The structure of tagetitoxin

Abstract

Based on detailed analysis of newly acquired NMR data, we show that the previously revised structure of tagetitoxin is incorrect. A new structure of tagetitoxin is proposed which is consistent with the NMR and MS data.

---

Introduction

Tagetitoxin is a toxin isolated from the plant pathogenic bacterium Pseudomonas syringae pv. tagetis.¹ It is known to cause chlorosis in young plant leaves, which has been attributed to inhibition of RNA polymerase in chloroplasts.² Tagetitoxin also inhibits bacterial RNA polymerase,² and is the only natural product known to inhibit eukaryotic RNA polymerase III in a specific manner.³ Recently, Yuzenkova et al. have shown that tagetitoxin neither affects the chemistry of RNA synthesis nor competes with the nucleoside triphosphate in the active centre.⁴ Instead, tagetitoxin increases the stability of the pre-translocated state of the elongation complex, thus slowing down addition of the following nucleotide.⁴

The first published structure of tagetitoxin by Mitchell and Hart in 1983 consisted of an eight-membered heterocycle with a sulfur atom (structure 1 in Fig. 1, molecular weight 435, C₁₁H₁₈NO₁₃PS).⁵

Fig. 1 Previously proposed structures of tagetitoxin.

It was found that heteroatomic components comprised of oxygen, nitrogen in an amine, phosphorus in a phosphate ester and sulfur. Investigations of the structure of tagetitoxin continued based on new MS and NMR data, after attempts to obtain crystals for X-ray analysis failed.⁵ In 1989, a revised bicyclic structure of tagetitoxin based on the 9-oxa-3-thiabicyclo[3.3.1]nonane ring system was proposed by Mitchell et al. (Fig. 1).⁶ FAB mass spectrometry showed (M + H)⁺ = 417.0361 (C₁₁H₁₈N₂O₁₁PS requires 417.0369) indicating that tagetitoxin has a molecular formula C₁₁H₁₇N₂O₁₁PS. Structure 2 was favoured, although the spectroscopic data did not rule out the closely related structure 3 (Fig. 1).⁶

In 2005, a crystal structure of the RNA polymerase from Thermus thermophilus with tagetitoxin bound to the active site was published by Vassylyev et al.⁷ Although the difference electron density map revealed an electron density attributed to tagetitoxin in this crystal structure, the structure of tagetitoxin was not investigated and a stereoisomer of structure 2 was used by Vassylyev et al. without further verification.⁷ In the same year, Gronwald et al. published their purification protocol and partial characterization of tagetitoxin.⁸ According to their analysis, the revised structure of Mitchell et al.⁶ is incorrect. Based on electrospray ionization mass spectrometry in 50% methanol : H₂O, Gronwald et al. reported that the molecular weight of tagetitoxin is 678, although the NMR spectrum of tagetitoxin published by them indicated that tagetitoxin contained additional peaks at 2.53 ppm and 1.75 ppm not observed previously by Mitchell et al.^5,6 Despite the ambiguity of the structure of tagetitoxin, several reports have been published to date, detailing synthetic approaches to tagetitoxin and its analogues with the basic bicyclic ring structure 2,⁹ though none of these has successfully delivered the full structure 2.

Here, we report the results of our analysis of NMR and MS data for tagetitoxin and show that neither of the published structures of tagetitoxin is correct. A new structure of tagetitoxin is reported which is in agreement with NMR and MS data.

Results and discussion

The sample studied was that originally isolated and purified by Mitchell.^1,5 In order to illustrate the purity of the compound studied, the proton NMR spectrum of tagetitoxin in D₂O is shown in Fig. 2. Note that additional peaks of smaller intensity appeared in ¹H NMR spectrum of tagetitoxin kept in D₂O solution over 4–6 weeks (see Fig. S2 in ESI†), suggesting that tagetitoxin gradually decomposes in aqueous solutions. Although most of the spectral features in Fig. 1 resemble those observed in the ¹H NMR spectrum of tagetitoxin by Gronwald et al.,⁸ no peaks are observed at 1.75 and 2.53 ppm. Similarly, no ¹³C peak was observed at 181.45 ppm. These observations suggest the material studied by Gronwald et al.⁸ was less pure compared to that extracted by Mitchell et al.^5,6 From the analysis of the MS data obtained in this work (see ESI† for full details), no species with the molecular weight of 678 were found, which is reported by Gronwald et al.⁸ as a revised molecular weight of tagetitoxin.

