13C, 15N CPMAS NMR and GIAO DFT calculations of stereoisomeric oxindole alkaloids from Cat's Claw (Uncaria tomentosa)

ARTICLE IN PRESS Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

Contents lists available at ScienceDirect

Solid State Nuclear Magnetic Resonance journal homepage: www.elsevier.com/locate/ssnmr

13

C, 15N CPMAS NMR and GIAO DFT calculations of stereoisomeric oxindole alkaloids from Cat’s Claw (Uncaria tomentosa)

Katarzyna Paradowska a, Micha" Wolniak a, Maciej Pisklak a, Jan A. Glin´ski b, Matthew H. Davey b, Iwona Wawer a, a b

Faculty of Pharmacy, Medical University of Warsaw, 02-097 Warsaw, Banacha 1, Poland Planta Analytica, LLC, Danbury, CT, USA

a r t i c l e in f o

a b s t r a c t

Article history: Received 19 June 2008 Received in revised form 19 August 2008 Available online 22 October 2008

Oxindole alkaloids, isolated from the bark of Uncaria tomentosa [Willd. ex Schult.] Rubiaceae, are considered to be responsible for the biological activity of this herb. Five pentacyclic and two tetracyclic alkaloids were studied by solid-state NMR and theoretical GIAO DFT methods. The 13C and 15N CPMAS NMR spectra were recorded for mitraphylline, isomitraphylline, pteropodine (uncarine C), isopteropodine (uncarine E), speciophylline (uncarine D), rhynchophylline and isorhynchophylline. Theoretical GIAO DFT calculations of shielding constants provide arguments for identification of asymmetric centers and proper assignment of NMR spectra. These alkaloids are 7R/7S and 20R/20S stereoisomeric pairs. Based on the 13C CP MAS chemical shifts the 7S alkaloids (d C3 70–71 ppm) can be easily and conveniently distinguished from 7R (dC3 74.5–74.9 ppm), also 20R (dC20 41.3–41.7 ppm) from the 20S (dC20 36.3–38.3 ppm). The epiallo-type isomer (3R, 20S) of speciophylline is characterized by a larger 15 N MAS chemical shift of N4 (64.6 ppm) than the allo-type (3S, 20S) of isopteropodine (dN4 53.3 ppm). 15 N MAS chemical shifts of N1–H in pentacyclic alkaloids are within 131.9–140.4 ppm. & 2008 Elsevier Inc. All rights reserved.

Keywords: Oxindole alkaloids Uncaria tomentosa Cat’s Claw Vilcacora Solid-state NMR 13 C CP MAS NMR 15 N MAS NMR Shielding constants GIAO DFT

1. Introduction Uncaria tomentosa (Cat’s Claw, Unaˆ de Gato, Vilcacora) is a large, woody vine with hook-like thorns that resemble the claws of a cat. It is indigenous to the Amazon rainforest and other tropical areas of South and Central America. The vine has been used medicinally by native tribes for at least 2000 years in treating inflammation, arthritis, bone pain, asthma, deep wounds, and cancer [1]. Since the early 1990s Cat’s Claw has been used in Peru and Europe as an adjunctive treatment for diseases that target the immune system. Numerous investigations have been carried out to isolate and determine its bioactive components. Over 50 compounds have been identified including oxindole alkaloids [2], ursane type pentacyclic triterpenes, ursolic and quinovic acid derivatives, sterols [3–5] and procyanidins [6]. The pentacyclic oxindole alkaloids: mitraphylline, isomitraphylline, pteropodine, isopteropodine, speciophylline and uncarine F have been especially associated with immunomodulatory and cytotoxic activities [7,8]. In Europe, the measurement of isopteropodine (uncarine E), the most immunostimulating alkaloid, is used for the purpose of standardization of Cat’s Claw products [9]. The content

of the alkaloids in the commercial extracts should be standardized to 1.3–1.75% by weight, out of which 97% should constitute pentacyclic alkaloids. The presence of alkaloids is most frequently tested by chromatographic methods [10–12]. The absolute stereochemistry has been determined by single crystal X-ray crystallography for pteropodine, isopteropodine [13], mitraphylline, speciophylline and rhynchophylline [14]. In 1993, eleven natural heteroyohimbine-type pentacyclic oxindole alkaloids, isolated from Uncaria species, were analyzed by means of 1H, 13C NMR, including 2D NMR correlation experiments [15]. These stereoisomeric alkaloids have also been studied by 15N NMR spectroscopy [16]. The present study deals with the characterization of the most abundant alkaloids of U. tomentosa by solid-state NMR technique. Their structures are illustrated in Fig. 1 and stereochemical data are listed in Table 1. To our knowledge, this is the first report of 13C and 15N CPMAS NMR chemical shifts for Uncaria oxindole alkaloids. The study provides theoretical calculations of shielding constants using the DFT/GIAO approach.

2. Experimental  Corresponding author. Fax: +48 22 5720 950.

E-mail address: [email protected] (I. Wawer). 0926-2040/$ - see front matter & 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.ssnmr.2008.10.002

Pentacyclic oxindole alkaloids are usually isolated following a procedure published previously [17]. Cat’s Claw alkaloids were

ARTICLE IN PRESS K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

203

Fig. 1. The structure of pentacyclic (a) and tetracyclic (b) alkaloids.

