Aβ2M (β2-microglobulin-related) amyloidosis is a frequent and serious complication in patients on long-term dialysis. Partial unfolding of β2-m (β2-microglobulin) may be essential to its assembly into Aβ2M amyloid fibrils in vivo. Although SDS around the critical micelle concentration induces partial unfolding of β2-m to an α-helix-containing aggregation-prone amyloidogenic conformer and subsequent amyloid fibril formation in vitro, the biological molecules with similar activity under near-physiological conditions are still unknown. The effect of various NEFAs (non-esterified fatty acids), which are representative anionic amphipathic compounds in the circulation, on the growth of Aβ2M amyloid fibrils at a neutral pH was examined using fluorescence spectroscopy with thioflavin T, CD spectroscopy, and electron microscopy. Physiologically relevant concentrations of laurate, myristate, oleate, linoleate, and mixtures of palmitate, stearate, oleate and linoleate, induced the growth of fibrils at a neutral pH by partially unfolding the compact structure of β2-m to an aggregation-prone amyloidogenic conformer. In the presence of human serum albumin, these NEFAs also induced the growth of fibrils when their concentrations exceeded the binding capacity of albumin, indicating that the unbound NEFAs rather than albumin-bound NEFAs induce the fibril growth reaction in vitro. These results suggest the involvement of NEFAs in the development of Aβ2M amyloidosis, and in the pathogenesis of Aβ2M amyloidosis.
In Aβ2M (β2-microglobulin-related) amyloidosis or DRA (dialysis-related amyloidosis), a common and serious complication in long-term haemodialysis patients , the deposition of Aβ2M amyloid fibrils in the tissue causes carpal tunnel syndrome and destructive arthropathy with cystic bone lesions [2,3]. Intact β2-m (β2-microglobulin) constitutes almost all amyloid fibrils deposited in the synovial membrane of the carpal tunnel [4,5]. Although the retention of β2-m in the plasma appears to be prerequisite , and other factors, such as the age of the patient, the duration of haemodialysis, and the type of dialysis membrane used, may also be involved [7–9], the mechanism of the deposition of these amyloid fibrils is not fully understood.
Because Aβ2M amyloid deposition is observed predominantly in the cartilaginous and tendinous tissues [10,11], the specific interaction between β2-m and the extracellular-matrix molecules in these tissues, such as type II collagen, GAGs (glycosaminoglycans) and PGs (proteoglycans), may cause Aβ2M amyloid deposition. We reported that various types of GAGs and PGs stabilize the Aβ2M amyloid fibrils and inhibit their depolymerization at a neutral pH . Furthermore, we reported that some GAGs, especially heparin, dose-dependently enhanced the 2,2,2-trifluoroethanol-induced fibril growth at a neutral pH . Partial unfolding of natively folded amyloid precursor proteins such as β2-m and transthyretin may be essential to their assembly into amyloid fibrils both in vitro and in vivo [14–16]. However, the biological molecules that induce partial unfolding of these fibrils and subsequent amyloid fibril formation, under the near-physiological conditions in vitro, are still unclear. To understand the molecular pathogenesis of Aβ2M amyloidosis, the biological molecules that induce amyloid fibril formation by affecting the conformation and stability of β2-m and amyloid fibrils need to be identified.
Several research groups reported that various lipid molecules induce the conformational change of various amyloid precursor proteins, as well as initiating the amyloid fibril formation in vitro [17–21]. SDS around the CMC (critical micelle concentration) unfolds the compact structure of β2-m to an α-helix-containing aggregation-prone amyloidogenic conformer and stabilizes the fibrils, resulting in the growth of Aβ2M amyloid fibrils at a neutral pH . Given the structural similarity to SDS (Figure 1) and the possible contribution to other forms of human amyloidosis [20,21], NEFAs (non-esterified fatty acids) could be candidates for the biological molecules inducing Aβ2M amyloid fibril formation. NEFAs, representative anionic amphipathic compounds in the circulation, are produced de novo by adipocytes or by lipolysis of plasma triacylglycerol in chylomicrons/very-low-density lipoproteins, transported by serum albumin, and incorporated into muscle cells as an important energy source, or into adipocytes to form re-esterified triacylglycerol [23–25]. Their concentrations range from 0.1 to 2 mM and turn over very rapidly (half-life<5 min) [23–25]. Since serum albumin plays a major role in the transport of NEFAs in the circulation , hypoalbuminaemia observed in patients with end-stage renal disease  may impair the metabolism of NEFA and influence the manifestation of DRA. Gellermann et al.  reported that amyloid fibrils derived from a subcutaneous node of a patient with Aβ2M amyloidosis are associated with NEFAs constituting more than 10% of the associated lipid.
