By Nestor B. Nestor and George A. Karam, Department of Exploratory Medicinal Sciences, Pfizer Central Research, Groton, CT

The molecular weight of biological macromolecules is critical to understanding the structure, function, and regulation of such molecules; it is also critical in the characterization of protein reagents and therapeutics. Improvements in mass spectrometry technologies in recent years have allowed accurate characterization of the covalent structure of a protein, but its aggregation state is just as critical to understanding the molecule's function.

Many proteins form oligomers in solution, and such quaternary structure is important to its function. Some proteins also tend to form "nonspecific" aggregates in solution under certain conditions, which can have deleterious effects on functional activity. Commonly used techniques for the determination of aggregation in solution, such as size-exclusion chromatography or analytical ultracentrifugation, are either indirect, time-consuming, or lack automation. Light scattering is a rapid way to measure solution molecular weight directly, with reasonable quantities of protein and in an automated fashion.

Calculating Solution Molecular Weight
The high sensitivity and excellent signal-to-noise ratio of the Agilent 1100 Series fluorescence detector (FLD) has allowed us to measure the intensity of Rayleigh scattering (scattering of light the same wavelength as the incident beam) and to calculate the solution molecular weight of small proteins and their aggregates. The intensity of scattered light (I_LS) is proportional to the product of the concentration of the scattering species (C) and their molecular weight (Mw)¹:

I_LS C^.Mw

Setting the FLD to the same excitation and emission wavelengths allows the Rayleigh light scattering intensity to be measured. A value in the high visible is usually chosen, because the protein must not absorb at the wavelength selected. If the extinction coefficient () of the protein is known or can be predicted,² the concentration can be determined using the diode array to measure A₂₈₀. If such information is not known, the Agilent 1047 refractive index detector can be used. The proportionality of the above relationship is defined by system constants such as the wavelength used, the detector responses, the buffer system used, and the chemical nature of the scattering species.¹

If non-glycosylated proteins are to be studied, a calibration curve can be constructed using commercially available standard proteins.³ It is also important to note that determining molecular weight from I_LS measured at one angle (90 in the FLD) is accurate only for proteins and biopolymers that have a conformation with a radius ~10 nanometers. For molecules of this size, scattering intensity becomes angle-dependent, and values measured at more than one angle are needed to allow data to be extrapolated to zero angle.⁴

Separation by Size-Exclusion Chromatography
Rayleigh light scattering can be carried out in a batch fashion, in which the concentration and scattering intensities of the sample are read in cuvettes. However, the technique is best employed on-line after separation of the sample by size-exclusion chromatography, primarily for two reasons. First, smaller species are separated from high Mw aggregates or contaminants such as dust. Because light scattering is extremely sensitive to large particles (see Figure 1), the presence of any large species (even in very small amounts) can skew the results. Second, the sample is separated from any unknown or unmeasurable buffer components (such as salts that were dried down with a lyophilized sample or oxidized DTT) and measured in the mobile phase of known composition.

Figure 1. The ability of light scattering to determine molecular weight can be qualitatively illustrated by comparing the FLD signal peak height (which is the intensity of light scattered, or ILS) and concentration detector peak heights of different proteins. A mixture of thyroglobin, IgG, ovalbumin, and myoglobin was separated by size-exclusion chromatography. The concentration of the eluting proteins is represented by the A280 (blue trace). Light scattering was carried out using the Agilent 1100 Series FLD set to Ex = 633 nm and Em = 633 nm (red trace). The FLD signal was aligned in time with the A280 signal to adjust for the delay volume between the detectors. The ratio of ILS to A280 decreases with elution time, implying that the larger molecules are eluting first (as is expected with SEC). The sensitivity of light scattering to large species is also clearly in the region of the chromatogram between exclusion limit of the column (approximately 10.5 min) and the first A280 peak (approximately 11.9 min). There is a significant increase in ILS, but only a small increase in the A280, indicating the presence of very large aggregates.

The authors (N. B. Nestor on the left) in their Laboratory at Pfizer.

