Structure, Stability, and Aggregation of β-2 Microglobulin Mutants: Insights from a Fourier Transform Infrared Study in Solution and in the Crystalline State

Abstract

β-2 microglobulin (β2m) is an amyloidogenic protein involved in dialysis-related amyloidosis. We report here the study of the structural properties of the protein in solution and in the form of single crystals by Fourier transform infrared (FTIR) spectroscopy and microspectroscopy. The investigation has been extended to four β2m mutants previously characterized by x-ray crystallography: Asp⁵³Pro, Asp⁵⁹Pro, Trp⁶⁰Gly, and Trp⁶⁰Val. These variants displayed very similar three-dimensional structures but different thermal stability and aggregation propensity, investigated here by FTIR spectroscopy. For each variant, appreciable spectral differences were found between the protein in solution and in single crystals, consisting in a downshift of the main β-sheet band and in better resolved turn and loop bands, indicative of reduced protein secondary structure dynamics in the crystalline state. Notably, the well-resolved spectra of the β2m crystalline variants enabled us to identify structural differences induced by the single amino acid mutations. Such differences encompass turn and loop structures that might affect the stability and aggregation propensity of the investigated β2m variants. This study highlights the potential of FTIR microspectroscopy to acquire useful structural information on protein crystals, complementary to the crystallographic analyses.

Introduction

β-2 microglobulin (β₂m) is the light chain of class I major histocompatibility complex; it is composed of 99 residues, organized in a classic β-sandwich structure, consisting of two β-sheets locked together by a disulphide bond (1). β₂m in vivo aggregation is responsible for dialysis-related amyloidosis (2), a pathological state that affects patients undergoing extended hemodialysis periods. Although no pathogenic β₂m mutations have been so far described, many β₂m mutational studies have been performed to elucidate which molecular properties affect β₂m aggregation. In particular, several mutants have been produced and characterized to investigate the role of the protein D β-strand, and of the loop comprised between the D and E strands (3–8). Although crystallographic analyses of such mutants showed strong overall conservation of the wild-type (WT) protein structure, the mutants display quite different stabilities and aggregation propensities (9).

To gain new insights into the correlation between the structural changes induced by single amino acid mutations and β₂m conformational stability and aggregation propensity, a comprehensive biophysical approach that enables studying the protein both in solution and in crystalline state may be particularly relevant. To this aim, Fourier transform infrared (FTIR) spectroscopy allows to analyze also highly scattering protein samples under different environmental conditions, thus enabling the comparison of secondary structures in solution—under native or fibrillogenic conditions—and in the crystals (10–14). In fact, although protein x-ray crystallography is an established investigation method providing atomic resolution three-dimensional (3D) structure, it may hold some drawbacks. In addition to the requirement for single-well diffracting crystals, the 3D structures provide limited information on protein dynamics, and may be locally affected by crystal packing contacts. Thus, it may be helpful to develop complementary biophysical tools (15) able to investigate key protein properties in solution and in the crystalline state in a rapid and noninvasive way. To this goal, FTIR spectroscopy has been widely used to analyze protein secondary structures, because the amide I band—due to the C=O stretching vibration of the peptide bond—is sensitive to the protein backbone conformation (10–12,16). Moreover, the possibility to couple an infrared (IR) microscope to a FTIR spectrometer allows collecting IR absorption spectra from a small sample area, down to 20 μm × 20 μm (17), a size suitable for the analysis of most protein crystals. Indeed, previous applications of FTIR spectroscopy to protein crystal analysis, reported in the literature, showed the possibility to study the backbone secondary structures, allowing structural comparisons for the selected protein in solution and in the crystals (13,18,19).

In this work, we studied five β₂m single-site variants (WT protein, D53P, D59P, W60G, and W60V mutants) whose 3D structures have been previously determined by x-ray crystallography (1,3–6). In particular, we applied FTIR spectroscopy to these proteins in solution and as single crystals, to detect the protein fine structural and dynamics variations that may result from single amino acid mutations in the two protein states. To this aim, we first optimized a protocol for the collection of high-quality micro-FTIR spectra from protein single crystals. By this FTIR approach, we found for each β₂m variant tighter structures and increased structural order in the crystalline state compared to the protein in solution. Moreover, the crystal spectra displayed better resolved components, due to turn and loop structures that allowed detection of fine structural differences among the examined β₂m variants.