The chemical structures of SDS and NEFAs analysed in the present study
Here, we investigated the effect of several NEFAs on the growth of fibrils at a neutral pH using fluorescence spectroscopy with ThT (thioflavin T), CD spectroscopy and electron microscopy.
MATERIALS AND METHODS
r-β2-m (recombinant human β2-m) was expressed and purified using the Escherichia coli expression system as described . Laurate (sodium salt) and SDS were obtained from Nacalai Tesque (Kyoto, Japan). Myristate, palmitate, stearate, oleate, linoleate (all sodium salts), PA (L-α-phosphatidic acid) from egg yolk lecithin and HSA (human serum albumin) (essential fatty acid free) were obtained from Sigma Chemical. DSPC (1,2-distearoyl-sn-glycero-3-phosphocholine) was obtained from Avanti Polar Lipids (Alabaster, AL, U.S.A.). NPN (N-phenyl-1-naphthylamine) was obtained from Wako Pure Chemical Industries (Osaka, Japan). The chemical structures of SDS and NEFAs analysed in this study are illustrated in Figure 1.
Preparation of Aβ2M amyloid fibrils and seeds
Aβ2M amyloid fibrils used for the growth reaction were prepared from the patient-derived Aβ2M amyloid fibrils by the repeated growth reaction at pH 7.5 with r-β2-m, as described elsewhere . We first obtained F1 fibrils by the growth of S0 seeds. F1 and S0 are fibrils of generation 1 and seeds of generation 0 (i.e. patient-derived fragmented fibrils) respectively. Next, S1 seeds were prepared by the extensive sonication of F1 fibrils. By repeating the algorithmic protocol nine times, F9 fibrils were obtained from S8 seeds and S9 seeds from F9 fibrils.
Preparation of the solutions of NEFAs
NEFAs were solubilized or suspended in distilled water to yield concentrations of 10–20 mM. The suspensions of myristate, palmitate and stearate were sonicated using an ultrasonic disruptor (UD-201, Tomy, Tokyo, Japan) equipped with a microtip (TP-030, Tomy) immediately before assaying. NEFAs were also used at ratios found in the serum [30,31]: 20 mM palmitate, stearate, oleate and linoleate were mixed in volume proportions of 3:1:3:1.5 (NEFA mixture A, MixA), and palmitate, stearate and oleate at 3:1:3 (NEFA mixture B, MixB). In Figure 9, unsaturated NEFAs (i.e. linoleate and oleate) were peroxidated before assaying. The reaction mixture, containing 10 mM linoleate or oleate, 0 or 0.05 mM FeSO4, 0 or 0.5 mM desferal, 50 mM Tris/HCl (pH 7.5) and 0.05% NaN3, was incubated at 37 °C for 50 h (linoleate) or 12 days (oleate). As controls for peroxidation reaction, part of the mixture was immediately frozen without incubation and stored at −80 °C before assaying (Figure 9, columns A, C, E and G). The hydroperoxidated derivatives of NEFAs were determined by the Methylene Blue method  and a commercial chromogenic assay kit (Determiner LPO, Kyowa Medex, Tokyo, Japan).
Preparation of phospholipid liposomes
PA and DSPC were suspended in 10 mM Tris/HCl, pH 7.5, at a concentration of 8 mM. Each suspension was frozen and thawed five times, then emulsified by passing through a polycarbonate filter (pore size 0.1 μm) using a Mini-Extruder (Avanti Polar Lipids) at 60 °C.
