Albumin: Biochemical properties and therapeutic potential

Human serum albumin (HSA) is an abundant multifunctional non-glycosylated, negatively charged plasma protein, with ascribed ligand-binding and transport properties, antioxidant functions, and enzymatic activities.1 It is synthesized primarily in the liver and is thought to be a negative acute-phase protein. Physiologically, albumin is responsible for maintaining colloid osmotic pressure and may influence microvascular integrity and aspects of the inflammatory pathway, including neutrophil adhesion and the activity of cell signaling moieties. Clinically, albumin has been employed as a plasma expander in many patient populations, although the evidence from meta analyses2, 3 and the recently published SAFE investigation4 suggests it does not afford a survival benefit over crystalloid solutions when administered to the critically ill. However, studies of albumin usage as a volume expander and albumin dialysis therapy in patients with liver disease have led to some encouraging results. This review aims to highlight current thinking regarding albumin therapy in the critical care and hepatological setting and also discusses other potential therapeutic applications for its use based around the complex biochemistry of this multifunctional plasma protein. Potential contraindications are also discussed. HSA, human serum albumin; NO, nitric oxide; ROS, reactive oxygen species; RNS, reactive nitrogen species; COP, colloid oncotic pressure. Albumin normally accounts for over 50% of total plasma protein content, being present at concentrations of approximately 0.6 mmol/L. HSA is a small (66 kd) globular protein composed of 585 amino acids, with few tryptophan or methionine residues but an abundance of charged residues such as lysine, and aspartic acids and no prosthetic groups or carbohydrate. X-ray crystallography has shown albumin to possess a heart-shaped tertiary structure, but in solution HSA is ellipsoid. Some 67% of the tertiary structure of HSA is composed of α-helices. Indeed, the protein is composed of 3 homologous domains (I-III), each containing two sub-domains (A and B) composed of 4 and 6 α-helices respectively. The sub-domains move relative to one another by means of flexible loops provided by proline residues, which helps accommodate the binding of an array of substances, as does the flexibility provided by domain-linking disulfide bridges. Figure 1 depicts the tertiary structure with bound fatty acids. HSA contains 35 cysteine residues, most of which form disulfide bridges (17 in all), contributing to overall tertiary structure. However, it also contains 1 free cysteine-derived, redox active, thiol (-SH) group (Cys-34), which accounts for 80% (500 μmol/L) thiols in plasma. The thiol moiety of Cys-34 is reactive and capable of thiolation (HSA-S-R) and nitrosylation (HSA-S-NO), processes that are thought to contribute to several in vivo functions. Tertiary structure of albumin showing the binding of seven archidonic acid ligands is depicted. Illustration obtained from the RCSB protein data bank PDB ID:1gnj by David S Goodsell Scripps Research Institute. Primary reference source: Petitpas I, Gruene T, Bhattacharya AA, Curry S. Crystal structures of human serum albumin complexed with monounsaturated and polyunsaturated fatty acids. J Mol Biol 2001;314:955. Physiologically, HSA exists predominantly in a reduced form (that is, with a free thiol, HSA-SH) and is known as mercaptoalbumin. However, a small but significant proportion of the albumin pool exists as mixed disulfides (HSA-S-S-R); where (R) represents low-molecular-weight, thiol-containing substances in plasma, chiefly cysteine and glutathione. Mixed disulfide formation also increases as part of the aging process (reviewed by Droge5) and during disease processes characterized by oxidative stress. Dimer formation is also theoretically possible (HSA-S-S-ASH), but in practice is unlikely to occur in vivo because of stearic interference. However, this process is known to occur ex vivo on purification and storage and therefore may have implications related to certain aspects of the therapeutic use of albumin. HSA binds many endogenous and exogenous compounds, including fatty acids, metal ions, pharmaceuticals, and metabolites, with implications for drug delivery and efficacy, detoxification, and antioxidant protection. Several low- and high-affinity ligand-binding sites have been identified on HSA, the first of which to be identified (termed site I and II) are responsible for the binding of most pharmaceuticals that interact with the protein. Sites I and II are located in different domains and exhibit differential, but not always exclusive, ligand-binding affinities. Site I tends to bind relatively large heterocyclic compounds or dicarboxylic acids. A diverse array of unrelated compounds bind with high affinity to various locations within this site, indicating adaptability. Moreover, this site is large and able to bind bulky endogenous substances, including