The marriage of these two techniques also allows some additional insights into the characteristics of the protein. Comparisons of the absolute molecular weight measurement from light scattering with the radius determined from size-exclusion chromatography can provide insights into shape, such as whether the molecule or aggregate is globular or more extended in conformation.⁵ Front-end separation of the sample can also be accomplished with other forms of chromatography.^6,7

The Analytical Strategy
Our system is comprised of an Agilent 1100 Series liquid chromatograph, including degasser, autosampler, UV-Vis diode-array detector (DAD), FLD, and an Agilent 1047 refractive index (RI) detector. To determine I_LS, the FLD was set at 633 nm for both excitation and emission. Concentration was determined from the product of the A₂₈₀ and the protein's extinction coefficient (expressed in mg/mL/A₂₈₀) as predicted by sequence or from the RI. A YMC-Pack Diol-200 column was used for the separation of aggregates, with phosphate-buffered saline (PBS) as the mobile phase. Noise in the light scattering baseline is often caused by even small amounts of column shedding, so flow rates were adjusted gradually, and the system was allowed to run (recycling the buffer into the reservoir) during the night before the experiments were performed.

We have constructed a calibration curve using ribonuclease A (RNAse A, see Figure 2a), bovine serum albumin monomer (BSA, Figure 2b), and aldolase (Figure 2c). Small amounts of dimer can be detected in the RNase A and BSA and in some heterogeneous material in the aldolase that eluted later than the main peak. We have chosen to use peak height for calculation, but peak area may also be used. The system calibration constant was determined from the slope obtained by plotting the protein's sequence (or mass spectrometry) molecular weights versus I_LS/C (Figure 3). Separate constants were determined for using A₂₈₀ or RI as the concentration detector. The solution Mw, and therefore the aggregation state, of novel proteins can then be calculated by dividing the I_LS/C by the system calibration constant.

Figure 2. RNAse A (a), BSA (b), and aldolase (c) were individually run on size-exclusion chromatography to separate any aggregation states. The concentration of the eluting material was determined from the A280 signal (blue trace) peak height and the protein's extinction coeffecient (expressed in mg/mL/A280). Light scattering data (red trace) was collected as described in Figure 1 and used to construct a calibration curve (Figure 3). The dimer present in the BSA (b) was also used in the calibration.

Figure 3. Calibration curves showing a plot of the molecular weight versus ILS/C for RNAse A (Mw = 13,690), BSA (Mw = 66,432, BSA dimer (Mw = 132,864), and aldolase (Mw = 156,843). Concentration was determined either from the A280 peak height and the protein's extinction coefficient (blue trace) or from the RI peak height (red trace). The solution Mw of proteins can be determined by simply dividing their ILS/C by the slope of the above line (0.2423 if A280 is used or 1.8695 if RI is used).

The method presented here is linear over the molecular weight range of proteins typically under study in most biotechnology and molecular biology laboratories—from peptides to antibodies. The amount of sample required is similar to that of SEC (typically 10 to 50 micrograms), and the analysis time is the same (approximately 30 minutes), with nominal time required for data analysis. This direct method for determining solution Mw offers increased accuracy and/or speed over most methods, using a detector that is multifunctional and cost-effective.

References
1. Wyatt, P. J. Anal. Chim. Acta 272, 1-40 (1993).
2. Pace, C. N., Vajdos, F., Fee, L., Grimsley, G., and Gray, T. Protein Science 4,
2411-23 (1995).
3. Wen, J., Arakawa, T., and Philo, J.S. Anal. Biochem. 240,155-166 (1996)
4. Wyatt, P. J. J. Chrom. 648, 27-32 (1993).
5. Potschka, M. Anal. Biochem. 162, 47-64 (1987).
6. Mhatre, R., Krull, I. S., and Stuting, H. H. J. Chrom. 502, 21-46 (1990).
7. Mhatre, R. M. and Krull, I. S. J. Chrom. 591, 139-148 (1992).