Fig. 2 ¹H NMR spectrum of tagetitoxin in D₂O at 293 K (600 MHz). Integral intensities are shown between the chemical shift axis and the spectrum. The most significant impurity peaks are observed as doublets at 3.21 ppm with 4.0 Hz splitting (integral intensity 0.15) and at 2.09 ppm with 5.0 Hz splitting (integral intensity 0.18).

In 1989, the revised structure 2 (Fig. 1) was deduced based on the analysis of ¹H and ¹³C NMR spectra, ¹H NOEs and the COLOC spectrum for ¹H–¹³C long-range correlations.⁶ The latter is expected to provide information similar to that from the HMBC spectrum, although it is significantly less sensitive than HMBC, which is usually used for identification of 2 or 3 bond correlations between ¹H and ¹³C nuclei. The HMBC spectrum shown in Fig. 3, as well as the values of long-range ⁿJ_CH couplings (Table 1), revealed several correlations which allowed us to rule out structures 1–3 shown in Fig. 1. In particular, some of the disagreements are as follows:

(1) A cross-peak is observed for the C11–H8 pair which is in disagreement with structure 2 with six bonds between C11 and H8.

(2) A strong cross-peak C10–H2′ is in disagreement with structure 3 with four bonds between C10 and H2′. In principle, ⁴J_CH correlations can be observed in HMBC spectra, however, the value of J_C10H2′ coupling derived from the HMBC-JC spectrum is 5.0 Hz, which cannot be attributed to a ⁴J_CH coupling.

(3) Cross-peaks are observed for C7–H2 (J_CH = 5 Hz) and C7–H2′ (J_CH = 3 Hz), which are in disagreement with all three structures shown in Fig. 1, with four bond separation between C7 and H2.

(4) Dihedral angle between C4 and H6 is ∼180° in structure 2, while only a weak HMBC cross-peak is observed in the HMBC spectrum. From HMBC-JC, the value of J_C4H6 is 1.4 Hz.

Fig. 3 ¹H–¹³C HMBC NMR spectrum of tagetitoxin in D₂O at 293 K (600 MHz).

Table 1 Experimental values of long-range ¹H–¹³C coupling constants (ⁿJ_CH, Hz) of tagetitoxin in D₂O. The signs of coupling constants measured using HSQC-HECADE are included in brackets. The calculated values [at the DFT B3LYP/6-311+G(2d,p) IEFPCM(H₂O) level of theory] for individual conformers, as well as averaged values, 〈ⁿJ_CH〉, over two conformers, 4-chair (75%) and 4-twisted-chair (25%), are shown

	Exper. ^nJCH/Hz	4-Chair Calc. ^nJCH/Hz	4-Tw.-chair Calc. ^nJCH/Hz	4 Calc. 〈^nJCH〉/Hz
C1–H7	1.3	1.20	2.05	1.41
C1–H8	2.4	−2.07	−2.46	−2.17
C1–H5	∼0	−0.62	−0.88	−0.69
C1–H6	∼0	0.16	−0.01	0.12
C2–H7	1.5	2.82	0.02	2.10
C4–H2	2.3	−2.32	−1.63	−2.15
C4–H2′	4.1	−4.02	−2.20	−3.56
C4–H13	0.8	0.53	0.48	0.51
C4–H6	1.4	0.79	−0.10	0.56
C5–H7	(+)1.1	1.28	1.02	1.22
C5–H8	(+)0.9	0.72	0.41	0.64
C5–H6	(+)0.2	0.17	0.95	0.37
C6–H5	(−)0.7	−0.14	−0.01	−0.11
C6–H7	(−)5.6	−5.44	−5.65	−5.49
C6–H8	(+)8.0	7.26	6.32	7.02
C7–H2	5.0	6.98	−0.10	5.17
C7–H2′	3.0	2.44	5.99	3.35
C7–H5	(+)6.1	5.53	5.31	5.47
C7–H6	(−)2.7	−2.36	−1.60	−2.17
C7–H8	(−)1.1	−2.12	−1.96	−2.08
C8–H2	1.4	1.39	−0.17	0.99
C8–H5	(+)6.2	6.35	6.40	6.36
C8–H6	(+)0.3	0.50	0.22	0.43
C8–H7	(−)0.4	−1.12	−0.44	−0.95
C10–H2	1.2	1.35	1.64	1.42
C10–H2′	5.0	7.03	1.54	5.63
C11–H8	1.5	1.86	2.89	2.12
C11–H5	2.7	2.12	1.95	2.07
C12–H13	6.0	−5.51	−5.52	−5.51
C12–NH	3.7	4.57	5.14	4.72
rmsJ/Hz	—	0.72	1.50	0.52

