Panigot MJ, Robarge MJ, Curley, Jr. RW. Virtual Coupling of Pyran Protons in the ¹H NMR Spectra of C- and NGlucuronides: Dependence on Substitution and Solvent. .
AAPS PharmSci. 2001; 3 (1): article 4. DOI: 10.1208/ps030104

Virtual Coupling of Pyran Protons in the ¹H NMR Spectra of C- and NGlucuronides: Dependence on Substitution and Solvent.
M.J. Panigot,¹  M.J. Robarge,²  and R.W. Curley, Jr.² 

¹Department of Chemistry, Arkansas State University, State University, AR 72467
²Division of Medicinal Chemistry and Pharmacognosy, The Ohio State University, Columbus, OH 43210

Correspondence to:
R.W. Curley, Jr.
Tel: 614-292-7628
Fax: 614-292-2435
Email: curley.1@osu.edu

Submitted: July 24, 2000; Accepted: January 8, 2001; Published: January 17, 2001

Keywords:  1H NMR, Glucuronides, Breast cancer, Chemoprevention, Virtual coupling

Abstract

We have observed that certain C- and Nglucuronides prepared as intermediates for breast cancer preventives demonstrate non-first order ¹H NMR spectra that are not the result of impurities or degradation but are instead due to virtual coupling in the pyran proton network. This virtual coupling shows the expected dependence on solvent and field strength and, more importantly, on the nature of the C-1 substitution. Although the hybridization of the atom bonded to C-1 may play a role, it appears that steric and/or electronic factors, which have the effect of increasing Δν/J for H-3 and H-4, are critical for eliminating the spectral complexity. These observations, which appear to be fairly general, suggest that this phenomenon should be considered when addressing the purity of pharmaceutical agents containing these types of structural units.

Introduction

The O-glucuronide metabolites of retinoic acid and certain of its natural and synthetic analogues have been suggested to be biologically active forms of the parent molecule.¹ As a class, these retinoids regulate epithelial tissue differentiation and show utility in treating dermatological diseases as well as promise for the treatment and prevention of cancer.² Because of the relative chemical and metabolic instability of these glucuronides, we have been synthesizing C- and N-glucuronosyl analogues of some of these metabolites in an effort to improve the activity of these compounds and/or to determine whether these metabolites are active themselves or are hydrolyzed to the active parent retinoid.³ Thus, we have prepared C-glucuronosyl analogues 1 and 2 (Figure 1) of the O-glucuronide 3 of the semisynthetic retinoid N-(4-hydroxyphenyl)retinamide. Our results suggest these compounds show promise as mammary tumor chemopreventive agents.^4,5

In the course of synthesizing 1, selective PtO₂-mediated oxidation^6,7 of the 6-hydroxymethyl group of glucosylbenzene⁴ followed by esterification and acetylation produced a product 5 that showed unusual complexity in the ¹H NMR spectrum in the region of the pyran ring protons. This was true for all resonances except that assigned for the H-1 proton. Since the Adams' catalyst that promoted oxidation had not to our knowledge been previously employed for the oxidation of C-glycosyl compounds into their glucuronide analogues, and given that this aryl-C-glycoside contains a tertiary carbon and benzylic ether unit (carbohydrate position 1), both of which may be prone to oxidation, we were concerned that other products might have been produced during the reaction that would compromise the purity of the materials and hence the validity of bioactivity assays performed with them.

After careful chromatographic purification and recrystallization of 5 to apparent homogeneity, while its ¹H NMR spectrum remained unchanged, other available spectroscopic evidence (¹³C NMR, IR, and MS) was consistent with a single compound assigned the structure 5. The possibility that the complexity of the ¹H NMR spectrum resulted from long-range virtual ¹H-¹H coupling was thus considered.⁸ Spin simulation of the spectrum using PANIC (Parameter Adjustment in NMR by Iterative Calculation) appeared to confirm this explanation.

Prompted by the report of Saito et al⁹ on their observation of virtual ¹H-¹H coupling in glucuronosyl moieties within O-disaccharides and their conjugates, we wish to report our interesting observations of similar phenomena in C- and N-glucuronosyl compounds, which appears to depend on the structure of the pyran C-1 substituent and the solvent employed in NMR measurements. This observation of deceptively complex spectra appears to be surprisingly general and should be considered when evaluating the purity, including the stereochemical purity, of potential pharmaceutical agents containing these structural units.