bilirubin and porphyrins. By contrast, site II (also known as the indole-benzodiazepine site) is smaller and less flexible in nature, because binding is more stereo-specific. Importantly, additional high-affinity binding areas are present within HSA for some drugs and compounds that do not conform to either site I or II. Furthermore, the binding domains of some substances such as digitoxin and the bile acids remains to be elucidated; for a concise review on ligand-binding, see Kragh-Hansen et al.6 Cysteine-34 binds drugs including cisplatin, D-penicillamine, and N-acetyl-cysteine.6 Covalent interactions (thiolations) also occur with endogenous, low-molecular-weight, thiol-containing substances via disulphide bridge formation. Higher oxidation states of cys-34 can also occur, resulting in the formation of sulfenic, sulfinic, or sulfonic acid residues (Fig. 2), although levels seen in normal plasma (bovine) are low.7 Both endogenous and exogenous nitric oxide (NO) are known to interact with cys-34 via electrophilic addition of the nitrosonium ion (NO+). Indeed, until recently NO was thought to circulate in plasma primarily as an S-nitroso HSA adduct and to possess vasodilatory properties, augmented by NO transfer to low-molecular-weight thiols.8 However, recent in vitro and in vivo studies indicate that levels of s-nitroso-albumin that form under biologically relevant conditions in normal plasma are in the low nanomolar range (<10 nmol/L) and that several other reaction products of NO contribute to the NO plasma sink.9 It is less clear to what extent HSA contributes to NO binding in vivo under pathological conditions, or to what extent the availability of catalysts and or other NO-derived species impacts on s-nitrosolation. Furthermore, recent studies using a targeted s-nitrosoglutathione reductase murine model have demonstrated the importance of nitroso-thiol turnover in endotoxic shock.10 Further studies are required to determine HSA's role under such circumstances. Key reactions are summarized in Fig. 3. Scheme gives an overview of the steps involved (highlighted in blue) for the nitrosylation of Cys-34 of human serum albumin (HSA). Nitric oxide (NO) requires an electron accepting catalyst (reactive transition metal ion, or metal-containing proteins) to favor such reactions. Steps leading to Cys-34 oxidation and thiolation are highlighted in red. Formation of higher oxidation states of HSA are also shown. RSH, glutathione or free cysteine; Alb, albumin. Scheme depicts antioxidant (highlighted in blue) and the pro-oxidant potential (highlighted in red) of human serum albumin (HSA). The potential of iron and copper ions to catalyze the formation of the extremely aggressive and damaging hydroxyl radical · OH (the Fenton reaction) is shown. The potential ability of HSA to redox cycle these metal catalysts exacerbates this pro-oxidant response when these metals have access to Cys-34, in other words, when they are not bound at protected sites on this protein or elsewhere. As shown, such metal salts also can propagate membrane lipid peroxidation directly if stable lipid peroxides are already present. Nitric oxide and bilirubin binding may provide an indirect (supportive) antioxidant response attributable to albumin, as both compounds have reported lipid-phase antioxidant function. The N-terminal portion of HSA (N-Asp-Ala-His-Lys-) binds Cu, Ni, and Co ions with high affinity, whereas Au, Ag, and Hg ions bind to cysteine-34 (reviewed in Kragh-Hansen et al.6) HSA is also the major Zn binding protein in plasma, although there is some debate as to the nature of and location of its binding site.11 HSA has also been reported to possess a relatively weak, nonspecific, latent iron-binding capacity.12 This is, however, unlikely to be of significance under normal circumstances in plasma, because the specific, high-affinity, iron-binding protein transferrin binds all low-molecular-mass ferric iron. Aerobic metabolism is energy efficient. However, whereas oxygen-containing end products of these processes are relatively innocuous, many intermediates thereby formed are potentially, or directly, extremely reactive in nature. Such reactive oxygen species (ROS) can inflict damage on molecules, leading to the accumulation of toxic end products and cellular dysfunction or damage. Normally, the body uses protective (i.e., antioxidant) and reparative systems that limit the effects of oxidative stress. An antioxidant is any substance that when present at low levels significantly diminishes or prevents the oxidation of an oxidizable substrate, and may be dietary, constitutive, or inducible in origin. Primary antioxidants prevent ROS formation and include the iron-binding antioxidant transferrin. Secondary antioxidants scavenge pre-formed ROS. Examples include ascorbate and superoxide dismutase. For the definitive text on ROS in biology, see Gutteridge and Halliwell.13 Reactive nitrogen species (RNS) are nitrogen-centered species analogous to ROS. Evidence indicates that such species are formed