---

Furthermore, the 1D NOESY spectrum with selective excitation of methyl protons at 2.01 ppm showed a negative exchange enhancement at 2.16 ppm with the integral intensity ratio 68 : 1 for singlets at 2.01 and 2.16 ppm in the ¹H NMR spectrum (Fig. S3 in ESI†). Such a slow exchange at room temperature between two sites with unequal populations is characteristic for an amide group NHCOMe, but not for OCOMe shown in structures 1–3 (Fig. 1). ¹H NMR spectrum recorded in H₂O : D₂O (9 : 1) showed a singlet at 8.47 ppm (Fig. S13 in ESI†), which is in agreement with the presence of the NHCOMe group. In addition, the ¹H–¹⁵N HMBC spectrum in D₂O (Fig. S20 in ESI†) showed a correlation for the methyl protons with the ¹⁵N signal at 140.5 ppm, in agreement with the expected ¹⁵N chemical shift for a secondary amide in the range 110–160 ppm (relative to liquid NH₃).

Based on mainly ¹H–¹³C HMBC correlations and the values of long-range J_CH couplings a new structure was derived shown in Fig. 4. The above noticed disagreements (1)–(4) for structures 1–3 were verified for structure 4:

(1) The cross-peak observed for C11–H8 pair is due to ³J_CH coupling in 4;

(2) The cross-peak C10–H2′ is due to ³J_CH coupling in 4.

(3) Cross-peaks observed for C7–H2 (J_CH = 5 Hz) and C7–H2′ (J_CH = 3 Hz) are due to ³J_CH coupling in 4.

(4) Assuming a chair conformation of the six membered ring, the dihedral angle between C4 and H6 is ∼60° in structure 4, in agreement with the measured value of ³J_C4H6 = 1.4 Hz based on the Karplus-type relationship for ³J_CH couplings.

Fig. 4 The proposed revised structure of tagetitoxin based mainly on the analysis of ¹H–¹³C HMBC correlations and the values of long-range J_CH couplings. The atom numbering used corresponds to that in 2 (Fig. 1).

In a similar fashion, we have analysed the measured values of all the vicinal ³J_CH couplings, which show good agreement with structure 4. There are relatively few J_HH couplings in tagetitoxin. Nevertheless, the large value of the ³J_H6H7 = 12.2 Hz is in favour of the trans fusion of two cycles with both protons occupying axial orientations. Furthermore, from the measured signal enhancements in 1D NOESY spectra (Table 2), the NOE is relatively small for the H6–H7 pair (0.5%) compared to, for example, H5–H6 (1.4–1.5%) or H7–H8 (1.2–1.3%). Combined with the values of vicinal couplings (³J_H5H6 = 4.1 Hz and ³J_H7H8 = 7.8 Hz), these NOEs are in favour of the trans configuration of protons H6 and H7, the cis configuration of protons H5 and H6 and the cis configuration of protons H7 and H8. A very small enhancement (0.1%) observed for the H6–H8 pair is in agreement with their trans configuration in the five-membered ring.

Table 2 NOE Enhancements (in %) from 1D NOESY experiments. No NOEs were observed on selective excitation of methyl protons. The measured distances (in Å) in DFT-optimised geometries of 4-chair/4-twisted-chair conformations are shown in brackets

	“Touched” protons
	H2′	H2	H7	H5	H8	H6
H2′	—	5.4 (1.77/1.77)	0.8 (2.37/3.92)			0.1 (3.95/3.73)
H2	5.6 (1.77/1.77)	—				0.6 (4.12/2.30)
H7	1.1 (2.37/3.92)		—	0.1(3.72/3.74)	1.2 (2.26/2.21)	0.5 (3.04/3.03)
H5			0.1 (3.72/3.74)	—		1.5 (2.44/2.45)
H8			1.3 (2.26/2.21)		—	0.1 (3.76/3.80)
H6		0.7 (4.12/2.30)	0.5 (3.04/3.03)	1.4 (2.44/2.45)	0.1 (3.76/3.80)	—