Materials and Methods

Fourier-transformed ¹H NMR spectra were obtained on sample solutions in glass 175 x 5 mm sample tubes (Wilmad; Buena, NJ). Spectra were collected for 20 mg/mL solutions at 250, 400, 600, and 800 MHz on AC250 or DPX250, DRX400, DMX600, and DMX800 instruments, respectively (Bruker Instruments; Billerica, MA). Samples were dissolved in CDCl₃, CD₂Cl₂, acetone-d₆, benzene-d₆, DMSO-d₆, CD₃OD, pyridine-d₅, and tetrahydrofuraN-d₈ as appropriate (Cambridge Isotope Laboratories; Andover, MA) and spectra referenced to the residual protio solvent (relative to TMS) in the deuterated solvents. Spectra were collected at ambient temperature using 90° pulse widths and transformed after exponential multiplication (LB = 0.2 Hz). Spectral simulation (see Table 1) was performed using PANIC version 840419 implemented on an ASPECT 3000 computer (Bruker Instruments).

The compounds studied were prepared as previously published.^3,10,11 Entries 2 and 9 (Table 2) were prepared by methods identical to those used for entries 1 and 10 using the appropriate Grignard reagents, while entries 17 and 18 were prepared by methods identical to those used in entry 19 using acetyl and benzoyl chloride respectively.

The 5 used in this study was prepared as previously described.³ The 250 MHz ¹H NMR spectrum of this compound in CDCl₃, in the region of the pyran protons, is shown in Figure 2. The surprising complexity of this spectrum, which is still present at 400 MHz (but is reduced at 600 MHz and and eliminated at 800 MHz), and the possibility that it arose from virtual coupling between H-2 and H-5, led us to simulate the spectrum using PANIC, as is also shown in Figure 2. The chemical shifts and calculated coupling constants derived from simulating the spectrum of 5 are shown in Table 1. For this simulation, the apparent couplings constants J_1,4,, J_1,5, J_2,4, and J_2,5 are sufficiently small that they can be set to zero and a satisfactory simulation can be obtained. Nonetheless, the H-2 and H-5 nuclei appear to show the observed complexity by virtue of being coupled as X parts of ABX spectra to H-3 and H-4, which themselves form a strongly coupled AB system with Δν/J = 0.82 at 250 MHz. As might be expected, this phenomenon can be eliminated by recording the ¹H NMR spectrum of 5 in different solvents. As also shown in Figure 2, the spectrum of 5 in acetone-d₆ can be analyzed as first order,with Δν/J for H-3 and H-4 now being 2.96.

Interestingly, our chemistry to further elaborate 5 to 1 produced intermediates that show virtual coupling that depends on both the nature and site of aromatic ring substitution. Nitration of 5 produced a 3:2 mixture of isomers 6 and 7, which were difficult to separate.³ In one instance, small quantities of pure 6 and 7 were obtained by preparative TLC. Their 250 MHz ¹H NMR spectrum in CDCl₃ showed virtual coupling comparable to that of 5 for 6 but not to that of 5 for 7 (Data not shown). Reduction of the nitroaromatic isomer mixture produced the readily separable O- and p-anilines 8 and 9.³ In this instance, the para substituted aniline 8 also shows strong virtual coupling that was not simulated but appears likely to result from the even smaller Δν/J_3,4 ratio (Figure 3). For the ortho regioisomer 9, this virtual coupling observed for 5 and 8 is also absent. Homonuclear decoupling and NOE difference spectra established that H-2 in 9 has moved substantially downfield to 5.61 ppm. More importantly, the chemical shift of H-3 and H-4 has reversed relative to 5 (5.37 and 5.29 ppm respectively) and Δν/J_3,4 has increased to 1.91, which appears to be sufficient to eliminate this coupling phenomenon.

