Coupling Constants

The enrichment of oligosaccharides with ‘^C was initially used as a method for verifying '^C chemical shifts, with early work by Walker et al (1976), Gorin and co-workers (1973; 1974; 1975), Bundle et al. (1973), and Koch and Perlin (1970), demonstrating the utility of isotopically labelled compounds in making unambiguous assignments, which had previously been difficult. Interestingly NMR on 50% fully enriched glucose was performed in 1969 (Perlin, 1969), however, no long range carbon-carbon couplings were resolved due to the significant broadening.

.OH

“OH

With relatively easy incorporation of a single label at C l of monosaccharides, by the cyanohydrin reaction, interest in accurate measurement of carbon-carbon couplings was initiated, because, in contiast to the fully labelled monosaccharides, these long range couplings are directly resolvable in ’^C NMR spectra (data reviewed by Krivdin and Della,

1991).

Marshall and Miller showed that three bond C-C-C-C couplings in a variety of aliphatic carboxylic acids follow a modified Karplus relationship (Marshall and Miller, 1973). However, although C-O-C-C couplings are believed to follow a Karplus relationship, no actual parametrisation has been possible, limiting the analysis of inter-glycosidic couplings to a qualitative approach.

Since the introduction of the single ^^C-label, NMR methods of determining have been proposed for natural abundance samples. The earliest of these methods used the INADEQUATE pulse sequence (Bax et a l, 1981), in which the undesirable "parent" signal aiising from molecules containing only a single isolated carbon-13 spin are suppressed, in order to reveal the pure satellite spectrum. The inherent low sensitivity of this method is due to only one molecule in 10'’ having the requisite isotopic composition in natural abundance samples. Transfer of proton magnetisation to caibons via INEPT/DEPT sequences increase sensitivity of the INADEQUATE experiment resulting in a threefold

Chapter 3: Theoretical and Practical Aspects 100

increase in signal to noise ratios (Sorensenef al., 1982; Kessler^/ a l, 1985; Lee and Morris, 1986; Podkorytov, 1990), whilst measurement of carbon-carbon splittings to quaternaiy caibons have been proposed by utilising relayed experiments, such as C-relayed HC INEPT (Kessler et a l, 1985) and DEPT C-C relay (Kawabata and Fukushi, 1994). Keller and Vogele (1986) developed a proton detected variant of the INADEQUATE experiment, allowing measurement of ^Jcc values.

These proposed methods are relatively insensitive and require large quantities of material, which is obviously unfeasible for complex carbohydrates. However, now that fully labelled proteins/peptides aie being produced, there has been a resurgence in interest in ^Jcc- Measurement of these long range cai'bon-carbon coupling constants from ID spectra is

obviously impracticable for proteins, due to the large number of signals and complexity of each resonance by multiple carbon-carbon couplings. Therefore, a variety of new experiments have been proposed for use with uniformly labelled samples, including the 2D and 3D quantitative J correlation method (Bax et a l, 1992; Bax et a l, 1994a; Bax et a l,

1994b), and the 3D HCCC-ECOSY method (Schwalbe et a l, 1993).

Work by the Serriani group has concentrated on the measurement of ^Jcoc couplings, with the prediction of the magnitude of the values due to conformation of the electionegative groups about a given carbon (Serianni et a l, 1996; Church, et a l 1996). This method known as the “sum projection rule” comes from a non Karplus relationship between the magnitude of the coupling constant and angle, and is based on the obseiwations made from the measurements of ^Jch. Unfortunately this only allows information about the angle ({) to be determined, as by definition of the sum projection rule (figure 3.16) the angle \|/ cannot affect the magnitude of the inter-glycosidic ^Jcoc, but Serianni and co-workers have been able to give an insight into the conformation about the glycosidic linkage (Duker and Serianni, 1993), and furanose conformations (Wu et a l, 1992; Church et a l, 1997). This approach, however, is limited by the fact that the sugars are only singly labelled, and are only concerned with ^Jcoc which can only be estimated by an empirical rule. In contrast, measurement of ^Jcocc with an appropriate Karplus curve would give a more practical approach in estimating the torsion angles (]) and \|/.

Chapter 3: Theoretical and Practical Aspects 101

For the general application of inter-glycosidic ^Jcocc values to the conformation of oligosacchaiides, there are two basic requirements. First, a Kaiplus equation is required, and second, fully labelled oligosacchaiides need to be generally available. The latter problem is being addressed with the commercial availability of fully labelled monosaccharides, which can be used to chemically and enzymatically synthesise oligosaccharides. The problem at present is that there are limited Karplus relationships paiametrised for carbon-carbon coupling constants. (C) psi (V) (H) no projection sum phi CX

cx

Figure 3.16 - Sum projection rule for ^Jcoc about oligosaccharide linkages. The sum projection rule involves an inspection of the angles each electronegative group on the C-O-C pathway makes with an axis trans to the 0-C bond (0). The value of the projection sum is then a summation of cos(0) over all contributions. Hence, in the case for t|f, there are no electronegative groups, therefore no projection sum.