---

Protons of the methylene group in tagetitoxin are labelled as 2 and 2′ (Fig. 4). The methylene proton with the cis configuration relative to proton H6 is denoted as H2 (the high-frequency CH₂ signal in the ¹H NMR spectrum), while the other methylene proton with the trans configuration relative to H6 is denoted as H2′ (the low-frequency CH₂ signal in the ¹H NMR spectrum). Thus, in a chair conformation of the six-membered ring with the axial orientation of H6, such a definition of H2 and H2′ corresponds to the equatorial orientation of H2 and the axial orientation of H2′. Relatively strong NOE (0.6%) was observed for proton pair H2–H6 (Table 2), which led to a consideration of a twisted chair conformation for the six-membered ring. Both chair and twisted-chair conformations were included into our computational analysis and the final lowest energy conformations derived from DFT M06-2X/def2-TZVP geometry optimisations are shown in Fig. 5. The free energy of the twisted-chair conformation relative to that of the chair conformation is +1.22 kcal mol⁻¹. On the assumption of a two-site fast exchange (in the NMR timescale) between chair and twisted-chair conformations, the predicted populations by DFT M06-2X/def2-TZVP calculations are 89% and 11% for chair and twisted-chair conformations, respectively. A more reliable estimate of the conformational populations was achieved using experimental values of 30 long-range J_CH couplings and predicted values of corresponding coupling constants in chair and twisted-chair conformations at the DFT B3LYP/6-311+G(2d,p) level of theory (Table 1; regarding the performance of B3LYP calculations for predictions of J couplings, see ref. 10). The populations of conformers derived from this analysis were 75% and 25% for chair and twisted-chair conformations, respectively.

Fig. 5 Geometries of 4-chair and 4-twisted-chair conformations derived from DFT M06-2X/def2-TZVP calculations. One of the carboxylic protons of structure 4 (Fig. 4) is delocalized between COO⁻ and OPO₃H⁻ groups in both conformations.

No HMBC correlations were observed for C1–H5 and C1–H6 pairs separated by two and three bonds, respectively, in structure 4. DFT calculations confirmed that the expected values of the corresponding ^2,3J_CH couplings are indeed small, e.g., −0.62 Hz and 0.16 Hz in the 4-chair conformation shown in Fig. 5 (−0.88 Hz and −0.01 Hz in the 4-twisted-chair conformation).

In order to determine orientations of substituents in position 4, we have used weak NOEs observed for the amide NH proton with H7 in H₂O + D₂O (9 : 1) solution, as well as the fact that the NOE for the NH–H2′ pair is significantly stronger than that for the NH–H2 pair (∼4.4 times based on the volume integration of the corresponding cross-peaks; the volume integration ratio for the cross-peaks of the amide proton NH with H7, H2, H2′, Me was 1.0 : 2.3 : 10.1 : 9.2; Fig. S13 in ESI†). Furthermore, the cis orientation of the NHCOMe group relative to proton H7 was confirmed by the analysis of vicinal J_CH couplings of the adjacent carboxylic carbon based on the Karplus-type relationship: ³J_C10H2 = 1.2 Hz and ³J_C10H2′ = 5.0 Hz. These agree well with the DFT predicted values of 1.4 and 5.6 Hz on the assumption of the equilibrium between 4-chair (75%) and 4-twisted-chair (25%) conformations (Fig. 5 and Table 1).

The determination of orientations of substituents in position 1 required consideration of both alternatives in DFT calculations and the analysis of ³J_CH couplings of the carboxylic carbon C11 with protons H5 and H8. Structures of 5-chair and 5-twisted-chair, in which the orientations of N⁺H₃ and COOH are interchanged at C1 compared to 4, are shown in Fig. S1 (ESI†). The free energy of the 5-twisted-chair conformation relative to that of the 5-chair conformation is +0.81 kcal mol⁻¹. On the assumption of a two-site exchange between chair and twisted-chair forms, the predicted populations by DFT M06-2X/def2-TZVP calculations are 81% and 19% for 5-chair and 5-twisted-chair conformations, respectively. From the analysis of experimental values of 30 long-range J_CH couplings and predicted values of corresponding coupling constants in 5-chair and 5-twisted-chair conformations at the DFT B3LYP/6-311+G(2d,p) level of theory (Table S3†), the populations of conformers were 78% and 22% for 5-chair and 5-twisted-chair conformations, respectively. However, the rms deviation between experimental and calculated couplings is 1.24 Hz for the conformational equilibrium 5-chair/5-twisted-chair, compared to 0.52 Hz for the conformational equilibrium 4-chair/4-twisted-chair. The predicted values for ³J_C11H8 and ³J_C11H5 were 3.7 and 0.3 Hz for the 5-chair/5-twisted-chair equilibrium, which are in disagreement with the experimental values of 1.5 and 2.7 Hz. In the case of the 4-chair/4-twisted-chair equilibrium, the predicted values for ³J_C11H8 and ³J_C11H5 were 2.1 and 2.1 Hz. Thus, the cis configuration of the phosphate and carboxylic groups at C8 and C1, respectively, can be deduced based on the analysis of experimental and calculated J_CH couplings.

