Keywords

1 Introduction

Despite normally being a small soluble protein in the circulation or in complex with membrane proteins on cell surfaces, β2-microglobulin (β2m) is prone to form amyloid under specific conditions in vitro as well as in vivo. Thus, β2m is the fibrillating culprit in Aβ2m, a systemic type of amyloidosis that is first and foremost associated with chronic hemodialysis in a clinical syndrome named dialysis-related amyloidosis (DRA). Due to its small size (Mr = 11,729) β2m is present in the glomerular filtrate of the normal kidney and β2m was originally discovered as a component present in the urine of patients with tubular proteinuria, i.e. patients with defects in the reabsorption of β2m in the proximal tubules of the kidneys (Berggård 1965; Berggard and Bearn 1968). While the DRA syndrome was defined earlier (Assenat et al. 1980; Charra et al. 1984; Kuntz et al. 1984; Warren and Otieno 1975) it was not until 1985 that it was discovered that β2m is the major constituent of the amyloid deposits in osteoarticular tissues encountered in long-term hemodialysis patients with DRA symptoms (Gejyo et al. 1985; Gorevic et al. 1985, 1986).

In addition to its importance for patients with DRA β2m is also an easily accessible model protein for in vitro studies of protein amyloidogenesis. Many of these aspects have been covered in several recent reviews (Drueke and Massy 2009; Eichner and Radford 2011; Heegaard 2009; Yamamoto et al. 2009). While Chap. 8 of this book focuses on the experimental aspects of β2m misfolding and fibrillogenesis in vitro the present chapter is focused on the molecular pathology of β2m under in vivo conditions where β2m amyloidosis is encountered.

2 β2-Microglobulin in Physiology

β2m is present on the surface of all nucleated cells where it is the invariant chain in complex with and stabilizing membrane-anchored peptide- and lipid-presenting polypeptide chains (MHC class I and CD1 complexes) and Fc-receptors. It is also found as a conformationally less restricted (Hodkinson et al. 2009; Verdone et al. 2002) free monomer in blood and other biofluids, including synovial fluid (Gobezie et al. 2007), at low concentration (1.5–3 mg/ml (128–256 nM) in serum) (Karlsson et al. 1980). The normal physiological function, if any, of the freely circulating β2m is unknown.

2.1 β2-Microglobulin in the Healthy Organism

β2m associates non-covalently with MHC class I and CD1 complex heavy chains during synthesis to form MHC molecules that present peptides of 8–10 residues processed from intracellular proteins for cytotoxic T-lymphocytes (Doherty and Zinkernagel 1975) or antigenic lipids for natural killer T-cells in the case of the CD1 complex (Gumperz 2006; Kang and Cresswell 2002). β2m is required for efficient transport of nascent MHC class I heavy chains to the cell surface (Williams et al. 1989; Zijlstra et al. 1990). β2m is also present in association with other types of MHC Class I-like cell receptors, such as the IgG Fc receptor (Praetor and Hunziker 2002) and the HFE hemochromatosis membrane protein (Feder et al. 1997). β2m is critical for the stability and function of these receptors but is not directly involved in their binding of ligands (Otten et al. 1992).

A variable number of MHC class I molecules are expressed on cell membranes from different cells. In the case of human lymphocytes it has been estimated that there are 105–106 β2m-molecules/cell (Bjorck et al. 1979). When MHC is degraded, the MHC-associated β2m is released to the circulation (Cresswell et al. 1974). This results in a quite constant production of free (monomeric) β2m at a level of 0.13 mg/h ´ kg body weight under normal conditions. More than 99.9 % of this circulating β2m is filtered through the kidney glomeruli into the proximal tubules where it is reabsorbed by pinocytosis or by binding to the endocytic multiligand receptor megalin. Subsequently, it is degraded by the tubular cells (Karlsson et al. 1980; Saito et al. 1994). As a result, less than 0.1 % of the glomeruli-filtered β2m is excreted in the urine under normal conditions. A half-time for free β2m in the circulation of only about 60 min can be estimated on the basis of its serum concentration and production rate and the normal glomerular filtration rate and discounting exchange with cell-surface associated β2m. These parameters fluctuate only minimally under normal physiological conditions (Karlsson et al. 1980). In renal disorders (e.g., cadmium poisoning) where proximal tubular function is destroyed and tubular proteinemia ensues, the GFR and β2m serum concentrations are normal, but β2m is lost in the urine. Urine from patients suffering from defects in the function of proximal tubules is therefore a convenient source of β2m that can be purified to homogeneity by a combination of size-exclusion chromatography and chromatofocusing (Nissen et al. 1987, 1997).