Because both the O-nitrophenyl and O-aminophenyl isomers 7 and 9 fail to show the virtual coupling present in 5, 6, and 8, which bear a C₂-symmetric substituent at C-1, it seems plausible that this lack of virtual coupling results from steric interactions of the O-substituent with the axial H-1 or H-2 protons. This results in a different favored rotamer about the C-1-Ar bond and/or causes subtle changes in the conformation of the pyran ring, changes that have the effect of increasing Δν/J_3,4. In support of this concept, none of the ortho nitro or amino C-benzyl analogues 10 or 11³ (which we required for the preparation of 2) that have an interposed methylene unit show evidence of virtual coupling in the 250 MHz ¹H NMR spectra in CDCl₃ (see Table 2 for a summary of the compounds we investigated to determine whether the phenomenon is observed). That other more subtle influences such as electronics may also play a role is suggested by inspection of the spectrum of the O-tolyl analog 12, which we prepared serendipitously during efforts to synthesize 2.¹⁰ In the CDCl₃ ¹H NMR spectrum of 12, the H-2, H-3, and H-4 resonances overlap extensively, unlike any of the other compounds reported here. However, the H-5 resonance at 4.16 ppm shows some evidence of much less extensive virtual coupling than for 5, implying that the impact of the O-methyl substituent is insufficient to change Δν/J_3,4 enough to eliminate virtual coupling under these spectroscopic conditions. Furthermore, we observed that the 1-β-azido glucuronide 13 we previously prepared¹¹ demonstrated virtual coupling in the ¹H NMR spectrum in CDCl₃, which is nearly identical to that of 5. This coupling is absent at 400 MHz and in the 250 MHz acetone-d₆, benzene-d₆, CD₂Cl₂, CD₃OD, pyridine-d₅, and tetrahydrofuraN-d₈ DMSO-d₆ spectra of 13 and also in the CDCl₃ spectrum of the amine prepared by reduction of 13 as well as its acylated derivatives.¹¹ Once again, linear, symmetrical azide substitution results in virtual coupling while reduction products do not show this property, suggesting, perhaps, that the hybridization of the C-1 attached atom may play a role in causing this phenomenon. However, as shown in entry 9 of Table 2, the spherically symmetrical, sterically undemanding methyl substituted compound also demonstrates this virtual coupling. Thus, with the limited set of examples explored here, while those with atoms with sp²-like character bonded to C-1 demonstrate this coupling, steric and electronic effects from the C-1 substituent are likely to be more important contributors to the complexity of the observed spectra than is hybridization.

It might be expected that homonuclear decoupling experiments would allow elimination of this observed virtual coupling in many instances. In the present case, this is only a partially successful strategy because the phenomenon is driven by the small value of Δν/J _3,4 and thus selective irradiation of H-3 or H-4 is not possible. As shown for compound 13 in the Appendix, irradiation of H-5 and H-2 (4.1 and 4.95 spm respectively) still leaves some significant evidence of a noN-first order spectrum. More successful in this case is the impact of raising the temperature on spectral appearance (also see Appendix). Interestingly, we have observed this virtual coupling for C- and N-glucuronides only when samples are dissolved in CDCl₃. Thus, it appears that in this solvent a unique pyran ring conformation and fortuitous ¹H chemical shifts create the observed phenomenon. Given the high volatility of CDCl₃, limits are placed on routine use of elevated temperature experiments. Nonetheless, raising the temperature for 13 in CDCl₃ by 20°C above ambient clearly alters spectral appearance in a manner consistent with movement toward a first order spectrum.

Thus, as in some β-D-glucopyranosuronate systems,⁹ certain C- and N-glucuronides can show surprisingly complex ¹H NMR spectra. These appear to be the result of long-range virtual coupling and are not caused by the presence of isomer mixtures at C-1 or in substitution of the aromatic ring in C-aryl glucuronides. The phenomenon shows sensitivity to substituents at the O-position of C-aryl glucuronides, but this is observed strongly only when the O-positions are unsubstituted. Both solvent and field strength dependences are observed. Changing the solvent from CDCl₃ to other solvents causes a greater chemical shift dispersion, thereby removing virtual coupling effects in these ¹H NMR spectra. By increasing the spectrometer magnetic field, the value of Δν/J becomes sufficiently large to no longer exhibit virtual coupling effects. The relatively high frequency with which this spectral phenomenon is observed in these types of structural units suggest it should be considered when the purity of potential pharmaceutical agents containing these structural units is in doubt based on ¹H NMR analysis.

Support of this work by a grant from the National Cancer Institute (CA49837) is gratefully acknowledged. The 600 and 800 MHz ¹H NMR spectra were recorded by Dr. C.E. Cottrell at The Ohio State University Campus Chemical Instrument Center. We thank Ms. Joan Dandrea for the preparation of this manuscript.

1.   Mehta RG, Barua AB, Olson JA, Moon RC. Effects of retinoid glucuronides on mammary gland development in organ culture. Oncology. 1991;48:505-509.
PubMed  

2.   Hill DL, Grubbs CJ. Retinoids and cancer prevention. Annu Rev Nutr. 1992;12:161-181.
PubMed  

3.   Panigot MJ, Humphries KA, Curley RW, Jr. Preparation of 4-retinamidophenyl- and 4-retinamidobenzyl-C-glycosyl and C-glucuronosyl analogues of the glucuronide of 4-hydroxyphenylretinamide as potential stable cancer chemopreventive agents. J Carbohydr Chem. 1994;13:303-321.
PubMed  

4.   Curley RW, Jr, Abou-Issa H, Panigot MJ, Repa JJ, Clagett-Dame M, Ashafie G. Chemopreventive activities of C-glucuronide/glycoside analogs of retinoid-O-glucuronides against breast cancer development and growth. Anticancer Res. 1996;16:757-764.
 