Table 1 Characterization of stereochemistry of pentacyclic 1–5 and tetracyclic 6, 7 alkaloids

1 2 3 4 5 6 7

Alkaloid

Type

Mitraphylline Isomitraphylline Pteropodine (Uncarine C) Isopteropodine (Uncarine E) Speciophylline (Uncarine D) Rhynchophylline Isorhynchophylline

Normal Normal Allo Allo Epiallo Normal Normal

D/E ring 3S 3S 3S 3S 3R 3S 3S

purified from a commercial extract of U. tomentosa provided by Advanced Nutra Inc., Redding, CA, lot # AN C9800F2. Typically, the extract was suspended in water and treated with an excess of sodium carbonate. The alkaline solution was then quickly extracted with chloroform. The chloroform layer was then washed with 3% sulfuric acid. In this operation, the alkaloids, but not neutral components dissolved in the aqueous phase. After neutralization of the acid the alkaloids were again extracted into chloroform and recovered from it by evaporation in vacuo. The mixture of alkaloids was purified into individual components using centrifugal partition chromatography (CPC), model FCPC1 from Kromaton Technologies Inc., France. Details of the purification will be published elsewhere. The 1H and 13C chemical-shift spectra were recorded for a CDCl3 solution on a Bruker DSX-500 spectrometer. Standard pulse programs from Bruker library were used for COSY, TOCSY, HMQC and HMBC experiments. The 2D 1H–13C correlations were performed using the phase-sensitive gradient-selected (PFG) inverse technique; the HMBC experiment was optimized for J ¼ 5 Hz. Chemical shifts are reported in ppm relative to internal TMS. Cross polarization (CP) magic angle spinning (MAS) solid-state 13 C and 15N NMR spectra were recorded at 100.6 and 40.6 MHz, respectively on a Bruker DSX-400 WB MHz spectrometer; powder samples packed in a 4 mm ZrO2 were spun at 10 kHz (13C) and 3.5 kHz (15N). For the acquisition of 13C spectra a contact time (tCP) of 2 ms, a repetition time (tR) of 8 s and spectral width of 25 kHz were used, and 400–600 scans were accumulated. 13C chemical shifts were calibrated indirectly through the glycine CO signal recorded at 176.0 ppm, relative to TMS. Dipolar dephasing (DD) pulse sequence (with 50 ms delay time (tD) inserted before acquisition) was used to observe selectively the non-protonated carbons. Two-dimensional phase adjusted spinning sideband (2D PASS) experiments (introduced by Anzutkin et al. [18,19]) were performed at a spinning rate of 2 kHz, 1000 scans for each spectrum were accumulated. The 13C dii parameters were calculated using the WINMAS program.

7R 7S 7R 7S 7S 7R 7S

15S 15S 15S 15S 15S 15S 15S

19S 19S 19S 19S 19S – –

20R 20R 20S 20S 20S 20R 20R

trans trans cis cis cis – –

Accumulation of the natural abundance 15N MAS NMR spectra used 1H decoupling, a 10 s recycle delay and a contact time (tCP) of 5 ms; typically, 3200 scans were accumulated in overnight experiments. Chemical shifts for 15N were calibrated indirectly on glycine resonance d15N 10.0 ppm, and referenced to nitromethane d15N ¼ 380.2 ppm (liquid NH3 d15N ¼ 0). The geometry of alkaloids was adopted from crystallographic data of: 1 (CCSD MUTZEG), 3 (CCSD refcodes QIKGOG and KIKXIL), 4 (CCSD refcode QIKGUM), and 6 (CCSD refcodes MUTYUV, NIFDUB). Since the positions of hydrogen atoms in X-ray crystal structures are inaccurate, energy minimization was performed on the hydrogen atoms of the crystal structures using the semi-empirical PM3 method (from HyperChem 7.0 [20]). A full structure relaxation was subsequently performed when the geometry was optimized at density functional theory (DFT) level using procedures implemented in the GAUSSIAN-03 package [21]. The final low-energy structures were used for calculations of NMR shielding constants. Shielding tensors were computed using the GIAO-DFT (gauge-independent atomic orbital) with the 6-311G** basis set approach. Gaussian-03 calculations provide absolute shielding values, which were assigned such that s33Xs22Xs11 (d11Xd22Xd33), the latter component being along the direction of greatest shielding. The shielding constants are frequently converted to NMR chemical shifts by calculating reference compounds. The systematic offset of these data by several ppm makes such recalculation somewhat arbitrary, but the values of theoretical chemical shifts are easy to compare with experimental ones. The calculations of carbon shielding constants were performed for a set of compounds, including CH3NH2, CH3OH, CH3CN or urea [22,23]. The results depend on the method and basis set applied, as illustrated for tetramethylsilane: sTMS ¼ 183.2 ppm, B3LYP/6-311++(3df,3dp)] [24], sTMS ¼ 202 ppm, GIAO CPHF 6-31G(d,p) [25]. The absolute shielding values calculated for peptides were converted to chemical shifts relative to the absolute shielding of liquid TMS of 184.1 ppm [26]. Theoretical isotropic 13C chemical shifts diso and the principal elements of chemical shift tensor (CST) dii are given relative to

