SEARCH

SEARCH BY CITATION

Keywords:

  • amyloidosis;
  • cleaved β2-microglobulin;
  • human β2-microglobulin;
  • NMR;
  • protein conformation

Abstract

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Cleavage of the small amyloidogenic protein β2-microglobulin after lysine-58 renders it more prone to unfolding and aggregation. This is important for dialysis-related β2-microglobulin amyloidosis, since elevated levels of cleaved β2-microglobulin may be found in the circulation of dialysis patients. However, the solution structures of these cleaved β2-microglobulin variants have not yet been assessed using single-residue techniques. We here use such methods to examine β2-microglobulin cleaved after lysine-58 and the further processed variant (found in vivo) from which lysine-58 is removed. We find that the solution stability of both variants, especially of β2-microglobulin from which lysine-58 is removed, is much reduced compared to wild-type β2-microglobulin and is strongly dependent on temperature and protein concentration. 1H-NMR spectroscopy and amide hydrogen (1H/2H) exchange monitored by MS show that the overall three-dimensional structure of the variants is similar to that of wild-type β2-microglobulin at subphysiological temperatures. However, deviations do occur, especially in the arrangement of the B, D and E β-strands close to the D–E loop cleavage site at lysine-58, and the experiments suggest conformational heterogeneity of the two variants. Two-dimensional NMR spectroscopy indicates that this heterogeneity involves an equilibrium between the native-like fold and at least one conformational intermediate resembling intermediates found in other structurally altered β2-microglobulin molecules. This is the first single-residue resolution study of a specific β2-microglobulin variant that has been found circulating in dialysis patients. The instability and conformational heterogeneity of this variant suggest its involvement in β2-microglobulin amyloidogenicity in vivo.

Abbreviations
β2m

β2-microglobulin

CE

capillary eletrophoresis

cK58-β2m

β2-microglobulin cleaved after lysine-58

dK58-β2m

β2-microglobulin with lysine-58 deleted

DRA

dialysis-related amyloidosis

ΔN3-β2m

β2-microglobulin devoid of N-terminal tripeptide

FID

free induction decay

The conformational behavior of β2-microglobulin (β2m) is of interest because this molecule is involved in dialysis-related amyloidosis (DRA) [1,2]. This condition, somehow induced by long-standing dialysis or renal insufficiency, is characterized by fibrillation and precipitation of β2m in osteoarticular tissues. Under normal conditions, β2m is a soluble plasma protein and also part of the MHC class I complexes on the surface of nucleated cells. It has become clear that this compact, seven β-stranded protein is conformationally unstable after cleavages and truncations, and that even intact β2m may, to a minor extent, adopt an alternative conformation at physiological pH [3]. Amyloid fibril formation from β2m in vitro requires nonphysiological conditions with respect to pH and ionic strength, the presence of divalent metal ions, or some of the truncations/deletions that have been reported to be present in β2m extracted from amyloid lesions [4–6]. The study of the behavior of β2m and β2m variants is relevant not only for DRA, but also for understanding common pathways of fibril formation in amyloidotic conditions such as Alzheimer's disease, transthyretin amyloidoses, immunoglobulin fragment amyloidosis, or some of the many other types of amyloidoses [7].

We have previously characterized two β2m variants, the first obtained by cleavage after Lys58 (cK58-β2m), and the second by further deletion of the same residue (dK58-β2m) (Fig. 1). It was shown that the concerted action of activated complement C1s and carboxypeptidase B cleaves β2m after Lys58, leading to cK58-β2m, and removes the same residue to generate dK58-β2m [8]. This limited proteolysis attacking a susceptible peptide bond residing in the loop between β-strands D and E of β2m (Fig. 1) increases the conformational heterogeneity of the cleaved β2m compared with the wild-type (wt) molecule [9,10]. The dK58-β2m variant may occur in vivo and has been reported to be generated in sera from patients with inflammation pathologies, cancer, and renal insufficiency [11–13]. Additionally, we recently showed, using dK58-β2m-specific antibodies, that dK58-β2m circulates in the blood of many dialysis patients [14].

image

Figure 1.  Structures of β2-microglobulin (β2m) and β2m variants. (A) View of the 20 best-fitting solution structures of wild-type β2m based on NMR restraints and tethered molecular dynamics. For the sake of simplicity, only the backbone is drawn, apart from the side chain of Lys58, which is highlighted. Designation of β-strands A–G is indicated. The local trace thickness corresponds to the spatial spreading over the best overlap of the structural family ensemble. Only the first members of the solution structure families were considered. Drawn with molmol[34]. (B) NMR-based solution structure of monomeric β2m (pdb entry: 1JNJ) in a ribbon drawing. The Lys58 residue (in red) and the Cys25 and Cys80 residues (yellow) connected by a disulfide bridge are shown in the backbone trace. Drawn with weblabviewerpro 3.7. (C) Schematic drawing of the variants of β2m generated by limited proteolysis of the wild-type molecule. From the single-chain wild type, a heterodimeric molecule (cK58-β2m), in which the two chains are connected by a disulfide bridge, is generated by cleavage between the Cys residues. The further trimming (removal of Lys58) of cK58-β2m generates the dK58-β2m variant.

