For decades, the complexity of albumin has been researched extensively, yet many manufacturers and users of the protein have treated it more as a commodity. Because albumin has been readily available, suppliers and purchasers alike have frequently relied on more obvious measures of “purity” and other minimal release criteria to make their decisions. If a lot does not perform well in practice, the typical supplier’s response has been to investigate the manufacturing process for deviations, then correct them if found and make another lot. This is a costly strategy for both manufacturer and purchaser — especially if the “corrected” lot still doesn’t work. So what’s the answer? We can begin with an appreciation of albumin’s heterogeneity, then determine the effects of purification processes on its properties, and correlate those properties to the product application.
Albumin is the most abundant serum protein. It serves several functions in vivo: e.g., binding and transport of fatty acids, hormones, and metal ions; maintenance of osmotic pressure and pH; and binding of exogenous toxins and products of lipid oxidation (1). Over time, development of large-scale purification methods have translated those functions into diagnostic, cell culture, and microbiological applications. It is important to note, however, that purification procedures can promote molecular changes and thereby add to the already complex nature of albumin. That increase in heterogeneity can cause unpredictable performance in vitro. Consequently, it is important for suppliers to understand how the process modifies the end product — and for buyers to consider the potential effects that process-induced modifications can have on the performance of their process. Relying strictly on published product specifications is often insufficient when making a decision to purchase.
PRODUCT FOCUS: ALL BIOLOGICS
PROCESS FOCUS: PRODUCTION
WHO SHOULD READ: MANUFACTURING, PROCESS DEVELOPMENT, QA/QC
KEYWORDS: CONTAMINATION, MEDIA, ANALYTICAL METHODS, TSE, RAW MATERIALS
THE ALBUMIN MOLECULE
Bovine and human albumins are carbohydrate-free monomeric proteins composed of three homologous domains (2,3). Estimated molecular weights are 66,267 and 66,439 Daltons, respectively (3). Their amino acid sequences are ~80% homologous, with bovine albumin composed of 582 amino acid residues and human albumin 585. Each contains 16.3–16.5% nitrogen (3). There are 35 cysteine residues involved in 17 intramolecular disulfide bridges, which contribute to overall conformation and stability (3).
A Cys34 residue is positioned at the solvent interface of each albumin molecule and has a free sulfhydryl (SH) group that is prone to oxidation and mixed disulfide formation. Depending on the age of an albumin preparation (and the method by which it was purified), slight differences have occurred in the SH/albumin molar ratios and extinction coefficients. For example, Janatova found a correlation between the SH/albumin ratio and the extinction coefficient: 6.67 for a freshly prepared albumin and 6.82 for an aged mercaptalbumin preparation with an SH/albumin molar ratio of 0.65 and 0.72, respectively (4). In vivo, the free sulfhydryl group is susceptible to oxidation by cysteine or glutathione, forming mixed disulfide bonds (2).
Fatty acids represent 80% of all the lipids bound to albumin. These bound fatty acids affect its conformation, electrophoretic mobility, and isoelectric point (pI). As a defatted albumin molecule is resaturated with fatty acids, its volume increases, dipole moment decreases, electrophoretic mobility increases, and pI decreases. It is estimated that 1–2 moles of long-chain fatty acids are bound to circulating albumin with a maximum fatty acid/albumin molar ratio of six — the same value that can be achieved in vitro (3). The first and second binding sites have the highest affinity and the greatest influence on the protein structure. However, Chen found the average fatty acid content of several commercially available bovine albumins to be ~0.5 mol/mol (5).
The following fatty acids are normally bound to albumin in vivo and are listed here in descending order of their affinity for albumin: oleate, stearate, linoleate, palmitate, and arachidonate. In addition to fatty acids, other ligands have varying affinity for albumin. Among its significant ligands are calcium, copper, zinc, prostaglandin, cortisol, testosterone, bilirubin, hematin, and thyroxine (3). Given that many of those ligands remain bound to purified albumin, it may be advisable for purchasers to consider measuring the relevant ones to ensure their protein’s optimal performance in specific applications. For example, Spector found that even trace amounts of fatty acids bound to human albumin will affect its ability to interact with drugs, dyes, detergents, and other potential binding ligands (6).
Another feature of albumin is its propensity to polymerize. It is common to find commercially available lyophilized albumins with high–molecular-weight polymers. That polymerization is traceable to specific steps in the manufacturing process that cause protein unfolding: acidic pH conditions (7,8), exposure to a source of energy (e.g., high temperatures (8,9), radiation (9,10), and dehydration caused by an organic solvent (11).
For our purposes, we divide heterogeneity into two categories: “macro” and “micro.” We define macroheterogeneity as the presence of other serum proteins and protein fragments (peptides) in a final albumin product and microheterogeneity as molecular differences. The method chosen to fractionate albumin will dictate the extent of added heterogeneity imposed on its already complex nature. However, detailed analysis that would include information on that added heterogeneity is generally not provided by suppliers or available as standard information. Depending how the albumin is to be used, it may be critical to have such information on nonstandard product attributes: e.g., fatty acid profile and concentration (short to long chain), hormone profile, specific protein or polypeptide contaminants, mole ratio of SH/albumin, and so on.
