Development of a nornicotine-reduced flue-cured tobacco line via EMS mutagenesis of nicotine N-demethylase genes

ABSTRACT

Substantial progress had been made in reducing nornicotine accumulation in burley tobacco, as nornicotine is a precursor of the carcinogen N-nitrosonornicotine (NNN). Three members of the CYP82E2 family encoding nicotine N-demethylase (NND) have been reported to be responsible for the majority of nicotine demethylation that forms nornicotine in burley tobacco. We had obtained a nonsense mutant of each NND member in flue-cured tobacco from an ethyl methanesulfonate (EMS)-mutagenized population. In this study, we developed dCAPS markers for each nonsense mutation. Using marker-assisted selection, NND mutants were crossed with each other to generate a triple mutant GP449. In line with previous reports, the triple knockout caused significantly decreased levels of nornicotine and NNN in flue-cured tobacco. With the decreased nornicotine, the nicotine level was expected to accumulate. However, the nicotine level in GP449 was significantly decreased to 72.80% of wild type. Realtime RT-PCR analysis showed that the nicotine reduction was correlated with inhibited expression of nicotine biosynthetic pathway genes. The triple mutant and dCAPS markers can be utilized to develop new flue-cured tobacco varieties with lower levels of nornicotine and NNN.

Introduction

Alkaloids are nitrogen-containing, low molecular-weight compounds. Nicotine is the principal alkaloid in cultivated tobacco (Nicotiana tabacum L.), accounting for more than 90% of the total alkaloids, and the remaining alkaloids are primarily nornicotine, anabasine, and anatabine.^1,2 Nornicotine reacts with nitrite to form N-nitrosonornicotine (NNN) during leaf senescence and curing.³ NNN is the most potent carcinogenic agent in tobacco and had been associated with many types of cancer.⁴ Therefore, suppression of nornicotine biosynthesis would reduce NNN production, which would be beneficial for tobacco harm reduction.

Nornicotine is produced by nicotine demethylation catalyzed by nicotine N-demethylase (NND). To date, four NND genes mediating nicotine to nornicotine conversion have been cloned from burley tobacco: CYP82E4, CYP82E5, CYP82E10, and CYP82E21.^5-10 CYP82E21 encodes an active demethylase, but it is only expressed in the ovary and may contribute little to nornicotine formation.⁷ The nicotine to nornicotine conversion rate (NCR) in a mutant of CYP82E4 (e4/e4/E5E5/E10E10) was about 2.2% and is much lower than that in control plants, indicating that CYP82E4 is the major NND for nicotine demethylation in burley tobacco. Consistent with this observation, no significant difference was found between single mutant of CYP82E10 (E4E4/E5E5/e10e10), a double mutant of CYP82E5 and CYP82E10 (E4E4/e5e5/e10e10), and control plants. However, introducing the e5e5e10e10 mutation into the e4e4 background resulted in even lower conversion rate, indicating minor roles for CYP82E5 and CYP82E10.⁶ The NNN levels decreased significantly in a triple mutant (e4e4e5e5e10e10) with less nornicotine.^6,11 This triple mutant was further used to develop harm-reduction burley varieties, which have been shown to selectively reduce NNN levels in cigarette tobacco filler and mainstream smoke.^12,13 However, the functions of NND genes and their effects on nornicotine levels in flue-cured tobacco remain to be clarified.

In this study, we identified single mutants for CYP82E4, CYP82E5 and CYP82E10 genes in an ethyl-methyl sulfone (EMS) mutagenic population of flue-cured tobacco (N. tabacum cv. Yunyan 87) and generated triple mutant plants using dCAPS markers. Consistent with previous reports, nicotine conversion was dramatically decreased in a triple mutant (e4e4/e5e5/e10e10). However, in contrast to these reports, we found that the flue-cured tobacco triple mutant had significantly decreased nicotine levels and down-regulated expression of nicotine biosynthetic pathway genes. This mutant line will serve as useful germplasm for harm-reduction breeding in flue-cured tobacco.

Materials and methods

Validation of mutation by sanger sequencing

Fresh leaf disks were collected from mutant plants. Total DNA was extracted using the DNeasy Plant Mini Kit following manufacturer’s instructions (Qiagen, Germany). Gene fragments harboring mutation sites were amplified with a high-fidelity Taq DNA polymerase with genomic DNA as the template. PCR products were purified for sequencing using a PCR purification Kit (Qiagen, Germany). Primers used are listed in Supplementary Table 1.