ARTICLE IN PRESS 204

K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

TMS and were obtained from a constrained linear regression: dii ¼ 182.0–0.98 sii. The RMS error for all of the shifts was found to be 4 ppm. Root-mean-square errors for carbon and nitrogen chemical shieldings at the B3LYP-6-311+G were determined to be 2.3 and 7.9 ppm, respectively [22]. The 15N chemical shifts are much more sensitive (due to the lone pair of nitrogen) to experimental conditions compared to the 13C ones. DFT GIAO calculations of the 15N CSA tensors were carried out in peptides [27,28], a comparison with the experimental values showed excellent agreement. The quantum chemical calculations were performed on computers at the Interdisciplinary Centre for Mathematical and Computational Modeling of Warsaw University.

3. Results and discussion 1

H and 13C NMR spectra of 1–7 measured in CDCl3 solutions were assigned using 2D COSY, HMQC and HMBC experiments, and the chemical shifts were found to be in agreement with those reported previously by Seki et al. [15]. The pentacyclic oxindole isomers can be also distinguished by comparing the 15N NMR chemical shifts. The most important problem is that the 13C and 15N NMR measurements in solution take several hours. During that time some alkaloids may undergo isomerization. In aqueous solutions most of these alkaloids isomerize within hours, creating an equilibrium of isomerized alkaloids whose ratios depend on pH and temperature. The kinetics and mechanisms of the isomerization process of pentacyclic oxindole alkaloids have been studied in detail [29,30]. In numerous cases, biological studies and measurements of particular alkaloid activity last more than a few hours. As a consequence, the longer lasting experiments involve studying the alkaloid isomer group (e.g. pentacyclic or tetracyclic oxindoles) [31]. Uncarine E and mitraphylline distinguish themselves from others in being the most resistant to isomerization. Therefore, hitherto unreported 13C CPMAS chemical shifts of the isomeric oxindole alkaloids were utilized to facilitate their characterization in the solid phase. It seemed interesting to check

whether solid-state 13C NMR data (isotropic and anisotropic chemical shifts and the parameters of spin dynamics) may help to distinguish between most of the isomers. Solid-state NMR was successfully applied to the study of isomers, organic polymorphs and solvates in a variety of chemical systems, as discussed by Harris [32]. Isomers may yield observably different NMR spectra, and in suitable cases even quantitative information on mixtures can be obtained from peak area ratios. The 13C CPMAS spectra were recorded for seven solid alkaloids, the pairs of compounds belong to the three types, normal: 1, 2, 6, 7, allo: 3, 4 and epiallo 5. The solid-state 13C spectra of 1 are illustrated in Fig. 2 and the spectra of 3, 4 and 6 in Fig. 3. 13C MAS chemical shifts measured for the solid pentacyclic and tetracyclic alkaloids are collected in Tables 2 and 3, respectively. The identification of the quaternary carbons was determined by means of DD experiments (see Fig. 2b). The resonances in the 13C MAS spectra of six alkaloids can be assigned directly by comparison with the solution data because single resonance for each carbon is observed; only 4 (i.e. uncarine E) exhibited polymorphism (Fig. 3a). The differences in 13C chemical shifts D ¼ dliquiddsolid reflect conformational changes and also changes of intermolecular interactions such as hydrogen bonding involving N1H and N4, OMe or CQO groups in the solids. Within these molecules there are five potential hydrogen bond acceptors and only one hydrogen bond donor at N1. According to available crystallographic (XRD) data, the N1H group forms intermolecular hydrogen bonds in all compounds. If the carbonyl group would be a hydrogen bond acceptor, one can expect a downfield shift of its resonance in the solid state spectrum compared to solution. However, for C2QO the differences (D) are small, less than 1 ppm. The chemical shifts of C22QO from ester groups are almost the same as those in solution. It suggests that carbonyl groups do not participate in the N1–HyOQC intermolecular interactions, in agreement with XRD data showing the N1–HyN4 bonds (in 1, 5 and 6) or the interactions of N1H with solvent (water, methanol, chloroform) molecules. In crystalline 1 weak N1–HyN4 bonding was found with an NyN distance of 3.284 A˚. There are also short contacts of

Fig. 2. 13C CP MAS spectra of mitraphylline (1) recorded at a spinning speed of 10 kHz: (a) standard (tCP ¼ 2 ms, tR ¼ 8 ms) and (b) dipolar dephased (tCP ¼ 2 ms, tR ¼ 8 ms, tD ¼ 50 ls).

