Functional Analysis of Nicotine Demethylase Genes Reveals Insights into the Evolution of Modern Tobacco*

Tobacco (Nicotiana tabacum L.) is a natural allotetraploid derived from the interspecific hybridization between ancestral Nicotiana sylvestris and Nicotiana tomentosiformis. The majority of cultivated tobacco differs from both of its progenitor species in that tobacco typically contains nicotine as the primary alkaloid, in contrast to its two progenitors that accumulate nornicotine in the senescing leaves. However, most, if not all, tobacco cultivars possess an unstable mutation, commonly referred to as the conversion locus, that when activated mediates the conversion of a large percentage of nicotine to nornicotine in the senescing leaf. We have recently identified CYP82E4, a tobacco nicotine N-demethylase gene whose expression was highly induced during senescence in plants that have converted, and CYP82E3, a closely related homolog that exhibited no nicotine N-demethylase activity. In this study, domain swapping and site-directed mutagenesis studies identified a single amino acid change that fully restored nicotine N-demethylase activity to CYP82E3. An examination of the N. tomentosiformis orthologs of CYP82E3 and CYP82E4 revealed that both are functional nicotine N-demethylase genes in N. tomentosiformis. Collectively, our results suggest that a single base pair mutation in CYP82E3 and transcriptional suppression of CYP82E4 played important roles in the evolution of the alkaloid profile characteristic of modern tobacco.

Pyridine alkaloids, such as nicotine, nornicotine, anabasine, and anatabine, have been detected in numerous species of the Nicotiana genus (1) where they are thought to function as herbivore deterrents (2). The production of pyridine alkaloids is elicited by herbivore attacks (3,4) through a jasmonic acidinduced signaling pathway (5,6). Several studies have demonstrated that the alkaloid composition within the Nicotiana genus is species-specific and highly variable, and in the majority of the species a single alkaloid predominates (1,7). For example, Nicotiana tomentosiformis primarily accumulates nornicotine, whereas in Nicotiana glauca anabasine is the major alkaloid (1). Interestingly, the alkaloid profiles of some Nicotiana are differ-ent in green versus senescing (yellow) leaves. For instance, in the green leaves of Nicotiana sylvestris nicotine is the primary alkaloid, but a substantial portion of nicotine is converted into nornicotine once the leaves senesce through air-curing (1).
In cultivated tobacco alkaloid composition can vary because of the presence of an unstable locus, termed the "converter locus" (8). In commercial tobacco leaves that harbor an inactive converter locus, the alkaloid pool typically contains about 95% nicotine and 3% nornicotine (7). These plants are frequently referred to as "nonconverters." In some of the individuals, however, the converter locus becomes activated, and a large portion of nicotine is metabolized to nornicotine (9) giving rise to plants termed "converters." Similar to N. sylvestris, nicotine to nornicotine metabolism in converter tobacco is induced during leaf senescence (10). The accumulation of nornicotine in converter plants is undesirable, because nornicotine is a biochemical precursor of NЈ-nitrosonornicotine, a tobacco-specific nitrosamine that has been shown to exhibit carcinogenic properties in laboratory animals (11)(12)(13). In addition, nornicotine has been recently recognized as an elicitor of protein glycation (14) and other biochemical processes that are detrimental to human health (15,16).
Nicotiana tabacum is a natural allotetraploid derived from the interspecific hybridization of ancestral N. sylvestris and N. tomentosiformis (17)(18)(19)(20)(21)(22), species that convert nicotine to nornicotine in senescing and green leaves, respectively (1). As a result, the emergence of the nicotine-accumulating (nonconverter) phenotype of N. tabacum from its two progenitors would have required the inactivation of nornicotine-accumulating traits originating from both ancestral parents. Mann et al. (8) suggested that the nicotine-accumulating phenotype of nonconverter tobacco was developed by the successive inactivation of the two dominant conversion loci donated by the nornicotine-containing parental species, but to date such loci have not been identified.
