Non-statistical 13C Fractionation Distinguishes Co-incident and Divergent Steps in the Biosynthesis of the Alkaloids Nicotine and Tropine*

During the biosynthesis of natural products, isotopic fractionation occurs due to the selectivity of enzymes for the heavier or lighter isotopomers. As only some of the positions in the molecule are implicated in a given reaction mechanism, position-specific fractionation occurs, leading to a non-statistical distribution of isotopes. This can be accessed by isotope ratio monitoring 13C NMR spectrometry. The solanaceous alkaloids S-(−)-nicotine and hyoscyamine (atropine) are related in having a common intermediate, but downstream enzymatic steps diverge, providing a relevant test case to: (a) elucidate the isotopic affiliation between carbon atoms in the alkaloids and those in the precursors; (b) obtain information about the kinetic isotope effects of as yet undescribed enzymes, thus to make predictions as to their possible mechanism(s). We show that the position-specific 13C/12C ratios in the different moieties of these compounds can satisfactorily be related to their known precursors and to the known kinetic isotope effects of enzymes involved in their biosynthesis, or to similar reaction mechanisms. Thus, the pathway to the common intermediate, N-methyl-Δ1-pyrrolinium, is seen to introduce similar isotope distribution patterns in the two alkaloids independent of plant species, whereas the remaining atoms of each target compound, which are of different origins, reflect their specific metabolic ancestry. We further demonstrate that the measured 13C distribution pattern can be used to deduce aspects of the reaction mechanism of enzymes still to be identified.

Members of the plant family Solanaceae produce a range of alkaloids derived from L-ornithine or L-lysine, several of which are exploited for their recreational and/or pharmaceutical properties, despite their toxicity. S-(Ϫ)-Nicotine, the principle alkaloid of Nicotiana tabacum L. and related species has a long history of use for recreational and ethnopharmaceutical applications. Considered a component of the defensive chemical array of the genus Nicotiana (1,2), nicotine was widely exploited as an insecticide, eventually being phased out as less toxic neonicotinoids became available. Within a closely related group of plants occurs the tropane alkaloids, of which atropine, hyoscyamine, and scopolamine (hyoscine) have a history of use in mystical ceremonies (3). These compounds are now extensively used in modern medicine, atropine and hyoscyamine as mydriatic agents, scopolamine as an anti-emetic for controlling travel sickness.
For both these alkaloid types, the biosynthetic pathways have been extensively investigated at the metabolic, enzymatic, and genetic levels (4,5). Despite superficial structural dissimilarities, the pyrrolidine moiety of nicotine and tropine (of which atropine or hyoscyamine is the tropoyl ester) have a common precursor in L-ornithine, and share four enzymatic steps to the intermediate N-methyl-⌬ 1 -pyrrolinium salt (Fig. 1). Thereafter, a series of diverse enzymes intervene to condense this intermediate with different groups to render more elaborated structures. A number of the enzymes involved in this pathway are well described: others remain poorly understood.
In both these alkaloids, the reactions by which N-methyl-⌬ 1pyrrolinium salt is further elaborated has resisted a number of attempts at characterization, despite its central role in the pathway. In the case of S-(Ϫ)-nicotine, a stereospecific decarboxylative condensation with a derivative of nicotinate leads to S-(Ϫ)-nicotine. An extremely weak "nicotine synthase" has been reported (6), but other authors could not repeat this finding (5). Although it has been suggested that this is possibly catalyzed by an enzyme of the berberine-bridge type (7), no details of the nature of this key enzyme nor of its mechanism are available.
In a parallel fashion, condensation with the C2 position of acetoacetate (probably activated as the CoA thioester) followed by decarboxylation (probably spontaneous) leads to tropinone (4). Similarly, despite many efforts, the enzymology of the conversion of N-methyl-⌬ 1 -pyrrolinium to tropinone remains elusive. Tropine is produced by a stereospecific oxidoreductase (8) and is esterified with phenyllactoyl-CoA thioester to form the alkaloid littorine, precursor to hyoscyamine and scopolamine (9 -12).