In addition to the measured values, we have also determined the sign of some of the J_CH couplings (Table 1). It is well known that ¹³C–¹H couplings over one and three bonds are positive, while those over two or four bonds (²J_CH or ⁴J_CH) are either positive or negative. Thus, if we know that the sign of ⁿJ_CH is negative, then the number of bonds between C and H cannot be three and using the absolute value of the coupling constant we could deduce whether it corresponds to ²J_CH or ⁴J_CH. We have used the HSQC-HECADE spectrum for sign determinations.¹¹ Note that the sign of only some of the ⁿJ_CH couplings are available from this spectrum (e.g., ⁿJ_CH correlations of quaternary carbons are not detectable).¹¹ Nevertheless, all the measured negative values (for spin pairs C6–H5, C6–H7, C7–H6 and C8–H7) can be attributed to geminal ²J_CH couplings, while ³J_CH couplings have a positive sign (for spin pairs C5–H7, C5–H8, C6–H8, C7–H5, C8–H5 and C8–H6). Thus, these results additionally support the sequence in which the corresponding C and H atoms are arranged. The signs of these couplings predicted by the DFT calculations were in agreement with the HSQC-HECADE measurements (Table 1).

The EASY-ROESY method was also used, which is known to provide accurate integration of cross-peaks for quantitative estimates.¹² We have analysed the observed rotational Overhauser effects (ROEs) using a simplified version of the growth rates method in order to estimate internuclear ¹H–¹H distances.¹³ The satisfactory performance of the simplified growth rates method has been demonstrated previously for cyclic organic compounds.¹⁴ The standard deviations for distance measurements were typically 10% of the corresponding mean values.¹⁴ Volume integrals of ROE cross-peaks for 6 proton pairs were measured for tagetitoxin (Table 3 and Fig. S14 in ESI†). Using r = 1.77 Å as the reference value for the geminal H2–H2′ pair, internuclear distances for other proton pairs were calculated using the r⁻⁶ dependence of ROEs.¹³ In Table 3, we compare experimental values with those from interatomic distances (r_i^calc, Å) derived from M062X/def2-TZVP-optimised geometries. The individual chair and twisted-chair conformers showed the rms deviations (rms_d, Å) of 0.62 Å and 0.71 Å, respectively, for five pairs of protons (excluding the reference geminal pair). Significantly improved agreement is observed with rms_d = 0.11 Å on considering a two-site fast exchange between 4-chair (75%) and 4-twisted-chair (25%), with the populations determined from the analysis of ⁿJ_CH couplings above.

Table 3 Internuclear distances (in Å) in 4 obtained from NMR ROE measurements in D₂O and DFT M06-2X/def2-TZVP calculations in H₂O with the IEFPCM solvation model. The rms deviations (rms_d, in Å) from the experimental NMR values are shown

Proton pair	NMR 〈r^exp〉^a (Å)	4-Chair r^calc (Å)	4-Twisted-chair r^calc (Å)	4 〈r^calc〉^b (Å)
a Uncertainties in experimental values were estimated using volume integrations of cross-peaks above and below the diagonal. b In calculations of averaged values of 〈r^calc〉, the calculated ROEs were weighted using populations of conformers 4-chair (75%) and 4-twisted-chair (25%).
2–2′	1.77	1.77	1.77	1.77
7–8	2.21 ± 0.02	2.26	2.21	2.24
5–6	2.31 ± 0.01	2.44	2.45	2.44
6–7	2.85 ± 0.02	3.04	3.03	3.03
2–6	2.76 ± 0.03	4.12	2.30	2.85
2′–7	2.43 ± 0.01	2.37	3.92	2.47
rmsd	—	0.62	0.71	0.11

---

The populations of chair (75%) and twisted chair (25%) conformers obtained from the analysis of ³J_CH couplings also agree well with the combined analysis of experimental and calculated ¹H and ¹³C chemical shifts. The methodology used here has been verified previously for cyclic organic compounds with known structures.^14,15 In particular, optimised geometries of 4-chair and 4-twisted-chair were used in GIAO B3LYP/6-311+G(2d,p) chemical shielding calculations. The conformationally averaged values of the isotropic shieldings 〈σ^calc(i)〉 were calculated. The averaged values of the isotropic shieldings were then converted into conformationally averaged values of chemical shifts, 〈δ^calc(i)〉, using both the slope and the intercept of the σ^calcvs. δ^exp plot, as described previously.¹⁴ From the results obtained (Table 4), the rms_δ values for 4 were 0.08 ppm (¹H) and 2.2 ppm (¹³C). For comparison, the rms_δ values for a closely related structure 5 (78% chair and 22% twisted chair) were 0.21 ppm (¹H) and 2.8 ppm (¹³C), showing high sensitivity of both ¹H and ¹³C chemical shifts to the change in the orientation of substituents. Overall, the relatively small values of rms deviations for chemical shifts (rms_δ¹H 0.08 ppm and ¹³C 2.2 ppm), together with the ROE analysis of interproton distances (rms_d 0.11 Å), further support the validity of structure 4 for tagetitoxin. The NMR-derived structure 4 of tagetitoxin was also consistent with the accurate mass measurements and gas-phase fragmentation patterns, full details of which are included in ESI.†