Another interesting question relates to the origin of the unstable conversion locus present in converter tobacco. In theory, either of the two progenitor species could have donated the conversion locus to tobacco, because nicotine to nornicotine conversion occurs in both N. sylvestris and N. tomentosiformis at one developmental stage. Classical genetic studies conducted by Mann et al. (8) demonstrated that the unstable conversion locus found in converter tobacco is allelic to the green leaf conversion locus of N. tomentosiformis and inherited as a dominant gene (8,9,23). The observation that the unstable conversion locus was originated from green-leaf converter N. tomentosiformis was surprising, because the reversion of the conversion locus to the ancestral type in tobacco confers senescing leaf conversion phenotype characteristic to N. sylvestris. The apparent inconsistency between the time of nornicotine production and the origin of the unstable conversion locus in tobacco suggested that evolution of the alkaloid profile of modern tobacco involved complex genetic changes.
We have previously isolated a closely related gene family, designated the CYP82E2 gene subfamily, whose collective transcript accumulation was up-regulated in converter versus nonconverter tobacco (24). Of these cDNAs, CYP82E4 was shown to mediate the metabolic conversion of nicotine to nornicotine when overexpressed in tobacco and yeast (24). Expression analysis of CYP82E4 revealed a Ͼ80-fold increase of transcript accumulation in the cured leaves of converter versus nonconverter plants demonstrating that CYP82E4 plays a central role in mediating nicotine N-demethylation in senescing leaves of tobacco (25). To distinguish between orthologous P450 genes from tobacco and N. tomentosiformis, an Ntab or Ntom prefix is attached, respectively, to the name of the genes in the remaining part of the paper (e.g. NtabCYP82E4 or NtomCYP82E4).
Because not all members of the NtabCYP82E2 gene subfamily exhibited nicotine N-demethylase (NND) 2 activity (24), the original goal of this study was to define the structural elements responsible for NND activity in the closely related P450 isoforms of the NtabCYP82E2 gene subfamily. Our results showed that a single base amino acid substitution in the coding region of another closely related isoform, NtabCYP82E3, is sufficient to confer NND activity to the encoded protein. DNA sequence and gene expression data also demonstrated that N. tomentosiformis possesses two active NND genes, NtomCYP82E3 expressed strongly in the green leaves and NtomCYP82E4 specifically up-regulated during senescence. We propose that during the evolution of modern tobacco, NtabCYP82E3 was impaired by a point mutation, and NtabCYP82E4 was inactivated by an unstable mutation resulting in the transcriptional suppression of the gene.

EXPERIMENTAL PROCEDURES
Plant Materials-Green-leaf converter SC58(C T C T ) tobacco line was generated by crossing an interspecific hybrid between N. tomentosiformis and Flue-cured tobacco variety 402 with Flue-cured tobacco cultivar SC58 followed by eight cycles of backcrossing of the progeny to the SC58 parent. In each generation, green-leaf converter individuals were selected and used in the subsequent cross with cultivar SC58. Finally, the conversion locus (C T ) was fixed to homozygosity by self-pollinating the heterozygous introgressed lines and identifying individuals whose progeny no longer segregated for the green-leaf conversion trait SC58 (8). Other tobacco cultivars included in the study were full-sib double haploid burley lines DH98-325-5 (nonconverter) and DH98-325-6 (senescing-leaf converter). All genotypes were originally grown for 30 days in a growth chamber and then transferred to a controlled environment greenhouse equipped with supplemental lighting providing a 14/10-h light/dark cycle. Leaf samples were collected 30 days after transplanting. For curing, tobacco leaves were dipped in 0.2% ethephon, dried for 30 min at room temperature, and stored in dark until they turned yellow.