In such cases, where enzymological approaches have proved inadequate, alternative means to probe the biochemistry are required. The development of molecular biological approaches has been fruitful (5,13,14) but is challenging where the enzymology is completely unknown. An alternative "retro-biosynthetic" approach involves the measurement and interpretation of the natural fractionation in 2 H or 13 C isotopes during the course of biosynthetic reactions. During the metabolic processes by which natural products are biosynthesized, isotopic fractionation 2 occurs, due to the sensitivity/insensitivity of the enzymes involved to the presence/absence of one or more heavy atoms in the reaction center. This is the kinetic isotope effect (KIE). 3 As a result, the position-specific isotope ratio in natural products is non-statistically distributed among the different positions. Of especial interest in the context of biosynthesis are the position-specific 13 C/ 12 C ratios, as these reflect the carbon skeleton(s) already present in the precursor molecule(s) and the biochemical processing each position has undergone at different steps in the pathway. Hence, a study of the position-specific 13 C/ 12 C ratios should enable the interpretation of the values observed in terms of the known biochemistry (15). This approach has proved remarkably accurate for primary metabolites and for a small number of specialized products (16 -18). However, accurate measurement of position-specific 13 C/ 12 C ratios has until recently been hampered by technical difficulties (see Ref. 19 for an expanded discussion of these), when suitable conditions for isotope ratio monitoring by 13 C NMR (irm-13 C NMR) spectrometry applied to medium-sized molecules have been established.
In a recent article (20) applying this tactic to a newly discovered natural product, we were able to propose a credible biosynthetic route based on the observed pattern of the site-specific 13 C/ 12 C ratios. However, in that study, no direct underlying biological chemistry is yet available to substantiate the proposed reaction scheme. We have now turned our attention to the natural solanaceous alkaloids, S-(Ϫ)-nicotine and tropine, to establish the degree to which observed positionspecific 13 C/ 12 C ratios can be related to known biochemistry to test three principal hypotheses.
1) Can we predict that similar reaction schemes should lead to similar isotope patterns? Isotopic fractionation should be associated with the KIEs intrinsic to the reaction mechanisms of the implied enzymes, rather than to the plant species involved. This can be tested by comparing the five carbon atoms derived from the common intermediate, N-methyl-⌬ 1pyrrolinium salt, that these alkaloids have in common in their biosynthetic schemes (Fig. 1).
2) Can we show that the differing origins of the other carbon atoms present that are derived from dissimilar origins, three in tropine and five in S-(Ϫ)-nicotine, do not display any particular common features?
3) Can the characteristics of the position-specific 13 C/ 12 C ratios help indicate mechanistic features of the enzymes that are as yet undescribed for these pathways?

Results
Suitable conditions were established (see "Experimental Procedures") for the acquisition of 13 C NMR spectra amenable to the calculation of position-specific 13 C/ 12 C ratios. Representative spectra are illustrated in Fig. 2.
Subjecting these spectra to total line shape curve fitting provided the area under the peak for each carbon atom i. From this, the molar fraction was calculated, which gives the extent to which the 13 C/ 12 C ratios diverge from a statistical distribution. By combining this with the global value for the whole molecule, ␦ 13 C g (‰), i.e. the deviation of the carbon isotopic ratio of the whole molecule relative to that of the international standard Vienna Pee Dee Belemnite, (V-PDB), values for the positional isotopic distribution, ␦ 13 C i (‰), can be calculated ( Table 1). The advantage of expressing isotopic deviation as ␦ 13 C i (‰) is that all molecular targets are related back to the same defined international reference.
For S-(Ϫ)-nicotine, the range of values obtained for the two samples is Ϫ17.3 to Ϫ52.3‰ and for tropine the range is Ϫ9.1 to Ϫ52.3‰, with standard deviations acceptable for irm-13 C NMR spectrometry (ϳ1‰; range 0.7 to 2.7‰). These ␦ 13 C i (‰) values are within the typical range for natural products from plants using a C 3 metabolism (17). Their relative position-specific distributions are illustrated in Fig. 4, which shows the extent to which each ␦ 13 C i (‰) value differs from the ␦ 13 C g (‰) value.