Our data and the experimental protocol devised show that FTIR microspectroscopy can reliably extract structural data from the analysis of protein crystals that also reflect the dynamics of the crystallized protein molecules.

Materials and Methods

Expression and purification

WT β₂m and the mutated variants were expressed and purified as previously reported (20).

FTIR spectroscopy of β₂m in solution: secondary structures, thermal stability, and aggregation kinetics

The different β₂m variants were dissolved in deuterated 50 mM phosphate buffer, 100 mM NaCl (pH 7.4), at a final concentration of 2.5 mg/ml. FTIR spectra were collected at 37°C, in transmission mode, using a Varian 670-IR spectrometer (Varian Australia, Mulgrave, Australia), equipped with a nitrogen-cooled mercury cadmium telluride detector, under accurate dry air purging. A sample volume of 20 μL was placed in a temperature-controlled transmission cell with two BaF₂ windows separated by a 150 μm Teflon spacer. Spectra were collected under the following conditions: 2 cm⁻¹ resolution, 25 kHz scan speed, 1000 scan coaddition, and triangular apodization.

For thermal stability experiments, the previous β₂m solutions (in deuterated 50 mM phosphate buffer, 100 mM NaCl, pH 7.4) were heated from 37°C to 100°C at a rate of 0.4°C/min, each transmission spectrum being collected every 3.6°C. The aggregation of β₂m WT and mutants was induced by the addition of 20% trifluoroethanol (TFE) to the protein solution (in deuterated 50 mM phosphate buffer, 100 mM NaCl, pH 7.4) and monitored measuring the FTIR spectra in transmission mode at different times of incubation at 37°C, over 24 h. The β₂m FTIR spectra were obtained after subtraction of solvent absorption, collected under the same conditions. Before analysis, the FTIR spectra were corrected, when necessary, for water vapor absorption. Second derivative analysis (21) was performed by the GRAMS-32 software (Galactic Industries, Salem, NH). To this aim, we applied the Savitzky Golay algorithm (third degree, 5-point smoothing), after an 11-point binomial smoothing of the measured spectra.

Crystallization and crystal handling

The purified β₂m variants were dissolved in distilled water at a concentration of 10 mg/ml and crystallized using the hanging drop technique as described earlier (4). The same crystal growth conditions were applied for all the β₂m variants (reservoir composition: 18–20% polyethylene glycol 4000, 20% glycerol, 0.2 M ammonium acetate, at pH 5.0–6.0, 20°C).

FTIR microspectroscopy of single β₂m crystals

For the IR analysis, single crystals (average size of 50–100 μm × 50–100 μm) and—in the case of D53P—homogeneous ensembles of thin needle-like crystals were shortly washed in a drop of D₂O (5–10 s) to exchange the water in the crystal with heavy water. The crystals were then transferred, by a cryoloop, on a BaF₂ IR transparent window inside a paraffin oil drop, to prevent evaporation. The aqueous solution around the crystals was carefully removed, and the paraffin oil drop containing the crystals was covered with a second BaF₂ window, to reduce the optical path.

FTIR spectra were acquired in transmission mode, between 4000 and 700 cm⁻¹, by means of a Varian 610-IR infrared microscope coupled to the Varian 670-IR FTIR spectrometer (both from Varian Australia Pty Ltd.), equipped with a mercury cadmium telluride nitrogen-cooled detector. The variable microscope aperture was adjusted to select single crystals (from ∼50 to 100 μm diameter). Measurements were performed at 2 cm⁻¹ spectral resolution; 25 KHz scan speed, triangular apodization, and by the accumulation of 512 scan coadditions. To compensate for buffer absorption, a background spectrum, taken in close proximity of the sample, was always collected immediately before each sample measurement. In this way, FTIR spectra with an excellent signal/noise ratio were obtained. Second derivative analysis was performed as described previously for the protein solution FTIR spectra. To assess that the measured spectra were not affected by crystal orientation, for each variant several randomly oriented crystals were examined and in all cases superimposable IR spectra were obtained.

Results

Crystal structures of the β₂m variants

We report here a systematic FTIR study of WT β₂m and of four mutants, whose 3D structures have been previously determined by x-ray crystallography at resolutions of 1.8–1.9Å (1,3–6). Because both WT β₂m and the mutants yielded fairly stable crystals, these samples appeared suitable for FTIR microspectroscopy measurements.