Growth assay of Aβ2M amyloid fibrils
The reaction mixture containing 0–60 μg/ml S9 seeds, 0–25 μM r-β2-m, 50 mM phosphate buffer (pH 7.5), 100 mM NaCl, 0.05% (7.7 mM) NaN3, 0 or 5 mg/ml (75 μM) HSA, and 0–10 mM SDS or NEFAs was prepared at room temperature (25 °C) and incubated at 37 °C without agitation as described elsewhere . Fluorescence spectroscopy with 5 μM ThT was performed as described previously .
SDS/PAGE of the reaction mixture
After fibril growth reactions in the presence or absence of HSA, aliquots (100 μl) of the reaction mixtures were centrifuged at 21500 g for 3 h at 25 °C. After removal of the supernatants containing soluble r-β2-m, the pellets containing fibrils were washed with 50 mM phosphate buffer (pH 7.5) including 100 mM NaCl, then centrifuged at 21500 g for 2 h at 25 °C. The pellets were dried in a centrifugation evaporator for 1 h at 37 °C. SDS sample buffer containing 8 M urea was added to the dried pellets. From both supernatants and precipitates, aliquots equivalent to 3.07 μl of the initial reaction mixtures, containing 1 μg r-β2-m in total, were subjected to SDS/PAGE (10% gels) as described by Schägger and von Jagow . After Coomassie Brilliant Blue staining, the density of monomeric r-β2-m band in each fraction was quantified by scanning densitometry, using the FluoroChem IS-8000 Advanced Fluorescence, Chemiluminescence and Visible Light Imaging System (Alpha Innotech, San Leandro, CA, U.S.A.) with AlphaEaseFC image processing and analysis software.
Measurement of the CMCs of NEFAs
The CMCs of NEFAs were determined by using the fluorescence probe NPN . Optimum fluorescence measurements of NPN bound to the micelle were obtained at the excitation and emission wavelengths of 340 and 430 nm respectively on a Hitachi F-4500 fluorescence spectrophotometer. The fluorescence of the reaction mixture containing 0–10 mM NEFAs, 50 mM phosphate buffer (pH 7.5), and 100 mM NaCl was first measured at 37 °C as a blank. After addition of 1 μM NPN, the fluorescence of the mixture was measured at 37 °C. NPN fluorescence intensities were plotted against NEFA concentrations (Figure 6). Two straight lines defining the fluorescence in an essentially aqueous environment and in the micellar environment were traced before and after abrupt increase in NPN fluorescence, and the point of intersection of these lines was used to define the CMC. To determine the Krafft point, i.e. the lowest temperature to form the micelle, the NPN fluorescence of the mixture containing 0.5 or 1 mM palmitate, stearate or myristate was measured at 37–65 °C.
Far-UV CD spectra of r-β2-m in the presence of oleate
Far-UV CD spectra (198–250 nm) of r-β2-m were recorded on a Jasco 725 spectropolarimeter (Jasco) at 25 °C as described previously . The reaction mixture contained 25 μM r-β2-m, 50 mM phosphate buffer (pH 7.5), 100 mM NaCl and 0–5 mM oleate. The results are expressed in terms of mean residue ellipticity. Secondary structure content was calculated using the CDPro software package . The spectral values, ranging from 198 to 240 nm, at 1 nm intervals, were used for the calculation by three programs, CONTINLL, SELCON3 and CDSSTR, included in the CDPro package. CD data with more than 800 V of HT (high tension) voltage were excluded from the calculation. Two different reference data sets supplied with the package were used for the analysis. Each of the three programs was run using these reference sets. As a consequence, six independent estimates were obtained for each experimental spectrum.
Other analytical procedures
Aβ2M amyloid fibrils were observed by electron microscopy as described previously . Protein concentrations of r-β2-m and Aβ2M amyloid fibrils were determined by the method using bicinchoninic acid  and a commercial protein assay kit (code 23235, Pierce). The purified r-β2-m solution quantified by UV absorbance (280 nm) was used as the standard. Concentrations of unbound oleate in the mixture containing HSA were determined by the method using ADIFAB (acrylodan-labelled intestinal fatty acid-binding protein) and a commercial assay kit (catalogue number ADI-1-200, FFA Sciences LLC, San Diego, CA, U.S.A.) . Statistical analysis was done by one-way ANOVA with the post-hoc test by Dunnett, and linear least-squares fit. A value of P<0.05 was considered statistically significant.