Construction and Analysis of the NtabCYP82E3/ NtabCYP82E4 Chimeras-Chimeric clones were generated by digesting the NtabCYP82E3 and NtabCYP82E4 cDNAs with restriction enzymes that recognize restriction sites conserved between the two cDNAs and replacing the resulting NtabCYP82E4 restriction fragment with the corresponding NtabCYP82E3 region. NtabCYP82E3/NtabCYP82E4 chimeras were subcloned into the pYeDP60 yeast expression vector, and the resulting constructs were introduced to yeast strain WAT11 as described (26). Yeast culturing, transgene induction, and preparation of yeast microsomes were performed as described previously (27). Microsomal fractions were incubated for 45 min at 23°C in 50 mM phosphate buffer, pH 7.1, containing 0.5 mg of microsomal protein, 2.45 M [pyrrolidine-2-14 C]nicotine (Moravek Biochemicals, Brea, CA), and 0.75 mM NADPH in a total reaction volume of 150 l. The reactions were arrested with 50 l of acetone, and 50 l of the reaction mixture was spotted on a 250-m Whatman K6F silica plate. Conversion of nicotine to nornicotine was quantified by localizing and integrating the radioactive traces of nicotine with a Bioscan System 400 imaging scanner. Site-directed mutagenesis was conducted with the QuikChange TM site-directed mutagenesis kit as described by the manufacturer (Stratagene, La Jolla, CA).
Isolation and DNA Sequence Analysis of the CYP82E3 and CYP82E4 Orthologs-Plant genomic DNA was extracted by grinding a 100-mg fresh leaf sample in 500 l of homogenization buffer (100 mM Tris-HCl, pH 8.8, 1 M KCl, 10 mM EDTA) using the FastPrep FP120 Cell Disrupter (Qbiogene, Inc. Carlsbad, CA). P450s were amplified in a PCR containing 1 l of leaf extract, 2 M each primer, 200 M each of dNTP, and 1.5 mM MgCl 2 in a 50-l final reaction volume. The primers used for the isolation of the various genomic P450 fragments are listed in Table 1. DNA amplification was performed in 30 cycles of denaturing at 95°C for 15 s, primer annealing at 55°C for 30 s, and primer extension at 72°C for 30 s. For the allele-specific amplification of the CYP82E3 orthologs, a 61°C annealing temperature was used. PCR products were visualized on an ethidium bromide-stained 1% agarose gel and directly cloned into the pGEM Easy cloning vector (Promega Corp., Madison, WI). DNA sequence of the clones were determined by the chain termination method (28). Sequences were aligned using the ClustalW algorithm (29).
Isolation of NtomCYP82E3 and NtomCYP82E4 cDNAs-Total RNA was extracted from plant leaves using the TRIzol reagent following the manufacturer's instructions (Invitrogen). Purified RNA was treated with RNase-free DNase (TURBO DNA-free; Ambion, Austin, TX). First strand cDNA was synthesized using 5 g of total RNA using the StrataScript first-strand synthesis system (Stratagene, Cedar Creek, TX). PCR amplification, cloning of the PCR products, and DNA sequencing of the P450 cDNAs were performed as described for the isolation of the genomic P450 fragments. The nucleotide sequences of the primers used in the PCR are listed in Table 1.
Kinetic Analysis of NtomCYP82E3-and NtomCYP82E4-mediated Nicotine Metabolism-Yeast microsomes were prepared as described previously (27). Microsomal protein concentrations were determined by the Bradford protein assay (Bio-Rad). Nicotine concentration ranges used in the study were 3-467 and 0.6 -13 M for NtomCYP82E3 and NtomCYP82E4, respectively. Kinetic constants were determined by plotting enzyme activity versus substrate concentration and fitting the Michaelis-Menten equation to the data by nonlinear least square method using Sigma Plot version 4.0. Significant differences between parameters were analyzed by one-way analysis of variance (PROC GLM, SAS, 2000).
Real Time-PCR Analysis-To measure the transcript abundance of the CYP82E3 and CYP82E4 orthologs, RT-PCR was employed using SYBR Green I fluorescence chemistry (30). A serial dilution of the P450 cDNAs clones was used as the templates for generating a calibration curve. The RT-PCR mixture contained 2.5 mM MgCl 2 , 125 M each dNTP, 0.5 M each primer, 0.5ϫ SYBR Green I, 0.5 g of cDNA (or 1 l of reference plasmid), and 1.25 units of platinum Taq polymerase (Invitrogen). The nucleotide sequences of the allele-specific P450 primers (E3RTF, E3RTR, E4RTF, and E4RTR) are listed in Table 1. RT-PCR was performed according to the following protocol: 95°C for 2 min; 35 cycles of 95°C for 30s, 55°C for 30 s, 72°C for 50 s followed by final extension at 72°C for 5 min in a Bio-Rad iCycler thermocycler (Bio-Rad) using a 220-bp fragment of the glyceraldehyde-3-phosphate dehydrogenase cDNA as an internal standard. P450 cDNA concentration was calculated based on the transcript-specific calibration curve normalized to the internal standard. The purity of PCR products was determined by melting-curve analysis as described previously (31). DNA sequence of the amplified cDNA fragments was confirmed by ligating the RT-PCR products into the pGEM-T Easy vector and sequencing 20 randomly selected clones of each cDNA fragment.

