Analytical approaches for protein-carbohydrate interaction studies

1.2 Analytical approaches for protein-carbohydrate interaction studies

An essential step preceding the characterization of protein-ligand intermolecular interactions is the primary structure analysis of proteins, which can be performed by mass spectrometry, proteolytic peptide mapping and Edman sequencing [50]. Additional characterization of the secondary and tertiary structure of proteins, especially the determination of disulfide bonds and the identification of post-translational modifications (PTMs), provides important information for the study of protein-ligand complexes. Secondary and tertiary structure analysis of proteins may be performed through various techniques, such as IR spectroscopy [51], circular dichroism spectroscopy (CD) [52, 53], hydrogen-deuterium exchange-mass spectrometry (HDX-MS), tertiary structure-specific chemical modification followed by mass spectrometry [54, 55], crosslinking followed by mass spectrometry, X-ray crystallography and nuclear magnetic resonance (NMR).

After careful characterization of the interacting partners, structural analysis of protein-ligand complexes may be carried out through different methods, each with its advantages and disadvantages in terms of purity and amount of sample required, sensitivity, specificity and speed of analysis [56-58]. These approaches may be divided in: (i) structural methods, e.g. X-ray crystallography [45-47, 59], NMR [13-15]; (ii) biophysical methods, e.g. isothermal titration calorimetry (ITC) [16-18], surface plasmon resonance (SPR) [60-62], quartz crystal microbalance (QCM), surface acoustic waves (SAW) [22, 23]; (iii) biochemical methods, e.g. enzyme-linked immunosorbent assay (ELISA), enzyme-linked lectin assay (ELLA) [63], inhibition of hemagglutination (HIA) [64].

The binding of carbohydrates to proteins, such as lectins, is a most important type of biological interaction, with important functions in cellular recognition processes, intracellular regulation pathways, and immunological reactions [25-29].

The three-dimensional structures of some carbohydrate-binding proteins, free and in complex with carbohydrates, have been solved by X-ray crystallography or nuclear magnetic resonance spectroscopy (NMR). X-ray crystallography is a powerful

technique for the structure analysis of proteins and protein-ligand complexes, capable to provide atomic coordinates of an entire molecular assembly in the solid crystalline state [65]. Crystal complexes with glycans have been defined for a number of plant and animal lectins, bacterial toxins, and enzymes that bind carbohydrates [31, 32]. In order to accurately characterize protein-glycan interactions, well-resolved crystal structures (2-2.5 Å) must be obtained. A sufficient resolution is often difficult or even impossible to attain and, in general, obtaining good quality crystals suitable for structure and interaction analysis of protein-glycan complexes is a tedious and time-consuming process. Crystallization of the sample is a critical step which requires large amounts of high purity protein and the optimization of a wide range of conditions (pH, temperature, salts, protein concentration). Large carbohydrates are often difficult to crystallize and too flexible to yield sufficient electron density. A crystal structure may be obtained for the protein in the absence of the ligand and a potential binding site may be predicted using modeling software and comparison with available structures of homologous proteins [33, 34].

Only a small number of proteins and protein-glycan complexes form crystals that diffract well and yield a crystallographic structure [66]. Furthermore, the conformation adopted by both ligand and protein in a crystal might differ from the conformation preferred in solution. To overcome these problems, structural information on proteins in solutions, under physiological conditions, may be obtained by small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) [67, 68]. SAXS and SANS are powerful albeit low-resolution techniques which may be employed to monitor the influence of various experimental conditions on the tertiary and quaternary structures of proteins [67, 69-71]. This aspect is especially important for carbohydrate binding proteins such as lectins which form oligomers.

Both SAXS and SANS may provide complementary information to X-ray crystallography, require overall smaller amounts of sample and may be applied for the analysis of dilute solutions, without restrictive pH and ionic strength constraints.