2.2 β2-Microglobulin Structure

A detailed review of β2m conformational structure and dynamics is provided by Chap. 7 by Esposito et al. Here, the features that are most relevant and important for the generation of β2m amyloid in vivo are summarized.

β2m is a 99 residue protein and has a molecular weight of 11,729 Da. It folds natively into a seven-stranded antiparallel β-sandwich consisting of two β-sheets linked covalently by one intramolecular disulfide bond between Cys25 and Cys80. One sheet contains four strands (A, B, E and D) and the other sheet three (C, F and G) (Fig. 19.1). Native β2m as well as amyloid β2m are highly structured, tightly packed proteins. Strands B and F, connected by the disulfide bond, are the most protected part of the protein. Strands A, C, E, and G are moderately protected, leaving the D-strand connecting the two β-sheets of the β2m sandwich as the most exposed part of the monomeric native protein (Villanueva et al. 2004). The D-strand interacts with the α-heavy chain in the MHC class II cell membrane complex (Khan et al. 2000; Saper et al. 1991). The loop between the D and E strands is a strained but dynamic part of the molecule that may switch between several conformations (Okon et al. 1992; Verdone et al. 2002).

Fig. 19.1
figure 1

Schematic β2m structure (PDB: 1DUZ) (Saper et al. 1991). The A-G β-strands are labelled and the DE-loop with the Lysine-58 cleavage point is marked by the arrow

3 β2-Microglobulin in Pathology

Levels of plasma β2m increase in lymphoproliferative disorders, in chronic renal failure (with decreased filtration), in inflammation and infection and in other conditions with high cell turnover (Wu 2006). A relationship between tumor burden in certain lymphoproliferative disorders, particularly multiple myeloma, and serum levels of β2m thus has been noted. β2m is lost in the urine in conditions where proximal tubular function is impaired.

While elevated plasma β2m levels are absolutely necessary for the development of DRA it is not sufficient since increased circulating levels of β2m where the kidney function is normal e.g. in chronic immunoinflammatory conditions, in infections and hematopoietic malignancies (Bartl et al. 1989; Bhalla et al. 1985; Bourantas et al. 1999) do not lead to β2m amyloidosis. In agreement with this there are no signs of β2m amyloidosis in engineered mice models with highly increased human β2m in the circulation (Zhang et al. 2010). A sustained increase in β2m concentration in the circulation combined with specific abnormalities (e.g. inflammation and oxidative stress (Annuk et al. 2005; Terawaki et al. 2004)) encountered in renal insufficiency and/or the process of hemodialysis thus are necessary for the development of DRA.

4 β2-Microglobulin in Dialysis-Related Amyloidosis

The incidence of DRA is seemingly declining with the increased use of high-flux dialysis, with changes in dialysate composition, and as the components of dialysis machines that come into contact with patient blood (dialysis membranes and tubings) have become more biocompatible (Fujimori 2011; Schwalbe et al. 1997; Winchester et al. 2003).

4.1 The Clinical Syndrome

DRA results from the deposition and accumulation of β2m amyloid (Floege and Ehlerding 1996; Gejyo et al. 1985; Gorevic et al. 1985). Deposition primarily takes place in osteoarticular tissue. The clinical manifestations of DRA include erosive and destructive osteoarthropathy, destructive spondyloarthropathy, cystic bone lesions, carpal tunnel syndrome, and other neuropathies (Kiss et al. 2005) (Fig. 19.2) leading to joint pain, decreased function, and fractures.