RESULTS
The C330W Amino Acid Substitution Confers NND Activity to the NtabCYP82E3 cDNA-We have demonstrated previously that a group of P450 isozymes, designated the CYP82E2 gene subfamily, play a major role in the conversion of nicotine to nornicotine (24). We isolated five unique members of the CYP82E2 gene subfamily, of which only one isoform, NtabCYP82E4, was shown to encode an NND. NtabCYP82E3, the most closely related homolog to NtabCYP82E4, shares 95% predicted amino acid sequence identity with NtabCYP82E4 but conferred no NND activity when expressed in yeast or tobacco (24).
The high amino acid sequence homology yet distinctly different NND activity between NtabCYP82E3 and NtabCYP82E4 provided us with an opportunity to identify unique structural elements that bestow NND activity to NtabCYP82E4. Our hypothesis was that one or few of the 30 polymorphic amino acids between NtabCYP82E3 and NtabCYP82E4 represent catalytically important residues for NND activity in NtabCYP82E4. To identify segments of the coding region that selectively impart NND activity to NtabCYP82E4, we employed domain swapping, a strategy that creates a fusion polypeptide from two (or more) protein sequences and thus facilitates the identification of domains that control the expression of traits (e.g. function) differentially encoded in the parent molecules.
The high degree of sequence homology between NtabCYP82E3 and NtabCYP82E4 enabled us to identify three restriction sites (PmlI, SwaI, and EcoRV) located in the same position in both cDNAs (Fig. 1). Using these restriction sites as junction points, we replaced the 459-, 711-, and 1201-bp 5Ј-most coding region of NtabCYP82E4 with the corresponding gene fragments of NtabCYP82E3, yielding the P-K, S-K, and E-K hybrid constructs, respectively (Fig. 1). The chimeras were cloned into the yeast expression vector pYeD60, and the clones were used to transform the WAT11 yeast strain. In vitro yeast assays were used to quantify the NND activity of all three chimeras. The results showed that chimeras P-K and S-K retained NND activity, whereas E-K did not support nicotine demethylation, suggesting that an element required for NND activity is located between the SwaI and EcoRV restriction sites of NtabCYP82E4 (Fig. 1). To test this hypothesis, we replaced the SwaI-EcoRV fragment of NtabCYP82E4 with the corresponding region from NtabCYP82E3 (construct S-E) and tested the To further define amino acid residues essential for NND activity, site-directed mutagenesis experiments were conducted. The SwaI-EcoRV region of NtabCYP82E4 encodes 164 amino acids, 7 of which are polymorphic to NtabCYP82E3 (Fig. 2). The relatively few polymorphic amino acids encoded within the SwaI-EcoRV region of the NtabCYP82E4 and NtabCYP82E3 genes made it feasible to employ a mutation stacking approach for identifying catalytically important amino acids. Codons encoding all polymorphic amino acid residues in the SwaI-EcoRV fragment of the NtabCYP82E3 coding region were progressively substituted for the corresponding codons of the NtabCYP82E4 cDNA and expressed in yeast. Microsomal extracts isolated from yeast cells enriched in the single (K269N), double (K269N,I274Null), or triple (K269N,I274Null,T279A) mutant NtabCYP82E3 variants exhibited no NND activity. In contrast, strong NND activity was apparent when a microsomal extract containing the quadruple mutant (K269N,I274Null,T279A,C330W) NtabCYP82E3 protein was assayed. These results indicated that the 990C3 G nucleotide substitution, encoding the C330W amino acid mutation, was necessary to confer NND activity to the NtabCYP82E3 cDNA. To determine whether the C330W substitution alone was sufficient to bestow NND activity to NtabCYP82E3, we tested the C330W NtabCYP82E3 single mutation variant in the yeast microsomal assay for nicotine metabolism. The results of the microsomal assay showed that C330W NtabCYP82E3 actively catalyzed the nicotine N-demethylation reaction. Finally, we introduced a reciprocal point mutation (990G3C) in the NtabCYP82E4 coding region to confirm that the Trp 330 residue is essential for the NND function. As expected, the C330W substitution in NtabCYP82E4 resulted in a complete loss of NND activity as indicated by the yeast microsomal assay. Overall, these results showed that the C330W substitution is both necessary and sufficient to impart NND activity to NtabCYP82E3.