ARTICLE IN PRESS K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

205

Fig. 3. 13C CP MAS spectra of (a) uncarine E (4), (b) uncarine C (3) and (c) rhynchophylline (6) recorded at a spinning speed of 10 kHz (tCP ¼ 2 ms, tR ¼ 8 ms). Spinning sidebands are marked with asterisks. Table 2 13 C NMR chemical shifts (d, ppm) for pentacyclic oxindole alkaloids 1–5 in solution (CDCl3) and solid phase 1

C2 C3 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23

2

3

4

5

dsolution

dCPMAS

dsolution

dCPMAS

dsolution

dCPMAS

dsolution

dCPMAS

dsolution

dCPMAS

181.34 74.57 54.28 35.15 55.56 133.33 122.88 122.53 127.99 109.74 140.87 28.37 30.44 106.92 154.04 14.83 73.82 40.49 54.33 167.08 50.17

181.7 74.8 53.0 37.6 55.1 133.2 123.3 121.8 128.3 111.5 142.7 29.8 30.5 108.0 154.1 15.6 74.8 41.9 53.0 166.2 50.9

181.25 71.80 53.38 35.40 56.6 133.60 124.92 122.37 127.54 109.40 140.29 29.10 30.00 107.29 153.85 14.85 74.00 40.88 54.28 167.04 50.77

181.1 71.5 54.1 35.0 56.7 134.2 124.3 122.0 127.7 110.0 141.5 30.4 30.4 107.9 154.1 14.9 74.0 41.4 53.0 166.1 50.6

181.38 74.41 55.19 34.66 56.11 133.45 123.01 122.57 127.91 109.64 140.82 29.53 30.96 109.14 155.23 18.97 72.16 37.81 53.68 167.70 50.91

182.0 73.1 54.9 35.6 56.1 132.8 123.1 121.5 128.5 111.0 142.8 29.7 31.9 110.0 154.1 16.7 73.1 38.3 52.1 166.3 50.1

181.29 71.20 54.10 34.78 56.90 133.72 124.50 122.45 127.63 109.64 140.23 30.14 30.41 109.77 154.99 18.61 72.12 37.82 53.47 167.63 50.97

180.9 71.0/69.9 55.4/54.9 35.5 56.9/56.3 133.9 123.5 121.8 128.6/128.1/127.6 111.9/110.7 142.1 30.0 31.8 110.7/110.0 154.7/154.0 18.2/16.7 72.7 36.3 52.8 166.0 49.3/48.8/48.6

181.91 70.89 55.01 34.42 56.38 134.18 123.00 122.11 127.60 109.61 141.70 26.74 25.66 105.73 153.90 18.45 74.57 36.52 53.45 167.00 50.65

181.6 70.3 53.4 33.2 55.9 134.3 124.2 120.5 126.3 109.1 142.3 27.5 24.7 104.1 154.0 17.7 74.4 36.1 50.0 166.6 50.2

the type C2 ¼ O1yH–C12 and C2 ¼ O1yH–C21. In the crystals of 3, the water/methanol solvate [13], the water molecules form intermolecular bonds N1–HyOH2 and link the acceptor atoms. In

the crystals of the chloroform solvate of 3 [14], the N1–HyOQC bonds link molecules into chains. No signal of solvent (methanol, acetone or chloroform) has been detected in the 13C MAS spectra

ARTICLE IN PRESS 206

K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

Table 3 13 C NMR chemical shifts (d, ppm) for tetracyclic oxindole alkaloids 6 and 7 in solution (CDCl3) and solid phase 6

C2 C3 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 OMe

7

dsolution

dCPMAS

dsolution

dCPMAS

181.53 75.23 58.19 34.81 56.03 133.78 123.16 122.43 127.74 109.33 140.90 28.90 39.61 111.74 159.81 11.32 24.14 37.80 58.19 166 51.33 61.71

179.3 78.0 58.3 30.4 56.1 133.3 120.1 120.1 127.7 111.5 145.2 29.2 38.3 112.5 158.0 12.5 20.6 40.6 54.6 166.2 51.7 61.1

182.20 72.63 58.40 35.81 56.30 134.25 125.47 122.58 127.67 109.48 140.20 29.94/30.54 38.39 112.54 159.87 11.45 24.50 37.77 54.51 168/169 51.66/50.96 61.49

181.4 72.1 59.9 29.8 56.7 133.9 124.2 124.2 126.1 110.5 140.9 29.8 37.6 110.5 159.2 11.2 24.0 37.6 56.7 167.6 50.2 59.9

Table 4 13 C NMR shielding constants (siso, ppm) calculated (GIAO DFT) for oxindole alkaloids: 1, 3, 4 (A and B), 5 and 6 C

1

3

4A

4B

5

6

C2 C3 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 OMe OMe0 R2a

19.33 115.64 135.56 152.11 129.96 57.75 71.94 73.68 68.94 87.96 54.08 161.51 156.68 85.60 39.19 176.51 114.85 148.34 136.69 29.61 141.31

18.71 115.23 134.76 152.06 132.53 57.98 72.42 73.71 68.20 87.85 53.72 159.60 156.47 82.86 39.69 172.62 117.68 149.82 137.72 28.43 141.41

12.35 121.45 135.64 148.41 129.60 56.96 70.42 75.50 70.15 85.23 52.13 157.57 161.30 85.99 42.17 170.63 119.41 151.16 135.54 30.61 142.12

14.19 123.82 135.26 151.16 129.59 58.28 71.33 75.26 70.01 85.18 52.72 157.10 163.76 81.21 42.16 167.93 117.10 147.53 133.03 31.02 142.25

22.23 120.02 135.08 156.73 133.00 63.95 75.19 72.39 78.81 91.79 57.09 164.19 161.90 92.87 46.13 170.67 115.36 153.68 137.50 37.35 135.03