To easily relate the results of this study with previous data, we briefly recapitulate the main structural and biophysical features of the β₂m mutants analyzed here. In our previous studies we produced and characterized four single-residue mutants (Asp-53→Pro D53P mutant, Asp-59→Pro D59P mutant, Trp-60→Gly and Trp-60→Val, W60G and W60V mutants, respectively) (3–6), designed to map a protein region held to be critical for β₂m stability and aggregation properties. The crystallographic analyses showed that WT β₂m and D59P share virtually identical 3D structures; W60G and W60V also displayed very similar structures (Table S1 in the Supporting Material), the main structural differences between WT β₂m and the W60G/V mutants being located in the four-residue DE loop, which hosts the single-site mutation. Conversely, D53P showed specific structural features that concern the interactions of the AB loop with the rest of the molecule (Fig. 1), and the D-strand, which is shorter compared to the other mutants (6). In relation to what will be discussed below, it should also be noted that, despite their very similar crystal structures, WT β₂m and the four mutants display different fold stabilities (as measured by melting temperatures) and fibril aggregation properties in solution (3–6,9). The four mutants, together with WT β₂m, and their 3D structures, therefore provide a challenging test case to probe the sensitivity of the FTIR approach in detecting the fine molecular properties that relate to fold stability and β₂m aggregation. Moreover, the presence of proteins with different orientations within a single crystal—the β-sheets of neighboring molecules being oriented 45–60 degrees apart in the crystal packing (1,3–6)—makes negligible possible orientation effects in the IR spectra.

Figure 1.

Structure superposition for WT β2m (gray) and the D53P mutant (black). The main differences are concentrated in the AB loop and in the shorter D-strand. The structures of the W60G/W60V/D59P mutants are not shown, being very similar to the 3D structure of the WT protein. The main β-sheet strands are labeled according to the accepted β2m nomenclature. Asp-53, Asp-59, and Trp-60 are shown in sticks, in WT β2m.

FTIR analysis of β₂m thermal stability

The native secondary structure and thermal stability of WT β₂m and mutants were analyzed by FTIR spectroscopy. Fig. 2 shows the second derivatives of the FTIR absorption spectra (Fig. S1) of WT β₂m measured in deuterated 50 mM phosphate buffer, 100 mM NaCl (pH 7.4), in the 37–100°C temperature range (Fig. 2 A). The second derivative spectra, where the minima correspond to absorption maxima of the measured spectra, allow a better resolution to the spectral components arising from different protein secondary structures (21). In particular, the WT β₂m spectrum at 37°C displays a main component at ∼1636 cm⁻¹ that, together with the ∼1689 cm⁻¹ band, is due to the native antiparallel β-sheet structure of the protein (1), in agreement with previous FTIR studies in solution (22,23). Moreover, a weak and broad component is found around 1668 cm⁻¹ that can be assigned to β-turn structures and the peak at ∼1610 cm⁻¹ is due to amino acid sidechain absorption (10,23).

Figure 2.

Thermal stability of β2m variants studied by FTIR spectroscopy. (A) The second derivative spectra of WT β2m in deuterated phosphate buffer (50 mM, NaCl 100 mM, pH 7.4) at different temperatures, from 37°C (bold continuous line) to 100°C (bold dashed line), are reported. The arrows point to increasing temperatures. The spectrum taken at 37°C after thermal treatment (dotted line) is also reported. (B) The temperature dependence of the native β-sheet peak intensity at ∼1636 cm⁻¹, taken from second derivative spectra, is reported for WT β2m and the studied mutants. (C) The temperature dependence of the intermolecular β-sheet peak intensity at ∼1619 cm⁻¹, taken from second derivative spectra, is reported for WT β2m and the studied mutants.

