Early events in light chain aggregation at physiological pH reveal new insights on assembly, stability, and aggregate dissociation

Abstract

Early events in immunoglobulin light chain (AL) amyloid formation are especially important as some early intermediates formed during the aggregation reaction are cytotoxic and play a critical role in the initiation of amyloid assembly. We investigated the early events in in vitro aggregation of cardiac amyloidosis AL proteins at pH 7.4. In this study we make distinctions between general aggregation and amyloid formation. Aggregation is defined by the disappearance of monomers and the detection of sedimentable intermediates we call non-fibrillar macromolecular (NFM) intermediates by transmission electron microscopy (TEM). Amyloid formation is defined by the disappearance of monomers, Thioflavin T fluorescence enhancement, and the presence of fibrils by TEM. All proteins aggregated at very similar rates via the formation of NFM intermediates. The condensed NFM intermediates were composed of non-native monomers. Amyloid formation and amyloid yield was variable among the different proteins. During the stationary phase, all proteins demonstrated different degrees of dissociation. These dissociated species could play a key role in the already complex pathophysiology of AL amyloidosis. The degree of dissociation is inversely proportional to the amyloid yield. Our results highlight the importance and physiological consequences of intermediates/fibril dissociation in AL amyloidosis.

Introduction

More than 30 human proteins are involved in a group of pathological disorders collectively known as amyloidoses [1, 2]. Immunoglobulin light chain (AL) amyloidosis is one of the most complex forms of systemic amyloidosis. AL amyloidosis is characterized by the secretion of free immunoglobulin light chains due to an abnormal proliferation of monoclonal plasma cells [3]. Once in circulation, the excess immunoglobulin light chains misfold, and deposit as fibrillar aggregates in the extracellular space of vital organs, resulting in multiple organ failure and death [4, 5]. Despite significant advancements in the management and treatment of AL amyloidosis in the past twenty years [6], the disease remains fatal. AL amyloidosis is a heterogeneous disease both molecularly and clinically. Molecular heterogeneity in AL amyloidosis stems primarily from the combination of gene rearrangements and somatic hypermutation, making each pathogenic AL protein sequence unique [7, 8].

The molecular determinants of AL amyloidosis disease initiation and progression are still not fully understood. In vitro aggregation of immunoglobulin variable domain proteins from AL amyloidosis patients (herein called AL proteins for simplicity) follow a nucleation dependent polymerization process. The amyloid formation reaction is characterized by sigmoidal kinetics presenting three distinct phases viz- nucleation, elongation, and stationary phases [9, 10]. Our previous study characterizing the early events of amyloid formation was conducted at pH 2.0 because this is the only condition in which we have been able to form fibrils with the germline control protein κI O18/O8 [11]. Under acidic conditions, stable oligomeric intermediates are formed in the early stages of amyloid formation by AL-09. These oligomers rapidly rearranged to fibrillar forms without an apparent change in the concentration of monomers [11]. Based on our previous results, we were interested in characterizing the early events of fibril formation at physiological pH to identify aggregation commonalities and differences between the two solution conditions for the same group of AL proteins.

In this study, we present a systematic analysis of the early aggregation events at pH 7.4 for AL-09 and AL-12 using a battery of biophysical approaches. These proteins come from two different patients with cardiac amyloidosis. They share >90% sequence homology with the κI O18/O8 germline control (IGKV 1–33) (Figure S1A). Structurally, AL-12 retains the canonical dimer interface whereas AL-09 adopts an altered dimer interface with a 90° rotation with respect to the canonical dimer of κI O18/O8 germline protein [12, 13] (Figure S1B). In general, mutations that decrease the thermodynamic stability of AL proteins favour amyloid formation [14, 15].The Y87H mutation in AL-09 (responsible for the altered dimer conformation) and S65R mutation in AL-12 (affecting the global tertiary structure) [16, 17] play a critical role in the enhancement of aggregation. Because of this, our study includes restorative mutants AL-09 H87Y and AL-12 R65S and reciprocal mutant κI O18/O8 Y87H to investigate the effects of these mutations in the aggregation reaction.

Materials and methods

Chemicals

All chemicals and reagents used in the experiments were procured from Sigma-Aldrich unless specified otherwise. Milli Q grade water was used as a solvent to prepare different solutions.