18.20 114.49 134.99 151.99 129.40 56.60 71.96 73.57 69.15 88.28 54.43 159.37 145.66 81.65 35.11 182.51 159.61 148.31 128.99 31.97 141.59 133.07 0.9924

a

of 1–7, but the presence of crystalline water, especially in 3 and 4 cannot be excluded. The values of D are also indicative of rigid and flexible fragments of the molecule, the latter are expected to undergo larger changes. No significant changes of chemical shifts are observed for chiral carbons indicating that these fragments of structure are rigid. Generally, there is no big difference (D43 ppm) between solution and solid-state 13C chemical shifts of pentacyclic alkaloids. It suggests that the tricyclic skeleton (rings C, D, E) is rigid, and the conformations of these rings are probably similar in both phases. Therefore, 13C CPMAS data can be useful for the identification of solid pentacyclic alkaloids. Larger values of D are observed for tetracyclic alkaloids, in 7 only for C6 and C19. In 6 the differences of ca. 4 ppm appear for C3, C6, C13, C19 and C21 suggesting some conformational freedom. Molecular structures of alkaloids taken from X-ray diffraction data were further optimized by PM3 and DFT methods, and selected low-energy conformations were used for calculation of NMR shielding constants. The values of isotropic shielding constants for 1, 3–6 are given in Table 4. Numerous reports on magnetic shielding tensor calculations employing the GIAO DFT methods showed very good results [33]. The HF//B3LYP//LB3LYP methods are efficient, and DFT calculations are less time consuming than MP2 [34]. Systematic investigation of 13C NMR constants using the B3LYP/6-311+G(2d,p) showed [35] that the HF, BLYP and B3LYP geometries lead to mean absolute deviation of chemical shifts of 2.36, 5.80, and 4.43 ppm, respectively. The plots of 13C siso versus dCPMAS are linear (R240.99) and confirmed the hierarchy of resonances in the solid-state spectra. Theoretical calculations not only allow for reliable assignment of MAS NMR spectra but also provide arguments for selection of stereoisomers at asymmetric centers and solid-state conformations. The oxindole alkaloids have five chiral carbons, and proper identification of their stereochemistry is not an easy task. Additionally, crystallographic studies revealed polymorphism of 4 and 5. According to the X-ray data of Muhammad et al. [13], crystalline uncarine E has three molecules in the unit cell, and hydrogen bonds link them into chains. The sample of 4 studied by

0.9968

0.9980

0.9967b

0.9979b

Correlation between calculated shielding constants and measured for solid samples (Table 2). b The highest R2 values for particular set of dCPMAS.

0.9975 13

C chemical shifts

us also exhibited polymorphism, as indicated by several multiplet resonances in the 13C CPMAS spectrum, and thus more structural information was required. We have also attempted to answer whether the structure and crystal packing are the same as previously found? One can expect the splitting of C2 caused by the differences in the geometry of N1–HyO1QC2 hydrogen bonds since the N1yO1 distances are: 2.864, 2.873 and 2.894 A˚. However, this signal is not split. It is probable that the differences in hydrogen bonds are too small to significantly influence the shielding of C2 (although the resonance of N1H appears as doublet in 15N MAS spectrum of 4, see Fig. 4b). Likewise, the C–HyO interactions (C10–HyO2, C15–HyO1 and C20–HyO1) do not produce differences in chemical shifts and the signals of C10, C15 and C20 appear as singlets. In the 13C MAS NMR spectrum of solid 4 the resonances of C11, C3, C5, C17 and C18 are split indicating some variability in the conformation of ring A, C and E. The splitting of C11 into three lines and 1:2 doublets for C18, C17 and OMe suggests that three molecules are present in an asymmetric unit cell. The multiplet resonances can be assigned on the basis of shielding constants calculated for particular conformers. Inspection of XRD data [13] for 4 reveals that two molecules differ in the geometry of ring C and E (with C20 atom out of the plane by 0.65–0.68 A˚); the third one resembles the first but its intermolecular interactions are not the same. Therefore, the GIAO DFT calculations were performed for two conformers (4A and 4B) of similar geometry as those in the crystal. The selected low-energy conformers (shown in Fig. 5) differ mainly in the geometry of ring E (boat 4A or chair-like 4B). Their isotropic shielding constants siso (given in Table 4) were compared with the experimental dCPMAS. The correlation of calculated isotropic shielding with each set of resonances in the MAS spectrum is satisfactory (R240.99). However, the R2 criterion used to determine the quality of fit is not sensitive enough to make selection of polymorphic forms. The highest values of R2 obtained for fitting the data for 4A and 4B are 0.9969 and 0.9980, respectively.

ARTICLE IN PRESS K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

Fig. 4.

207

15

N MAS spectra of pentacyclic alkaloids 1 and 4 recorded at a spinning speed of 3.3 kHz (tCP ¼ 5 ms, tR ¼ 10 ms).

Fig. 5. The structure of two conformers of uncarine E (4A, 4B) optimized by DFT.