Development of dCAPS markers and maker-assisted mutant screening

The online program “dCAPS Finder 2.0” (http://helix.wustl.edu/dcaps/dcaps.html) was used to design dCAPS primers. Wild type and mutant sequences were input for analysis. The output primers introduced a mismatch nucleotide to couple a restriction enzyme recognition site to a SNP site of the mutated gene. Since the CYP82E2 family comprises several homologous members, a two-round PCR strategy was chosen to avoid the influence of homologous genes. The first-round of PCR was conducted using the sequencing primers for the target gene to obtain products harboring the mutation. Then, these PCR products were used as templates in second-round amplification with dCAPS primers. After recovery with a Gel Extraction Kit (Qiagen, Germany), the PCR fragments were digested with appropriate restriction enzymes and separated in 6% non-denaturing PAGE (220 V, 1.5 h) with silver staining for visualization. Primers and enzymes used here are listed in the Supplementary Table 1.

Quantitative RT-PCR analysis

RNA was isolated from root and leaf of mutant plants. Total RNA was extracted using the RNeasy Plant Mini Kit following manufacturer’s instructions (Qiagen, USA). RNA quantity was determined with a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA). About 2 μg of total RNA were used in DNase I digestion. Synthesis of first-strand cDNA was performed using Superscript III Reverse Transcriptase (Invitrogen) in a total volume of 20 μl. Quantitative real-time PCR (qRT-PCR) was performed using FastStart Universal SYBR Green Master (Rox) (Roche, Germany) with a LightCycler 480 II (Roche, USA). The actin gene was used as an internal reference for normalization. The relative expression levels were calculated as the ratio to control. Primers used are listed in Supplementary Table S1.

Test of mutant plants in field and pots

The field test was performed at Yanhe research station (Yuxi, China) in 2017. Three replicates, with 10 plants in each replicate, were grown. All plants were topped (removal of the flower) just before flowering. Two weeks after topping, all leaves from individual plant were collected for flue-curing. For experiments in pots, samples were collected from 10 plants grown in sand pots. Topping was done as described above. All leaves from each plant were collected 2 weeks after topping and dried in a 60°C oven for 3 d. Dried leaf samples were ground into powder for alkaloid quantification.

Alkaloid and NNN quantification

To prepare each sample, 0.5 g of dried leaf powder was placed into a 50 mL flask. Then, 5 mL of 10% (v/w) NaOH solution was added to each flask, and flasks were swirled for 15 min. Next, 20 mL of dichloromethane containing 1 μg/mL quinolone was added, followed by sonication for 60 min. After centrifuging at 5000 rpm for 5 min, 2 mL of the bottom dichloromethane layer were transferred through a microporous filter and collected filtrate was quantified for alkaloids with a 450GC-300MS system (Bruker, USA).¹³ For NNN quantification, 0.5 g of dried leaf powder was placed into a 50 mL centrifuge tube. After being spiked with 150 μL internal standard solution (1.0 μg/mL), 15 mL ammonium acetate solution (0.01 M) was added and sonicated for 30 min. After centrifugation (10,000 rpm, 2 min), the upper layer was filtered for LC-MS/MS analysis.¹⁴

Results

Validation of CYP82E4, CYP82E5 and CYP82E10 single mutants

We successfully identified CYP82E4, CYP82E5 and CYP82E10 single mutants from an EMS mutant population of N. tabacum cv.Yunyan87 (Y87).¹⁵ For further validation, genomic DNA was isolated from leaves, and then PCR products were amplified for each gene using corresponding primers and confirmed by Sanger sequencing. As shown in Figure 1, there is a C-to-T transition at 505th position of the cDNA sequence in the CYP82E4 mutant, which creates a premature stop codon (TGA) that terminates CYP82E4 translation. Similarly, a G-to-A transition at 392nd position and 687th position of CYP82E5 and CYP82E10, respectively, prematurely stop gene translation (Figure 1).

Figure 1.

Single nucleotide polymorphisms (SNPs) between wild type and mutant alleles in CYP82E4, CYP82E5, and CYP82E10. (a) forward sequence of CYP82E4 showing C-to-T transition. (b) forward sequence of CYP82E5 showing G-to-A transition. (c) forward sequence of CYP82E10 showing G-to-A transition. The SNP is indicated by a red arrow in the sequence alignment.