Download figure to PowerPoint

The conformations of cK58-β2m and dK58-β2m have not previously been probed at the single amino acid level and correlated with the solution stability of these molecules. We therefore here explore the structural features and stability of the Lys58-cleaved β2m variants compared with those of wt β2m by a combination of NMR spectroscopy, MS, and capillary electrophoresis (CE).

Results and Discussion

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Solution stability of cleaved β2m variants monitored by 1H-NMR spectroscopy

When β2m is modified by limited proteolysis cleaving the chain between the Cys25 and Cys80 residues, a heterodimeric molecule consisting of two chains held together by a disulfide bridge is generated. This molecule (cK58-β2m) is further processed in vivo to the dK58-β2m variant, which lacks the K58 residue exposed in the A-chain of cK58-β2m (Fig. 1) [11]. The behavior of cK58-β2m and dK58-β2m in solution was studied by a series of one-dimensional 1H-NMR spectra collected at different conditions of temperature and protein concentration. The stability of concentrated solutions (c. 0.3 mm) in the temperature range between 288 and 310 K was first investigated. The one-dimensional 1H spectra of cK58-β2m and dK58-β2m collected at 288 K (Fig. 2A) exhibit the typical resonance pattern of the folded protein, with a few resolved peaks in the aliphatic and aromatic regions. In particular, the upfield shifts of Val37, Ile35 and Leu23, due to the proximity of aromatic residues such as Tyr66, Phe30, Phe70 and Trp95 (Fig. 2B), are diagnostic of tertiary structure interactions in the hydrophobic core and represent a signature of the native fold of the β2m molecule (Fig. 2A, lower panel) [15]. When the temperature is increased in steps of five degrees up to 298 K, the lower solution stability of dK58-β2m compared to cK58-β2m is highlighted. While the latter at 298 K maintains a folded conformation, the variant devoid of Lys58 is less stable and undergoes slow unfolding and aggregation over time, as shown in Fig. 3. The unfolding is evidenced by the progressive loss of spectral spreading and the simultaneous growth of some main envelope at the typical frequencies of unfolded polypeptides (around 1 p.p.m.). The formation of large aggregates is suggested by the broadening linewidth and the related decrease of the overall integral value under equivalent NMR acquisition conditions. Over the −2/12 p.p.m. region, the spectra of dK58-β2m shown in Fig. 3 exhibit signal losses of 16% and 33%, respectively, corresponding to 10 and 41 h at 298 K. In the absence of overt precipitation, this suggests the formation of aggregates with substantially larger linewidths. The loss of stability and the formation of large, soluble aggregates in dK58-β2m solutions at 310 K over time were suggested previously by CE analyses, and evidenced by size-exclusion chromatography with light-scattering detection. In these experiments a well-defined aggregate formation with an aggregate size of about 50 nm or 5 × 106 g·mol−1 was noted [10]. No estimate of the aggregate dimensions by measurement of translational diffusion coefficients using diffusion-ordered 2D-NMR spectroscopy experiments [16,17] was possible in the present study, because the relatively low sample concentration (0.3 mm) prevented reliable exponential fitting of the experimental data.

image

Figure 2.   (A) One-dimensional 1H-NMR aliphatic (right) and aromatic (left) region of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) and β2m with Lys58 deleted (dK58-β2m), 0.3 mm, at 288 K and pH 7.4, and of wild-type β2m, 0.7 mm, at 310 K at pH 6.6, observed at 500 MHz. The upfield shift of Val37, Ile35 and Leu23, which is diagnostic of tertiary structure interactions in β2m native folding, is highlighted. (B) Representation of the β2m hydrophobic core and of the aliphatic residues giving rise to the most upfield-shifted methyls in the 1H-NMR spectrum. Val37, Ile35 and Leu23 (green) are placed in the shielding cone of aromatic rings (red). Only the most important residues are included in the plot. The plot was drawn using weblabviewerpro 3.7 (Accelrys Inc., San Diego, CA, USA).

Download figure to PowerPoint

image

Figure 3.  One-dimensional 1H-NMR traces of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) and β2m with Lys58 deleted (dK58-β2m) at 298 K and pH 7.4. At 298 K, cK58-β2m exhibits the typical folded protein spectrum, whereas dK58-β2m undergoes an unfolding–aggregation process that is monitored at 0, 10 and 41 h from the temperature setting. The intensity of the upfield-shifted resonances of Leu23, Ile35 and Val37 gradually diminishes, while the envelope around 1 p.p.m. increases. Simultaneous changes are observed in the aromatic region, involving a loss of signal dispersion. The overall integral value is reduced after 41 h.