The anisotropic elements of CST may add some information in structural analysis. For the majority of carbons the values of dii calculated for conformer 4A and 4B differ by 72 ppm (Table 4). Larger differences in d22 (up to 12 ppm) appearing for C18 and C19 reveal structural differences in the ring E, and may help to distinguish the conformers which differ in chemical shift of C18 (see Fig. 3). Therefore, anisotropic elements of CST have been measured. The PASS-2D spectrum of 4 recorded at 2 KHz represents a complex pattern. By proper data shearing it was possible to separate spinning sidebands for particular carbons. By employing a calculation procedure, the 13C chemical shift anisotropy parameters: d11, d22 and d33 could be determined for 10 carbons of the 23 in the molecule. The 13C principal elements of the CST are collected in Table 5. Of the several multiplets observed in the spectrum, only in the case of C18 was it possible to determine all elements. Although methyl carbon exhibits smaller anisotropy that aromatic carbon (and a smaller number of sidebands at 2 KHz), it appears to be sensitive to molecular conformation. The comparison of experimental tensor values with calculated ones clearly shows that the low-field set can be assigned to conformer 4B, whereas the upfield set better represents conformer 4A. Correlations of the experimental

isotropic and anisotropic shielding parameters and the respective theoretical shielding values for these two conformers are illustrated in Fig. 6. Thus, it is probable that such conformers exist in solid 4. Crystallographic data for 5 showed [13] that the asymmetric unit contains three independent molecules. Characteristic structural features include: ring C is twisted, ring D is a chair, ring E has a conformation in which C15, 16, 17 and O20 lie in a common plane, whereas atom C19 lies below and C20 lies above this plane. Molecules are linked by C2QO1yH–N1 bonds, and also by weak N1–HyN4 bonds. However, the 13C CPMAS spectrum of solid 5 does not show multiplet signals. Broader signals, for example of C2QO, probably arose from chemical shifts dispersion, e.g. due to the differences in hydrogen bonding. The splitting of 13C resonances due to residual dipolar interaction with neighbor quadrupolar 14N nuclei is usually not observed at 9.4 T. Shielding constants were calculated for the low-energy conformer of 5, and the correlation dCPMAS versus siso is satisfactory (R240.9975, Table 4); a high correlation coefficient may suggest that this conformation represents that in the studied solid sample of 5. The alkaloids 1 and 2 belong to the normal-type group, which is characterized by the trans relationships of the C/D and D/E ring

ARTICLE IN PRESS 208

K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

Table 5 The principal elements of chemical shift tensor (dii, ppm) of uncarine E (4) obtained from PASS 2D spectra (measured and best-fitting simulated 1D sideband patterns) and calculated using (GIAO DFT) for conformer A and B Measured values

C2 C3 C5 C6 C7 C8 C9 C10 C11 C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23

Calculated for 4A

Calculated for 4B

diso

d11

d22

d33

diso

d11

d22

d33

diso

d11

d22

d33

180.7

25975

16574

11875

133.3 123.3 121.7

20573 21973 22075

18674 15672 14374

974 572 273

141.7

22976

16775

2975

112.0

18774

11974

3073

187.3 68.2 53.5 39.9 59.5 139.4 126.4 120.9 126.6 110.5 146.0 28.8 26.0 107.3 156.1

281 103 94 74 81 215 224 226 231 195 238 44 37 189 253

155 60 58 44 58 177 145 131 141 127 152 36 28 99 142

126 41 9 2 39 27 10 6 8 9 48 6 14 33 72

18.3 16.7 72.7

4373 3274 9474

1772 2173 8274

573 372 4275

186.1 66.8 54.9 37.1 59.7 138.9 126.0 121.1 126.6 110.4 144.6 29.4 22.6 113.4 156.7 18.4

279 100 95 68 82 216 225 226 233 196 236 49 34 207 257 46

156 60 63 41 58 174 143 131 141 127 151 30 26 104 143 15

124 40 6 3 40 27 10 6 6 8 47 9 8 29 70 6

25474 8773

14573 5174

10273 972

16.2 70.8 37.2 53.4 169.0 46.5

28 93 52 83 266 79

27 90 35 57 132 59

6 30 25 20 109 2

72.6 40.5 55.7 168.0 46.1

99 60 92 266 78

77 41 58 128 59

42 20 17 110 2

167.0 49.0

Fig. 6. Correlations of the experimental and theoretical shielding parameters for conformers A and B of 4.

junctions, whereas the allo-type alkaloids are characterized by D/E cis relationship. So, in 3 and 4 the bulky substituent at C3 is located in the axial position to the D ring. The three studied pairs