Growth of Aβ2M amyloid fibrils by NEFAs at a neutral pH
S9 seeds incubated with 25 μM r-β2-m at pH 7.5 in the absence of NEFAs did not show a significant increase in ThT fluorescence (results not shown). In the presence of 1.0 and 5.0 mM laurate (Figure 2A), 1.0 and 5.0 mM myristate (Figure 2B) and 1.0 mM oleate and linoleate (Figure 2C), the fluorescence increased without a lag phase and proceeded to equilibrium after a 24–150 h incubation. As shown in Figures 2(E) and 2(F), the initial rate of the growth of S9 seeds was linear in relation to the seed concentration (0–60 μg/ml, r=0.997), as well as to the r-β2-m concentration (0–20 μM, r=0.998) in the presence of 1 mM oleate. Thus fibril growth in the presence of 1 mM oleate can be explained by first-order kinetics . The fluorescence signal increased with laurate and myristate concentration, peaking at 5 and 10 mM respectively (Figures 3A and 3B). The fluorescence signal increased with oleate and linoleate concentration, peaking at 1 mM before steadily decreasing above this level of NEFAs (Figures 3C and 3D). As shown in Figure 3(I), the fluorescence signal increased with SDS concentration, peaking at 0.5–1 mM before steeply decreasing above this level of SDS due to strong detergent activity . On electron microscopy, we observed fibril growth with the helical filament structure in the presence of 5 mM laurate (Figure 4B), 10 mM myristate (Figure 4C), 0.5 mM oleate (Figure 4D) and 0.5 mM linoleate (Figure 4E). After fibril growth reaction, the reaction mixtures were separated into supernatant and precipitate fractions by centrifugation, electrophoresed, and the density of monomeric r-β2-m band in each fraction was quantified by scanning densitometry. Oleate-dependent ThT fluorescence increase was well correlated with the increase in the amount of r-β2-m in the pellets, i.e. amyloid fibrils formed during the reaction (Figure 5A, r=0.99). No fluorescence increase was observed in the presence of 0.25 mM PA or DSPC liposomes (Figure 2G), 0–5 mM palmitate or stearate (Figures 3G and 3H). No fibril growth was observed in the presence of 2 mM palmitate or stearate (compare Figure 4A with 4H, and results not shown).
Growth of Aβ2M amyloid fibrils at a neutral pH
Effect of NEFA and SDS concentration on the growth of Aβ2M amyloid fibrils at a neutral pH
Electron micrographs of extended Aβ2M amyloid fibrils
SDS/PAGE analysis of the effect of oleate concentration on Aβ2M amyloid fibril growth
Although palmitate or stearate alone does not induce fibril growth, these species are abundant in the plasma [30,31]. To mimic the circulating NEFA composition in vivo, we prepared MixA and MixB. In the presence of 1.0 mM MixA or MixB (Figure 2D), the fluorescence increased without a lag phase and proceeded to equilibrium after a 24 h incubation. The fluorescence signal increased with MixA and MixB concentration, peaking at 1 mM before steadily decreasing above this level of NEFAs (Figures 3E and 3F). Under the electron microscope, fibril growth was observed in the presence of 1 mM MixA or MixB (Figures 4F and 4G).
Micelle formation of NEFAs
We estimated the CMC of laurate to be 1.8 mM from the point of intersection of two lines (Figure 6A). The CMCs of oleate and linoleate were estimated to be <0.1 mM (Figures 6B and 6C). Clear micelle formation of myristate, palmitate and stearate was not observed at 37 °C (Figure 6D). This may be explained by the observation that the Krafft points of myristate, palmitate and stearate were 43, 55–60 and >65 °C respectively (results not shown). Interestingly, clear micelle formation was observed with the CMC of <0.1 mM when palmitate and stearate were mixed with oleate and linoleate in volume proportions of 3:1:3:1.5 (Figures 6B and 6C, MixA). These data suggest that laurate, oleate, linoleate and MixA induce fibril growth above their CMCs (1.8, <0.1, <0.1 and <0.1 mM respectively) (Figures 3A, 3C, 3D and 3E). The fluorescence signal steadily decreased above 1 mM of oleate, linoleate, MixA and MixB (Figures 3C–3F). One possible explanation may be that these NEFAs exert weak detergent activity above 1 mM and denature amyloid fibrils formed. As for myristate, the fluorescence signal abruptly increased at 10 mM after a plateau over the concentration range of 0.5–5.0 mM. While we have no clear explanation for this finding, myristate and r-β2-m might form mixed micelles above 0.5 mM, and the structure of these might change at 10 mM.