Download figure to PowerPoint

Upon further increase of the temperature to 310 K, cK58-β2m eventually slowly undergoes the same unfolding–aggregation process as observed for dK58-β2m (data not shown). In accordance with earlier observations using other methods [10], this thermal transition is irreversible (data not shown).

A different behavior is found when obtaining a series of one-dimensional 1H spectra at 310 K using more dilute solutions of cleaved β2m variants (c. 0.05 mm). In contrast to the results at 0.3 mm, the unfolding–aggregation process at a concentration of 0.05 mm is very slow. This is indicated by only a minor decrease of the diagnostic upfield-shifted peaks of Leu40, Val37, Ile35 and Leu23, even after 4 days (Fig. 4). Nevertheless, a slight and continuous modification of the tertiary structure is evident from the slow overall drift of the resonance system with a pattern suggesting loss of conformational homogeneity. After some 60 h, for both cK58-β2m and dK58-β2m, the presence of shoulders within the monitored isolated peaks indicates the presence of two or more conformers in equilibrium (peak shoulders are indicated by asterisks in Fig. 4). Further evidence for conformational heterogeneity comes from several other envelope changes that appear when the spectra are superimposed (data not shown).

image

Figure 4.  Details of one-dimensional 1H-NMR traces of diluted β2-microglobulin (β2m) with Lys58 deleted (dK58-β2m) and β2m cleaved after Lys58 (cK58-β2m) solutions (0.05 mm) at 310 K and pH 7.4. At low concentration, the unfolding process is very slow, as indicated by an only very minor decrease of the intensity of the diagnostic peaks of Leu23, Ile35, Val37 and Leu40, even after some days. The presence of one or more conformational isomers, indicated by the resonance splitting of some isolated peaks (highlighted by asterisks), is particularly manifest in the spectra recorded after more than 100 h of incubation at 310 K, but may also be noticed after 60 h and to some degree in the very first recorded spectra (t = 0 h). The increasing splitting between the peaks assigned to Ile35 Hδ1 and Leu23 Hδ2, which is especially noticeable in the left panel, is consistent with a slow, continuous modification of tertiary structure, which takes place at 310 K in the dilute protein solution.

Download figure to PowerPoint

Protein aggregation monitored by capillary electrophoresis

In contrast to wt β2m, which is freely soluble in physiological buffers up to at least 10 mg·mL−1 (0.85 mm), the cleaved variants, in particular dK58-β2m, are prone to aggregation at high protein concentrations, especially at increased temperatures. Visible precipitation occurs over time at concentrations higher than 2 mg mL−1 (0.17 mm) for the dK58 variant; the cK58 variant is more stable. The aggregation behavior at different concentrations and temperatures was characterized by CE (Fig. 5). In these experiments, the changes in the amount of soluble material were followed over time. As shown in Fig. 5A, a 1 mg·mL−1 (0.09 mm) dK58-β2m solution incubated at increasing temperature initially exhibits a shift in the conformational equilibrium between the fast (f) and slow (s) species to more of the (s) species, which is believed to be a partly unfolded intermediate (as can be seen below). Subsequently, at higher temperatures, an irreversible loss of soluble material occurs. In Fig. 5B, an analysis of soluble material over time at a fixed sample temperature of 308 K at two different protein concentrations, 0.9 mg·mL−1 (0.08 mm) and 2.5 mg·mL−1 (0.22 mm), clearly show the loss of solubility in the higher-concentration solutions of both variants, whereas at lower concentrations both species have constant peak areas from 0 to 24 h. This dependence of the solution stability of cleaved β2m on its concentration is in agreement with the NMR results presented above.

image

Figure 5. Capillary electrophoresis separation of β2-microglobulin (β2m) with Lys58 deleted (dK58-β2m) incubated at different temperatures. (A) Separation profiles to show that sample temperature (indicated in the figure) influences the ratio between f and s conformers of dK58-β2m in CE. All CE experiments were performed at a constant capillary temperature of 278 K to preserve the distribution of conformers in the injected samples. Shown are overlayed electropherograms with time windows showing the s and f conformer peaks. Samples were 1 mg·mL−1 dK58-β2m electrophoresed at 278 K using 90 µA constant current after injection for 2 s. (B) Aggregation propensity of β2m cleaved after Lys58 (cK58-β2m) and dK58-β2m at different protein concentrations. Soluble material was monitored by CE as a function of incubation time. Samples of 0.9 mg·mL−1 (triangles) or 2.5 mg·mL−1 (circles) β2m variants (cK58, open symbols; dK58, filled symbols) were kept at 308 K, and aliquots (2 s injections of high-concentration samples and 4 s injections of low-concentration samples) were analyzed by CE performed at constant current of 80 µA, with the capillary cooling fluid maintained at 278 K. Samples also contained 0.2 mg·mL−1 of a marker peptide. Shown are the summed peak areas P (total area of f + s peaks) divided by the marker peak area M at different time points as a percentage of the initial value of P/M at the onset of the experiments where the sample temperature was brought from 278 K to 308 K.