Discussion
Following the successful development of the required protocols, it proved feasible to obtain position-specific 13 C/ 12 C ratios for all carbon atoms of the two target alkaloids. Thus, it is possible to exploit these to probe a number of aspects of the biosynthesis of these natural products.
To What Extent Are Isotope Ratios Determined by the Primary Precursor Molecules Giving Rise to the N-Methyl-⌬ 1 -pyrrolinium Salt?-In S-(Ϫ)-nicotine the carbon atoms that compose the pyrrolidine ring, C2Ј N , 4 C3Ј N , C4Ј N , and C5Ј N are derived from L-ornithine. This involves four known enzymatic steps ( Fig. 3) (5). Looking further back into the metabolic precursors, these carbons can be traced to citric acid, a component of the Krebs cycle, then to oxaloacetate and acetyl-CoA. There- fore, it can be deduced that the C2 OX and C3 OX of oxaloacetate become the C3Ј N and C2Ј N , respectively, in nicotine. Similarly, the C1 A and C2 A of acetate are incorporated via citrate into the C5Ј N and C4Ј N , respectively.
For tropine, by a similar argument, it can be deduced that the C2 OX and C3 OX of oxaloacetate become the C6 T and C5 T , respectively, whereas the C1 A and C2 A of acetate are incorporated into the C1 T and C7 T , respectively (Fig. 3). This ancestry will play a significant role in determining the position-specific 13 C/ 12 C ratios observed, as they will either essentially be retained, or will be modified due to KIEs in the enzymes involved in the pathway.
Because tropine is spectrally the simpler of the two target molecules, its isotope pattern will be analyzed first. As can be seen from Fig. 2A, the pairs of carbons showing the same degree of deshielding, C2 T ϩ C4 T , C1 T ϩ C5 T , and C6 T ϩ C7 T result in coincident peaks in the 13 C NMR spectrum. The impact of this on interpreting the 13 C/ 12 C ratios measured is, however, minimal, as the C1 T ϩ C5 T and C6 T ϩ C7 T pairs of positions are biosynthetically equivalent. Not only is N-methyl-⌬ 1 -pyrrolinium a symmetrical molecule, but it is derived from the symmetrical molecule, putrescine. As there is no evidence of metabolic tunneling between L-ornithine and N-methylputrescine (21), it can be presumed that methylation of putrescine can occur on either the 1-or 4-amino group with equal probability. Hence, the (C2 O ;C5 O ) and (C3 O ;C4 O ) positions in L-ornithine become indistinguishable in putrescine, hence in N-methylputrescine. Furthermore, tautomerism occurs around the quaternary nitrogen in N-methyl-⌬ 1 -pyrrolinium (22), so no distinc-  13

C distribution in natural tropine or natural S-(؊)-nicotine
a Mean of 10 spectral acquisitions obtained as two separate preparations from the same sample (5 spectra per preparation). b As these are independent samples, it is not appropriate to use a mean value.
Rather, so as to give an indicative value for the text discussion, an average of the two samples each of nicotine and of tropine is given.  tion is made as to at which carbon the substitution that leads to the elaborated molecules takes place. Considering the history of the four carbon atoms derived from N-methyl-⌬ 1 -pyrrolinium, the C6 T and C7 T have undergone no bond-forming reaction since the formation of citrate (Fig. 3). As isotopic fractionation is associated with bond forming and breaking, so the opposite is true: carbon atoms not undergoing reactions are likely not to be significantly fractionated. Hence, at these positions the chances for isotopic fractionation are minimal, and this is reflected in these positions having a relatively high ␦ 13 C i (‰) value: Ϫ11.9‰ (Fig. 4A, Table  1). The C(1 ϩ 5) T is, however, relatively depleted, having a ⌬␦ 13 C i (‰) 5 value 10.4‰ lower than the C(6 ϩ 7) T . This can be explained by two contributing factors. The C(1 ϩ 5) T is derived from the carboxyl group (C1 A ) of acetyl-CoA, which in eukaryotes is enriched relative to the methyl group in conditions in which commitment of acetyl-CoA to synthesis is significant (16,17,23). The degree of this varies. In lipids of eukaryotic origin (Saccharomyces cerevisiae) carbon positions derived from the C1 A are enriched relative to those from the C2 A by ϳ5‰ (23). Similarly in ethanol from C 3 plants, although here the difference is closer to 2‰ (24). In tropine, the impact of this is, however, compensated by the relative depletion at the C2 of citrate (C2 C ), and, based on estimated values for citrate where it is acting as a metabolic intermediate (16), a ⌬␦ 13 C i ϭ [C(1 ϩ 5) T -C(6 ϩ 7) T ] value of ϳ9‰ can be deduced. This is close to the observed differences in tropine.