Fig. 19.2
figure 2

Magnetic resonance imaging of humeral erosions in a 63-year-old man on hemodialysis for 31 years for chronic glomerulonephritis with biopsy-proven dialysis-related amyloidosis. Coronal T1-weighted MR image (470/12) shows osteolysis in superior-posterior humeral head, which communicates with joint (arrow). Low-signal-intensity tissue representing amyloid appears within lesion. Amyloid deposits are also visible within subdeltoid bursa between deltoid muscle and humerus (arrowheads). (Reproduced with permission from Fig. 4 in Kiss et al. 2005)

In addition, β2m amyloid deposits may be found in extra-osteoarticular tissues, including the heart, gastrointestinal tract, ovaries, liver, ureter and subcutaneous tissue (Buchholz et al. 1995; Kawano et al. 1998; Mogyorosi and Schubert 1999; Mount et al. 2002; Shimizu et al. 2003). Amyloid deposits outside the musculoskeletal system are often asymptomatic but may cause serious complications such as heart failure or gastrointestinal bleeding. Extensive deposition of β2m amyloid in the small blood vessels in the myocardium may cause calcification of the heart and cardiac dysfunction, thereby contributing to the occurrence of heart disease in long-term dialysis patients (Takayama et al. 2001).

The prevalence of DRA may also be influenced by patient age and duration of dialysis treatment (Jadoul 1998; Jadoul et al. 1997). Recent studies, however, fail to demonstrate a clear correlation between clinical manifestations (carpal tunnel syndrome) and duration of end-stage kidney disease and hemodialysis (Kwon et al. 2011). Furthermore, the type of hemodialysis, e.g. high-flux/low-flux dialyzers, membrane type, dialysate composition and purity, and biocompatibility of the material in contact with the biofluids are all factors that may influence the development of amyloid deposits (Baz et al. 1991; Campistol et al. 1999; Locatelli et al. 1999; van Ypersele et al. 1991) even though there are no clear effects on circulating oxidative stress parameters (Schneider et al. 2011). There is no direct correlation between the efficiency of β2m removal and the rate of progress of DRA (Danesh and Ho 2001; Kwon et al. 2011).

4.2 β2-Microglobulin Aggregation, Fibrillation, and Amyloidogenesis

A great number of in vitro conditions have been shown to induce amyloid fibrillation of β2m (cf. Chap. 8). These conditions include extremes of pH, amino acid substitutions, anionic detergent, addition of trifluoroethanol, ultrasonication, agitation, heat treatment, incubation with Cu2 + (Morgan et al. 2001; Ohhashi et al. 2005; Sasahara et al. 2007; Yamamoto et al. 2004a) and other factors (Heegaard 2009). Most of these conditions are unlikely to be encountered in vivo except for slightly decreased pH at inflammatory sites (Treuhaft and McCarty 1971), increased divalent cations (especially Cu2 +), Proline-32 cis-to-trans isomerisation, and Lysine-58 cleaved forms of β2m (cf. Table 19.1) (Eichner and Radford 2011; Heegaard 2009).

Table 19.1 Modifications reported on Aβ2m in vivo (based on Table I in Heegaard 2009)

Some of the fibrillation pathways do not necessarily involve major conformational changes (Calabrese et al. 2008; Eakin et al. 2006; Liu et al. 2011). This is in accord with the lack of correlation between thermodynamic stability and tendency to amyloid formation of β2m variants (Smith et al. 2003). Thus, highly conformationally unstable β2m variants may readily unfold and aggregate but do not in every case form amyloid. Amyloid fibrillation appears to be linked to the conformational dynamics of retained globular folds e.g. leading to domain swapped dimers as the building blocks in fibril formation (Hafner-Bratkovic et al. 2011; Liu et al. 2011; Sambashivan et al. 2005; Wahlbom et al. 2007; Domanska et al. 2011; Mendoza et al. 2011). Such mechanisms are compatible with limited structural rearrangements e.g. cis-to-trans isomerisation of the Pro-32 bond or rearrangement or pertubations of the DE-loop in the globular fold (Barbet-Massin et al. 2010; Colombo et al. 2011; Eichner and Radford 2009; Eichner and Radford 2011; Mimmi et al. 2006; White et al. 2009) and has recently also been shown to be a plausible mechanism involved in ΔN6β2m amyloid formation (Domanska et al. 2011). While increasingly supported by experimental data these mechanisms have not yet been conclusively tested with in vivo-generated material.