Cloning, expression, extraction and purification of recombinant κI O18/O8, κI O18/O8 Y87H, AL-09, AL-09 H87Y, AL-12 and AL12 R65S variable domain proteins

κI OI8/O8 germline DNA, also known as IGKV 1–33, was generated by mutating AL-103 cDNA (κI OI8/O8 sequence deposited under GenBank accession number EF640313) as described previously [12]. DNA sequences of the mutant proteins AL-09 (GenBank accession number AF490909) and AL-12 (GenBank accession number AF490912) belonging to the κI gene family were obtained from the cDNA of AL patients plasma cells exhibiting cardiac involvement as described previously [18]. DNA sequence of reciprocal mutant κI OI8/O8 Y87H and restorative mutants AL-09 H87Y and AL-12 R65S were generated by mutating the cDNA of κI OI8/O8, AL-09 and AL-12 by using QuickChange^™ multisite directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) [16, 17]. All proteins were expressed using the pET12a plasmid (Novagen, Madison, WI).

Proteins were expressed in Escherichia coli BL21 (DE3) gold competent cells (Agilent Technologies, Santa Clara, CA) as reported previously [12, 13, 19]. AL-09 was extracted from the insoluble fraction (using 6 M urea) while all other proteins (κIO18/O8, κIO18/O8 Y87H, AL-09 H87Y, AL-12, and AL-12 R65S) were extracted from the periplasmic space by breaking the cells through one freeze thaw cycle using PBS buffer at pH 7.4 as described previously [13, 16, 17]. Purification of all proteins was conducted on an AKTA FPLC (GE Healthcare) system with a HiLoad 16/60 Superdex 75 size exclusion chromatography (SEC) column using 10 mM Tris-HCl at pH 7.4 as mobile phase. Protein content verification and protein size verification, were done by UV-Spectrophotometry (Absorbance ratio at 260/280 nm of less than 1) and SDS-polyacrylamide gel electrophoresis (SDS-PAGE) respectively. Secondary and tertiary structure and qualitative assessment of the purified fractions was done using far UV-CD spectra and thermal unfolding/refolding experiments at 217 nm on a CD spectropolarimeter (JASCO J 810), as reported previously [10, 11, 20]. Protein concentration was determined by UV absorbance at 280 nm using extinction coefficient calculated from amino acid sequence as follows: ε=14890 M⁻¹ •cm⁻¹ for κI O18/O8 and AL-09 H87Y, and ε=13610 M⁻¹ •cm⁻¹ for κIO18/O8 Y87H, AL-09, AL-12, and AL-12 R65S.

Sample preparation for aggregation assays

Sample preparation was done as reported previously [21]. The purified proteins are centrifuged at 643,867×g for 3 h and 20 min in a Beckman Coulter (Optima L-100 XP) ultracentrifuge and filtered through 0.2 μM low protein binding membrane filter to remove preformed aggregates (if any) formed during the freeze/thaw process. The ultracentrifuged and filtered protein solution was used to prepare aggregation reaction mixtures consisting of 20 μM protein, 150 mM NaCl in phosphate buffered saline (PBS buffer) at pH 7.4. For the HPLC sedimentation assay, 1200 μL of the reaction mixture were placed in low protein binding microcentrifuge tubes. For the Thioflavin T (ThT) fluorescence assay, 800 μL of the reaction mixture were mixed with ThT (final concentration of 20 μM ThT). The reaction tubes were kept at 4°C as the t=0 aliquots were being evaluated using the different experimental approaches. Thereafter, the aggregation mixture was moved to an incubator kept at 37°C with orbital shaking set at 300 RPM to monitor the aggregation kinetics. We excluded sodium azide from our reaction mixture due to the artefacts this molecule caused during our previous chromatography experiments.

The protein concentrations calculated using the HPLC standard curve method differed slightly from protein concentrations calculated measuring the absorbance at 280 nm and using the extinction coefficient for each protein. The HPLC calculated concentrations of the AL proteins used in different experimental assays are reported in Table ST1.

Measurement and analysis of fibril formation kinetics

A. HPLC (High performance liquid chromatography) sedimentation assay

Quantification of unreacted soluble protein in the aggregation mixture was followed by periodically removing aliquots (~80 μL) from ongoing aggregation reactions followed by centrifugation at 160,000×g for 45 minutes and injecting two technical replicates of the supernatant (approx. 20 μL) into an analytical reverse phase column (Agilent Zorbax-C8 2.5 × 50 mm column). Protein samples were run against a gradient of acetonitrile (0–100%) in 0.1% TFA maintained at 300 μL·min⁻¹ flow rate in a Shimadzu 20 HPLC system. The wavelength of the UV detector was fixed at 280 nm. HPLC grade solvents (Water and Acetonitrile) used as mobile phase in the reverse phase HPLC system were obtained from Fisher Chemicals. Concentration of the unreacted soluble protein was determined from integrated peaks at absorbance 280 nm using standard curves individually acquired for each protein as described previously [21, 22].