An addition potential source of fractionation at the C(1 ϩ 5) T value (but not the C(6 ϩ 7) T value) is the two reactions involved in the final steps of the common pathway forming N-methyl-⌬ 1 -pyrrolinium, in both of which bond breaking/creation  Table 1 for absolute values. In these graphs the notation ⌬ g ␦ 13 C i (‰) indicates the difference at site i of ␦ 13 C i from the mean overall value of ␦ 13 C g (‰): i.e. ⌬ g ␦ 13 C i (‰) ϭ (␦ 13 C i -␦ 13 C g ).
occurs. The ring closure reaction (Fig. 3, step d), in which a C-N bond is formed, is considered spontaneous (13), the two compounds being in equilibrium. This step is unlikely to contribute a significant isotope fractionation as, even if there is an equilibrium isotope effect, the compound in either open or ring-closed structure is fully committed to the pathway. The previous step, in which a C-N bond is broken and a CϭO bond is formed is catalyzed by the enzyme N-methylputrescine oxidase (EC 1.4.3.6) (Fig. 3, step c). An intramolecular normal KIE associated with amine oxidase have been reported (25), so a fractionation during the deamination of N-methylputrescine would select in favor of isotopomers containing 12 C at the C1 T position, because it involves an sp 3 3 sp 2 conversion (26). This would lead to a 13 C depletion of the C-atom. However, as the commitment of N-methylputrescine to tropine biosynthesis is very high, this intermediate being a dedicated component of the pathway, isotope fractionation due to this enzyme is unlikely to be significant (see Ref. 17 for a detailed discussion of this phenomenon).
These arguments apply equally to the equivalent four carbon atoms of the pyrrolidine ring in S-(Ϫ)-nicotine biosynthesis (C2Ј N , C3Ј N , C4Ј N , and C5Ј N ). In this analyte, however, these four carbons are resolved in the 13 C NMR spectrum (Fig. 2B), which enables a more detailed interpretation (see below). Once again, the C3Ј N and C4Ј N , derived from citrate essentially without undergoing further reaction, are enriched relative to the C5Ј N and C2Ј N , which have undergone deaminative oxidation and C-N bond formation, respectively (Fig. 3). Furthermore, the relative depletion at the C(2Јϩ5Ј) N relative to the C(3Ј ϩ 4Ј) N of ϳ7‰ is in the same sense and similar size to this comparison in tropine (ϳ10‰). Hence, it can be concluded, (i) that the values obtained in the common compound N-methyl-⌬ 1pyrrolinium can be explained on the basis of the values in their distal precursors, and (ii) that this relationship is dictated primarily by the metabolic pathway, and is relatively independent of the plant species involved.
To What Extent Are Isotope Ratios Determined by the Primary Precursor Molecules Giving Rise to the Nicotinic Acidderived Moiety of S-(Ϫ)-Nicotine?-Nicotinic acid in plants is considered to be synthesized primarily from aspartate plus dihydroxyacetone phosphate/glyceraldehyde-3-phosphate (27). The incorporation of [3-14 C]aspartate into both the C2 N and C3 N of S-(Ϫ)-nicotine effectively demonstrated this origin for the pyridine ring (28). Considering that nicotinate is a predominant source of precursor for S-(Ϫ)-nicotine, which is a major sink for carbon, then some characteristics can potentially be traced back to the precursor pools. C2 N and C3 N are derived from the C1 ASP and C2 ASP , respectively, whereas the C4 N , C5 N , and C6 N are from dihydroxyacetone phosphate/glyceraldehyde-3-phosphate (Fig. 5). (Note that the interconversion of dihydroxyacetone phosphate and glyceraldehyde-3-phosphate in primary metabolism means that the C4 N and C6 N are of equivalent origin.) During the synthesis of nicotinic acid, all these positions are involved in reactions, although only the C3 N , C4 N , and C6 N positions are involved in bond formation. Again, therefore, it is these three positions that might be expected to show isotopic fractionation.