Generation of triple mutant using dCAPS makers

Using the point mutations in CYP82E4, CYP82E5, and CYP82E10, three dCAPS markers were designed and named E4_XhoI, E5_BalI, and E10_BamHI, respectively (Supplementary Table 1). As expected, the digested PCR products of each dCAPS marker could precisely identify wild type (WT), heterozygote and homozygote plants (Figure 2). To generate a heterozygous triple mutant (E4e4E5e5E10e10), successive crossing between single mutants and then double mutants were performed with the assistance of these dCAPS markers. The heterozygous triple mutant was self-pollinated, and homozygous triple mutants (e4e4e5e5e10e10) were identified in the progeny and designated as GP449. The genotype of GP449 was double-checked with Sanger sequencing (data not shown).

Figure 2.

Development of gene-specific dCAPS markers for CYP82E4 (a) CYP82E5 (b) and CYP82E10 (c). The wide type, homozygous mutant and heterozygous mutant of each gene were marked as A/A, a/a and A/a, respectively. The 1-kb plus DNA ladder (Takara) was used. All primers and restriction enzymes are listed in Supplementary Table 1.

Quantification of alkaloids and NNN in GP449

A field test was performed to evaluate the effect of introduced mutations on alkaloid content of flue-cured tobacco (Figure 3). The nornicotine level in GP449 (0.22 mg/g) decreased significantly compared to that in wild type (2.35 mg/g). As expected, the NNN level was also significantly reduced in GP449. The NCR in GP449 was 1.25%, which was 14.65% of that in wild type (8.53%). A similar result was observed in burley tobacco with an NCR only 7.69% of wild type in a triple mutant (e4e4e5e5e10e10).⁶ Therefore, CYP82E4, CYP82E5, and CYP82E10 are responsible for the majority of nicotine to nornicotine conversion in both burley and flue-cured tobacco. However, a relatively high proportion of NCR remained in flue-cured tobacco GP449, so there may be other genes, most likely from the CYP82E2 family, or pathways involved in nicotine conversion in flue-cured tobacco. We further hypothesized that inhibition of the demethylation step would result in nicotine accumulation. Surprisingly, the nicotine level in the GP449 triple mutant did not increase. On the contrary, the nicotine level in GP449 was only 72.80% of wild type (1.74% DW vs 2.39% DW of wild type) (Figure 3).

Figure 3.

Alkaloids and NNN content in mutants grown in the 2017 field experiment. Alkaloid and NNN contents are averages of three replicates, each replicate consists of 10 plants. NCR = [nornicotine/(nicotine+nornicotine)] × 100%. Error bars indicate standard errors. The asterisk indicates a significant difference compared with control (Y87) by the Student t-test (p < .05).

To confirm the nicotine decrease observed in the field trial, we measured alkaloid content in pot grown plants. Similar results were obtained, as the nicotine level in GP449 had decreased to 73.33% of wild type (0.22% DW vs 0.30% DW of wild type) and the NCR had also decreased to 32.05% of wild type (0.94% vs 2.92% of wild type) (Figure 4).

Figure 4.

Alkaloid content and NCR of mutants grown in pots. (a) Accumulation of nicotine and nornicotine in Y87 and GP449. Values are an average of 10 plants with standard errors. (b) NCR in Y87 and GP449. The asterisk indicates a significant difference compared with control (Y87) by the Student t-test (p < .05).

Expression analysis of nicotine pathway genes in GP449

The decrease in leaf nicotine could be caused by suppression of its biosynthesis in tobacco root. To test this hypothesis, we measured the expression of nicotine biosynthetic pathway genes in the triple mutant. As shown in Figure 5, the expression of NtPMT1a, NtA622, and NtQPT2 in GP449 was significantly reduced compared to that of control, indicating that the reduction of nicotine is correlated with inhibited expression of nicotine biosynthesis genes in root. We further assayed the expression of CYP82E4, CYP82E5, and CYP82E10 in the leaf and root of GP449. The results showed that the expression of CYP82E5 and CYP82E10 were significantly decreased in both of root and leaf, while CYP82E4 expression could not be detected in both tissues (Figure 5). Reduced expression of NND genes further implies that the nicotine decrease was caused by impaired nicotine biosynthesis.

Figure 5.