When the temperature was increased, the native β-sheet component at ∼1636 cm⁻¹ was found to be stable up to ∼61°C, and to rapidly decrease above this temperature, as shown in Fig. 2, A and B. The loss of native β-sheet was accompanied by the appearance of new components at ∼1619 and 1684 cm⁻¹, due to the formation of intermolecular β-sheet structure typical of protein aggregates (Fig. 2, A and C) (24). Such a transition was found to be irreversible, as indicated by the spectrum collected after cooling the sample down to 37°C, where the IR band of protein aggregates was still present and downshifted to ∼1615 cm⁻¹. This downshift indicates that sample cooling led to the formation of tighter intermolecular β-sheet structures (10,12). The spectral changes reported in Fig. 2 are due to thermal unfolding of the protein, because such spectral variations were not observed incubating WT β₂m at 37°C for 24 h in deuterated 50 mM phosphate buffer, 100 mM NaCl, pH 7.4. As shown in Fig. S1, under this last condition a small downshift of the native β-sheet band from 1636 to 1634 cm⁻¹ was observed, whereas a larger downshift from 1689 to 1682 cm⁻¹ was found for the high-wavenumber native β-sheet component. The different downshift of the two β-sheet bands after resuspension in deuterated buffer has been observed for other proteins (see for instance (25)).

We then compared the results obtained on WT β₂m with those of the four mutants to explore by FTIR the effects of the different single residue mutations on the protein thermal stability (see also Fig. S1 and Fig. S2). As a summary of all such measurements, Fig. 2, B and C, shows for each protein variant the temperature dependence of the IR bands due to the native β-sheets and the intermolecular β-sheets in protein aggregates, respectively at ∼1636 and ∼1619 cm⁻¹. This analysis showed that D59P was the less stable mutant, whereas W60G and D53P were the most stable ones, with a difference in thermal stability, between the two extremes, of the order of 20°C. Finally, the thermal stability of W60V was instead observed to be slightly lower than the WT protein. All such results are in good agreement with those previously provided by circular dichroism (CD) and Trp fluorescence of the β₂m mutants (6,9), with the exception of W60V. Indeed, W60V and the WT protein display indistinguishable thermal stabilities when probed by CD and by fluorescence spectroscopies (9), suggesting that the experimental conditions (e.g., protein concentration) required by the different biophysical approaches may in fact turn into differences in the measured thermal stabilities.

FTIR analysis of β₂m aggregation kinetics

As widely reported in the literature, β₂m can undergo amyloid aggregation under different in vitro conditions, such as low pH or TFE addition (26,27). We compared here the kinetics of TFE-induced aggregation of WT β₂m and of its four mutants through FTIR spectroscopy. To this aim, the protein samples were transferred in deuterated 50 mM phosphate buffer, 100 mM NaCl (pH 7.4) in the presence of 20% TFE and incubated at 37°C for 24 h directly within the IR sample cell. Fig. 3 A shows the second derivatives of the absorption spectra (see also Fig. S3) of WT β₂m collected at different times during the 24 h of incubation. In particular, at incubation time 0 the WT β₂m spectrum was similar to that of the untreated protein (see Fig. 2 A); however, in the presence of 20% TFE, the native β-sheet component at ∼1689 cm⁻¹ was found to be downshifted to ∼1683 cm⁻¹, and the main native β-sheet band at ∼1636 cm⁻¹ showed decreased intensity, suggesting a partial destabilization of the native β-sheet structure of the protein, as expected (28,29). During 24-h incubation, the native-like β-sheet component at ∼1636 cm⁻¹ was found to further decrease, whereas two new components appeared at ∼1615 and ∼1685 cm⁻¹, both due to the intermolecular β-sheet structure of the growing protein aggregates.

Figure 3.

TFE-induced aggregation of β2m variants monitored by FTIR spectroscopy. (A) The second derivative spectra of WT β2m in deuterated phosphate buffer (50 mM, NaCl 100 mM, pH 7.4) added with 20% TFE, at different times of incubation at 37°C. The arrows point to increasing incubation times. (B) The time dependences of the native-like β-sheet peak intensity at ∼1636 cm⁻¹ and of the intermolecular β-sheet peak intensity at ∼1615 cm⁻¹, taken from second derivative spectra, are reported for WT β2m. (C) The time dependence of the ratio between the intensity of the intermolecular β-sheet component (∼1615 cm⁻¹) and that of the native-like β-sheets (∼1636 cm⁻¹) is reported for each mutant. D59P aggregation kinetics is not reported, because it immediately aggregated within the test tube after the addition of TFE.