First, let us consider the three carbon positions showing the least relative depletion in the pyridine ring, C2 N , C4 N , and C5 N . The C5 N (␦ 13 C i ϭ Ϫ26.7‰) is from the keto group of dihydroxyacetone phosphate and undergoes an sp 2 3 transition in the formation of quinolinic acid. The C4 N is the most enriched position (␦ 13 C i ϭ Ϫ18.9‰) and, although this carbon is involved in the formation of a C-C bond with the C3 N in the formation of quinolinic acid and an sp 3 3 transition, its bonding thereafter is unchanged. In both of these positions, it is apparent that negligible fractionation has occurred, at least relative to the other three positions in the pyridine ring. The C2 N is involved in the decarboxylation of quinolinic acid and has a ␦ 13 C i ϳ15‰ and ϳ7‰ depleted relative to the C4 N and C5 N , respectively. Although decarboxylation can take place with negligible 13 C KIE on the ␣-carbon (29), it would appear that

. Schematic representation of the biosynthetic pathway to S-(؊)-nicotine and tropine indicating the origins of the carbon atoms.
Those carbon and nitrogen atoms involved in the formation of bonds and therefore the more likely to undergo isotope fractionation are indicated in italics. Carbon atoms on the final compounds are numbered according to the IUPAC nomenclature. For all other compounds, the numbering indicates the position in the analyte in which the given carbon will be found. Numbers in normal bold font refer to tropine; numbers in italic bold font refer to nicotine. AUGUST 5, 2016 • VOLUME 291 • NUMBER 32 fractionation is occurring here and that decarboxylation is probably a kinetically regulating step.

C Isotopic Links in Alkaloid Biosynthesis
The C3 N at ␦ 13 C i Ϸ Ϫ42.8‰ is ϳ8‰ depleted relative to the mean for the pyridine ring. This may be due to two additive effects. The first is the origin of this carbon: the C3 OX of oxaloacetate. This carbon is estimated to be markedly depleted (16) by ϳ11‰ relative to the other carbons of oxaloacetate (the immediate precursor of L-aspartate). Furthermore, C3 N is involved in two bond-forming and one bond-breaking reaction, each of which may lead to selection against 13 C in this position (see below), hence to relative depletion.
However, the most depleted position is the C6 N , which at ␦ 13 C i ϭ Ϫ52‰ is ϳ17‰ depleted relative to the mean for the pyridine ring. This position is involved in the formation of the CϭN bond created when it is joined to the imino group of iminoaspartate by quinolinate synthase (EC 2.5.1.72) (Fig. 5, boxed positions in the first row). Two mechanisms have been proposed for this enzyme, which differ essentially in the order in which the two bonds are formed: there is no experimental evidence to favor one over the other (30). However, in both cases the formation of the new bond involves the nucleophilic attack of either -NH 2 or ϭNH 2 ϩ on a ϾCϭO group. The degree of electrophilicity of the ϾCϭO is considered likely to have a major influence on the 13 C/ 12 C ratio at this atom (30). Based on our data, a strong normal KIE is indicated.