of alkaloids: 1/2, 3/4 and 6/7 are the stereoisomers at the spiro C7 position. The stereochemistry at the spiro center can be easily distinguished by the 13C MAS chemical shifts of C3 carbons: dC3 70–71 ppm in 7S and dC3 74.5–74.9 ppm in 7R. The C9 signals of 7S isomers 1 and 5 appear at lower fields by 1 ppm, compared with 7R spiro isomers (2, 6). The tetracyclic alkaloids 6 and 7 have the normal configuration; a distinct feature of these alkaloids is an additional OMe group. In the 13C NMR spectra of both tetracyclic alkaloids in solution either doublets or broadened resonances appear for C14, C15, C19, C22 and also for OMe and OMe0 . This suggests some intramolecular dynamic (involving ring D). The conformational equilibration is frozen in the solid-state and the 13C CPMAS spectra of solid 6 (see Fig. 3c) and 7 show one set of resonances. The X-ray diffraction measurements performed for rhynchophylline 6 [13] showed that the side chain is perpendicular to the ring plane. There is N1–HyN4 bonding between chains of molecules, with an NyN distance of 2.999 A˚. Correlation between experimental chemical shifts for 6 and calculated siso is quite good, and confirms the assignment of resonances in solid-state 13C spectrum. The signals of two methoxy groups are separated by 10 ppm; in pentacyclic alkaloids the signal of OMe (C23) appears at 49–51 ppm whereas that of OMe’ at 60–61 ppm. The two tetracyclic alkaloids can be easily distinguished considering the chemical shift of C3: d 78 ppm in 6 (7R) and d 72 ppm in 7 (7S). Although pentacyclic oxindole isomers (1–5) can be identified by comparing 13C CPMAS NMR spectral data, it seemed interesting to measure also 15N MAS chemical shifts. For solutions, the chemical shift of N1 was within 135.1–136.9 ppm, whereas dN4 ranged from 54.6 to 64.8 ppm [16]. The allo-type isomers (3S, 20S) exhibited larger chemical shift differences at the N4 nitrogen (8–10 ppm) compared to those observed for normal-type (3S, 20R). Therefore, the 15N MAS chemical shifts can be used as an additional tool to distinguish among stereoisomers. 15N MAS NMR chemical shifts for solid oxindole alkaloids have been measured for the first time. Although the measurements on natural 15N abundance are time consuming, the technique is nondestructive and the samples can be further used for biochemical studies. Unfortunately, the 15N MAS spectra of three alkaloids

ARTICLE IN PRESS K. Paradowska et al. / Solid State Nuclear Magnetic Resonance 34 (2008) 202–209

Table 6 15 N NMR chemical shifts for oxindole alkaloids in CDCl3 solution [29] and solid state Alkaloid

N1–H

N4

dsolution

dCPMAS

1 2 3 4

135.1 135.8 136.9 136.7

5

135.1

131.9 138.8 – 140.4 138.7 135.7

siso 120.4 – 123.1 A 118.9 B 117.9 122.2

dsolution

dCPMAS

64.7 62.8 56.9 54.6

63.2 60.8 – 53.3

64.8

64.6

siso

209

interactions (NHyN and or NHyO hydrogen bonds) in the solid alkaloids are weak and do not significantly influence 13C and 15N shieldings. The DTF GIAO calculations of shielding constants are helpful in determining of solid-state conformers. Analysis of CST suggests that polymorphs existing in solid 4 differ in the geometry of ring E.

205.5 – 191.8 A 188.2 B 192.4 198.5

Isotropic shielding constants are calculated using GIAO DFT.

Acknowledgment The calculations were performed in the Interdisciplinary Center for Mathematical and Computational Modeling (ICM) at Warsaw University under the computational Grant G14-6. References

could not be obtained because of insufficient amount of the sample. In the 15N CPMAS NMR spectrum of an oxindole alkaloid two narrow peaks due to the two functional groups, N1–H and N4 can be observed, as shown in Fig. 4. In the spectrum of polymorphic 4 the resonance of N1H is split; the difference of ca. 2 ppm confirms the presence of nonequivalent molecules, in agreement with 13C CPMAS spectra. Somewhat unexpectedly, 15N chemical shifts measured for solid samples appeared to be close to the respective solution values (Table 5). Weak interactions with solvent such as CDCl3 are replaced by hydrogen bonds NHyN and/or NHyO in the crystal. These interactions should be reflected in the 15N MAS spectra since hydrogen bond formation significantly influences the 15N chemical shift. Similar values of d15N in both phases can be explained by weak intermolecular interactions involving nitrogen atoms in the solid alkaloids. Weak N10 –HyN4 bonds, or lack of such interactions, is in agreement with the XRD data on oxindole alkaloids. The 15N MAS chemical shift of N1 H is in the range 131.9–140.4 ppm. Based on the shift of N4 of 53.3 ppm, the allotype isomer (3S, 20S) uncarine E (4) can be distinguished from the epiallo-type (3R, 20S) 5, as well as from the normal-types, characterized by larger chemical shifts (60.8–64.6 ppm). In order to support the 15N chemical shift assignments, the calculated (GIAO DFT) 15N isotropic shielding constants have been compared with the experimental chemical shifts for solution and solid-state. Successful correlations between experimental 15N chemical shifts and those calculated by HF and DFT methods were obtained previously [36,37]. The DFT approach with the BLYP functional and the 6-311G** basis set was able to reproduce the hierarchy of 15N NMR chemical shifts in the spectra of oxindole alkaloids (Table 5). Unfortunately, since d 15N MAS values for three alkaloids are missing from Table 5, only four values remain for comparison (Table 6). It is worth noting that the correlation between experimental and calculated 15N chemical shifts is reasonably good for alkaloids; such a relationship was constructed [16] for 49 15N chemical shifts, and the regression analysis delivered R2 ¼ 0.987. Therefore, 15N NMR spectra can be used in identification of alkaloids.