Another method for investigating tertiary and quaternary structures of proteins and protein complexes in solution is NMR spectroscopy. In NMR, structural

information may be obtained by use of heteronuclear single quantum coherence (HSQC) and the transferred nuclear Overhauser effect (TRNOE). HSQC involves the transfer of nuclear spin polarization (magnetization) from a proton to a directly-bonded second nucleus (¹⁵N or ¹³C). The magnetization is then transferred back to the proton for detection. The transfer occurs through J-coupling (through-bond dipole-dipole coupling). The 2D ¹H-^l5N HSQC spectrum of a protein correlates the ¹H chemical shift of the proton attached to the backbone amide nitrogen of each amino acid with the ¹⁵N chemical shift of the nitrogen. Ligand binding causes perturbations in the peaks' positions and by comparing the free protein and protein-ligand complex spectra the ligand-contacting amino acids may be identified [40, 41]. NOE is a distance-dependent (~r^-6 [72]), through-space transfer of nuclear spin polarization between protons in close contact (r < 5 Å [73]). Intramolecular NOEs, which arise in the bound conformation of a ligand, are transferred to the free ligand in solution through chemical exchange. TRNOE measurements allow the determination of proton-proton distances within a bound ligand, thus providing information on the bound-state conformation of the ligand [74-76]. TRNOE measurements are applicable to ligands that are weakly bound (KD > μM) and exchange with the free ligand faster than the cross-relaxation rate. For NOE measurements, like for X-Ray crystallography, the flexibility of the carbohydrates causes problems, e.g. if multiple conformations of a molecule co-exist in equilibrium in solution, the measured NOE intensities will yield time-averaged values of the existing conformations. The large excess of ligand required to maximize the number of exchange events may lead to nonequilibrium conditions and non-specific binding [77]. Intermolecular TRNOE cross-peaks, arising from magnetization transfer between protein protons and bound ligand protons, may be used to for mapping intermolecular contact sites [43, 48]. In addition to the structural information, NMR may also provide equilibrium constants by monitoring the chemical shift variation induced by increasing amounts of ligand (1D ¹H NMR titration) [49, 50].

Kinetic and thermodynamic data may also be obtained using ITC, QCM, SPR, and SAW. ITC evaluates the change in free energy resulting from binding of a glycan to a lectin [78]. Its advantages are the ability to monitor interactions without the need

for immobilization or chemical modification of the binding partners. Drawbacks of ITC include the requirement of high sample amounts (> 10 mg), solubility problems when low-affinity interactions are studied as well as the limited accuracy with which temperature changes may be determined. In ITC, small amounts of glycan are added to the protein solution at regular time intervals and the heat exchanged due to the complex formation is recorded and plotted as power (μcal∙s^-1) versus time (s). The data is integrated with respect to time yielding the titration curve, which represents the change in enthalpy (ΔH, Kcal∙mol^-1) as a function of molar ration [64]. This curve is then fitted to a theoretical binding model to yield the binding constant (K), binding enthalpy and interaction stoichiometry. The change in free energy and entropy may be determined from the Gibbs equation.

GHTSRTlnK (1)

Biosensor measurements (QCM, SPR, SAW) require the immobilization of one of the interaction partners on the surface of the sensor chip while the other one is passed in solution over the chip surface. The response is recorded as a function of time and the shape of the signal is independent of the biosensor type (Figure 5). The injection of analyte is associated with a signal increase due to the formation of the intermolecular complex (association curve). At the end of the analyte injection the biosensor signal decreases, due to the dissociation of the protein-ligand complex. In SPR the recorded response represents the change in the refractive index of the chip surface, probed with a laser beam, while QCM sensors exploit the piezoelectric effect of a quartz chip to which an alternating current is applied and measures the change in frequency of the quartz crystal resonator as the analyte binds to the immobilized molecule. The SAW method employed in the present work is also based on the piezoelectric effect of a quartz chip that enables the conversion of electronic signals in mechanical acoustic waves (called Love waves) [79]. The wave changes both in phase and amplitude in response to the binding of the analyte.

Figure 5. Typical sensorgram for several types of biosensors. The association curve (in blue) represents the signal increase due to the binding of the analyte in the mobile phase to the interaction partner immobilized on the biosensor chip. At the end of the injection the analyte flow is replaced by a buffer flow and the signal decreases due to the dissociation of the complex (dissociation curve, in red).

The main drawback of biomolecular interaction analyses is their inability to provide chemical structure information about the interacting biomolecules. A common problem of all biosensor methods employed in the study of protein-carbohydrate