To What Extent Do the Reactions Involving the Conversion of N-Methyl-⌬ 1 -pyrrolinium to Tropinone Influence the Observed Isotope Ratios in Tropine?-The elaboration of N-methyl-⌬ 1pyrrolinium to tropinone involves the introduction of the three carbons C2 T , C3 T , and C4 T and the creation of two new C-C bonds at the C1 T and C5 T positions (Fig. 5). These three carbons are derived from acetate, via acetoacetyl-CoA and 4-(1methyl-2-pyrrolidinyl)-3-oxobutyrate (probably as the CoA thioester) (31). On the basis of the above discussion of the relative enrichments in acetyl-CoA, the C3 T , derived from the carboxyl in acetoacetyl-CoA, might be expected to be slightly enriched relative to the C(2 ϩ 4) T if no other factors come into play. In fact, the ⌬␦ 13 C i ϭ [(C3 T )-(C(2 ϩ 4) T )] of ϳ13‰ implies a significant KIE associated with the formation of this moiety. This greater relative enrichment is, however, consistent with the characteristic strong 13 C KIE associated with aldol reactions of acetyl-CoA in the biosynthesis of natural products when there is a high commitment of acetyl-CoA (16). In these cases, a large depletion in the carbons derived from the methylene group is observed. Similarly, in citrate extracted from fruit juice, in which there is again a high commitment of acetyl-CoA to this end product, the C2 CIT is relatively depleted, by ϳ11‰ (16). Hence, it is probable that these positions in acetoacetyl-CoA are more depleted relative to the carboxyl positions than in acetyl-CoA.
In addition, the condensation of N-methyl-⌬ 1 -pyrrolinium with acetoacetyl-CoA is likely to have normal 13 C KIEs for both the formation of the C1 T -C2 T bond in the synthesis of 4-(1methyl-2-pyrrolidinyl)-3-oxobutyrate, and for the formation of the C4 T -C5 T bond created during the synthesis of the 8-azabicyclo[3.2.1]octan-3-ol ring. This should lead to a further relative depletion in the C(2 ϩ 4) T position, compatible with the observed values. Unfortunately, due to the coincidence of the C2 T and C4 T in the 13 C NMR spectrum ( Fig. 2A), it is not possible to assess whether the first or second bond formation has a major influence on the 13 C/ 12 C ratios observed.

To What Extent Do the Reactions Involving the Conversion of N-Methyl-⌬ 1 -pyrrolinium to S-(Ϫ)-Nicotine Influence the
Observed Isotope Ratios?-The elaboration of N-methyl-⌬ 1pyrrolinium to S-(Ϫ)-nicotine involves the creation of a new C-C bond linking the C3 N and C2Ј N positions, thus joining the nicotinate-derived moiety to the N-methyl-⌬ 1 -pyrroliniumderived moiety (Fig. 5). Once again, the symmetry of the N-methyl-⌬ 1 -pyrrolinium salt means that in this precursor the C2Ј N ϭ C5Ј N and C3Ј N ϭ C4Ј N . However, the values observed at the C2Ј N and C5Ј N are not equivalent (Fig. 4B, Table 1): the C2Ј N , the carbon at which the C-C bond with the nicotinate moiety is made, is found to be ϳ4‰ enriched relative to the C5Ј N . The observed difference needs therefore to be accounted for by another cause. It is proposed that this is due to an inverse 13 C KIE associated with the mechanism of the nicotine synthase reaction condensing the N-methyl-⌬ 1 -pyrrolinium-and nicotinate-derived moieties (Fig. 6). Although the mechanism of this reaction is not known, some suggestions can be made on the basis of the isotope fractionation, hence the intrinsic KIE of the reaction. The C2Ј N undergoes a change of state from a semiaromatic nature with delocalized electrons to a tertiary aliphatic carbon (Fig. 6). Our data indicate that this reaction favors the 13 CϭN ϩ over the 12 CϭN ϩ , as known for sp 2 3 sp 3 conversions (26). Such a KIE will lead to relative enrichment in the position becoming the C2Ј N and an equivalent relative depletion in the C5Ј N must occur. This is clearly seen in Fig. 4B. Furthermore, the C3 N position is strongly depleted, which would indicate a selection against 13 C at this position, compatible with a mechanism in which the C3 N undergoes an sp 3 3 sp 2 conversion with kinetic limitation (Fig. 6). Further support for this proposal needs to be sought using specifically labeled substrates. FIGURE 6. Schematic representation of the proposed mechanism for nicotine synthase, the last step on the biosynthetic pathway to S-(؊)-nicotine, based on the observed site-specific 13 C/ 12 C isotopic fractionation. Partial enzymatic steps are given as: e1, the formation of the bond between C2Ј N and C3 N in which the C2Ј N undergoes an sp 2 3 sp 3 with kinetic limitation leading to selection for 13 C at the C3 N position; e2, the decarboxylation at the C3 N position, in which the C3 N undergoes an sp 3 3 sp 2 with selection for 12 C at the C3 N position. Carbon atoms on the final compounds are numbered according to the IUPAC nomenclature. Numbers in normal bold font refer to tropine; numbers in italic bold font refer to nicotine.