N. tomentosiformis CYP82E3 Encodes a Functional NND-
The observation that a single C330W substitution confers NND activity to NtabCYP82E3 led us to the hypothesis that the inactivation of NtabCYP82E3 by the W330C mutation may have represented one of the genetic changes responsible for the evolution of nicotine-retaining N. tabacum from its nicotine N-demethylating ancestors.
To test this hypothesis, we attempted to amplify the CYP82E3 orthologs from N. tomentosiformis and N. sylvestris genomic DNA using NtabCYP82E3-specific primers. PCR products were detected only when genomic DNA isolated from N. tomentosiformis was used as template suggesting that in tobacco CYP82E3 originated from the N. tomentosiformis ancestral parent (data not shown). DNA sequence analysis of NtomCYP82E3 predicted a Trp amino acid residue at position 330 lending support to our hypothesis that NtomCYP82E3 encodes a functional NND gene (Table 2). In addition to the predicted C330W mutation, a second substitution (K118N) was also detected in NtomCYP82E3. The GenBank TM accession number of the genomic sequence of NtomCYP82E3 is EF042306. Heterologous expression of the NtomCYP82E3 cDNA in yeast, however, suggests that the K118N polymorphism is neutral given that NtomCYP82E3 also exhibits an NND activity similar to that displayed by the C330W NtabCYP82E3 mutant (Table 3).
NtomCYP82E4 Is a Second NND Gene in N. tomentosiformis-To further characterize the genes involved in nicotine N-demethylation in N. tomentosiformis, we sought to determine whether NtomCYP82E4 also encodes a viable NND protein.
NtomCYP82E4 was considered a likely NND candidate, because NtabCYP82E4 has been shown previously to mediate the nicotine to nornicotine conversion in senescing tobacco leaves (24).
To investigate the role of NtomCYP82E4 in nicotine N-demethylation, NtomCYP82E4 was isolated from N. tomentosiformis genomic DNA, and the nucleotide sequence of the gene was compared with that of NtabCYP82E4 derived from tobacco cul-   tivar DH98-325-6. Sequence comparison revealed only two differences between NtomCYP82E4 and NtabCYP82E4 at nucleotide positions 56 and 300, where NtomCYP82E4 contained A and T, and NtabCYP82E4 showed T and C, respectively ( Table  4). The 56A3 T nucleotide polymorphism encodes for a Phe 19 amino acid residue in tobacco and a Tyr 19 in N. tomentosiformis ( Table 4). The GenBank TM accession number of the genomic sequence of NtomCYP82E4 is EF042307. Similar to NtomCYP82E3, NtomCYP82E4 also exhibited NND activity when expressed in yeast (Table 3). To compare catalytic competence of NtomCYP82E3 and NtomCYP82E4 for nicotine N-demethylation, we determined the kinetic parameters of the two proteins. Microsomal fractions isolated from yeast cells transformed with the NtomCYP82E3 or NtomCYP82E4 cDNAs catalyzed the conversion of nicotine to nornicotine, whereas microsomes expressing the empty plasmid did not produce nornicotine. Although both proteins exhibited high affinity toward nicotine, the binding affinity and V max values of NtomCYP82E3 were about 5 and 7 times higher than that of NtomCYP82E4, respectively (Table 3).