CMC of NEFAs determined by fluorescence spectroscopic analysis with NPN
Effect of oleate on the conformation of r-β2-m
In the presence of 0–0.2 mM oleate, far-UV CD spectra of r-β2-m exhibited a positive peak at 202 nm and a negative peak at 221 nm immediately after the addition of oleate at 25 °C (Figure 7A). In the presence of 0.5–2.0 mM oleate, the spectra exhibited a transition with the ellipticities of the positive peaks decreasing dose-dependently. In the presence of 5 mM oleate, the spectra exhibited a largely disordered and partially helical state. These spectra are similar to those of r-β2-m in the presence of 0–5 mM SDS as previously reported . The secondary structure estimation indicated that the fractions of β-sheet and β-turn decreased with the increase in oleate concentration, while the fractions of α-helix and random structure increased (Figures 7C and 7D). After the incubation at 37 °C for 24 h, the spectra were almost unchanged in the presence of 0–1 mM oleate (Figure 7B). Thus the data in Figures 3(C), 6(B), 6(C) and 7 suggest that in vitro, a significant amount of a partially unfolded amyloidogenic conformer of r-β2-m for the fibril growth reaction may be formed by the interaction between r-β2-m and the oleate micelles.
Far-UV CD spectra of r-β2-m in the presence of oleate
Effect of HSA on the growth of Aβ2M amyloid fibrils in the presence of NEFAs
The long-chain NEFAs, including palmitate, stearate, oleate and linoleate, are primarily transported by plasma HSA . We examined whether NEFAs can induce fibril growth in the presence of HSA. In the absence of HSA, the fluorescence signal increased with oleate concentration, peaking at 1 mM before steadily decreasing above this level of oleate (Figure 8A). In the presence of HSA at 5 mg/ml (75 μM, one-tenth of the physiological plasma concentration), the fluorescence signal increased sigmoidally with oleate concentration, beginning to increase at 1 mM and reaching a maximum at 3 mM. A similar concentration dependency was observed with 0–5 mM MixA, in both the presence and absence of HSA (Figure 8B). Oleate-dependent ThT fluorescence increase was well correlated with an increase in the amount of r-β2-m in the pellets, i.e. amyloid fibrils formed during the reaction (Figure 5B, r=0.97).
Effect of HSA on the growth of Aβ2M amyloid fibrils in the presence of NEFAs at a neutral pH
X-ray structural analysis revealed seven binding sites of long-chain NEFAs in HSA and three of them exhibited high affinity . Thus 75 μM HSA can bind up to 525 μM NEFAs, and fibril growth could theoretically be observed above this concentration of NEFAs. Using ADIFAB , we measured unbound oleate concentrations in the mixture containing 75 μM HSA. Unbound oleate concentrations were 2.4±0.6 nM, 7.2±0.5 nM, 76±1.1 nM, 187±12 μM, 442±19 μM, 907±99 μM, 1.44±0.16 mM and 3.82±0.27 mM (means±S.D., n=3) for total oleate concentrations of 0.1, 0.2, 0.5, 1, 1.5, 2, 3 and 5 mM respectively. At 1 mM total oleate, unbound oleate concentration increased by three orders of magnitude as compared with that at 0.5 mM oleate (76±1.1 nM compared with 187±12 μM). Above 1 mM, the increase in unbound oleate concentration was correlated with the oleate-dependent ThT fluorescence increase and the amount of r-β2-m in the pellets, i.e. amyloid fibrils formed during the reaction (Figure 5B). Note that values at total oleate concentrations above the binding capacity (525 μM) are semi-quantitative, since dilution of the mixtures during measurement shifts the oleate–HSA binding equilibrium. On the other hand, values below the binding capacity are independent of dilution [39,42]. These results indicate that unbound NEFAs rather than HSA-bound NEFAs induce the fibril growth reaction in vitro.