Table 4 Experimental and calculated ¹H and ¹³C chemical shifts in 4 (δ, ppm). Optimised geometries from M062X/def2-TZVP IEFPCM(H₂O) were used in GIAO B3LYP/6-311+G(2d,p) IEFPCM(H₂O) chemical shift calculations. The rms deviation (rms_δ, in ppm) and the largest residual deviation (Δ^max, in ppm)between the calculated and experimental values are shown

Proton	δ ^expH (ppm)	〈δ^calcH〉^a (ppm)	Carbon	δ ^expC (ppm)	〈δ^calcC〉^a (ppm)
a Calculated chemical shifts [δ^calc(i) = (σ^calc(i) − b)/a] were determined using the slope [a(^1H) = −1.14 and a(^13C) = −0.98] and the intercept [b(^1H) = 32.31 ppm and b(^13C) = 177.66 ppm] derived from the least squares fittings [σ^calc(i) = aδ^exp(i) + b].
2	3.25	3.23	1	71.2	73.1
2′	2.98	2.87	2	33.1	36.2
5	4.46	4.39	4	85.6	82.8
6	5.13	5.15	5	72.9	72.8
7	3.48	3.65	6	79.8	80.4
8	4.73	4.72	7	43.2	44.8
13	2.01	2.02	8	77.0	78.4
			10	174.4	174.4
			11	171.2	170.1
			12	173.8	174.3
			13	22.9	17.7
rmsδ	—	0.08	rmsδ	—	2.2
Δ ^max	—	0.17	Δ ^max	—	−5.2

---

Experimental

NMR spectroscopy

Purified tagetitoxin was received in non-crystallised solid form from Robin Mitchell.^1,5,61H and ¹³C NMR spectra were recorded on a Bruker Avance III 600 MHz NMR spectrometer equipped with a 5 mm cryoprobe (¹H 600.13 MHz and ¹³C 150.90 MHz). These spectra showed no change from the data of Mitchell et al.⁶¹⁵N and ³¹P NMR spectra were recorded on a Bruker Avance III 400 MHz NMR spectrometer equipped with a 5 mm ¹H–¹³C–¹⁵N–³¹P probe (¹H 400.13 MHz, ¹⁵N 40.55 MHz and ³¹P 161.98 MHz). Data acquisition and processing were performed using standard TopSpin software (versions 2.1 and 3.2). ¹H and ¹³C chemical shifts were calibrated indirectly, using dioxane shifts in D₂O (¹H 3.75 ppm, ¹³C 67.19 ppm). ¹⁵N and ³¹P NMR chemical shifts were calibrated using ¹⁵N₂-urea dissolved in DMSO-d₆ (77.6 ppm relative to liquid NH₃) and 85% H₃PO₄ (0 ppm). Unless otherwise specified, NMR measurements were carried out at 293 K. Temperature calibration was carried out using a sample of 99.8% deuterated MeOD in a 5 mm NMR tube. In addition to standard 1D and 2D spectra, additional techniques were employed for measuring long-range J_CH couplings, including HMBC-JC¹⁶ and HSQC-HECADE.¹¹

One- and two-dimensional NOE measurements were undertaken for establishing spatial proximities of protons.¹³ Standard pulse sequences and those with the elimination of strong interference caused by zero-quantum coherence were employed.¹⁷ 2D EASY-ROESY spectra were also acquired. The main advantage of this experiment is that artifacts due to J-couplings are minimised. It has also been shown to yield reliable intramolecular distances without a sample-specific setup.¹²

Calculations

Initial structures for quantum-mechanical calculations were built and optimized using PCMODEL (version 8.5).¹⁸ The MMX force field was used for energy evaluations.^18,19 Relaxed grid search (RGS) analysis¹⁸ was carried out for each conformer considered using PCMODEL. RGS is a systematic method, which involves creation of a large number of starting configurations and mapping out the shape of the potential energy surface. In this method the rotatable bonds of interest are first identified. The calculation starts by evaluating the energy when all the rotatable bonds are set to 180°. The bonds are then rotated sequentially and all the structures are minimized and sorted based on their total energy, with any duplicate configurations removed. Since the total number of energy evaluations can be very large (usually several hundreds or thousands depending on the number of rotatable bonds), the energies of conformers were calculated using molecular mechanics method and the MMX force field.