The kinetics of in vitro generation of amyloid-like fibrils displays characteristic traits in which a lag phase precedes a period of exponential growth. During this lag phase the early aggregates or nuclei are generated, and the subsequent rapid fibril elongation corresponds to the exponential growth period. The majority of ex vivo β2m amyloid (i.e. material isolated from clinical specimens and studied in the laboratory) appears to consist of aggregated full-length, wild-type β2m (Gorevic et al. 1986). However, natively folded β2m is not prone to self-aggregate at neutral pH even at vastly supraphysiological concentrations (Eakin and Miranker 2005; Kad et al. 2001; Myers et al. 2006) and, as mentioned above, in vivo fibrillogenesis will not be initiated simply by the presence of high β2m plasma concentrations. It is likely that the assembly of β2m monomers into bona fide amyloid fibril requires encounters between partially unfolded molecules. Under native conditions this is extremely rare, and possibly explains why the generation of amyloid in vivo is a process that occurs over prolonged periods of time, usually several years.

4.2.1 β2m Post-Translational Modifications

The conformational dynamics of β2m has been studied using H/D-exchange experiments monitored by e.g. mass spectrometry (Heegaard et al. 2005; Hodkinson et al. 2009; Jørgensen et al. 2007). These experiments show the capability of β2m to undergo transient regional unfolding-refolding under physiological conditions and also document the increased rate of unfolding—even though the molecule has the same overall conformation as the wild-type (Mimmi et al. 2006)—of a cleaved variant of β2m termed ΔK58-β2m in which Lys58 has been proteolytically removed (Heegaard et al. 2005). This post-translationally modified β2m has been found in the circulation of many (20–40 %) hemodialysis patients (Corlin et al. 2005), and serum levels are markedly higher in patients dialyzed with the less biocompatible, copper containing low-flux membranes (Cuprophane), than in those treated with synthetic high-flux membranes. ΔK58-β2m is thought to be generated in vivo as a consequence of the activation of the complement system by cleavage of β2m by the serine protease C1s between Lys58 and Asp59 in the DE-loop followed by the rapid removal of Lys58 by carboxypeptidase B activities (Nissen et al. 1990). The removal of Lys58 changes the protein from a single chain molecule with an intrachain disulphide bond to two chains held together by an interchain disulphide bond. The two chains consist of residues 1–57 and 59–99 of native β2m. The removal of Lys58 only results in minor global conformational changes (Mimmi et al. 2006) but a positively charged residue that otherwise could contribute to electrostatic repulsion between β2m molecules is lost. This makes the AA59–70 region of the molecule more hydrophobic and aggregation-prone. Also, the molecule is considerably more conformationally unstable as shown by capillary electrophoresis experiments (Heegaard et al. 2002). Thus the cleavage of β2m induces significantly accelerated (about 10 times) rates of cooperative unfolding at physiological temperature, increasing molecular aggregation and the ability to generate fibrils with amyloid features (Heegaard et al. 2005).

Other posttranslationally modified species e.g. β2m with deletions of N-terminal residues (Esposito et al. 2000) (Table 19.1) are also of interest for the in vivo situation because catalytic amounts (1 %) are sufficient to induce conversion of native β2m into Aβ2m in vitro (Eichner and Radford 2011). A catalytic effect was also shown for fibrils of Lys-58 cleaved β2m formed in the presence of heparin sulfate (Corlin et al. 2010). N-terminally truncated β2m has been found to be a significant constituent of Aβ2m fibrils ex vivo (Stoppini et al. 2000, 2005) (cf. below). The ΔN6 truncation of β2m (Bellotti et al. 1998; Linke et al. 1987, 1989; Stoppini et al. 2000, 2005) results in structural rearrangements of the protein rendering the N-terminal region more disordered at the same time as the β-bulge of strand D, the successive loop, and strand E all adopt a less flexible conformation. The ΔN6β2m-variant has been shown to have a reduced free energy of stabilization as compared to native β2m and to possess an enhanced tendency to precipitate and self-aggregate (Esposito et al. 2000). At pH lower than 7, ΔN6β2m generates amorphous aggregates and short fibrillar structures, and at neutral pH this variant possesses the ability to further extend ex vivo β2m fibrils. Moreover, examination of a putative role of collagen on β2m fibril formation showed that the presence of Δ N6β2m together with collagen induced the generation of β2m amyloid-like fibrils (Canale et al. 2011; Relini et al. 2006).