B. ThT fluorescence assay

ThT fluorescence from the aggregation mixture was monitored on a plate reader (Analyst AD; Molecular Devices, Sunnyvale, CA) using an excitation wavelength of 440 nm and an emission wavelength of 480 nm as previously reported [21]. Two aliquots (~260 μL technical replicates) from the aggregation reaction (containing 20 μM ThT) were placed on a black 96-well polystyrene plate (Greiner, Monroe, NC) at specific time points to measure the fluorescence intensity. After the measurements, aliquots were placed back into the microcentrifuge tube to resume the incubation at 37°C with orbital shaking set at 300 RPM. The measurement process was repeated until the apparent change in fluorescence intensity plateaus over a period of at least 72 h.

C. Kinetic data analysis and figure presentation:

The disappearance of monomers in solution (obtained from the sedimentation assay) and the emission fluorescence signal (obtained from the ThT fluorescence assay) were plotted as a function of time. The t₅₀ value was obtained by fitting each independent kinetic trace from the sedimentation assay and ThT assay to a sigmoidal function (Boltzman function in Origin software package) as previously reported [21].

$$y=\frac{A_1-A_2}{1+e^{\left(x-x_0\right)/dx}}+A_2$$ (Equation 1)

where A₁ is the initial fluorescence value, A₂ is the final fluorescence value, x₀ is the center (t₅₀ value), dx is defined as the time constant.

The t₅₀ values as well as kinetic parameters (from the sedimentation and ThT fluorescence assays) of all proteins reported in this study are an average of at least three independent experiments (biological replicates: n=3 or more). The representative kinetic data presented for each protein in figures 1, 2, 3 (A and B) are the result of one independent experiment (2 technical replicates). The representative data includes color-coded bars in the figures of kinetic data analysis based on concentration of monomers (█ – nucleation, █ – elongation, █ – stationary, and █ – dissociation phase). The approximate time limits of each of these phases for the different AL- proteins are presented in Table ST2.

Figure 1.

Composite kinetic data analysis of monomer disappearance, Thioflavin T fluorescence, and structural variations determined by intrinsic fluorescence and CD of κIO18/O8 (A, C and E) and κIO18/O8 Y87H (B, D and F) during aggregation. Shown are (A and B) the representative kinetic traces obtained from sedimentation assay to calculate the % monomer remaining in the reaction (■) vs ThT fluorescence assay (●) (For information regarding number of replicates [technical and biological], see Methods section). The highlighted area represents the different phases (based on concentration of monomers): [█ – nucleation, █ – elongation, █ – stationary, and █ – dissociation phase]. The colour coded dotted lines represents the exact time points when aliquots from the ongoing aggregation reaction were analysed by EM. The brown box highlights the aggregate dissociation in these proteins. (C and D) Intrinsic tryptophan fluorescence assay representing changes in the ${\lambda }_{max}^{em}$ and emission intensity during aggregation. (E and F) Secondary structural information obtained from far UV-CD spectra during aggregation.

Figure 2.

Composite kinetic data analysis of monomer disappearance, Thioflavin T fluorescence, and structural variations determined by intrinsic fluorescence and CD of AL-09 (A, C and E) and AL-09 H87Y (B, D and F) during aggregation. Shown are (A and B) the representative kinetic traces obtained from sedimentation assay to calculate the % monomer remaining in the reaction (■) vs ThT fluorescence assay (●) (For information regarding number of replicates [technical and biological], see Methods section). The highlighted area represents the different phases (based on concentration of monomers): [█ – nucleation, █ – elongation, █ – stationary, and █ – dissociation phase]. The colour coded dotted lines represents the exact time points when aliquots from the ongoing aggregation reaction were analysed by EM. The brown box highlights the aggregate dissociation in these proteins. (C and D) Intrinsic tryptophan fluorescence assay representing changes in the ${\lambda }_{max}^{em}$ and emission intensity during aggregation. (E and F) Secondary structural information obtained from far UV-CD spectra during aggregation.