Macroheterogeneity: Depending on the method used to fractionate albumin, different protein contaminants will be observed in the purified product. For example, measurable amounts of IgG and transferrin are frequently found in cold-ethanol–fractionated albumin. Additionally, because ethanol precipitation is not especially precise, the concentration of those contaminants will vary from lot to lot. IgG and transferrin are present in heat-shock–purified albumin, although in significantly lesser amounts. In addition to such macromolecules, peptides are also potential contaminants.
Microheterogeneity: Because albumin’s conformation will affect its capability to bind ligands, manipulation of the SH/albumin ratio can affect its binding capability. Several studies have shown that free SH groups can either form mixed disulfides with cysteine or glutathione or catalyze internal disulfide rearrangements.
An example of albumin’s heterogeneity with respect to free SH groups is described in an anion-exchange chromatography study of freshly prepared albumin. As the albumin eluted over a linear salt gradient, its fractions ranged from a high of 0.8 to a low of 0.07 mol of free SH/mol albumin (4). The range in free SH comes from the formation of three types of mixed disulfide bonds. Two (those formed with cysteine or glutathione) are found in the native molecule, and the third follows long-term storage of purified albumin. The first two bonds are relatively resistant to reduction. Furthermore, the free SH group is even susceptible to oxidation during dialysis and/or lyophilization.
Intermolecular disulfide bond formation is optimal at pH 3.3 and catalyzed by the divalent copper ion (Cu+2) (7). A study with human albumin showed that the Cu+2 ion will form complexes with albumin monomers, dimers, and tetramers (12). Given this capability of Cu+2 to enhance albumin polymerization, users should be aware of trace amounts of the ion present in albumin and reagents used when working with it. If it is critical to maintain the protein’s native monomeric conformation, then its SH groups can be reacted with l-cysteine to form mixed disulfides and stabilize monomer content. This is particularly important when albumin is used as a chemistry or molecular-weight standard.
Far (170–250 nm) and near (250–350 nm) ultraviolet (UV) spectra provide information on protein secondary and tertiary structures. Peptide bonds absorb light within the far-UV range, whereas SH bonds and aromatic amino acids absorb light in the near-UV range. A far-and near-UV profile correlates to the amount of each chromophore found in a given protein (13,14). Gaber used fluorescence emission at 300–440 nm, UV absorption at 200–300 nm, infrared absorption at 30–1,000 m, and light scattering to report changes in the molecular properties of a dilute solution of gamma-irradiated bovine albumin (15). The study showed that increasing irradiation intensity quenched emission intensity, altering secondary protein structure, reducing molecular weight, and changing the light-scattering properties of the protein. Those data are consistent with alterations caused by irradiation: e.g., protein fragmentation (15).
pH and N-F Transition: Microheterogeneity also can be measured by determining a protein’s isoelectric point and electrophoretic mobility. Each protein has a unique pI, defined as the pH at which its total net charge is zero. Albumin’s pI varies by method of measurement and preparation, but it is normally 4.8–5.3 (7). Different study results appear to be related to three factors: fatty-acid content, protein crystallization, and sulfhydryl content (7). Whether the protein is positively or negatively charged depends on its degree of protonization (loss or gain of protons, H+). Above its pI, the protein’s charge is negative; below the pI, its charge is positive.
Changes in pH can alter a protein’s conformation, resulting in a molecular transition. Foster coined the term “N–F transition” to describe the change in albumin’s electrophoretic migration due to a low-pH molecular transition as either “native” or “fast.” Albumin’s F form migrates ahead of its N form. The difference in mobility is attributable to an expansion of the albumin domains caused by internal molecular charge repulsion forces as pH drops below the protein’s pI. This transition was noted by Foster as the albumin’s pH environment ranged from 4.5 to 3.5. The expansion was observed as N to F and eventually from F to E (a fully expanded molecule). Foster also observed changes in the tertiary structure of albumin in the pH range of 7–9. Those molecular transitions were monitored spectrophotometrically at 313 nm (7).
Foster also reported additional evidence of albumin’s heterogeneity, finding that commercially available albumins did not bind hydrophobic ligands nearly as well as albumin purified from fresh serum (7). In fact, Freeman states that only four purification procedures leave the molecule intact: electrophoresis, (NH4)2SO4 precipitation, gel filtration, and ion-exchange chromatography (16).
What Does It Mean to You?
Years of research clearly demonstrate the complexity of the albumin molecule and begin to explain why performance variability has been observed in commercially available albumins. Native albumin is constructed in such a way as to perform a set of specific functions in vivo. Intended or unintended alterations to the molecular structure caused by manufacturing techniques have the potential to alter the way the protein functions in vitro. Part 2 of this article (March 2010) will examine the major albumin processing methods and how they can affect the heterogeneity of the protein.