in vivo; some, such as nitric oxide, contribute to various biological signaling responses. Others, however, are powerful oxidants and nitrating species capable of damaging biomolecules; antioxidant protection also limits the damage inflicted by RNS. Several such antioxidant functions have been ascribed to HSA. HSA in plasma, or bovine serum albumin in artificial systems, provides protection from lipid peroxidation propagated by inorganic ROS generated from xanthine oxidase/hypoxanthine.14 However, thiol oxidation occurs, indicating the cys-34 moiety to be the source of the antioxidant protection afforded. In more recent studies, hydrogen peroxide (H2O2) and the RNS peroxynitrite (ONOO−) have been shown to oxidize cys-34 to a sulfenic acid derivative (HSA-SOH).15 This is subsequently converted to a disulfide with the potential to be redox cycled to mercapto-albumin (HSA-SH), thereby restoring antioxidant function (Fig. 2). Increased ROS and RNS formation have been implicated as contributory factors in the onset and progression of critical illness.16 Albumin may provide effective extracellular scavenging antioxidant protection under such circumstances. Thus, albumin supplementation has been shown to replenish extracellular thiol status in patients with sepsis by means other than that which would be expected on purely stoichiometric grounds.17 Moreover, such supplementation was shown to improve thiol-dependent antioxidant protection in plasma obtained from patients with acute lung injury and to be associated with decreased levels of oxidative markers (protein carbonyls),18 although there was no difference in survival rates between groups. Persistent hypoalbuminemia is also associated with peroxidation of erythrocyte membranes in patients undergoing chronic hemodialysis, indicating that HSA protects against lipid oxidation.19 In vitro studies have shown that bovine serum albumin scavenges neutrophil-derived ROS, including hydrogen peroxide, superoxide, and hypochlorous acid.20 Inflammatory cell-derived oxidants contribute to oxidative stress during acute inflammation and the consequences thereof. HSA could potentially reduce such effects through scavenging antioxidant actions in humans, which may, also through modifying redox balance, regulate cell signaling moieties active in mediating pro-inflammatory forces (Fig. 3). In vitro, albumin has been shown to offer antioxidant protection against the oxidative effects of carbon tetrachloride and uremic toxins,21, 22 findings with implications for both hepatic and renal failure. HSA may provide a supportive antioxidant role in vivo, through its ability to bind and transport substances with known antioxidant function, specifically, bilirubin and NO, which are effective lipid phase antioxidants23, 24 (Fig. 3). Bilirubin may also protect albumin from oxidant-mediated damage.25 Heme is thought to possess pro-oxidant properties through the redox properties of iron. HSA is an effective heme-binding protein.26 Once bound to albumin, such pro-oxidant properties are decreased, indicating an antioxidant function,27 although under physiological circumstances the heme-binding plasma protein hemopexin provides most of this form of antioxidant protection.28 Free, or loosely bound, redox-active transition metal ions (low molecular mass) are potentially extremely pro-oxidant, having the ability to catalyze the formation of damaging and aggressive ROS from much more innocuous organic and inorganic species (Fig. 3). In strictly biological terms the 2 most important such metals are iron and copper. In specific circumstances (certain disease states and poisoning), these metal ions can become free of constraints, which normally limit and control their reactivity. By virtue of its high-affinity copper-binding site, HSA limits copper-catalyzed oxidative damage to other biomolecules by directing damage toward the albumin molecule itself in a sacrificial fashion.29 In similar fashion, HSA can limit damage caused by accidental biological contamination by redox active metal ions such as vanadium, cobalt, and nickel. Although HSA iron-binding is weak and nonspecific, it may offer antioxidant protection when other specific protective stratagems become overwhelmed, such as under conditions of iron overload or pronounced hemolysis (Fig. 3). Evidence indicates that accessible thiol groups can signal inflammatory cell regulatory changes dependent on their redox state.30 Thus, 25% albumin has been shown to modulate neutrophil/endothelial cell interactions after shock and resuscitation and to attenuate lung injury.31 Furthermore, HSA augments intracellular glutathione levels and influences activation of the ubiquitous transcription factor nuclear factor-kappa B using both in vitro and in vivo protocols.32 Moreover, several recent studies using a rat model of hemorrhagic shock have indicated that the type of resuscitation fluid administered greatly influences proinflammatory responses, including lung apoptosis and rates of neutrophil activation. Plasma