Effect of NEFA oxidation status on the growth of Aβ2M amyloid fibrils
Since oxidative stress markers are elevated in haemodialysis patients, supporting the hypothesis that these individuals are under increased oxidative stress , we examined whether the oxidation status of NEFAs may influence their activity on fibril growth. Unsaturated NEFAs (e.g. linoleate and oleate) are peroxidated in vivo to form hydroperoxide, lipid free radicals or thiobarbituric acid reactive substance . Since linoleate is the most abundant polyunsaturated fatty acid in mammals, its lipid peroxidation products dominate among the peroxidated lipids [30,31,44]. When linoleate was preincubated alone (Figure 9A, column B), 44% of linoleate was peroxidated, possibly due to a trace amount of contaminated metal ions, and the fluorescence signal decreased significantly to 53% of the control value. When linoleate was preincubated with Fe2+ (Figure 9A, column D), a similar amount (44%) of linoleate was peroxidated, but the fluorescence decreased significantly to 19% of the control value, possibly due to the change in the composition of peroxidation products. Preincubation of linoleate with desferal, a metal ion chelator, inhibited the peroxidation of linoleate and recovered the fibril growth activity (Figure 9A, columns F and H). Irrespective of preincubation, addition of desferal significantly increased the fluorescence signal, possibly due to the suppression of oxidation during the reaction (Figure 9A, columns E–H). When Fe2+ or desferal was added after preincubation, the effect on the growth reaction was small (Figure 9A, compare columns B with I and J, and D with K).
Effect of NEFA oxidation status on the growth of Aβ2M amyloid fibrils at a neutral pH
Some NEFAs and their mixtures at micellar concentrations (100 μM order) induce the growth of Aβ2M amyloid fibrils at a neutral pH (Figures 2–6). These NEFAs are structurally similar to SDS. Moreover, like SDS, oleate partially unfolded the compact structure of β2-m to an aggregation-prone amyloidogenic conformer (Figure 7). Thus NEFAs may exert an amyloidogenic effect by interacting with β2-m between their alkyl chain and the hydrophobic side-chains of β2-m, as well as between their negatively charged carboxyl group and the positively charged side chains of β2-m. We speculate that the basic structure shared by SDS and NEFAs, i.e. one alkyl chain and one negatively charged hydrophilic group, is essential to the amyloidogenic effect because (i) synthetic detergents with positive (i.e. dodecyl trimethylammonium chloride), ampholytic (i.e. lauryl sulfobetain) and non-ionic (i.e. Triton X-100) hydrophilic groups did not induce the fibril growth at a neutral pH , (ii) lysophospholipids composed of an alkyl chain with one negative hydrophilic group (i.e. lysophosphatidic acids and lysophosphatidylglycerols) induced the fibril growth, whereas those containing ampholytic hydrophilic groups (i.e. lysophosphatidylethanoleamines and lysophosphatidylcholines) did not , and (iii) liposomes of phospholipids with two fatty acid chains and one negative (i.e. PA) or ampholytic (i.e. DSPC) hydrophilic group exhibited no amyloidogenic effect at a neutral pH (Figure 2G).
Biological molecules and metal ions involved in the pathogenesis of DRA
Relini et al.  reported that in temperature and pH conditions similar to those occurring in peri-articular tissues in the presence of flogistic processes, β2-m fibrillogenesis takes place in the presence of fibrillar collagen, indicating that collagen plays a crucial role in β2-m amyloid deposition under physiopathological conditions, suggesting an explanation for the strict specificity of DRA for the tissues of the skeletal system. Myers et al.  reported high yields of amyloid-like fibrils with wild-type β2-m, when assembly is seeded with fibril seeds formed from recombinant protein at pH 2.5, stabilized by the addition of heparin, serum amyloid P component, apolipoprotein E, uraemic serum, or synovial fluid, and concluded that the physiologically relevant factors enhance fibrillogenesis by stabilizing fibril seeds, thereby allowing fibril growth by rare assembly competent species formed by local unfolding of native monomers. Miranker's group demonstrated that, at 200 μM, Cu2+ destabilizes β2-m and promotes de novo fibre formation from monomeric β2-m in the presence of 0.5–1 M urea at 37 °C and neutral pH [48,49]. This indicates that Cu2+ contributes to the initiation step of amyloid fibril formation by destabilizing a native conformation of β2-m.
