Figure 6G). Furthermore, we determined the expression of senescence-related marker genes NtCP1 (SAG12 homolog in tobacco) and NtRBCS in leaf 5. As shown in Figures 6H,I, the expression of NtCP1 was lower and NtRBCS was higher in two mutants. These data indicated that NtNAC080 acts as a positive regulator of leaf senescence in tobacco.