albumin was found to be the least proinflammatory of the fluids utilized.33 The formation of sulfenic acid residues by cys-34 oxidation15 also may be a key factor determining signaling responses, because recent evidence indicates that such groups impact on cellular signaling functions, reviewed in Poole et al.34 Paradoxically, and in common with other redox active antioxidant substances, albumin can display pro-oxidant properties, through its ability to redox cycle/recycle transition metal ions such as iron and copper from the less reactive (ferric/cupric) to more pro-oxidant (ferrous/cuprious) states (Fig. 3). Thus, a recent study has shown that copper/HSA could become pro-oxidant after fatty acid binding and subsequent cys-34 oxidation.35 Iron has much less binding affinity for HSA and is more likely to be recycled as a free agent able to catalyze damaging ROS formation at sites distant from HSA. Such an action is, therefore, potentially more deleterious. Indeed, HSA administration was reported recently to be adversely associated with a decline in iron-binding antioxidant protection in patients with acute lung injury,18 an effect thought to be related to the redox cycling of iron. Intravenous albumin therefore may be inadvisable in circumstances when pronounced extracellular iron mobilization or overload are complicating factors. In healthy adults, albumin synthesis occurs predominantly in polysomes of hepatocytes (10-15 g/day) and accounts for 10% of total liver protein synthesis. Relatively small amounts of albumin are hepatologically stored (<2 g), the majority being released into the vascular space. Approximately 30%-40% of albumin synthesized is maintained within the plasma compartment. The remaining pool is located within tissues such as muscle and skin. Studies of radiolabeled albumin catabolism in normal healthy young adult males indicate a variation of half-life of between 12.7 and 18.2 days (mean, 14.8 days),36 although a dynamic exchange between plasma and the interstitium occurs. Albumin leaks from plasma at a rate of 5%/hour and is returned to the vascular space at an equivalent rate through the lymphatic system. Synthesis is a constant process, regulated at both transcriptional and posttranscriptional levels by specific stimuli, but change in interstitial colloid oncotic pressure is thought to be the predominant regulatory influence.37 Albumin homeostasis is maintained by balanced catabolism that is not well characterized, occurring in all tissues. However, most albumin (40%-60%) is degraded in muscle, liver, and kidney. Plasma hyperalbuminemia is rare, whereas hypoalbuminemia is a feature of a variety of pathological processes, including liver disorders, cancer, and severe sepsis. Ascites formation is a common complication of cirrhosis and contributes to vascular hypoalbuminemia. Perturbations in hepatic vascular control are thought to be responsible for ascites formation, although the precise mechanisms remain a matter of debate (reviewed in Arroyo38). However, cirrhosis in the advanced stages is also characterized by protein wasting and hence albumin depletion. The reasons for such dysfunction remain unresolved, but nutritional, metabolic, and hormonal abnormalities and uncharacterized responses to the release of bioactive substances including chemokines may contribute to hyopalbuminemia (reviewed in Tessari39). The capillary bed is known to be hyperpermeable in patients with sepsis, thereby leaking albumin. However, the extent to which altered liver biosynthesis and rates of catabolism contribute to the plasma albumin deficiency seen in sepsis and critical illness remains uncertain. Indeed, the half-life of HSA in patients with hypoalbuminemia supported with total parenteral nutrition is only 9 days, although rates of catabolism are normal.40 Moreover, catabolism actually may be decreased while the extracellular pool is increased, suggesting that HSA may be spared or protected.41 Studies in healthy individuals given endotoxin, as well as in the critically ill, indicate that albumin synthesis increases under these circumstances, even though hypoalbuminemia is common.42, 43 By contrast, in animal models of sepsis and endotoxemia, decreased rates of liver albumin synthesis at the expense of acute phase protein synthesis is detectable regardless of nutritional support.44 Further studies in humans are required to resolve this controversy. HSA is a relatively small protein that accounts for some 75% of protein molecules in plasma in healthy individuals. Because of its disproportionate contribution to the plasma protein pool, albumin is also responsible for approximately 75% of plasma colloid oncotic pressure (COP). Oncotic pressure becomes osmotic pressure as the negative charges surrounding the protein molecules attract sodium, thus holding water. Its remaining contribution to COP is due to the Donnan effect attributable to its overall negative charge. Moreover, HSA may influence directly vascular integrity by binding in the interstitial matrix and