Posttranslational modifications in the form of truncations and cleavages thus significantly decrease β2m solubility at physiological conditions and are likely to occur in vivo. Measurements suggest that up to 30 % of the β2m in amyloid fibers extracted from ex vivo deposits are present as N-terminally modified or truncated forms while Lys-58 cleaved β2m could not be detected in the same extracts (Bellotti et al. 1998; Floege and Ehlerding 1996; Giorgetti et al. 2007; Linke et al. 1986). Conversely, Lys-58 cleaved β2m has been demonstrated in the circulation of many dialysis patients while N-terminally truncated β2m variants have only been demonstrated in amyloid deposits.

4.2.2 β2m Pro32-Isomers and Conformers

In native, wild-type β2m the peptide bond preceding Pro32 attains the less common, thermodynamically unfavorable cis configuration (Jahn et al. 2006). Engineered mutations changing Pro32 to glycine or alanine, causing the peptide bond to stay in the trans configuration, result in markedly increased rates of amyloid fiber growth and in a repacking of the hydrophobic core of the protein wherein the backbone changes in the BC loop affect the conformation of strand D and the DE loop (Jahn et al. 2006; Eakin et al. 2006). Thus, a β2m folding intermediate in which Pro32 is kept in trans configuration does not bind the MHC heavy chain well (Esposito et al. 2008) and is much more prone to aggregation than the wild-type β2m molecule (Jahn et al. 2006). Since the isomerization of this bond is the slowest step in the folding of native β2m (Eakin et al. 2006; Eichner and Radford 2009; Kameda et al. 2005), a population of protein intermediates in the non-native trans configuration may exist and be of significance in fibril formation. Interestingly, chelation of Cu2 + by His31 promotes Pro32 isomerization (Eakin et al. 2002).

A β-bulge in this D strand is straightened out in infrequently encountered β2m structures and this may favour oligomerization by edge strand docking (Richardson and Richardson 2002) possibly facilitating amyloid formation (Trinh et al. 2002). However, recent data do not support a role for the β-bulge in protecting against amyloid formation (Azinas et al. 2011).

4.2.3 β2m-Cu2 + Interactions

Many proteins are known to have specific binding sites for metal ions, and binding of ions can significantly alter the properties of the protein. β2m binds metal cations, in particular Cu2 +, and this binding will, under the proper experimental conditions, result in the generation of fibrous β2m aggregates (Eakin et al. 2002). The binding of Cu2 + to β2m is specific, while other divalent cations such as Ca2 + and Zn2 + bind nonspecifically (Morgan et al. 2001). Studies have indicated that binding of Cu2 + to β2m lower the energy barrier of transition (Deng et al. 2006) to a partially unfolded state. The binding of Cu2 + (chiefly by His31) promoting these local molecular rearrangements has also been proposed to enhance the cis-trans isomerization at Pro32 leading to a partial unfolding of the hydrophobic core (Eakin et al. 2004) and thus increasing conformational instability (De Lorenzi et al. 2008). From a pathophysiological point of view, the relevance of Cu2 + to β2m amyloidogenesis in DRA is evident, not only because of the former use of the Cuprophane membranes containing copper, but also since a dialysis patient is exposed to more than 100 L of dialysate during each dialysis session (Cheung et al. 2006). The dialysate contains copper to a maximum level of 1.6 mM (Vorbeck-Meister et al. 1999), being within a factor of two of the measured affinity of Cu2 + and β2m (Eakin et al. 2002).

4.2.4 β2m-amyloid Seeding

Addition of nucleating seeds (preformed β2m fibrils) to a solution of native β2m will result in extensive fibrillation by elongation of the existing fibrils. In a study by Jones et al. (2003), a range of peptides of β2m were examined for their ability to assemble into amyloid-like fibrils in vitro and for their ability to act as seeds for full-length β2m. The study showed that only peptides having a sequence corresponding to strand E in wild-type β2m (E peptides) were capable of fibrillation, even when studied over a wide pH range. These fibrils also had the ability to seed the formation of fibrils in a solution of full-length β2m. The E peptides contain a very large number of aromatic amino acids, i.e., the aromatic residues Phe56, Trp60, Phe62 and Tyr63 which are all important for hydrophobic interactions of β2m with the heavy chain in the MHC class I complex (Saper et al. 1991). The importance of the DE-loop region is further underscored by a substantial number of experiments showing the effect of mutations and posttranslational modifications in the D-strand and DE-loop region (AA50–63) (Azinas et al. 2011; Santambrogio et al. 2010) including other nucleation/elongation experiments strongly implicating this aromatic-rich region in amyloidogenetic interactions (Platt et al. 2008). The region was shown directly by 2D NMR measurements to be the longest contiguous region involved in intermolecular contacts in low pH-generated fibrils (Debelouchina et al. 2010) and is also central in the putative amyloidogenic structure involved in domain-swapped propagating dimers (see below).