Download figure to PowerPoint

MS analysis of global conformation by amide hydrogen (1H/2H) exchange

We have previously shown that native wt β2m and dK58-β2m undergo transient cooperative unfolding, evidenced by a correlated isotopic exchange of amide hydrogens [10]. This type of exchange mechanism (EX1) leads to the appearance of distinct bimodal isotopic envelopes in the mass spectra. The lower mass peak of this envelope represents the population of molecules that has not yet undergone cooperative unfolding; the higher mass peak represents the population of molecules that has been in the unfolded state and thus undergone correlated exchange. To investigate the structural stability of the folded states of wt β2m, cK58-β2m and dK58-β2m, the exchange kinetics of the folded populations were determined at 298 K (Fig. 6). At this temperature, a gradual mass increase with exchange time is observed for the lower-mass population. This is due to the noncorrelated exchange mechanism, which in structural terms can be explained by small-amplitude fluctuations within the protected core. The noncorrelated isotopic exchange kinetics shown in Fig. 7 was determined by the mass difference of the lower-mass populations relative to the fully deuterated control. Fig. 7 shows that at the shortest deuteration period (t = 0.5 min), all three proteins contain the same number (i.e. 32; this number is also displayed in Fig. 6) of 1H atoms not yet exchanged for deuterium. This indicates that an identical number of protecting hydrogen bonds exists in the folded states of wt β2m, cK58-β2m and dK58-β2m. Furthermore, the cleaved variants, cK58-β2m and dK58-β2m, exhibit very similar noncorrelated exchange kinetics (Fig. 7). This indicates that the stability of the hydrogen bond network that confers protection against isotopic exchange is almost identical for these proteins. However, with prolonged incubation this network appears to be slightly more stable in wt β2m than in the cleaved species (Fig. 7).

image

Figure 6.  Global amide hydrogen (1H/2H) exchange analysis of the folded conformations of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) and β2m with Lys58 deleted (dK58-β2m) at 298 K in deuterated NaCl/Pi. The proteins were incubated pairwise in deuterated NaCl/Pi buffer. After various periods of deuteration, isotopic exchange was quenched by acidification. Subsequently, the samples were desalted at quench conditions and analyzed by ESI-MS. Shown are the ESI mass spectra of a mixture of cK58-β2m and dK58-β2m obtained after various deuteration periods (given in minutes in the figure) at 298 K. Left panel: deconvoluted ESI mass spectra. Right panel: ESI mass spectra of the m/z region with the [M + 9H]9+ ions. The spectra obtained at t = 0 min (i.e. lowest traces) were obtained from 1H2O. Ox, Met99-oxidized species.

Download figure to PowerPoint

image

Figure 7.  Noncorrelated exchange kinetics of the folded conformations of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) (triangles), β2m with Lys58 deleted (dK58-β2m) (crosses), and wild-type (wt) β2m (circles). Shown are mass shifts (expressed as loss of protected protiated residues to adjust for differences in chain lengths) at 298 K as a function of time incubated in deuterated NaCl/Pi.

Download figure to PowerPoint

Thus, in accordance with the NMR results, the global conformation of wt β2m appears to be conserved in the cleaved forms of β2m. Note that in these experiments only the slowly exchanging hydrogens are monitored. Thus, contributions from amide hydrogens in loop regions and from new termini generated in the cleaved variants are not expected to affect the exchange count.

Two-dimensional NMR characterization

The detailed interpretation of the 1H-NMR spectra of β2m variants is based on the parent spectra of wt β2m obtained at different temperature and pH values [15,18] and requires the checking and redetermination of most of the resonance assignments of the molecule under investigation according to the standard methodology, i.e. going through scalar and dipolar connectivity patterns for each amino acid residue [19]. This work could be almost entirely completed for cK58-β2m, but only partially for dK58-β2m. The difficulty with both variants, particularly dK58-β2m, is due to their thermal lability (unfolding and aggregation). This prevented the use of optimal temperatures (e.g. 310 K) to improve data quality with concentrated samples (e.g. 0.5 mm). Increasing the temperature up to 320 K, whenever possible, generally improves the NMR data quality for 10–15 kDa proteins by reducing linewidths and thus favoring spectral analysis. As a compromise in the present study, the two-dimensional TOCSY and NOESY spectra of cK58-β2m were obtained at 298 K, whereas the best results with dK58-β2m were generated at 310 K by working with a very dilute sample (0.05 mm).