Both NtomCYP82E3 and NtomCYP82E4 Correlate with the Green-leaf Converter Phenotype-A distinguishing feature of the converter phenotype between N. tomentosiformis and tobacco is the accumulation of nornicotine in the green, nonsenescent leaf of N. tomentosiformis. To investigate whether a correlation exists between NtomCYP82E3 and/or NtomCYP82E4 and the green-leaf converter phenotype of N. tomentosiformis, we extended our sequence analysis of the CYP82E3 and CYP82E4 orthologs to nonconverter individuals of the Flue-cured variety SC58 and the green-leaf converter SC58(C T C T ) tobacco line. SC58(C T C T ) was developed by crossing an interspecific hybrid between N. tomentosiformis and Flue-cured variety 402 with cultivar SC58, followed by eight backcrosses to the SC58 recurrent parent (8). During each generation, selection was made based on the dominant green-leaf converter trait originating from N. tomentosiformis. Theoretically, SC58(C T C T ) plants should contain less than 0.5% of the N. tomentosiformis haploid genome, including the selected DNA fragment containing the greenleaf converter locus (C T ).
If the NtomCYP82E3 and/or NtomCYP82E4 genes are associated with the green-leaf converter phenotype, one would expect to find these genes within the SC58(C T C T ) line. DNA sequence analysis of the CYP82E3 and CYP82E4 orthologs showed that both NtomCYP82E3 and NtomCYP82E4 were present in SC58(C T C T ) (Tables 2 and 4). As expected, the NtabCYP82E3 and NtabCYP82E4 genes were detected in all pure tobacco genotypes tested, including SC58, DH98-325-5, and DH98-325-6 (Tables 2 and 4). These results suggest that the presence of the NtomCYP82E3 and NtomCYP82E4 genes correlates with NND activity in the green leaves of SC58(C T C T ) tobacco.
Allele-specific primers based on the mutant-defining polymorphism at position 990 were also effective in distinguishing the origin of the specific CYP82E3 alleles in the materials tested. PCRs were performed using reverse primers specific for C 990 or G 990 in combination with an intron-specific forward primer. Accumulation of PCR products was evident when the C 990 allele-specific primer was used to amplify the NtabCYP82E3 fragment from the SC58, DH98-325-5, and DH98-325-6 plants or when the G 990 primer was used with SC58(C T C T ) and N. tomentosiformis DNA (Fig. 3). In contrast, no DNA amplification was detected when the alternative allele-specific primers were added. These results provide additional confirmation that the CYP82E3 variant found in SC58(C T C T ) originated from N. tomentosiformis.
Expression Analysis of the CYP82E3 and CYP82E4 Orthologs in DH98-325-6 and SC58(C T C T ) Tobacco-Discovery of the N. tomentosiformis orthologs of CYP82E3 and CYP82E4 in the introgression line SC58(C T C T ) suggest that the NND genes per se may be responsible for mediating the green-leaf converter phenotype common to N. tomentosiformis and SC58(C T C T ). To better understand the regulatory roles of the CYP82E3 and CYP82E4 orthologs in nornicotine biosynthesis, we measured the transcript accumulation of NtabCYP82E3/NtabCYP82E4 and NtomCYP82E3/NtomCYP82E4 in green and senescing leaves of tobacco lines DH98-325-6 and SC58(C T C T ), respectively, using an allele-specific RT-PCR strategy. NtabCYP82E3 and NtabCYP82E4 were weakly expressed in the green leaves of cultivar DH98-325-6 ( Table 5). Consistent with our previous observations, however, NtabCYP82E4 transcription was sharply induced in the cured leaves. No enhancement of NtabCYP82E3 expression was observed in cured leaves of DH98-325-6 ( Table 5).
Transcript accumulation of NtomCYP82E3 and NtomCYP82E4 was markedly different in line SC58(C T C T ).