4.2.5 Structure of β2m Amyloid Fibrils

All amyloid fibrils, despite having rather different morphologies, share a common cross-β spine structure (Jahn et al. 2010). Albeit not fully understood, the propensity of a protein or peptide to form amyloid fibrils is dependent on an interplay between secondary structure, charge, sequence and hydrophobicity.

Experiments using low pH and high salt have shown that in the early aggregation processes β2m monomers first assemble into dimers and tetramers, then as β2m aggregates over time, changes in secondary structure can be observed indicative of the emergence of antiparallel intermolecular β-sheet structures (Fabian et al. 2008). In a study using H/D exchange of amide protons combined with NMR analysis the core of the β2m amyloid fibril was mapped. The result was that in addition to the regions protected from exchange in the native monomeric β2m, the residues in the native loops also become highly protected in the fibrillar state indicating an increase in the hydrogen bond network in the fibrils, leaving only the N- and C-terminal ends unprotected from exchange. Also, the CD spectrum of β2m amyloid fibrils showed increased β-sheet content, supporting the suggestion that both the native loops and the native β-strands are transformed into β-sheets in the β2m amyloid fibrils (Hoshino et al. 2002).

The structure of β2m amyloid fibrils has been suggested to involve the assembly of six protofilaments, arranged in pairs of three protofilaments, wherein each protofilament is build from globular subunits in a dimer-of-dimers packing (Mendoza et al. 2011; White et al. 2009). Domain-swapped dimers are also at the heart of a model for the β2m fibril core formation stabilized by disulfide exchange leading to intermolecular disulfide bridges. This leads to a “steric zipper” arrangement with the AA54–59 segment most likely involved and thereby forming the β-spine with retention of the globular features of the other parts of the protein that participates in the amyloid fibril structure (Liu et al. 2011). In another model tetramers are formed by dimer-dimer interactions involving D- and G-strands from the two different dimer-units (Mendoza et al. 2011) while D-D interactions were found unlikely to form dimer interfaces (Mendoza et al. 2010).

Three-dimensional domain swapping is emerging as a common mechanism for amyloid fibril formation (Ecroyd et al. 2010; Hafner-Bratkovic et al. 2011) and entails disulfide exchange to get covalently linked dimer β2m building blocks forming the steric-zipper spine of amyloid fibrils (Fig. 19.3). This is consistent with the accelerating effect of reducing agents on Aβ2m formation in vitro at physiological pH in some studies (Liu et al. 2011) and with the fiber polymorphism characteristic of various amyloids. Disulfide rearrangement is well known to occur in vivo e.g. in immunoglobulin arm rearrangements (Liu et al. 2010) and in some amyloidogenic proteins (Knaus et al. 2001; Nilsson et al. 2004). The precise mechanisms associated with this model in vivo and its relevance for DRA remain to be settled. Domain swapping and stabilization by disulfide exchange have not been shown in ex vivo2m, and some studies actually show that thiol compounds (reductants) inhibit β2m amyloid fibril formation at neutral pH (Yamamoto et al. 2008). Also, while there is no doubt that chronic kidney disease and hemodialysis lead to a proinflammatory and oxidatively stressed environment in the circulation (possibly further enhanced by oxidizing Cu2 +-ions) (Lee et al. 2011) the factors associated with chronic kidney disease and hemodialysis that would be especially conducive to breaking intramolecular disulfide bonds in the core of a tightly packed globular protein remain to be characterized.