C Isotopic Links in Alkaloid Biosynthesis
What Can Be Deduced About the Introduction of the N-Methyl Group Based on the Observed Isotope Ratios?-In addition to having the ⌬ 1 -pyrrolinium moiety in common, the N-methyl group in both tropine and S-(Ϫ)-nicotine is derived from the common source, S-adenosylmethionine (AdoMet) and is introduced by the enzyme putrescine N-methyltransferase EC 2.1.1.53) (32). In both compounds, this carbon is found to be depleted relative to the other putrescine-derived positions, with which it should directly be compared (rather than with the whole molecule). In tropine, the extent of depletion is the major difference between the two samples: relative to the putrescine-derive part: ϳ38 and ϳ22‰ in TRI-1 and TRI-3, respectively. In nicotine, similarly, when the comparison is made only with the putrescine-derived atoms, then the relative depletion is ϳ12‰, rather less than in tropine but still well below the average of the other carbon atoms in this moiety.
This relative depletion in the 13 C/ 12 C isotope ratio of the N-methyl group appears to be characteristic for AdoMet-derived methylation in natural products, previously recognized for O-methyl groups (33,34). The position-specific analysis of 13 C/ 12 C isotope ratios of N-methyl group has only recently been investigated by using irm-13 C NMR and so far only for a very few alkaloids. Nonetheless, comparison can be made with the purine alkaloids, caffeine and theobromine (35), in which the N-methyl group is AdoMet-derived (36). In caffeine extracted from coffee the mean of the three N-methyl groups is on average ϳ36‰ depleted relative to the other carbon atoms, whereas in theobromine extracted from cocoa, the mean of the two N-methyl groups is on average ϳ23‰ below that of the other carbon atoms (35). These values suggest that relative depletion in the N-methyl group is a general phenomenon reflecting the biochemistry of methyl group insertion. The degree of depletion varies, but in all four alkaloids so far analyzed, including tropine and S-(Ϫ)-nicotine (Fig. 4), it is marked. This also parallels the relative depletion in a number of O-methyl groups, for example, ferulic acid (33). In tramadol, the N-Me 2 is also depleted relative to the part of the molecule considered to be derived from L-lysine by ϳ12‰ (20), which encouraged us to consider AdoMet as its probable origin. The O-methyl group of tramadol was also relatively depleted, although only by ϳ7‰. Overall, then, the depletion found in tropine and S-(Ϫ)-nicotine compares favorably with that in other N-and O-methyl groups, supporting the suggestion that this is a general phenomenon related to AdoMet metabolism (37).
The application of irm-13 C NMR to the study of compounds of known biosynthetic pathways has made possible the interpretation of the origin of the ␦ 13 C i (‰) values obtained for two alkaloids, S-(Ϫ)-nicotine and tropine. Generally, we can conclude that the majority of the carbon positions show an isotope fractionation that is dominated by the position-specific 13 C/ 12 C isotope ratios in the primary precursors that are exploited by the pathway. Thus, where moieties are incorporated that have other functions in the cell, for example, acetate via acetyl-CoA or acetoacetyl-CoA, then isotopic fractionation against the cellular pool can occur, as these precursors fulfill many other metabolic roles. The strength of this link will reflect the extent to which the primary precursors are committed to the synthesis of the specialized compounds, offering access to improved understanding of the primary/specialized metabolite interface in plant metabolism.