TABLE 4
Polymorphic nucleotides and the encoded amino acids in the orthologous CYP82E4 genes isolated from various tobacco genotypes and N. tomentosiformis The abbreviations used are as follows: N, nucleotide; AA, amino acid.
a Nucleotide numbering is based on cDNA sequences. b The C/T polymorphism at N 300 does not cause change in the amino acid sequence. N. tomentosiformis (N.tom). C/G letters indicate the nucleotide residue at N 990 complementary to the allele-specific reverse primers E3-Cys (left) and E3-Trp (right) . Lanes 1-4 show the PCR products generated by the amplification of purified NtomCYP82E3 and NtabCYP82E3 templates, respectively. The abbreviation used is as follows: 325-6, DH98-325-6.

FIGURE 3. Allele-specific PCR amplification of CYP82E3 orthologs isolated from various tobacco cultivars and
Expression of NtomCYP82E3 was high in the green leaves but rapidly declined during senescence (Table 5). Although transcript accumulation of NtomCYP82E4 in the green leaves of SC58(C T C T ) plants was about 30-fold higher than that of NtabCYP82E4 in the pure tobacco line, it was still about 20-fold lower than that of NtomCYP82E3. Amplification specificity of the cDNAs was confirmed by DNA sequence analysis of 20 randomly selected clones derived from each RT-PCR product. In all cases, the clones yielded the identical DNA sequence to the predicted cDNA fragments confirming that the PCRs were allele-specific (data not shown). These results demonstrate that although both NtomCYP82E3 and NtomCYP82E4 are expressed in the green leaves of SC58(C T C T ), the expression of NtomCYP82E3 clearly predominates. Furthermore, the transcription of NtomCYP82E4 originating from N. tomentosiformis is strongly inducible by senescence in the SC58(C T C T ) line, similar to the manner in which NtabCYP82E4 is induced in converter tobacco lines. Finally, the expression of both NtabCYP82E3 and NtabCYP82E4 in green leaves is lower than that of their N. tomentosiformis orthologs.

DISCUSSION
Domain swapping and site-directed mutagenesis experiments were performed to determine the structural basis for the catalytic activity of NtabCYP82E4. These studies revealed that a single amino acid substitution (C330W) was necessary and sufficient to confer NND activity to NtabCYP82E3. Amino acid alignments of P450s isolated from various organisms, including bacteria, fungi, plants, and animals, indicated strong conservations of Phe, Trp, or Tyr at the position homologous to Cys 330 of NtabCYP82E3 (data not shown), suggesting that the presence of an aromatic amino acid residue at this location is important for P450 function. Therefore, the predicted Cys 330 residue in NtabCYP82E3 is likely to cause functional impairment rather than alternative substrate specificity in NtabCYP82E3.
The observation that a single amino acid mutation could bestow NND activity to NtabCYP82E3 and the fact that NND genes are considered prime candidates for constituting the converter loci characterized decades ago by Mann et al. (8) prompted us to sequence the CYP82E3 ortholog from N. tomentosiformis. A comparison of the genomic DNA sequence of NtabCYP82E3 and its N. tomentosiformis ortholog revealed that NtomCYP82E3 contained a C330W and a K118N predicted amino acid substitution (Table 2), and heterologous expression of the NtomCYP82E3 cDNA in yeast demonstrated that NtomCYP82E3 encoded a functional NND (Table 3). In vitro nicotine metabolism assays demonstrated that NtomCYP82E4 also possessed NND activity ( Table 3). The observation that NtomCYP82E3 and NtomCYP82E4 were located in the SC58(C T C T ) introgressed line supports the hypothesis that NND genes represent the converter loci described by genetic studies (Tables 2 and 4 and Fig. 3). The finding that the SC58C T C T plants, which were selected based on their green-leaf converter phenotype, contain the senescence-inducible NtomCYP82E4 gene in addition to the constitutively expressed NtomCYP82E3 indicate that NtomCYP82E3 and NtomCYP82E4 are physically linked, an observation consistent with the genomic organization of other closely related P450 isoforms that are often arranged in tandem repeats (32). Because of the close genetic linkage among the NND genes in N. tomentosiformis, nicotine conversion appears to be inherited as a single Mendelian trait as observed by several investigators (8,9,23).