The assignment lists (supplemental Tables 1 and 2) indicate an overall conservation of the resonance frequencies with respect to the corresponding wt values and thus confirm the retention of the main features of the native structure in both variants. The backbone Hα chemical shifts of cK58-β2m and dK58-β2m (wherever assignments were available) were compared to the corresponding values of the wt protein, as shown in Fig. 8. As expected, the largest deviations of Hα chemical shifts of cK58-β2m are found in proximity to the cleavage site, more specifically in fragments 56–58 and 59–64, i.e. at the opened loop D–E, and at the start of strand E [15]. Interestingly, similar deviations are also found in fragment 26–35, i.e. at the end of strand B and at loop B–C, which faces the D–E region. According to the well-established correlation between Hα chemical shifts and secondary structure in polypeptides [20], the shifts of cK58-β2m Hα resonances compared to wt β2m reflect changes in the backbone arrangement within the D–E as well as in the B–C loop. Thus, two opposite changes of secondary structure are found at the end of strand D and the beginning of strand E, i.e. a further loss in D and a stabilization in E of the local β-structure geometry. Compared with wt β2m, the cK58-β2m molecule is thus most conformationally different in the cleavage site region (D–E loop), with additional involvement of the adjacent residues of strands D and E, and the facing residues of loop B–C. Unfortunately, this analysis could not be extended to dK58-β2m, because of the ambiguous assignment of residues from these regions of the molecule.

image

Figure 8.  Two-dimensional NMR study of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) and β2m with Lys58 deleted (dK58-β2m) at pH 7.4. The assigned backbone Hα chemical shifts of 0.3 mm cK58-β2m at 298 K, and of 0.05 mm dK58-β2m at 310 K, are compared with the corresponding values of the wild-type (wt) species obtained at 310 K and pH 6.6. The ΔδHα values (p.p.m.) are reported as (Δvariant − Δwt). Residue labels are omitted in regions where the resonance assignment was ambiguous.

Download figure to PowerPoint

Conformational heterogeneity of cK58-β2m and dK58-β2m

The whole NMR dataset for cK58-β2m and dK58-β2m revealed the occurrence of at least two different conformers for each molecule. These conformers were undergoing slow exchange on the chemical shift timescale. Examination of the two-dimensional maps obtained with concentrated cK58-β2m at 298 K showed a generalized resonance doubling at the locations and to the extent reported in Fig. 9. The features of the pattern of the second conformer resemble the features of a minor monomeric intermediate occurring along the β2m-refolding pathway that was named I2 and initially identified by Chiti et al. in wt β2m [3]. The I2 conformer was subsequently also detected in real-time NMR experiments [21]. Further analysis of other amyloidogenic β2m variants, and in particular of the species devoid of the N-terminal tripeptide, ΔN3-β2m, has shown that the I2 conformer is in equilibrium with the fully folded species [21,22]. This indicates that it can be precisely identified through NMR characterization. In the case of ΔN3-β2m, the observation of resonance doubling for the side chain signals of residues Val9, Ser11, Leu23, Val37 and Ala79, which are all close to one or more aromatic residues in the cluster of Tyr26, Tyr66, Phe70, Tyr78 and Trp95 [15], strongly suggested that I2 corresponds to a slightly destabilized fold that has the overall conformation of wt β2m, but exhibits a looser packing of its hydrophobic core. This interpretation was recently challenged by Kameda et al.[23], who reported evidence in favor of a slow transcis isomerization of Pro32 during refolding of β2m. Whatever the origin of the conformational equilibrium that gives rise to the slow refolding step of β2m, the proposed correspondence of the second form observed in the cK58-β2m spectra with the I2 conformer identified in ΔN3-β2m is based on the similarity of the resonance doubling patterns of the two variants. This analogy is visualized in Fig. 10, where details of NOESY spectra are shown. In spite of the different conditions of temperature and pH, the close similarity of the patterns is readily appreciated. The excellent resolution of the resonances in Fig. 10 could not be exploited for quantitation of the relative concentrations of the two forms because, in general, NOESY cross-peak amplitudes are determined by the actual motional characteristics of the connected nuclear pairs, and thus may differ between distinct conformers [24]. Many other resonance doublings were observed (the most relevant are reported in Fig. 9), all consistent with the expected pattern of the I2 intermediate that was unambiguously recognized in previous studies of other β2m variants [21,22]. The best estimate of the equilibrium populations of the fully folded and I2 forms for cK58-β2m at 298 K was obtained by using, for each conformer, the pair of TOCSY connectivities assigned to Val37 Hγ1–Hγ2. Taking into account the partial overlap of the specific cross-peaks, the resulting relative amount of I2 at 298 K was 19 ± 9% of the total protein. The occurrence of an I2 intermediate in equilibrium with the main species was also deduced from the dK58-β2m NMR spectra, although the lower resolution made it necessary to rely more on peak shape distortion than on actual peak separation (Figs 4 and 11B).

image

Figure 9.  Resonance doublings in β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) indicating conformational heterogeneity. The proton chemical shifts of the alternative conformer (I2) of cK58-β2m are compared with the corresponding values of the natively folded form of cK58-β2m in the graph. The two conformers are recognized in TOCSY and NOESY maps, obtained at 298 K and pH 7.4. The ΔδH values (p.p.m.) are reported as (δI2 − δN), where N stands for the natively folded form. Only the most relevant deviations are shown. Resonance doubling observed elsewhere was less pronounced in terms of Δδ.