subendothelium and reducing the permeability of these layers to large molecules, and indirectly through its scavenging properties. HSA has long been used to combat vascular collapse in severely ill patients, customarily as an iso-oncotic (4%-5%) solution for intravascular volume expansion; and as an hyperoncotic (20%-25%) solution for the maintenance of fluid balance between compartments and the restoration of COP. More recently, the benefits of HSA in supporting the critically ill have been debated, not least because it is expensive compared with synthetic colloid solutions. Concerns have been raised regarding the distribution of albumin in states of altered capillary permeability, such as sepsis. However, clinical trials have confirmed that albumin remains a potent volume expander compared with crystalloid solutions even under these circumstances.45 However, a recent study finding that bolus administration of 20% HSA to patients with sepsis led to a significantly faster decline in plasma albumin compared with healthy controls raises questions about longer-term efficacy of albumin administration.46 Furthermore, in these states of altered capillary permeability, the formation of edema (such as pulmonary edema) is governed more by hydrostatic pressure than COP.47 Colloid administration to patients with acute lung injury does not worsen pulmonary edema and may in fact reduce edema-producing forces.48 A variety of meta-analyses have produced conflicting results regarding the risk and benefits of albumin administration to critically ill patients. Specifically, the 1998 Cochrane Injuries Group concluded that albumin administration may increase the risk of death,49 whereas a larger meta-analysis published subsequently found no difference in outcome.2 Systematic reviews have attempted to elucidate potential reasons as to why albumin would adversely affect outcome, but have failed to identify the mechanism.50 However, in one of the largest clinical trials ever conducted in critically ill patients, the SAFE study randomized 7,000 critically ill patients needing fluid resuscitation to receive either 4% albumin or normal saline. There were no differences between groups in survival, measures of morbidity, and organ dysfunction, nor in length of stay in the intensive care unit or hospital.4 Is albumin neither beneficial nor harmful in a therapeutic sense in the critically ill, despite its wide range of potentially significant properties? We suggest that uncertainty remains for a number of reasons. First, the wide range of patients with critical illness that have been studied to date may conceal specific populations in which HSA may be definitely beneficial or harmful. Indeed, HSA is increasingly recognized as a "niche" drug, evidenced by its ability to significantly improve outcomes in patients with cirrhosis complicated by spontaneous bacterial peritonitis.51 This is further supported by subgroup analysis of the SAFE study,4 which has indicated that albumin may have varying effects in different patient groups. From these data, patients with traumatic brain injuries may actually be harmed by albumin administration (relative risk for death = 1.62, P = .009), whereas sepsis patients may derive benefit (relative risk = 0.87, P = .06). Second, the use of albumin in all trials published to date has concentrated on its volume-expanding effects. The total dose administered may therefore be inappropriate if other, non-circulatory properties such as those relating to antioxidant or anti-inflammatory capacities are the desired end point. Such properties might account for any disease-specific (i.e., niche) effects of HSA therapy. Third, albumin may have valuable properties when used as an adjunct with other therapies, as in hypoproteinemic patients with acute lung injury and acute respiratory distress syndrome, where the combination of HSA and furosemide therapy has been shown to improve fluid balance, oxygenation, and hemodynamics.48 The mechanisms responsible for the benefits observed in acute lung injury and acute respiratory distress syndrome are incompletely understood and may be specific to albumin compared with synthetic colloids. There is no clear evidence that pulmonary edema may be influenced by one colloid more than another.47 The ability of HSA to bind many drugs, sometimes irreversibly, is a complicated issue that may impact on combinational therapies.52, 53 Furthermore, binding of biologically active moieties such as NO (NO+), levels of which become elevated during critical illness, may influence the ability of HSA to bind other ligands.54 Historically albumin was used in patients with cirrhosis for vascular volume maintenance, because of its oncotic properties. As theories regarding the nature of vascular control and ascites formation developed, and with a better understanding of the use of diuretics and other management strategies, the use of albumin for the treatment of this condition declined. However, it is now apparent that the volume-expanding properties of albumin, in