Fig. 19.3
figure 3

Schematical model of β2m amyloidogenesis in vivo in chronic hemodialysis patients. Upper panel (based on Liu et al. 2011) shows the disulfide exchange involved in domain-swapped β2m dimer formation and the proposed structure of a runaway domain-swapped oligomer with inter-dimer disulfide bonds indicated. Lower panel shows a schematic of amyloid formation by β2m. Three stages are envisioned: (A), normal conditions, monomeric β2m displays conformational fluctuations (depicted as closed and open domain forms) but is present at normal concentrations; (B) Increased concentrations in renal disease lead to formation of β2m dimers possibly stabilized by intermolecular disulfide bonding between swapped domains (cf. upper panel); (C) Association, organization and consolidation on collagen of modified and/or dimerized β2m and then further oligomerzation with runaway dimerization as proposed by Eisenberg et al. (Liu et al. 2011) on the seeding surface of collagen-attached β2m

4.3 Accessory Molecules

In the attempt to elucidate the early fibrillating events in DRA a number of co-factors have been studied for their ability to elicit and facilitate β2m fibrillogenesis (Table 19.2). Glycosaminoglycans (GAGs) are long unbranched polysaccharide chains consisting of repeating disaccharide units. Heparan sulfate is a glycosaminoglycan which is expressed on cell surfaces and binds non-covalently to a variety of proteins (Jackson et al. 1991). In many of the amyloidoses heparan sulfate has been shown to be a universal component of amyloid (Jackson et al. 1991; Magnus et al. 1991; Snow et al. 1987, 1988, 1991; Snow and Kisilevsky 1985; Young et al. 1989, 1992) and thus this GAG has been proposed to play an active role in amyloid generation by promoting fibrillogenesis (Castillo et al. 1998; Cohlberg et al. 2002; Goedert et al. 1996) rather than being passively accumulated. Heparan sulfate and heparin (which is commonly administered as an anticoagulant during hemodialysis treatment) have been shown to promote β2m amyloid formation in vitro (Borysik et al. 2007) and to exert a stabilizing effect on such β2m fibrils (Myers et al. 2006; Yamaguchi et al. 2003; Yamamoto et al. 2004b). Both uremic serum and synovial fluid also have amyloid-enhancing effects in seeding experiments (Myers et al. 2006). Interestingly, a mouse human β2m model with vastly supraphysiological circulating β2m levels did not show β2m-amyloid formation even in the cases where preformed amyloid fibril seeds were injected (Zhang et al. 2010). This again shows that the specific uremic environment of kidney failure patients is an indispensable prerequisite for the triggering of Aβ2m.

Table 19.2 Factors with β2m amyloid-enhancing or inhibiting effects that may be encountered or exploited in vivo

Collagen fibers, which are found in the joint environment, have also been shown to promote β2m fibrillogenesis (Relini et al. 2006) and could be relevant for DRA pathology. The positively charged collagen molecules are proposed to act as an immobilized surface on which the β2m molecules bind and become oriented in a fashion facilitating fibril formation. Also apolipoprotein E (ApoE) which is a cholesterol transport protein has been suggested to be of relevance for amyloidogenesis, since it has been found to be ubiquitously co-localized with amyloid deposits in both systemic and localized amyloidoses. However, reports are contradictory showing both promoting and inhibitory effects of ApoE on fibril formation (Naiki et al. 1997; Wisniewski et al. 1994).

Serum amyloid P component (SAP) is a common component of extracellular matrix in the microfibrillar mantle of elastic fibers and in the glomerular basement membrane. SAP binds to amyloid in a calcium-dependent manner and is a universal constituent of all amyloid deposits, comprising up to ~ 15 % of the amyloid tissue mass (Skinner et al. 1980). SAP itself is highly resistant to proteolysis (Kinoshita et al. 1992) and possesses the ability to also prevent proteolysis of the amyloid fibrils to which it binds (Tennent et al. 1995), thereby possibly contributing to their persistence in vivo. Due to its specificity and affinity for amyloid deposits SAP may be used for radioimaging in vivo of some systemic amyloidoses (Pepys 2006).

