Applied Photophysics, Ltd.

In this study, far-UV CD spectra of eight different mammalian serum albumins were measured repeatedly using automated CD spectroscopy. Two independent methods of normalizing the CD data were used to eliminate the need for accurate knowledge of protein concentration or extinction coefficient. The normalized far-UV data, representative of secondary structure, were compared to determine if there were statistically significant differences between samples. The two normalization methods agreed in every case, increasing confidence in the results. 

The eight proteins (~1 mg/mL) were dissolved in buffer (20 mM phosphate, 150 mM NaF, pH 7.4). Aliquots of sample and buffer were pipetted into a microtiter plate. Three independent sets were loaded permitting triplicate measurements and the determination of standard deviations. Samples were transferred using a fixedprobe robot from the microplate to a 0.1 mm pathlength flow-cell. CD and absorption spectra were measured simultaneously and four repeats for each of the 36 wells used were recorded. The experiment ran unattended overnight, using a Chirascan-plus automated CD spectrometer. CD and absorption data were processed by automated averaging and then subtraction of buffer baselines. Averaging across repeat samples gave standard deviations which were propagated through all subsequent calculations. 

Figure 1: Difference-CD and difference G-factor spectra, hamster SA vs. human SA.

Human serum albumin (HSA) was chosen as the reference protein to which the seven others were compared. CD spectra were normalized to that of HSA using the sum of the absolute CD values across all wavelengths; a second method of normalization involved calculating the G-factor spectra (1). Difference-CD and difference-G-factor spectra were plotted with their associated two-sigma error bars: representative spectra from the set are shown in Figure 1. Where there is systematic excursion from zero and in particular outside the 2-sigma (95% confidence) limits, a significant difference in secondary structure is indicated. Confidence in these results is high because the control experiment shows no systematic excursion and because the independent methods of normalization agreed in every case. 

Figure 2: Control experiment - difference-CD and difference Gfactor spectra, human SA (0.2 mm cell) vs. human SA (0.1 mm cell).

This model study shows that it is relatively straightforward to identify very subtle differences in protein conformation without accurate knowledge of protein concentration or extinction coefficient. Establishing whether or not the higher structures of proteins are similar is becoming increasingly important with the advent of generic biotherapeutics. The automation of CD spectroscopy opens up applications that were previously inaccessible and, in this example, offers a critical insight into subtle conformational differences in similar proteins. 

 21 Mole Business Park, Leatherhead, Surrey, United Kingdom

tel. +44 (0)1372 386537; Fax +44 (0)1372 386477

Articles

In this study, far-UV CD spectra of eight different mammalian serum albumins were measured repeatedly using automated CD spectroscopy. Two independent methods of normalizing the CD data were used to eliminate the need for accurate knowledge of protein concentration or extinction coefficient. The normalized far-UV data, representative of secondary structure, were compared to determine if there were statistically significant differences between samples. The two normalization methods agreed in every case, increasing confidence in the results.

Comparison of Secondary Structures of Similar Proteins Using Automated CD

This study describes a new automated technique for measuring protein conformation by CD spectroscopy and how changes can be followed in different pH environments. The study confirms the robustness of the technique and superb sample-to-sample reproducibility. The improvements in productivity enable experiments to be undertaken that were previously thought to be too time consuming or laborious.

Correct protein function is critically dependent upon conformation — if a protein is not correctly folded, it will not execute its function or worse, it may be toxic. Circular dichroism spectroscopy (CD) is an exquisitely sensitive probe of protein conformation, but it is often perceived as a technique limited to more specialized applications. This is to some extent because, up until now, CD spectrometers have remained resolutely manual in operation which has now been overcome by new automation technology. 

Solutions of BSA, sodium phosphate, and sodium citrate were used to prepare a series of buffered protein samples from pH 2.2 to pH 8. The samples were prepared directly in a 96-well plate and each protein solution was paired with its buffer. Buffers and samples were transferred to the spectrometer using a fixed-probe x, y, z robot, and CD and absorption spectra measured simultaneously and in duplicate. Each complete measurement, including washing, took 7.5 min to perform. 

Each protein CD and absorption spectrum was baseline-corrected using its paired buffer (Figure 1). 

Figure 1: Buffer-subtracted CD and absorption spectra for BSA (260 nm â¥ Î» â¥ 188 nm) from pH 2.2 to pH 8.			

Plotting the raw data as a contour map gives an immediate semi-quantitative feel for the stability of the protein. Using quantitative methods, such as the global fitting of a model to the data, make it clear even to the untrained eye where regions of conformational stability exist (Figure 2). A two-state model was used to describe the data and although it is known that this model may be inadequate to describe the behavior of serum albumins completely (1), it is good enough to show the potential of the method. End-point spectra for high- and low-pH were calculated as was the pKa of the transition between them. 

Figure 2: Calculated conformational stability map (CD contours). The loss of structure below pH 4.2 indicates clearly that the protein conformation changes at low pH.			

The automated sample preparation and measurement of CD and absorption spectra have been shown to be reliable, giving quantitative results with minimum effort using robust global analysis techniques. Excellent reproducibility and sample economy together with the ability to carry out an unattended measurement at a rate of 192 samples per day considerably improves the productivity of CD measurements. These improvements offer the opportunity to study relative protein stabilities with ease, for example, in protein biotherapeutic development and clone selection procedures. 

(1) D.C. Carter and J.X. Ho, "Structure of Serum Albumin" 

, Lipoproteins, Apolipoproteins and Lipases, Academic Press 1994, Ed. Verne Schumaker. ISBN: 0-12-034245-6. p. 153–203. 

201-205 Kingston Road, Leatherhead, Surrey, United Kingdom 

tel. +44 1372 386537; fax +44 1372 386477 

This study describes a new automated technique for measuring protein conformation by CD spectroscopy and how changes can be followed in different pH environments.

Applied Photophysics, Ltd.

Articles

Comparison of Secondary Structures of Similar Proteins Using Automated CD

Mapping Protein Conformational Stability Using Automated CD