Download figure to PowerPoint

image

Figure 10.  Details of two-dimensional NMR NOESY maps of β2-microglobulin (β2m) cleaved after Lys58 (cK58-β2m) (left) and β2m devoid of the N-terminal tripeptide (ΔN3-β2m) (right), recorded at 500 and 800 MHz, respectively. The intraresidue connectivities Hε1–Hδ1 (top) and Hε1–Hζ2 (bottom) of Trp95 are indicated for the natively folded form (N) and for the I2 form, in equilibrium under the chosen experimental conditions.

Download figure to PowerPoint

image

Figure 11.  Conformational heterogeneity of cleaved β2-microglobulin (β2m) by NMR and CE. (A) CE analysis of β2-microglobulin cleaved after Lys58 (cK58-β2m) and β2m with Lys58 deleted (dK58-β2m) kept at 298 K and separated at 283 K. Samples (2.5 mg·mL−1) were injected for 2 s and analyzed at a constant current of 90 µA in 0.1 m phosphate, pH 7.4, with a capillary temperature setting of 283 K. The f and s conformers are indicated. (B) Details of two-dimensional NMR TOCSY spectra of dK58-β2m and cK58-β2m obtained at 500 MHz, and β2m devoid of the N-terminal tripeptide (ΔN3-β2m) recorded at 800 MHz. The specific intraresidue connectivities Hδ*–Hz of Phe70 arising from the natively folded form (N) and I2 intermediate (I2) are labeled.

Download figure to PowerPoint

Two conformers of the β2m variants cleaved at Lys58 were also detected by CE as previously reported [9] and are shown in Figs 5A and 11A. The precise nature of the slow conformer peak (labelled ‘s’ in Fig. 5A and 11A) could not be unequivocally determined in these experiments. The two populations observed in CE separations of samples kept at 298 K gave, for the slow-migrating conformer, concentrations of 38% ± 2% for cK58-β2m and 30% ± 4% for dK58-β2m (triplicate experiments ± SD), relative to the total peak area, independently of the total β2m concentrations used (examples are shown in Fig. 11A). CE separations are accomplished at low temperature in 10–12 min. Thus, solution states are sampled under dynamic conditions where the conformers are being separated from each other, whereas NMR spectra record steady-state solution distributions. Such differences in experimental conditions may explain the differences in the relative concentration estimates for the two conformers. However, both the NMR and CE approaches strongly support the notion of conformational heterogeneity of the cK58-β2m and dK58-β2m variants.

Conclusions

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Although cK58-β2m is unlikely to have a long lifetime in vivo, where the exposed Lys58 is rapidly cleaved off by endogeneous carboxypeptidase B activity [25], this variant was included in our study because it is more stable in solution than dK58-β2m and thus more accessible to analysis. It has very similar characteristics in all the MS and CE analyses. However, we found for both β2m variants that the protein concentrations required for high-resolution NMR spectroscopy were detrimental to their stability in solution. The two cleaved β2m species have a pronounced propensity to undergo temperature-dependent unfolding and aggregation. In addition, the data show the occurrence of conformational heterogeneity in cK58-β2m and dK58-β2m solutions, which is consistent with their thermal lability. Despite these difficulties, detailed characterization of the conformational states of the cK58-β2m and dK58-β2m variants has now been accomplished, and has made it possible, by reference to the NMR pattern of the ΔN3 variant of β2m, to identify a minor conformational species that also exists in the conformational equilibrium of the cleaved β2m variants. This conformer is a monomeric intermediate (I2) occurring on the β2m-refolding pathway. These findings are consistent with the existence, in addition to the folded conformation, of a less abundant form with amyloidogenic features, which has also been suggested by CE experiments [9,10,22].

The unfolding processes that are observed by NMR with temperature increase may be driven by the seeding probability, which increases with protein concentration and brings about increased recruitment of monomers or small oligomers onto the surface of soluble large aggregates. The occurrence of large, well-defined aggregates has been recently demonstrated by size exclusion chromatography of dK58-β2m incubated at 310 K [10].