The kinetics of TFE-induced misfolding and aggregation is illustrated in Fig. 3 B, where the time dependence of the native-like β-sheet intensity (∼1636 cm⁻¹) and that of the intermolecular β-sheet (∼1615 cm⁻¹) are reported. Both display a typical sigmoidal curve characterized by a short lag phase, and reach a plateau in ∼10 h. It should be noted that WT β₂m aggregation (at 20% TFE), which in these measurements occurs directly within the IR sample cell, was found to be 10-fold faster than that occurring—under similar buffer and solute conditions—in the test tube, as monitored by thioflavin T fluorescence in the presence of preformed seeds (3–5). We explain such a different kinetic behavior as a result of a higher protein concentration used for the IR measurements. We should also consider that the high interfacial surface in the FTIR sample cell—two BaF₂ windows separated by a Teflon spacer of 150 μm—might promote the formation of early protein assemblies, speeding up β₂m aggregation (30,31). Notably, this unexpected effect allowed us to follow rapidly the aggregation process that, otherwise, would have required much longer times.

Following the same approach, we investigated the effects of 20% TFE on the four β₂m mutants, and, as a result, each was characterized by specific aggregation kinetics (Fig. 3 C and Fig. S3 and Fig. S4). First, just after the 20% TFE addition, the spectra of the β₂m mutants displayed slight variations in the peak position of the native β-sheet components at ∼1636 and ∼1689 cm⁻¹, when compared to the untreated proteins (Fig. S3 and Fig. S4). Notably, the D59P mutant was found to aggregate immediately in the test tube after the TFE addition, contrary to what was observed for the WT protein and the other mutants under investigation (Fig. S3 and Fig. S4). To compare the aggregation kinetics induced by TFE on all the β₂m variants, we report in Fig. 3 C the time dependence of the intensity ratio between the intermolecular β-sheet band at ∼1615 cm⁻¹ and the native-like β-sheet at ∼1636 cm⁻¹, to provide information on protein aggregation and misfolding in a concise way. In particular, the aggregation kinetics of the W60V mutant was found to be similar to that of WT β₂m, being faster at the beginning of the process, but reaching a slightly lower level of aggregate plateau. The W60G mutant was characterized by the lowest aggregation propensity: after 24 h the native β-sheet band around 1636 cm⁻¹ was still present, contrary to what was observed for the other variants (Fig. S3 and Fig. S4). Finally, despite its high thermal stability, the aggregation kinetics of D53P was found to be similar to that of the WT protein. As discussed in the following section, the differences in aggregation propensity detected by FTIR spectroscopy for the β₂m variants are generally found to be in agreement with the results reported in the literature (3,6).

FTIR analysis of WT β₂m in solution and in the crystals

To shed more light on the relationship between the contained structural differences observed in the crystal structures of β₂m single-site mutants and the resulting variability in thermal stability and aggregation propensity, we explored the secondary structure features of the different β₂m species in the crystalline state using FTIR microspectroscopy. To this aim, we first optimized a crystal handling protocol for IR analysis, which may be of general value and can in principle be applied to crystals of other proteins, as reported in the Supporting Material.

Fig. 4 A shows the FTIR absorption spectrum of a single WT β₂m crystal in the amide I and amide II regions. To collect the spectrum, single crystals were washed in D₂O and then transferred in paraffin oil, to avoid solvent evaporation and crystal dissolution. The presence of an intense amide II band indicates that D₂O washing replaced bulk H₂O, without important Hydrogen/Deuterium exchange of the amide backbone groups involved in secondary structures (11,12). Furthermore, the flat baseline in the 2000–1800 cm⁻¹ spectral range indicates that the contribution of residual H₂O to the absorption spectrum is negligible, allowing a reliable assignment of the amide I band components to β₂m secondary structures.

Figure 4.

FTIR microspectroscopy of WT β2m single crystals. (A) FTIR absorption spectrum of a single WT β2m crystal measured in paraffin oil after D₂O washing, reported in the 2000–1470 cm⁻¹ spectral region. The optical image of a single crystal is shown in the inset. (B) The second derivative of the absorption spectrum of WT β2m crystal (reported in A) is compared with that of the WT protein in solution.