4. Conclusions The 13C CPMAS spectra of alkaloids of remarkably good quality were obtained in less than an hour. The standard 13C spectra exhibited clearly resolved signals for all 21 (or 22) carbon atoms of these molecules. The solid-state 13C and 15N resonances have chemical shifts almost the same as their liquid state counterparts, enabling easy identification of particular stereoisomers. These results indicate that the pentacyclic rings structure of alkaloids is relatively rigid, without much conformational freedom and results in the averaging of chemical shifts. Intermolecular

[1] K. Keplinger, G. Laus, M. Wurm, M.P. Dierich, H. Teppner, J. Ethnopharmacol. 64 (1999) 23. [2] M. Shamma, R.J. Shine, I. Kompis, T. Sticzy, F. Morsingh, J. Poisson, J.L. Pousset, J. Am. Chem. Soc. 89 (1967) 1739. [3] R. Aquino, F. De Simone, F.F. Vinceri, C. Pizza, E. Gacs-Baitz, J. Nat. Prod. 53 (1990) 559. [4] R. Aquino, V. De Feo, F. De Simone, C. Pizza, G. Cirino, J. Nat. Prod. 54 (1991) 453. [5] R. Aquino, N. De Tommasi, F. De Simone, C. Pizza, Phytochemistry 45 (1997) 1035. [6] C. Goncalves, T. Dinis, M.T. Batista, Phytochemistry 66 (2005) 89. [7] I. Lemaire, V. Assinewe, P. Cano, D.V.C. Awang, J.T. Arnason, J. Ethnopharmacol. 64 (1999) 109. [8] M. Sandoval, R.M. Charbonnet, N.N. Okuhama, J. Roberts, Z. Krenova, A.M. Trentacosti, M.J. Miller, Free Radical Biol. Med. 29 (2000) 71. [9] G. Laus, D. Keplinger, J. Chromatogr. A 662 (1994) 243. [10] A. Van Ginkel, Phytother. Res. 10 (1996) 18. [11] M. Ganzera, I. Muhammad, R.A. Khan, I.A. Khan, Planta Med. 67 (2001) 447. [12] P. Montoro, V. Carbone, D. Quiroz Jde, F. De Simone, C. Pizza, Phytochem. Anal. 15 (2004) 55. [13] I. Muhammad, I.A. Khan, N.H. Fischer, F.R. Fronczek, Acta Crystallogr. C 57 (2001) 480. [14] G. Laus, K. Wurst, Helv. Chim. Acta 86 (2003) 181–187. [15] A. Seki, H. Takayama, N. Aimi, S.I. Sakai, D. Ponglux, Chem. Pharm. Bull. 41 (1993) 2077. [16] I. Muhammad, D.C. Dunbar, R.A. Khan, M. Ganzera, I.A. Khan, Phytochemistry 57 (2001) 781. [17] H. Wagner, B. Kreuzkamp, K. Juric, Planta Med. 5 (1985) 419. [18] O.N. Antzutkin, S.C. Shakar, M.H. Levitt, J. Magn. Reson. Ser. A 115 (1995) 7. [19] O.N. Antzutkin, S.C. Shakar, M.H. Levitt, J. Magn. Reson. 135 (1998) 144. [20] Hyperchem, Release 7.0, Hypercube Inc. Toronto, Canada, 2002. [21] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A. Montgomery Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian, 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, P.Y. Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Inc., Wallingford CT, 2004. [22] D.B. Chesnut, Chem. Phys. Lett. 380 (2003) 251. [23] F. Blanco, I. Alkorta, J. Elguero, J. Magn. Reson. Chem. 45 (2007) 797. [24] A. Gryff-Keller, J. Mol. Struct. Theochem. 587 (2002) 171. [25] T. Z˙o"ek, K. Paradowska, D. Krajewska, A. Ro´z˙an´ski, I. Wawer, J. Mol. Struct. 646 (2003) 141. [26] J. Birn, A. Poon, Y. Mao, A. Ramamoorthy, J. Am. Chem. Soc. 126 (2004) 8529. [27] J.R. Brender, D.M. Taylor, A. Ramamoorthy, J. Am. Soc. 123 (2000) 914. [28] A. Poon, J. Birn, A. Ramamoorthy, J. Phys. Chem. B 108 (2004) 16577. [29] G. Laus, D. Bro¨ssner, G. Senn, K. Wurst, J. Chem. Soc. Perkin Trans. 2 (1996) 1931. [30] G. Laus, J. Chem. Soc. Perkin Trans. 2 (1998) 315. [31] M. Wurm, L. Kacani, G. Laus, K. Keplinger, M.P. Dierich, Planta Med. 64 (1998) 701. [32] R.K. Harris, Analyst 131 (2006) 351. [33] D.B. Chesnut, Chem. Phys. Lett. 380 (2003) 251. [34] W. Hieringer, F. Della Sala, A. Goeorling, Chem. Phys. Lett. 383 (2004) 115. [35] Y. Zhang, A. Wu, X. Xu, Y. Yan, J. Phys. Chem. A (2007) 176. [36] E. Kleinpeter, L. Hilfert, A. Koch, J. Phys. Org. Chem. 13 (2000) 473. [37] M.B. Ferraro, V. Repetto, J.C. Facelli, Solid State Nucl. Magn. Reson. 10 (1998) 185.