5.   Abou-Issa HM, Alshafie GA, Wong MF, Clagett-Dame M, Repa JJ, Sikri V, Curley RW, Jr. Chemopreventive activity of a C-glucuronide analog of N-(4-hydroxyphenyl)retinamide-O-glucuronide against mammary tumor development and growth. Anticancer Res. 1999;19:999-1004.
PubMed  

6.   Heyns K, Paulsen H. Selective catalytic oxidation of carbohydrates, employing platinum catalysts. Adv Carbohydr Chem. 1962;17:169-211.
 

7.   Wong MF, Weiss KL, Curley RW, Jr. Recent improvements towards the synthesis of the C-glucuronosyl cancer chemopreventive (β-D-glucopyranosyluronate)-4-retinamidophenylmethane. J Carbohydr Chem. 1996;15:763-768.
 

8.   Musher JI, Corey EJ. Virtual long-range spiN-spin couplings in nuclear magnetic resonance (N.M.R.): The linear 3-spin system and qualitative implications of higher systems. Tetrahedron. 1962;18:791-809.
 

9.   Saito S, Sasaki Y, Furomoto T, Sumita S, Hinomoto T. Virtual 1H-1H spiN-spin coupling in a linear five-spin system on the pyramose rings of some glucuronides. Carbohydr Res. 1994; 258:59-75.
 

10.   Panigot MJ, Curley RW, Jr. Reaction of glycosyl halide with benzyl Grignard reagents: unexpected O-tolyl alkylation of tetra-O-acetylglucopyranosyl bromide and direct synthesis of (β-glycosyl)phenylmethanes. J Carbohydr Chem. 1994;13:293-302.
 

11.   Robarge MJ, Repa JJ, Hanson KK, Seth S, Clagett-Dame M, Abou-Issa H, Curley RW, Jr. N-Linked analogs of retinoid O-glucuronides: potential cancer chemopreventive/chemotherapeutic agents. Bioorg Med Chem Lett. 1994;4:2117-2122.
 

Additional Spectra for Table 2, Entry 1 (5) and Entry 15 (13)

Click below to view spectra and data parameters or download the PDF of just the Appendix at the following URL: http://www.aapsj.org/articles/ps0301/ps030104/ps030104_Appendix.pdf.

¹H phenylglucuronide @250 MHz in CDC13 - (Spectra )- ( Data Parameters)

¹H phenylglucuronide in CDC13 - (Spectra )- ( Data Parameters)

¹H phenylglucuronide in CDC13 @250 MHz - (Spectra )- ( Data Parameters)

¹H phenylglucuronide in CDC13 - (Spectra )- ( Data Parameters)

PHGLUC at 600 MHz 4/27/99 - (Spectra )- ( Data Parameters)

PHGLUC at 600 MHz 4/27/99 - (Spectra )- ( Data Parameters)

PHGLUC 800 MHz - (Spectra )- ( Data Parameters)

PHGLUC 800 MHz - (Spectra )- ( Data Parameters)

¹H azidogluc in CDC13 @ 250 MHz - (Spectra )- ( Data Parameters)

¹H of 1-beta-azidoglucuronide in CDC13 - (Spectra )- ( Data Parameters)

¹H azidogluc @ 250 MHz in DMK-d₆ - (Spectra )- ( Data Parameters)

¹H azidogluc @ 250 MHz in C₆D₆ - (Spectra )- ( Data Parameters)

¹H azidogluc @ 250 MHz in MeOH-d₄ - (Spectra )- ( Data Parameters)

¹H azidogluc @ 250 MHz in pyr-d₅ - (Spectra )- ( Data Parameters)

¹H azidoglucuronide in CD2C12 - (Spectra )- ( Data Parameters)

¹H azidoglucuronide in THF-d8 - (Spectra )- ( Data Parameters)

¹H azidoglucuronide at 250MHz in CDC13 undecoupled - (Spectra )- ( Data Parameters)

¹H azidoglucuronide at 250MHz in CDC13 decoupled at 4.1ppm - (Spectra )- ( Data Parameters)

¹H azidoglucuronide at 250MHz in CDC13 decoupled at 4.9ppm - (Spectra )- ( Data Parameters)

¹H azidogluc in CDC13 at 25 degrees - (Spectra )- ( Data Parameters)

¹H azidogluc in CDC13 at 35 degrees - (Spectra )- ( Data Parameters)

¹H azidogluc in CDC13 at 45 degrees - (Spectra )- ( Data Parameters)