Expression analysis of the CYP82E3 and CYP82E4 orthologs demonstrated that NtomCYP82E3 is highly expressed in the green leaves and NtomCYP82E4 is highly induced during leaf senescence. Although variations in the expression profiles of the NND genes are theoretically possible between N. tomentosiformis and SC58(C T C T ), the similar expression patterns of NtomCYP82E4 in SC58(C T C T ) and NtabCYP82E4 in converter tobacco confirms the notion that NtabCYP82E4 represents the unstable locus, derived from N. tomentosiformis, that is responsible for the genetic conversion of tobacco (Table 5).
Collectively, these results support the model proposed in Fig. 4. In N. tomentosiformis at least two active NNDs, NtomCYP82E3 and NtomCYP82E4, are located on the chromosomal region defined by the C T locus. These genes likely  arose by the duplication of an ancestral NND gene followed by sequence diversification. At the onset of amphiploidy, both progenitor species donated a set of conversion factors to tobacco. Because the acquired conversion loci included NtomCYP82E3 and NtomCYP82E4, ancestral tobacco had likely emerged as a green-leaf converter plant. During the evolution of tobacco, mutations at several conversion loci affected the alkaloid composition of the plant. These lesions include a stable small or large scale mutation, at the N. sylvestris-originating C S locus, whose regulation appears to mirror that of the senescence-inducible CYP82E4 gene, and the W330C mutation at the C T locus originating from N. tomentosiformis. In addition to the stable mutations, an unstable mutation in the N. tomentosiformis-derived CYP82E4 locus led to the inactivation of this NND gene by transcriptional suppression. Similar to NtabCYP82E4, the transcription of NtabCYP82E3 was also greatly reduced. These and possibly other mutations affecting additional conversion loci gave rise to nicotine-producing modern nonconverter tobacco, and the transcriptional reactivation of the NtabCYP82E4 locus played a major role in the emergence of converter tobacco. This model also suggests the mechanism by which the conversion locus derived from the green-leaf converter N. tomentosiformis conferred the senescing-leaf converter phenotype to converter tobacco. Because the C T locus donated by N. tomentosiformis includes both the constitutively expressed NtomCYP82E3 and senescence-induced NtomCYP82E4 genes, the inactivation of NtabCYP82E3 in tobacco restricted nornicotine production to the senescing leaves. Our model can be extended to explain the differences in the nicotine alkaloid profiles of other Nicotiana species. Sisson et al.
(1) classified numerous species of the Nicotiana genus as nicotine, green leaf nornicotine, or curing-specific nornicotine accumulators. One possibility is that nicotine accumulators possess a dysfunctional CYP82E3 and CYP82E4, green leaf nornicotine accumulators carry a functional CYP82E3 with or without CYP82E4, and curing-specific nornicotine accumulators contain a functional CYP82E4 and a dysfunctional CYP82E3.
The driving force behind the evolution of nicotine-accumulating tobacco remains an open question. High nicotine and low nornicotine tobacco plants could have been selected by man at the beginning of tobacco use. However, Wernsman and Matzinger (10) challenged this hypothesis arguing that man could not have selected nicotine-accumulating tobacco because of the presence of the two conversion loci donated by the N. sylvestris and N. tomentosiformis. Given our results that a minimum of three mutations was required for the evolution of nicotine-accumulating tobacco, the selection of these plants by planned breeding alone appears to be even more unlikely. Nicotine-containing tobacco could have also arisen through natural selection, because nicotine exhibits superior insecticidal activity to nornicotine in several insects (33)(34)(35). Alternatively, it is possible that the adaptive values of nicotine and nornicotine are equal, and nicotine N-demethylation is a dispensable function in tobacco. Interestingly, in the Repandae section of the Nicotiana genus nornicotine serves as a precursor for N-acyl-nornicotine, a compound that exhibits 1000-fold higher insecticidal activity against nicotine-resistant Manduca sexta than nicotine (36 -38). Other sections of Nicotiana, however, lack the acyltransferase enzyme necessary for N-acyl-nornicotine biosynthesis; therefore, nornicotine production provides no adaptive advantage in these species (36). Similar to members of the Repandae section, an early progenitor of Nicotiana could have benefited from nornicotine production, but the diminished adaptive value of the compound could have led to the loss of NND function in tobacco.