In some cases, the RGS derived structures were further optimized via semi-empirical PM6²⁰ calculations using Gaussian 09. The reaction field method IEFPCM²¹ was used to account for water solvent effects in PM6 calculations.

All quantum mechanical calculations were carried out using Gaussian 09.²² For geometry optimizations using density functional theory (DFT), the M06-2X²³ functional with def2-TZVP basis set was used.²⁴ The performance of M06-2X functional has been compared extensively to other DFT methods and MP2.^23,25 Its superior performance has been illustrated in a comprehensive review article by Zhao and Truhlar,²⁵ in which they have included comparisons of M06-2X with SCS-MP2 and B2PLYP-D. The choice of the def2-TZVP basis set is dictated primarily by the presence of sulfur and phosphorus atoms in tagetitoxin. At the DFT level the def2-TZVP basis set has been shown to produce results that are not too far from the DFT basis set limit.²⁴ For optimization of structure parameters, the def2-TZVP errors in bond lengths are typically smaller than 1 pm and that in bond angles are smaller than 1°.²⁴ The ultrafine numerical integration grid (with 99 radial shells and 590 angular points per shell) was used in our M06-2X/def2-TZVP geometry optimisations, combined with the “verytight” convergence condition (requesting the root-mean-square forces to be smaller than 1 × 10⁻⁶ Hartree Bohr⁻¹). Additional frequency calculations were also undertaken in order to verify that the optimized geometries correspond to true minima. The reaction field method IEFPCM²¹ was used to account for water solvent effects. NMR chemical shieldings and J couplings were computed at the B3LYP/6-311+G(2d,p) level using the GIAO method.²⁶ Water solvent effects were used in all the quantum mechanical calculations using the reaction field method IEFPCM.²¹

Conformationally averaged interatomic distances from the QM calculations were determined in a way similar to that used in NMR measurements: (i) internuclear distances (r_i) for pairs of hydrogen atoms were measured in each conformer i; (ii) a quantity equal to r_i⁻⁶ was calculated as a measure of the expected NOE in each conformer, η_i; (iii) the sum of p_ir_i⁻⁶ was calculated, where values of populations p_i were derived from the analysis of experimental long range J_CH couplings using their QM-predicted boundary values in each conformer i; (iv) using r = 1.77 Å as the reference H2–H2′ distance for geminal protons, internuclear distances for other proton pairs were calculated using the η ∼ r⁻⁶ relationship.¹³

Acknowledgements

We thank University College London (UCL) for the provision of computational facilities. The authors acknowledge the use of the UCL Legion High Performance Computing Facility (Legion@UCL), and associated support services, in the completion of this work.