Figure 3.

Composite kinetic data analysis of monomer disappearance, Thioflavin T fluorescence, and structural variations determined by intrinsic fluorescence and CD of AL-12 (A, C and E) and AL-12 R65S (B, D and F) during aggregation. Shown are (A and B) the representative kinetic traces obtained from sedimentation assay to calculate the % monomer remaining in the reaction (■) vs ThT fluorescence assay (●) (For information regarding number of replicates [technical and biological], see Methods section). The highlighted area represents the different phases (based on concentration of monomers): [█ – nucleation, █ – elongation, █ – stationary, and █ – dissociation phase]. The colour coded dotted lines represents the exact time points when aliquots from the ongoing aggregation reaction were analysed by EM. The brown box highlights the aggregate dissociation in these proteins. (C and D) Intrinsic tryptophan fluorescence assay representing changes in the ${\lambda }_{max}^{em}$ and emission intensity during aggregation. (E and F) Secondary structural information obtained from far UV-CD spectra during aggregation.

Circular dichroism (CD) spectroscopy

Secondary structural analysis of all proteins was conducted on a CD spectropolarimeter (JASCO J 810) with a 0.2 cm path-length quartz cuvette. The temperature was controlled within ±0.01°C using a Peltier system. After protein purification, all the proteins samples (20 μM) prepared in 10 mM Tris HCl buffer at pH 7.4 were analysed by far UV-CD spectra and thermal unfolding/refolding experiments following the ellipticity at 217 nm over a temperature range of 10–80°C as reported previously [10, 16, 17]. During the aggregation experiments, all protein samples (20 μM) prepared in phosphate buffered saline at pH 7.4 with 150 mM NaCl (aggregation reaction mixture) were investigated by taking an aliquot (~ 600 μL) at predetermined time interval to record the variation in secondary structure. At least two independent CD experiments (biological replicates) were performed for every protein aggregation reaction. After the far UV-CD measurements, the reaction mixture was placed back into the microcentrifuge tube to resume the incubation at 37°C with orbital shaking set at 300 RPM.

Fluorescence spectroscopy

The intrinsic Tryptophan fluorescence assay was performed on a modular research grade spectrofluorometer (Photon Technology International, Inc.) equipped with FeliX32I analysis software. The fluorescence measurements were conducted by taking an aliquot (~ 600 μL) from the aggregation reaction and placing the sample in a 4 mm × 4 mm quartz cuvette at specific time points followed by selectively exciting the samples at 295 nm (2.5 nm excitation slit width) and recording the emission (5 nm emission slit width) over 310–450 nm. The maximum emission wavelength (${\lambda }_{max}^{em}$) at each time point was subsequently analysed to ascertain the variation in the global structure of the proteins during the aggregation reaction. At least two independent intrinsic fluorescence experiments (biological replicates) were performed for every protein aggregation reaction. After the fluorescence measurements, the reaction mixture was placed back into the microcentrifuge tube to resume the incubation at 37°C with orbital shaking set at 300 RPM.

Analytical size exclusion chromatography (SEC)

The size exclusion chromatography studies on all the proteins during the aggregation reactions were performed on a Superdex 200 10/300 GL analytical size exclusion column (separation range of 10,000–600,000 Daltons) attached to an AKTA FPLC system (GE Healthcare). An aliquot (~200 μL) from the ongoing aggregation reaction was injected into a 150 μL loop, eluted in phosphate buffer (0.05 M monobasic sodium phosphate, 0.05 M dibasic sodium phosphate and 0.15 M sodium chloride) at pH 6.8 (as recommended by the manufacturers) at a constant flow rate of 0.40 ml•min⁻¹. The wavelength of the detector was fixed at 280 nm. The molecular weight calibration curve was acquired prior to the aggregation experiments by injecting Gel Filtration Standard (Bio-Rad laboratories) and Gel Filtration Markers (Sigma Life Science) (Figure S2 and Table ST3). Molecular weight and oligomeric states were estimated from elution volume (V_e) values using the calibration curve equation as previously described [9, 10, 13].

Transmission electron microscopy (TEM)

Aliquots (~8 μL) from the aggregation reaction at different time points were placed on a 300 mesh copper Formvar/carbon coated grid (Electron Microscopy Science, Hatfield, PA) and allowed to adsorb for 90 s. The grid was washed with Milli Q H₂O, stained with freshly filtered (2% w/v) uranyl acetate for 30 s and then again washed with Milli Q H₂O. During the entire process of staining and washing, the excess of water and uranyl acetate were blotted from the sides with filter paper. Grids were analysed on a Philips Technai T12 transmission electron microscope (FEI, Hillsboro, OR) at 80 kV.