4.4 Treatment of β2-microglobulin Amyloidosis

Treatment of amyloidosis is difficult. The general strategy is to treat symptoms secondary to organ damage and to reduce the production of amyloid by restricting the production of precursor molecules (Pepys 2006). In the case of Aβ2m there is no curative treatment except for reconstitution of renal function by renal transplantation (Pepys 2006). This will halt DRA progression but does not revert established β2m amyloid deposits (Labriola et al. 2007; Mourad and Argiles 1996) that must be treated by surgery. It is worth noting that amyloid deposits are not static entities but rather go from a more reversible state when newly formed to much more stable irreversibly consolidated fibril formations over time (Kardos et al. 2011). Dye molecules and GAG analogues have been used experimentally based on the notion of inhibiting interactions with accessory molecules. Screening of sulfonated small molecule, tetracycline analogies, and other antibiotics-based potential fibrillogenesis inhibitors (Giorgetti et al. 2011; Regazzoni et al. 2011) has identified promising compounds that may divert aggregation into non-amyloidogenic pathways (Woods et al. 2011). Also tetracycline (Giorgetti et al. 2011) and proteins such as α2-macroglobulin (Ozawa et al. 2011a) and compounds stabilizing native protein structures have been launched, especially for treating the transthyretin amyloidoses (Adamski-Werner et al. 2004; Sacchettini and Kelly 2002) but have not yet been explored in DRA. In DRA most work is directed at reducing disease progression by depleting the precursors of β2m amyloid by optimized dialysis including Cu2 +-free dialysis membranes (Miura et al. 1992; van Ypersele et al. 1991).

Experimental therapies may be derived from preliminary reports of amyloid fibril destruction by laser irradiation of thioflavin T-fibril complexes which has been shown in vitro both for keratoepithelin and β2m-derived peptide fibrils (Ozawa et al. 2011b; Ozawa et al. 2009). The clinical implementation of fibril destruction by amyloid dye-excitation, however, is far from straightforward.

5 Conclusions and Future Directions

The list of factors that have been shown to influence the conformation of intact β2m is very long. In DRA-patients, the increased serum concentration is important, but there is no simple correlation between DRA-severity and circulating β2m concentration (Gejyo et al. 1986), and β2m-amyloid has not been observed in other diseases where there are sustained, elevated β2m concentrations. Furthermore, many of the in vitro conditions that are highly favorable for amyloid formation from normal β2m such as very low pH and very high ionic strength are not encountered in vivo. Some of the factors that may be relevant enhancers and inhibitors of β2m amyloidogenicity in vivo are listed in Table 19.2. In this regard the catalysis of amyloidogenic conformations by divalent cations, especially Cu2 + and its relationship to Pro-32 trans-to-cis conversion, at neutral pH (Eakin et al. 2002, 2004, 2006) is very interesting. Most, if not all, amyloid proteins have affinities for divalent cations, and metal ions have been proposed as a triggering factor in Alzheimer’s disease (Atwood et al. 2000) even though the evidence linking Cu2 + to enhancement or inhibition of amyloid formation by amyloid β-peptides is ambiguous (Pedersen et al. 2011). Metals are also involved in Parkinson’s disease (Uversky et al. 2001), immunoglobulin light chain (Davis et al. 2001), and prion protein amyloidogenesis (Jackson et al. 2001). Furthermore, the presence of a subfraction of cleaved β2m-species in the amyloid deposits (Linke et al. 1986, 1987, 1989; Stoppini et al. 2000) and of another cleaved variant in the circulation of hemodialysis patients—especially those treated with Cu2 +-containing, complement-activating dialysis membranes and the demonstration of the decreased conformational stability in vitro of such truncated variants (Corazza et al. 2004; Esposito et al. 2000; Heegaard et al. 2005) together with their catalytic effects on amyloidogenesis (Eichner and Radford 2011) provide strong indications of the pathogenetic importance of post-translationally modified β2m for the development of β2m amyloidosis in dialysis patients.

The possibility that several factors may interplay in different ways in vivo and yet lead to the β2m amyloidosis syndrome DRA is unlikely since the syndrome is limited to such a well-defined group of patients. In that regard it is an interesting model disease for all types of amyloid.

With a normal half-life in solution of only one hour it is likely that monomeric β2m may be regarded as a conformationally unstable waste product with unfavorable exposure of hydrophobic side chains in the solvent exposed DE-loop which is otherwise designed to be stabilized by fitting into the MHC class I receptor complex in which β2m fulfils its proper physiological function.