References

1. R. E. Mitchell and R. D. Durbin, Physiol. Plant Pathol., 1981, 18, 157–168.
2. D. E. Mathews and R. D. Durbin, J. Biol. Chem., 1990, 265, 493–498.
3. (a) T. H. Steinberg, D. E. Mathews, R. D. Durbin and R. R. Burgess, J. Biol. Chem., 1990, 265, 499–505; (b) L. P. Wu, J. Pan, V. Thoroddsen, D. R. Wysong, R. K. Blackman, C. E. Bulawa, A. E. Gould, T. D. Ocain, L. R. Dick, P. Errada, P. K. Dorr, T. Parkinson, T. Wood, D. Kornitzer, Z. Weissman, I. M. Willis and K. McGovern, Eukaryotic Cell, 2003, 2, 256–264.
4. Y. Yuzenkova, M. Roghanian, A. Bochkareva and N. Zenkin, Nucleic Acids Res., 2013, 41, 9257–9265.
5. R. E. Mitchell and P. A. Hart, Phytochemistry, 1983, 22, 1425–1428.
6. R. E. Mitchell, J. M. Coddington and H. Young, Tetrahedron Lett., 1989, 30, 501–504.
7. D. G. Vassylyev, V. Svetlov, M. N. Vassylyeva, A. Perederina, N. Igarashi, N. Matsugaki, S. Wakatsuki and I. Artsimovitch, Nat. Struct. Mol. Biol., 2005, 12, 1086–1093.
8. J. W. Gronwald, K. L. Plaisance, S. Marimanikkuppam and B. G. Ostrowski, Physiol. Mol. Plant Pathol., 2005, 67, 23–32.
9. (a) T. Sammakia, T. B. Hurley, D. M. Sammond, R. S. Smith, S. B. Sobolov and T. R. Oeschger, Tetrahedron Lett., 1996, 37, 4427–4430; (b) B. R. Dent, R. H. Furneaux, G. J. Gainsford and G. P. Lynch, Tetrahedron, 1999, 55, 6977–6996; (c) M. Ioannou, M. J. Porter and F. Saez, Chem. Commun., 2002, 346–347; (d) M. Ioannou, M. J. Porter and F. Saez, Tetrahedron, 2005, 61, 43–50; (e) J. R. H. Plet and M. J. Porter, Chem. Commun., 2006, 1197–1199; (f) A. J. P. Mortimer, A. E. Aliev, D. A. Tocher and M. J. Porter, Org. Lett., 2008, 10, 5477–5480; (g) J. R. H. Plet, A. K. Sandhu, M. Sehailia and M. J. Porter, Synlett, 2009, 3258–3262; (h) A. J. P. Mortimer, J. R. H. Plet, O. A. Obasanjo, N. Kaltsoyannis and M. J. Porter, Org. Biomol. Chem., 2012, 10, 8616–8627; (i) H. Yamada, M. Adachi and T. Nishikawa, Chem. Commun., 2013, 49, 11221–11223.
10. (a) A. E. Aliev and D. Courtier-Murias, J. Phys. Chem. B, 2007, 111, 14034–14042; (b) A. E. Aliev, Z. A. Mia, M. J. Busson, R. J. Fitzmaurice and S. Caddick, J. Org. Chem., 2012, 77, 6290–6295; (c) A. E. Aliev and D. Courtier-Murias, J. Phys. Chem. B, 2010, 114, 12358–12375.
11. W. Koźmiński and D. Nanz, J. Magn. Reson., 1997, 124, 383–392.
12. C. M. Thiele, K. Petzold and J. Schleucher, Chem. – Eur. J., 2009, 15, 585–588.
13. (a) D. Neuhaus and M. P. Williamson, The Nuclear Overhauser Effect in Structural and Conformational Analysis, Wiley-VCH, New York, 2nd edn, 2000; (b) T. Claridge, High-Resolution NMR Techniques in Organic Chemistry, in Tetrahedron Organic Chemistry Series, Pergamon Press, Oxford, 1999, vol. 19; (c) M. Reggelin, H. Hoffman, M. Köck and D. F. Mierke, J. Am. Chem. Soc., 1992, 114, 3272–3277.
14. A. E. Aliev, Z. A. Mia, H. S. Khaneja and F. D. King, J. Phys. Chem. A, 2012, 116, 1093–1109.
15. (a) R. Jain, T. Bally and P. R. Rablen, J. Org. Chem., 2009, 74, 4017–4023; (b) C. R. Mooney, M. E. Sanz, A. R. McKay, R. J. Fitzmaurice, A. E. Aliev, S. Caddick and H. H. Fielding, J. Phys. Chem. A, 2012, 116, 7943–7949.
16. A. Meissner and O. W. Sørensen, Magn. Reson. Chem., 2001, 39, 49–52.
17. M. J. Thrippleton and J. Keeler, Angew. Chem., Int. Ed., 2003, 42, 3938–3941.
18. K. E. Gilbert, PCMODEL, version 8.5, Serena Software, Bloomington.
19. M. F. Schlecht, Molecular Modeling on the PC, Wiley-VCH, New York, 1998.
20. J. J. P. Stewart, J. Mol. Model, 2007, 13, 1173–1213.
21. (a) E. Cances and B. Mennucci, J. Math. Chem., 1998, 23, 309–326; (b) M. Cossi, N. Rega, G. Scalmani and V. Barone, J. Comput. Chem., 2003, 24, 669–681.
22. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, Ö. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian 09, Revision D.01, Gaussian, Inc., Wallingford CT, 2009.
23. (a) Y. Zhao and D. G. Truhlar, Theor. Chem. Acc., 2008, 120, 215–241; (b) Y. Zhao and D. G. Truhlar, J. Chem. Theory Comput., 2008, 4, 1849–1868.
24. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297–3305.
25. Y. Zhao and D. G. Truhlar, Chem. Phys. Lett., 2011, 502, 1–13.
26. J. R. Cheeseman, G. W. Trucks, T. A. Keith and M. J. Frisch, J. Chem. Phys., 1996, 104, 5497–5509.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ob02076j

---