Chapter 3: Theoretical and Practical Aspects 102

3.3.3.1 Karplus parametrisation

Parametrisation of a Karplus relationship for use with carbohydrates ideally requires experimental ^Jcc values distributed over a wide range of C-O-C-C dihedral angles. On grounds of sensitivity such coupling constants must be determined for compounds with a high degree of ^^C emichment, and in this regai'd the infoimation that can be obtained from readily available compounds is limited. In monosaccharides only two dihedral angles are available, 180° for C1-0-C5-C6, and 60° for C1-0-C5-C4. The measurement of the latter coupling constant is hampered by a simultaneous and opposite coupling pathway, C1-C2- C3-C4, and the two contributions can not be readily separated resulting in a zero coupling (Krivdin and Della, 1991).

Dihedral angles of +60°/-60°, 180°, and 109° in C-O-C-C fragments can be obtained in two simple sugar derivatives, namely methyl 4,6-0-(l-methylbenzylidene)-a-D-glucopyranoside (I) and 2,3-0-isopropylidene-1,6-anhydro-D-mannopyranose (II), see Table 3.1. Compound I allows measurement of ^Jcc values for +60°, -60° from C4 and C6 to the methyl carton of the methylbenzylidene protecting group, respectively. The value of the dihedral angle, -110°, between C l and the central carbon of the isopropylidene protecting group in compound II, was confirmed by x-ray crystallography (Dr. P. Lightfoot, 1997).

Measurement of the ^Jcc values were obtained from the ID carton spectra with selective carton decoupling. From the data collated on monosaccharides (reviewed by Krivden and Della, 1991) it is apparent that the magnitude of the three bond coupling constant between Cl and C6 is dependent on the anomeric configuration, as is seen for glucose (Table 3.1). It is therefore important to realise that any Karplus parametrisation will only be appropriate to a particular linkage type. Using the data obtained from the spectra, the ^Jcc values listed in Table 3.1 were fitted to a Karplus relation of the form ^Jcc=Acos^0 + Bcos0 + C, giving rise to curves appropriate for a and p linked carbohydrates (figure 3.17). It is appreciated that with the limited experimental data used in fitting, this parametrisation is at best a semi- quantitative, and is probably unreliable in the region O°<0>6O°. However, it is worthy of note that the predicted average ^Jcc value over 360° rotation (2.55Hz) is in reasonable agreement with the measured ^Jcc value for the freely rotating dihedral angle C2-Cl-0-Me in methyl-4,6-0-(l-methylbenzylidene)-a-D-glucopyranoside (Table 3.1). In any event, the

Chapter 3: Theoretical and Practical Aspects 103

experimental data-points lie within the region of the curve that is lilcely to be populated by trans-glycosidic ^Jcc values, and thus should predict these values with reasonable accuracy.

Table 3.1 - Three-bond coupling constants measured in model compounds and utilised for the parametrisation of a Karplus curve for the C-O-C-C fragment.

Compound Coupling Angle (deg.) 'Jcc (Hz)

a-D-Glucose ^Jci-0-C5-C6 180 3.3 p-D-Glucose ^Jci-0-C5-C6 180 3.8 Me ^Jci-0-C5-C6 ^JMe-C-04-C4 180 60 3.1 2.4 3 OhT ' ^lMe-C-06-C6 60 1.9

OMe _{^Jc2-Cl-01-Me} Av/ 2.8

(I)

OH yc ^Jci-C2-02-C7 109^ 0.6

Me Me

(II)

1 Averaging over 360°

Chapter 3: Theoretical and Practical Aspects 104 6- 5 - 4 - 2- 60 30 120 0 90 150 180

Karplus Curve appropriate for a linked carbohydrates Karplus Curve appropriate for p linked carbohydrates Experimental Data Points

Angle (deg)

Figure 3.17 - Karplus Curve for ^Jcocc» parametrised using the data in Table 3.1. Experimental data-points (❖) were fitted to the function ^Jcc = Acos^G + BcosG +C, giving rise to the constants A = 3.9, B = 1.2, and C = 0.6 for a-linked caibohydrates, and A = 4 .4, B = 1.1, and C = 0.5 for P- linked carbohydrates.

Measurement of long range ^Jcocc in ID spectra with selective decoupling in uniformly labelled oligosaccharides is not practical because of the number of carbon-carbon couplings experienced by each carbon. Therefore, measurement of inter-glycosidic ^Jcc values in uniformly labelled oligosaccharides have been measured using the 2D (Bax et aL, 1992) and 3D LRCC experiments (Bax et a l, 1994a,b). The long range coupling constant is derived by a ratio of the volumes of these cross-peaks and the “reference” HSQC-type correlations in the following manner:

V cc _2nT1 ■tan-1

where Vjr is the volume of the long range (Ha,Ca+n) cross-peak, is the reference cross- pealc (Ha,Ca) volume, and T is a delay tuned to evolve "Jcc whilst refocusing ^ Jcc-

Chapter 4

Three Dimensional Structure and Dynamics of the