Fig. 4 B compares the second derivative spectrum of a WT β₂m single crystal in paraffin oil with that of β₂m in D₂O solution, both in the amide I region. The second derivative spectrum of the WT β₂m crystal displays evident features differing from those of the protein in solution. First, the main absorption due to the native antiparallel β-sheets is downshifted from ∼1636 cm⁻¹ in solution to ∼1628 cm⁻¹, with a shoulder around 1638 cm⁻¹, in the crystal. This downshift suggests a stronger H-bonding in the β-sheets of the protein in the crystalline state, implying a tighter structure that could reflect reduced dynamics of the crystallized β₂m protein compared to that in solution (10,12,24). Notably, the peak position of the ∼1628 cm⁻¹ band was found to be unaffected by D₂O washing of the crystal, as verified by washing the β₂m crystals in distilled H₂O instead of D₂O (data not shown). This result suggests that, compared to the solution spectrum, the downshift of the ∼1628 cm⁻¹ band is due to intrinsic structural properties of the crystallized protein. Such a hypothesis is also supported by the well-resolved spectral component at ∼1670 cm⁻¹, due to turn absorption (11), which is narrower than that of the protein in solution, and by the presence of a new component at ∼1649 cm⁻¹. The assignment of the latter band, not detected in the protein solution spectrum, is not unequivocal because several secondary structures —such as α-helix, random coils and loops—display absorption bands in this spectral region (11,32–34). Nevertheless, taking into account that no changes in peak position were observed when WT β₂m crystals were measured in H₂O (data not shown) and that α-helices are absent in the protein structure, we tentatively assigned this band to protein loops, also considering its reduced band width. Finally, the two bands at ∼1689 and ∼1681 cm⁻¹ can be also assigned to intramolecular β-sheet structures (11). The presence of narrower components and of new bands, unresolved in the solution spectra, suggests an increased structural order and reduced dynamics (10) of the involved protein residues in the crystal. The sharper spectral features in the crystal, compared to the protein in solution, can be related to the water interaction with amide I oscillators, which reduces the spectral effects of their coupling and will give broader bands in solution.

Taken together, the previous results indicate that WT β₂m in the crystalline state is characterized by a higher structural order than in solution, which might result from reduced dynamics of β-sheets, turns, and loops.

FTIR comparisons of β₂m mutants in solution and in the crystalline state

Following the approach described for WT β₂m in the previous section, we also characterized the structural properties of the four β₂m mutants in the crystalline state and compared them with those of the corresponding proteins in D₂O solution. Fig. 5 A shows the second derivatives of the absorption spectra (see also Fig. S5) of the four β₂m mutants measured in paraffin oil as single crystals, in addition to the WT protein. We should first note that—as discussed for the WT protein—for each mutant we observed important differences between the crystalline state (Fig. 5 A) and the proteins in solution (Fig. 5 B). Indeed, the main β-sheet component displayed a downshift from ∼1636 cm⁻¹ in solution to 1631–1627 cm⁻¹ in the crystals of all the mutants. Moreover, all the mutant crystal spectra are characterized by better resolved components due to turn and loop structures, compared to the solution spectra, again indicating more restrained protein dynamics. Interestingly, the comparison among the different β₂m mutant spectra highlights differences linked to the specific residue mutations that are more evident in the crystal than in the solution spectra. In particular, the D59P IR response both in solution and in the crystal is almost identical to that of the WT protein, in agreement with the comparison of their crystal structures (Table S1) (5). For the W60V mutant, although the IR response of the crystalline protein is similar to that of WT β₂m, the solution spectrum displays minor differences, involving the β-sheet band observed at ∼1689 cm⁻¹ in the WT protein that downshifts to ∼1685 cm⁻¹ in W60V.

Figure 5.

Comparison between β2m variants in single crystals and in solution. (A) The second derivative spectra of single crystals of β2m WT and mutants, measured in paraffin oil after D₂O washing, are reported in the amide I region and compared with (B) the same proteins in solution. A schematic peak assignment is also reported.