The NMR data clearly indicate that the overall folding pattern of the cleaved β2m variants is very similar to that of the wt protein. In fact, distinct differences in the conformation of the variants are confined to the cleavage site region (the D–E loop) with additional involvement of the adjacent residues of strands D and E and the facing residues of loop B–C. Accordingly, the noncorrelated amide hydrogen (1H/2H) exchange experiments indicate that only slightly increased protection is conferred by the hydrogen bonds in wt β2m.

The reduced thermostability of β2m cleaved at Lys58 occurs despite an overall native-like folding and stems from a single cleavage in a rather mobile and exposed loop region [15,21]. This cleavage is known to be mediated by complement enzymes that may be activated during inflammation. To substantiate a relationship between the molecular destabilization characterized here and the formation of amyloid in vivo, these β2m molecular variants should be present in amyloid lesions from patients. Conversely, given the propensity of β2m for specific cleavage, the pheomenon may be part of a physiological system marking β2m for clearance in the circulation and possibly failing in amyloidosis. In any case, the results reported here provide a further basis for understanding the link between in vivo stability and the amyloidogenicity of conformationally unstable β2m variants.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Protein purification

β2m cleaved at Lys58 (cK58-β2m) and with an additional deletion of residue 58 (dK58-β2m) (Fig. 1) were derived from wt β2m purified from a pool of urine from nephropathy patients, as described previously [12]. cK58-β2m and dK58-β2m were generated by treating purified wt β2m with activated complement C1s in the presence or absence of a carboxypeptidase B inhibitor, as previously described [11]. Molecular masses of the purified proteins were determined by MS on a Mariner ESI-TOF biospectrometry workstation (Applied Biosystems, Foster City, CA, USA) and were in agreement with the theoretical masses of 11729.2 Da (wt β2m), 11747.2 Da (cK58-β2m), and 11619.0 Da (dK58-β2m). The purified proteins were kept in NaCl/Pi (137 mm NaCl/2.7 mm KCl/1.5 mm KH2PO4/6.5 mm Na2HPO4, pH 7.4) at −20 °C until used.

Nuclear magnetic resonance (NMR) spectroscopy

1H-NMR spectra of cK58-β2m and dK58-β2m were obtained at 500.13 MHz with a Bruker Avance spectrometer on approximately 0.3 mm and 0.05 mm samples dissolved in NaCl/Pi at pH 7.4. Deuterium oxide (Cambridge Isotope Laboratories, Andover, MA, USA, 99.9 atom percentage D) was added (5% by volume) for frequency lock purposes. The concentrated dK58-β2m sample (0.3 mm, 3.5 mg·mL−1) was obtained from a 2 mg·mL−1 solution by centrifugal ultrafiltration in 5 kDa cut-off vials. The cK58-β2m samples (0.3 mm and 0.05 mm) and the diluted dK58-β2m sample (0.05 mm) were filtered before transfer into the NMR tube using 0.22 µm-pore syringe filters (Millipore, Bedford, MA). The temperature effect on solution stability was monitored over time by observing the concentrated proteins at 288, 293, 298 and 310 K. To probe for the effect of dilution, a series of (one-dimensional) NMR experiments was performed with 0.05 mm solutions at 310 K over 1 week. To assign resonances, two-dimensional TOCSY [26] and NOESY [27] spectra were recorded for both cK58-β2m and dK58-β2m. Different temperatures between 288 and 310 K were explored to collect data with the concentrated samples until a folded protein conformation appeared to be conserved. Typical two-dimensional acquisition schemes included: solvent suppression by excitation sculpting [28], 1–1.5-s steady-state recovery time, mixing times of 38–50 ms for TOCSY and 150 ms for NOESY, and t1 quadrature detection by the time proportional phase incrementation method [29]. The spin-lock mixing in the TOCSY experiments was obtained with an MLEV17 [30] pulse train at γB2/2π = 7–10 kHz, sandwiched by two purging pulses of 0.75 ms. Acquisitions were performed over a spectral width of 8012.820 Hz in both dimensions, with matrix sizes of 1024–2048 points in t2 and 512 points in t1, and 128–256 scans for each t1 free induction decay (FID) (total maximum experiment duration was 47 h). In an effort to improve the two-dimensional data quality for dK58-β2m, a diluted sample (0.05 mm) was examined at 310 K. A NOESY spectrum was recorded at 800.13 MHz with a cryoprobe-equipped Bruker DRX spectrometer. The acquisition was performed over a spectral width of 12 820.513 Hz in both dimensions and with a mixing time of 150 ms. The total experimental time was c. 14 h for 2048 points in t2, 256 points in t1, and 128 scans for each t1 FID. The corresponding TOCSY experiment was performed at 500.13 MHz, using a DIPSI-2 isotropic mixing train lasting 28 ms [31], solvent suppression by excitation sculpting [28], 1 s steady-state recovery time and t1 quadrature detection by the echo–antiecho method [29]. The time needed to collect a signal intensity amenable to analysis was c. 61 h for 1500 points in t2, 256 points in t1, and 696 scans for each t1 FID. Apodizations by Gaussian multiplication in t2 and shifted (72°) square sinebell in t1 were applied for processing using the Bruker software. In general, however, data processing and analysis were performed using Felix (Accelrys Inc., San Diego, CA) software with shifted (60–90°) square sinebell apodization and zero filling (up to 2048 × 1024–2048 real points). All spectra were referenced on the Leu23 CδH3 resonance at −0.58 p.p.m.