Once a major commitment of the substrate/product cascade to the final product is made, overall fractionation must be expected to have a more limited impact, as there is no branching taking place (17). Nevertheless, some intra-molecular fractionation can be observed when C-C bond formation/ breaking takes place. In contrast, isotopic fractionation is minimal for those carbon atoms that undergo no bond-forming reactions. These position-specific observations make possible deductions as to the putative reaction mechanism involved, as in the present case of the nicotine synthase condensation.
Isotope Ratio Monitoring by Mass Spectrometry-The global value for the whole molecule, ␦ 13 C g (‰), is the deviation of the carbon isotopic ratio R s relative to that of the international standard Vienna Pee Dee Belemnite, (V-PDB), R v-PDB . It is determined by isotope ratio monitoring by mass spectrometry and calculated from, where the value of R V-PDB is 0.0111802 (38).
Isotope Ratio Monitoring by 13 C NMR Acquisition Conditions-Prior to preparation of samples for irm-13 C NMR, purity was confirmed to be superior to 98% by recording a 1 H NMR spectrum. Tropine-free base from Fluka (TRI-3) was recrystallized from diethyl ether before use.
For the analysis of nicotine, 250 l of S-(Ϫ)-nicotine was homogenized in 500 l of acetonitrile-d 3 . To this was added 100 l of a solution of relaxation agent Cr(Acac) 3 (0.1 M) prepared by dissolving 10.5 mg of Cr(Acac) 3 in 300 l of acetonitrile-d 3 . Under these conditions, T 1 max ϭ 3.74 s. Spectral acquisition was with AQ ϭ 1.0 s for 152 scans, giving a signalto-noise ratio of ϳ700.
For the analysis of tropine, 150 mg of tropine was dissolved in 600 l of benzene-d 6 . As T 1 max ϭ 1.53 s is relatively short in these conditions, relaxation agent was not required. Spectra acquisition was with AQ ϭ 0.95 s for 400 scans, giving a signalto-noise ratio ϳ550.
Quantitative 13 C NMR spectra were recorded at 100.6 MHz using a Bruker 400 Avance I NMR spectrometer fitted with a 5-mm 1 H/ 13 C dual ϩ probe. For S-(Ϫ)-nicotine, spectra were also measured on a Bruker 400 Avance III NMR spectrometer fitted with a 5-mm BBFO probe. The temperature of the probe was set at 303 K for nicotine and 313.2 K for tropine (to decrease the viscosity of the sample). The offsets for both 13 C and 1 H were set at the middle of the frequency range for each analyte. Inverse-gated decoupling was applied and the repetition delay between each 90°pulse was set at 10 ϫ T 1 max of the analyte to avoid the nuclear Overhauser effect and achieve full relaxation of the magnetization. The decoupling sequence used adiabatic full-passage pulses with cosine square amplitude modulation ( 2 max ϭ 17.6 kHz) and offset independent adiabaticity with optimized frequency sweep (39). Each measurement consisted of the average of five independently recorded NMR spectra.
Spectral Data Processing-To obtain S i , the area under the 13 C-signal in the irm-13 C NMR spectrum for C-atom in position i, curve fitting based on a total line shape analyses (deconvolution) was carried out with a Lorentzian mathematical model using Perch Software (Perch TM NMR Software). In this procedure, line shape parameters are optimized in terms of intensities, frequencies, line width, and line shape (Gaussian/ Lorentzian, phase, asymmetry) by iterative fitted to a minimal residue. For each target compound, spectral phasing and line fitting was performed using the same parameters. The positional isotopic distribution, f i /F i , was then calculated from the values of S i , hence the ␦ 13 C i (‰) value, as described previously (19,40,41). For S-(Ϫ)-nicotine, due to the large range of its spectral shifts, a correction was made to data obtained on the Avance I NMR spectrometer using the data from the Avance III NMR spectrometer (see Ref. 42 for a detailed explanation). It should be noted that, because the coupling constants J( 13 C-15 N) are of the order of 10 Hz, no correction for these is required as they are included within the curve fitting applied to the peak.
Author Contributions-R. J. R. and G. S. R. designed the study. K. M. R. and V. S. conducted most of the experiments. R. J. R., G. S. R., and P. P. analyzed and interpreted the data. R. J. R. and K. M. R. wrote the paper.