combination with other therapeutic approaches, is of clinical benefit to patients with cirrhosis, thereby reducing the renal impairment that complicates liver disease38 (Table 1). Recently albumin has been used not as a medium for administration, but as a part of a hemodialysis regimen in patients with hepatic failure, the so-called molecular adsorbent recirculating system (MARS, for a review see Sen et al.55). MARS uses an albumin-containing dialysate that is regenerated by dialysis against buffered bicarbonate solution subsequent to carbon and anion exchange column treatment to combine the removal of toxins normally cleared by the kidneys with those removed by the liver. MARS has been used to treat liver dysfunction and failure in more than 4,000 patients over the last 4 years56 and has been shown to improve renal function and hemodynamics, and to decrease brain edema and hepatic encephalopathy (reviewed in Mitzner et al.57). HSA avidly binds toxins, including bilirubin, copper ions, and protein breakdown products, substances that accumulate in primary liver diseases including cirrhosis,58 hepatitis C infection,59 and Wilson's disease.60 There are also indications for the treatment of secondary liver disease and during posttransplantation complications. The ability of MARS to remove other toxins and pro-inflammatory stimuli such as lipopolysaccharides, chemokines, and lipid peroxidation end-products, xenobiotics and free heme/hemoglobin, may have implications for limiting the inflammatory response. Ample evidence exists of enhanced NO production during both chronic and acute liver failure61 that may relate to adverse clinical events such as renal impairment and circulatory dysfunction, including hepatopulmonary syndrome. NO is specifically bound by HSA at position cys-34, MARS treatment may therefore also help modulate circulatory NO levels during liver failure, thereby protecting against the cascade of other organ failures that accompany acute liver disease. However, as previously stated, the levels of s-nitroso-albumin found in healthy plasma subjected to relevant physiological exposure with NO are low. Caution should therefore be used when interpreting results of MARS trials in this context, with the need for the true extent of s-nitrosylation to be determined under such pathological circumstances. There may well be implications for hepatopulmonary syndrome, although studies in this area are somewhat limited. However, the MARS system has been used to treat a small number of patients with ARDS.62 Recognized contraindications to albumin therapy include a known allergy to albumin and states in which fluid overload could be harmful (decompensated congestive heart failure or hypertension, severe anemia, etc.). Administration of certain colloids may be contraindicated in specific patient populations, such as dextrans or starches in patients with clinically significant coagulopathies, or hetastarch to patients with severe sepsis.63 Whereas albumin has similar colligative properties to these synthetic colloids, the adverse effect profile is less cumbersome, although it could be pro-oxidant under defined circumstances with as yet unknown consequences. Finally, HSA from different manufacturers may differ markedly in terms of the types of metals bound to it and in levels of oxidation.64 Albumin employed for clinical use therefore may differ markedly from endogenous HSA. Such batch differences may influence biochemical properties, and HSA can thereby vary in its ability to influence adhesion molecule expression from endothelial cells in culture.65 Furthermore, dimer and polymer formation on storage may contribute to rare instances of allergic reactions being seen,66 and vanadium contamination of commercial batches of HSA has been shown to adversely influence renal function in patients with coronary heart disease.67 Table 2 describes theoretical/potential implications associated with intravenous albumin administration. A more complete description of possible adverse effects from colloid administration can be found in a recent review.68 Human serum albumin (HSA) has many physiological and biochemical properties that render it relevant to many aspects of the disordered vascular and cellular functions that characterize the critically ill; particularly those afflicted by the systemic inflammatory response syndrome—sepsis, severe sepsis, and septic shock. The use of albumin as a routine volume expanding agent in the intensive care setting cannot be justified in terms of a mortality or morbidity advantage over crystalloid solutions. However, albumin may be beneficial in specific clinical circumstances, such as in patients with cirrhosis complicated by SBP, and its potential to modulate the inflammatory response is, we suggest, worthy of further exploration. The authors thank the British Heart Foundation, The Dunhill Medical Trust, The Garfield Weston Trust and the David Boston Trust for their support, and the U.S. National Institutes of Health for support via grant HL-67739.