In the development of an active pharmaceutical ingredient (API), it is important to design a formulation that provides adequate stability over long term storage using the recommended handling conditions. A liquid formulation is an easy and economical way to handle the product during manufacture and is very convenient for the end user. However many proteins prove difficult to formulate in a stable solution and often require lyophilisation. Recombinant Human albumin (Recombumin) has been developed to aid in the development of liquid formulations for unstable API's.
Recombumin is produced by Novozymes Biopharma UK Ltd, and is a Saccharomyces cerevisiae yeast- derived protein. This is structurally identical to the human serum albumin (HSA). Recombumin itself is formulated at pH 7 and has been shown to have a shelf life of greater than 5 years at 5°C .
This application note studies the stability of Recombumin over the range of pH conditions which API's may be formulated, and compares the results obtained by dynamic light scattering (DLS) with size exclusion chromatography (SEC) data kindly provided by Novozymes Biopharma UK.
This is one of several application notes that use Recombumin or Albucult, both recombinant proteins, to explore how light scattering techniques can be used to evaluate formulation stability [2, 3, 4].
Recombumin and Albucult samples were kindly provided by Novozymes Biopharma UK.
The Recombumin sample was diluted from its 20% stock solution to 10mg/ml in the following 25mM ionic strength buffers; citrate pH 3, 4, 5, and 6, phosphate pH 6, 7, and 8, and borate pH 8, 9, and 10.
The ionic strength of each sample was adjusted with sodium chloride to be approximately equivalent to 0.9% w/v sodium chloride. The samples had been sterile filtered in 2mL aliquots and stored at 5°C for about 6 months.
SEC data were provided by Novozymes Biopharma UK. DLS measurements were made on a Zetasizer Nano ZS using a detection angle of 173°. All measurements in this study were taken at a temperature of 25°C with at least 3 repeat measurements on each sample taken to check for result repeatability.
The SEC results are shown in figure 1. The neutral and alkaline samples all exhibit the presence of very low percentages of dimers, less than 2%. For the samples within the acidic formulated pH range, the results showed an increase in the proportion of oligomer species; here it is % dimer that is displayed. The acidic pH sample range demonstrated a lower sample recovery from the column compared with the alkaline samples.
Other aliquots of the same pH samples as measured by SEC were measured with DLS at a similar time point. The results display a very similar trend, with an increase of Z-average size at acidic pH, figure 2.
The Z-average size derived from the cumulants analysis  is the intensity weighted mean size of the sample. As the amount of light a molecule scatters is proportional to its diameter to the power 6 (I α d6), this dependence on size is the basis of the DLS technique's inherent sensitivity to aggregates, and the reason for its ability to detect the presence of only a few aggregate precursors in a sample.
The Z-average size shows a low and reasonably consistent value for the samples in the pH range of pH 6 to pH 9, with a hydrodynamic radius of 3.6nm, indicating that these samples are more stable than at the outer values of the pH range tested. The presence of oligomers and/or aggregates in the acidic formulated samples results in increased Z-average values, although the technique cannot quantify the proportion of oligomers or aggregates present.
Molecular weight can be converted into hydrodynamic size from well established relationships . The hydrodynamic size expected for a globular protein of 66kDa such as albumin is 3.57nm in radius which is within the range of sizes obtained from the measurements in this study.
The cumulants analysis also calculates a second parameter called the polydispersity index (PdI) which is influenced by the width of the size distribution of the sample. The PdI in the Zetasizer software ranges in value between 0 and 1, and typically for a monodisperse sample is below 0.05. Samples with a PdI of 0.08 often show only 1 peak in the distribution, although the peak has an increased width compared to monodisperse samples, indicating the presence of small oligomers/aggregates. %Polydispersity (%Pd) is sometimes reported in the literature, which is derived from the PdI according to the following equation:
%Pd = (PdI)0.5 x 100
This results in %Pd values of 28% for a PdI of 0.08 and 23% for a PdI of 0.05.
In the case of the Recombumin samples, the measured PdI values are shown in Figure 3. From this figure it is clear that the samples above pH 5 all have %Pd values of 28%Pd or less, which indicates that they are mainly monodisperse showing the presence of only a small proportion of oligomers. The acidic formulation of sample shows the presence of larger amounts of oligomers and aggregates as the %Pd values are >37%.
It is known that albumin undergoes several transitions that are pH dependent, forming oligomers . As the pH becomes more acidic the protein unfolds to the maximum extent allowed by its disulphide bonding structure. In this extended, unfolded state it is likely that hydrophobic areas of the protein once buried within the protein structure are exposed at the surface, and when these hydrophobic areas are in contact, they form aggregates. The DLS data of acidic formulated samples display the presence of aggregates and/or oligomers.
As the pH increases towards the alkaline region, it is known that albumin tends to undergo a transition which causes the molecule to unfold, and this flexibility may cause weak albumin dimers to dissociate so the albumin is likely to be more monomeric at these pH values . This is supported by both the SEC data and the DLS data as the dimer content detected by SEC is low and the Z-average and polydispersity of the samples are within the expected range for monodisperse albumin.
If figure 2 and 3 are compared, it is possible to see that the variations in size are most likely due to slight differences in the sample polydispersity as the Z-average and polydispersity follow a similar pattern. If these data are compared to the SEC data in figure 1, it is clear that DLS can detect subtle changes in the sample composition before the UV or RI detectors of a chromatography system.
In this application note, data was presented from measurements of Recombumin a recombinant human serum albumin, which was prepared in different pH buffers. The aim was to monitor the protein stability over the range of pH where API formulation typically occurs.
These data show how relatively fast measurements by DLS, about 2 minutes per sample, can provide a lot of information about the relative composition and stability of protein samples. The pre-screening by DLS can enable the user to select samples of interest before performing more time consuming measurements such as SEC and thereby minimize analysis time.
The thermal stability of these samples is explored in another application note .
 "Investigating the stability of recombinant human serum albumin under a range of pH conditions with both time and temperature"
Poster presentation by Mr Karl Nicholls, May 2010, BPI Europe 2010, Vienna - Austria
 MRK1617 Thermal stability Recombumin app note
 MRK1614 App note high conc protein data
 International Organization for Standardization, 1996, ISO13321, Methods for determination of particle size distribution part 8: photon correlation spectroscopy
 P Claes, A Kennedy and P Vardy "An on-line dynamic light scattering instrument for macromolecular characterization" in "Laser light scattering in biochemistry" SE Harding, DB Sattelle and VA Bloomfield (eds) pp 66-76 RSC (1992)