Amide hydrogen (1H/2H) exchange monitored by MS

Deuterated NaCl/Pi was prepared by lyophilization of protiated buffer followed by redissolution in D2O. To achieve full deuteration, the deuterated buffers were twice lyophilized and redissolved in D2O. Isotopic exchange was initiated by dilution (1 : 50) of the protiated protein solution with deuterated buffer, resulting in a final protein concentration of 20 µg·mL−1. Typically, 10 µL of wt β2m, cK58-β2m or dK58-β2m (c. 1 mg·mL−1 in protiated NaCl/Pi) was added to 490 µL of deuterated NaCl/Pi, pH 7.3 (value uncorrected for isotope effects). The proteins were incubated pairwise at equimolar concentrations at 25 °C in a thermomixer. At appropriate intervals, 50 µL aliquots were withdrawn and quenched by adding 2 µL of 2.5% trifluoroacetic acid, which lowered the pH to 2.2 (uncorrected value). The samples were stored in liquid N2 until analysed by ESI-MS in positive ion mode on a quadrupole time-of-flight mass spectrometer (Model Q-TOF 1; Micromass, Manchester, UK). The MS instrument was coupled to rapid desalting equipment, as described previously [10]. The total time for desalting and elution was less than 2 min. The solvent precooling coils, injector with loop, valve and microcolumn were immersed in ice–water slurry (0 °C) to minimize back-exchange with the protiated solvents. The desalting step mainly removes deuterium exchanged for labile hydrogens, i.e. hydrogen attached to N, O and S in the side chains, and not main chain amide hydrogens at acidic pH [32]. Thus, the mass increase observed after deuteration and desalting primarily reflects deuterium incorporated into the main chain amide groups. Back-exchange control experiments were performed to determine the inevitable deuterium loss that occurs during desalting under quench conditions (pH 2.2 and 0 °C), where the exchange kinetics of main chain amide hydrogens is very slow. Aliquots of fully deuterated wt β2m, cK58-β2m and dK58-β2m under quench conditions (pH 2.2 and 0 °C) were subjected to rapid desalting and they were observed to contain 88, 87 and 86 ±1 deuterium atoms, respectively. Since wt β2m, cK58-β2m and dK58-β2m contain a total of 93, 92 and 91 main chain amide hydrogens, respectively, approximately five deuterium atoms are back-exchanged with hydrogen atoms under quench conditions.

CE

A Beckman P/ACE 5010 instrument with sample cooling and UV detection facilities and placed in a cold room was used. Electrophoresis buffer was 0.1 m phosphate, pH 7.4. Detection took place at 200 nm and the separation tube was a 50 µm inner diameter uncoated fused silica capillary of 57 cm total length with 50 cm to the detector window. Separations were carried out at 80 or 90 µA constant current corresponding to a field strength of about 490 V·cm−1. The capillary cooling fluid and the samples were kept at the temperatures indicated. Samples (30 µL, 0.9 mg·mL−1 (0.08 mm) or 2.5 mg·mL−1 (0.22 mm) of cK58-β2m or dK58-β2m, both with 0.2 mg·mL−1 marker peptide (M) added), were protected against evaporation by 15 µL of overlayed light mineral oil (Sigma M-3516, St Louis, MO, USA) [33]. Injected sample volumes were approximately 2.3 or 4.5 nL (2 or 4 s pressure injection for 2.5 mg·mL−1 and 0.9 mg·mL−1 samples, respectively). Data were collected and processed with the beckman system gold software (Beckman, Fullerton, CA, USA). The capillary was rinsed after electrophoresis for 1 min with each of 0.1 m NaOH and water and then prerinsed for 2 min with electrophoresis buffer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

This work was supported by MIUR (COFIN 2003) and by Sygesikringen ‘danmarks’ forskningsfond, Apotekerfonden af 1991, The Danish Medical Reseach Council, Lundbeckfonden, and M. L. Jørgensen og Gunnar Hansens Fond. CarlsbergFondet is acknowledged for financial support to TJDJ. The advice of Professor V. Bellotti and the assistance of Dr A. Makek are gratefully acknowledged. A special acknowledgement is due to CERM, Florence (Italy), for the use of their 800 MHz NMR facility.

References

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Results and Discussion
  4. Conclusions
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information
FilenameFormatSizeDescription
FEBS_5254_sm_tableS1-S2.pdf155KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.