The FTIR spectral features of the W60G mutant are quite different from those of W60V, both in solution and in the crystals, opposite to what would be expected based on the common mutation site and on their closely similar 3D structures (Table S1). Indeed, although the IR spectrum of the W60G solution resembles that of WT β₂m, in the IR spectrum of the W60G crystal the main β-sheet band is broader and upshifted to ∼1631 cm⁻¹, whereas the high-wavenumber β-sheet component occurs again at ∼1689 cm⁻¹. Moreover, the ∼1681 cm⁻¹ peak was not detected in the W60G spectrum, whereas a new low intensity band was observed around 1677 cm⁻¹, assigned to β-sheet and/or turns (11). Considering that the crystal structure of the W60G mutant is almost identical to that of the WT protein (4), the FTIR spectral differences observed for the crystals might highlight fine structural rearrangements, beyond the resolution limits of the crystallographic analyses (e.g., changes in hydrogen bond strength) (10), accounting for their different conformational stability and aggregation propensity in solution (Figs. 2 and 3).

Concerning the D53P mutant we should first note that, although the spectrum of the protein in solution is almost identical to that of WT β₂m, the spectrum of the crystalline proteins is significantly different (Fig. 5). In particular, the main β-sheet component is broader and upshifted to ∼1631 cm⁻¹ in the mutant, whereas the high-wavenumber β-sheet component occurs at ∼1691 cm⁻¹. Moreover, major differences were found in the spectral region for turn, loop, and random coil absorption. Indeed, a broad component occurred at ∼1678 cm⁻¹ with a shoulder around 1667 cm⁻¹, which can be assigned to turns (11). Furthermore, the ∼1681 and 1670 cm⁻¹ bands that have been observed in the WT spectrum are not detected in the D53P spectrum, where instead a new component appeared at ∼1656 cm⁻¹, together with the ∼1649 cm⁻¹ band. These bands can be respectively assigned to turn/loop and loop structures (11,32–34). The FTIR results indicate that the peculiar features of D53P mainly concern rearrangements of turns and loops, in agreement with crystallographic data showing that the substitution of Asp-53 with Pro led to a break of part of the D-strand, which resulted in a rearrangement of the CD and AB loops (6).

Discussion

Our study presents a FTIR spectroscopy characterization of WT β2m and four single-residue mutants in solution, under native and fibrillogenic conditions, and in the crystalline state. The study allowed us not only to validate through unrelated approaches the stability and fibrillogenesis results previously reported on the selected DE loop β2m mutants (6,9), but also to draw methodological considerations on the application of FTIR to proteins in the crystalline state.

Thermal stabilities and aggregation trends, here reported for WT β2m and mutants, were generally found to be in good agreement with previous data (6,9); in particular, the aggregation propensity of the investigated β₂m variants displayed the following trend: D59P ≫ W60V ≥ WT ≈ D53P ≫ W60G. W60V is the only exception within such trend, because it had been earlier reported to display an aggregation propensity lower than WT β₂m (3). Such modest variability may be related to the different experimental procedures adopted to promote protein aggregation, as discussed previously. Indeed, in the FTIR analysis setup protein aggregation was 10-fold faster, and took place directly inside the sample cell without any requirement for seed addition, whereas in the thioflavin T fluorescence experiments aggregation occurred in a test tube after addition of preformed seeds (3).

The overall FTIR results for the protein variants in solution are in keeping with previous models of the effects of single amino acid mutations on β₂m thermal stability and aggregation propensity. In particular, the Asp-59→Pro mutation increases the DE loop rigidity and structural strain, leading to decreased thermal stability—relative to the WT protein—and high aggregation propensity (5). On the contrary, the conformational freedom coded by Gly-60 in the W60G mutant results in a stable conformation for the DE loop, high thermal stability, and low aggregation propensity (4). The substitution of the bulky aromatic Trp-60 with a smaller hydrophobic amino acid, in the W60V mutant, leads to decreased aggregation propensity (3). Finally, the Asp-53→Pro mutation, in the middle of the D-strand, surprisingly increases protein stability but does not play a protective role against amyloid aggregation (6).