Clinical relevance of NEFAs to Aβ2M amyloid deposition
We observed that unbound NEFAs rather than HSA-bound NEFAs induce fibril growth reaction in vitro in the presence of HSA at 75 μM (one tenth of the physiological plasma concentration) (Figure 8). Hypoalbuminaemia is associated with mortality in long-term haemodialysis patients and is primarily caused by reduced albumin synthesis due to a response to inflammation . On the other hand, in patients receiving regular haemodialysis, the administration of heparin results in an acute rise in plasma NEFA, largely from the activation of lipoprotein lipase . Wessel-Aas et al.  compared the NEFA pattern of uraemic and normal plasma at 0 and 15 min after heparinization before the start of haemodialysis. At 15 min, the NEFA:HSA molar ratio of uraemic plasma (median value 5.99, ranging from 2.35 to 7.88) increased significantly as compared with the value at 0 min (median value 2.11, ranging from 1.12 to 2.33, P<0.01) and the 15 min value of normal plasma (median value 2.54, ranging from 2.22 to 4.86, P<0.01). Importantly, HSA concentration of uraemic plasma (median value 0.44 mM, ranging from 0.28 to 0.57) was significantly lower than that of normal plasma (median value 0.76, ranging from 0.70 to 0.83, P<0.01). Richieri et al. [39,42] studied the equilibrium binding of NEFAs with HSA by measuring the equilibrium levels of unbound NEFAs at pH 7.4 and 37 °C. Under conditions observed in normal human physiology (NEFA:HSA molar ratio less than 2), the equilibrium unbound concentration of a NEFA mixture (palmitate:stearate:oleate:linoleate=27:11:41:21; similar to our MixA) is less than 15 nM. In contrast, at NEFA:HSA molar ratios of 5 and 5.5, the equilibrium unbound concentration increased dramatically to 0.8 and 1.3 μM respectively. These values are comparable to the CMC of MixA (10 μM order, Figure 6C), suggesting that NEFA micelles may be at least transiently formed in uraemic plasma during haemodialysis. In patients with end-stage renal failure, the degradation of β2-m in the kidney does not occur and the serum β2-m levels increase from 0.1 μM to more than 5 μM . Thus, during haemodialysis, freshly released unbound NEFAs may interact with a significant amount of stagnant β2-m and subsequent accumulation of amyloidogenic β2-m conformer may result in the manifestation of DRA after a long incubation period.
Effect of linoleate peroxidation on the Aβ2M amyloid fibril growth
As shown in Figure 9(A), peroxidation of linoleate decreased its ability to induce fibril growth, possibly because the introduction of a hydroperoxide residue into the alkyl chain  interrupts the hydrophobic interaction between linoleate and β2-m and the subsequent conformational change of β2-m. Although the plasma of haemodialysis patients contained about 50% more lipid hydroperoxides than plasma from control subjects , these modified lipids are usually present as a small fraction of total plasma lipids: one hydroperoxide group per 104 fatty acid side chains . Thus, the increased oxidative stress in the haemodialysis patients may not significantly affect the amyloidogenic activity of linoleate in vivo.
In conclusion, this study suggested that the interaction between β2-m and NEFAs may be involved in the manifestation of DRA. The overall profile of lipid metabolism in haemodialysis patients needs to be analysed to elucidate the molecular pathogenesis of DRA.
Thanks are due to R. Nomura, N. Takimoto and H. Okada for excellent technical assistance.
acrylodan-labelled intestinal-fatty-acid-binding protein
critical micelle concentration
human serum albumin
non-esterified fatty acid
recombinant human β2-m
This research was supported in part by Grants-in-Aid for Scientific Research (C) (K. H.), Exploratory Research (T. O.), Scientific Research (B) (H. N.), Scientific Research on Priority Areas ‘Life of Proteins’ and ‘Water & Biomolecules’ (H. N.), and 21st Century COE Program (Medical Sciences) (H. N.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and for Research on Specific Diseases (H. N.) from the Ministry of Health, Labour and Welfare, Japan.