Expression analysis of alkaloid-related genes. qRT-PCR was performed to measure gene expression in root (a) and leaf (b) of Y87 and GP449. NtActin was used as an internal control. The gene expression in Yunayan87 was set to 1, and gene expression in mutants is shown relative to this level. Error bars indicate the standard error of the means of more than three biological replicates. The asterisk indicates a significant difference compared with (Y87) by the Student t-test (p < .05).

Discussion

In agreement with previous studies of burley tobacco,^6,11 triple mutation of CYP82E4, CYP82E5, and CYP82E10 in flue-cured tobacco significantly reduced the NCR, suggesting that both burley and flue-cured tobacco share similar molecular mechanisms for nicotine conversion. However, important differences were also observed in this study. First, the major gene controlling the demethylation of nicotine in burley is CYP82E4, but we could not detect its expression in any flue-cured tobacco tissue (root, stem and leaf, Figure 5; senescing leaf, data not shown) tested in this study, suggesting that this gene is not involved in nicotine demethylation in flue-cured tobacco. Second, the relative NCR remaining in the triple mutant GP449 (14.65%) is much higher than that in corresponding burley tobacco mutant (7.69%), indicating mechanistic differences in the control of nicotine demethylation in these two types of cultured tobacco. In tobacco, there are many members of the CYP82E2 family. Recently, one member, CYP82E21, was found to possess demethylase activity even though it is unlikely to participate in nicotine demethylation because it is predominantly expressed in ovary.⁷ It is possible that other members are involved in nicotine demethylation specifically in flue-cured tobacco. Furthermore, a nicotine-independent nornicotine biosynthesis pathway is probably present in flue-cured tobacco. Putrescine can be oxidized by methylputrescine oxidase (MPO) to form demethylated pyrrolinium salt,¹⁶ which can be directly converted into nornicotine.¹⁷ In GP449, in contrast with other nicotine biosynthetic genes, the expression of NtMPO was not significantly reduced (Figure 5). This indicates nornicotine could be formed directly from putrescine via NtMPO catalyzation.

In this study, the nicotine levels were significantly decreased compared with control in the triple mutant GP449 (Figures 3 and 4), which is contrary to our expectation that nicotine should accumulate after its breakdown has been largely blocked. Further qRT-PCR analysis demonstrated simultaneously reduced expression of nicotine biosynthesis genes, indicating that reduced nicotine levels were not caused by mutations of these genes (Figure 5). Since EMS can induce many mutations in a genome, it would be possible to generate this phenotype if a key transcription factor related to nicotine biosynthesis was mutated in GP449. Furthermore, the methyl moiety of nicotine is derived from S-adenosylmethionine (SAM),¹⁸ which is the universal donor of a methyl group produced in the THF-mediated C1 metabolic pathway.^19,20 Methyl groups released by nicotine demethylation are recycled back into the C1 pool.²¹ All these studies clearly demonstrate a close relationship between nicotine metabolism and the C1 unit pool in tobacco. It has been reported that an increased C1 pool created by overexpression of S-adenosylmethionine synthetase gene (SAM-S) can stimulate nicotine biosynthesis in tobacco callus.²² Consistent with this, reduced expression of methylenetetrahydrofolate reductase gene (NtMTHFR), which may limit the C1 unit supply, resulted in depressed nicotine content.²³ In GP449, C1 unit supply was probably restricted due to the suppression of nicotine demethylation in the root, which could subsequently inhibit nicotine biosynthesis and thereby reduce nicotine transfer to tobacco leaf. However, the regulation of nicotine biosynthetic genes under C1 unit deficiency remains to be investigated.

Trait improvement in tobacco through breeding of EMS mutants requires several rounds of crossing (in the case of gene pyramiding) and backcrossing. To speed up the breeding process and reduce the labor, molecular markers are desired. In this study, we developed dCAPS markers for the mutated NND genes to accurately differentiate homozygous, heterozygous and wild type plants. Taking advantage of these markers, we obtained the triple mutant GP449 which can be used to develop low NNN flue-cured tobacco varieties in the future.

Funding Statement

This study was supported by the National Natural Science Foundation of China [31560080]; the grants from China Tobacco Company and Yunnan Tobacco Company [110201801027 (JY-04), 2016YN25, 2017YN03 and 2018530000241001].

Acknowledgments

We would like to thank Professor Ralph E. Dewey (North Carolina State University) for helpful suggestions.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.