To gain further insight and innovative data on the dynamics and on the secondary structure of the β2m variants, we applied FTIR microspectroscopy to the analysis of β2m single crystals. Such an approach was used on protein crystals only sporadically in the past because of the experimental requirements posed by FTIR data acquisition and by the protein crystal milieu: salts and/or organic precipitants strongly absorb in the IR range, but are fundamental for protein crystal stability. WT β2m and the DE loop mutants grow as properly shaped and stable crystals, and all their 3D structures were determined to mid-high resolution, thus representing a suitable test case for FTIR microspectroscopy. To this aim, we devised an experimental protocol that provided an excellent compromise between the quality of the FTIR spectra and protein crystal stability, as we verified for the β2m variants investigated here and for other unrelated protein crystals (see Materials and Methods and the Supporting Material, Additional information on the crystal handling protocol and on the stability of protein crystals). It is worth noting that FTIR spectrum collection, besides being fast, does not require diffraction quality crystals: thus, poorly diffracting crystals and bunches of thin/needle crystals are suitable samples for FTIR microspectroscopy. The FTIR spectra collected on β2m crystals display better resolved components compared to those collected in solution, highlighting increased structural details. Although in solution the five β2m variants spectra are almost identical, the FTIR analysis in the crystalline state disclosed small but significant spectral differences (Fig. 5), mainly assigned to loops and turns, likely underlying differences in structure and/or dynamics enhanced by the crystal contacts. Thus, the FTIR spectra collected on the crystalline proteins are highly resolved fingerprints of each β2m variant.

Among the FTIR spectra of crystals, those of D53P and W60G deviate most from the spectrum of the WT protein but, intriguingly, although the D53P crystal structure is reported somehow different from the WT β2m, the W60G is structurally very similar to the WT protein (Table S1). The insertion of a Pro at site 53, in the middle of the D-strand, breaks the regularity of the strand resulting in a rearrangement of the region between residues 45–60, in a very long CD loop and in a short D β-strand (6). As shown in Table S1 the D53P mutant, among the mutants here analyzed, is indeed the one showing the largest structural perturbation compared to the WT protein, and the FTIR spectrum rightly highlights such differences.

The crystal structure of the W60G mutant has been reported to be very similar to that of the WT β2m (Table S1), although its thermal stability is higher and the aggregation propensity is definitely lower ((4) and this work). The W60G FTIR spectrum measured on crystals, on the other hand, shows that loop and turn regions appear modified in their structure/dynamics relative to the WT protein, despite the fact that essentially the same crystal contacts take place in the two crystallized proteins. It is thus notable that, in parallel with bulk stability and aggregation properties observed for the protein in solution, fine molecular properties of the W60G mutant protein could be highlighted by FTIR on the crystalline β2m mutant. Thus, both spectra of crystalline D53P and W60G mutants show that FTIR crystals cannot only highlight fine structural rearrangements but also different dynamics of variants with otherwise extremely similar structure, as in the W60G mutant.

Trp-60 in WT β2m is characterized by unfavorable ψ and φ Ramachandran angles; this unfavorable backbone conformation is likely accepted in vivo thanks to a series of stabilizing intermolecular contacts that residue Trp-60 can achieve in the class I major histocompatibility complex. Such DE loop conformational strain is released in the W60G mutant and reflected by its improved thermal stability. On the contrary, the W60V mutant hosts a Cβ sidechain atom and shows thermal stability and amyloid propensity similar to the WT protein ((3) and this work). Despite the crystal structures of the W60G and W60V mutants are virtually identical (root mean-square deviation of 0.19Å calculated over the entire polypeptide backbone), the FTIR spectra of the two mutants are different and, remarkably, the spectrum of W60V resembles closely that of the WT protein, with which it shares the same thermal stability and similar amyloid propensity.

Concluding Remarks

The FTIR spectra collected on five crystalline β2m variants show that i), FTIR spectra of crystals bear in general more information compared with the spectra measured in solution (see Figs. 4 and 5), ii), β2m variants with contained structural rearrangements can be clearly fingerprinted using FTIR microspectroscopy, and finally iii), the spectra on crystalline samples indicate that FTIR microspectroscopy accounts for differences both in structure and in protein dynamics which, in the present case, relate well to the corresponding amyloid propensity of the examined β2m variants.

Therefore, FTIR experiments here reported for β2m suggest new potential applications of FTIR to different protein crystals. FTIR microspectroscopy may be used as a rapid screening method for ligands/mutations/crystallization conditions triggering conformational change(s) in the targeted protein, thus directing the subsequent collection and analysis of x-ray diffraction data. Considering that FTIR microspectroscopy is a noninvasive approach, this technique might allow monitoring in real-time reactions, interactions, and protein conformational variations within protein single crystals. In this view, FTIR microspectroscopy appears as a key complementary tool for the crystallographic analysis.