Results

Biophysical properties of the native proteins prior to aggregation experiments

The structural features of the natively folded κI O18/O8 (control), reciprocal mutant κI O18/O8 Y87H, the patient proteins (AL-09 and AL-12), and their corresponding restorative mutants (AL-09 H87Y and AL-12 R65S) in 10 mM Tris HCl buffer, pH 7.4 at 4°C were analysed by circular dichroism (CD) spectroscopy. All the proteins presented far UV-CD spectra with two minima at ~217 nm (indicative of β-sheet structure) and ~235 nm (due to absorbance by aromatic residues in the far UV region) as reported previously [19, 23, 24]. The thermal unfolding transitions of all the proteins tested at 217 nm at pH 7.4 were observed to be reversible with different degrees of hysteresis, as previously reported [9, 10, 17]. Prior to the aggregation experiments, the mono-dispersity and monomeric state of all AL proteins were confirmed by SEC and SDS-PAGE.

General overview of aggregation of AL proteins and its mutants at pH 7.4

In this study, we are making a distinction between general aggregation and amyloid formation based on our observations. Aggregation is defined by the disappearance of monomers and the detection of sedimentable intermediates we call non-fibrillar macromolecular (NFM) intermediates (previously called nebular meshes of non-fibrillar oligomers [11]) by transmission electron microscopy (TEM). Amyloid formation is defined by the disappearance of monomers, Thioflavin T fluorescence enhancement, and the presence of fibrils by TEM.

The aggregation reactions of all the AL proteins were prepared at 20 μM in PBS at pH 7.4. The kinetic traces obtained from the sedimentation assay for all AL proteins shows a sigmoidal decay in the concentration of monomers, suggesting a nucleation dependent polymerization. Further, the rate of aggregation of all proteins was observed to be very similar (shown as the t₅₀ value) by the sedimentation assay (Table 1). The t₅₀ values determined from the ThT fluorescence assay for κI O18/O8 Y87H, AL-12, and AL-12 R65S show very large errors in the rate of fibril formation and the ThT fluorescence enhancement, suggesting a stochastic aggregation reaction. These proteins also show a very low increase in ThT fluorescence intensity. κI O18/O8 did not present the expected ThT fluorescence enhancement (4-fold increase) to confirm amyloid formation in any of the independent replicates. AL-09 and AL-09 H87Y on the other hand, consistently demonstrated a very fast rate of aggregation by ThT assay with a sharp increase in ThT fluorescence intensity across all biological replicates (Table 1).

Table 1.

Amyloid Kinetics and Folding thermodynamic parameters of AL proteins at pH 7.4.

Proteins	t50 (Hours)		Tm (°C)	ΔGfolding at 4°C (kcal.mol−1)
	Sedimentation Assay	Thioflavin T Assay
κI O18/O8	62.4 ± 4.9 (n=5)	NR*	54.7 ± 0.3a	−6.1 ± 0.2a
κI O18/O8 Y87H	80.3 ± 13.9 (n=4)	192.2 ± 32.9 (n=4) (Very stochastic)	47.3 ± 0.4a	−4.6 ± 0.4a
AL-09	61.9 ± 14.6 (n=5)	45.9 ± 11.7 (n=3)	41.1 ± 1.0a	−3.5 ± 0.3a
AL-09 H87Y	62.1 ± 5.6 (n=5)	59.3 ± 4.0 (n=3)	54.6 ± 0.6a	−6.1 ± 0.3a
AL-12	67.8 ± 7.8 (n=3)	Rate could not be determined (n=3) (Highly stochastic)	46.3 ± 0.4b	−4.4 ± 0.8b
AL-12 R65S	81.5 ± 23.6 (n=3)	207.8 ± 76.8 (n=3) (Very stochastic)	49.0 ± 0.2b	−4.3 ± 0.9b

Table shows average t₅₀ values obtained from sedimentation and ThT assays for different proteins at 20 μM. “n” represents the number of biological replicates used to obtain the reported average t₅₀ values. The error in the t₅₀ values represents SEM.

^*NR− No Reaction;

^aData from [16];

^bData from [17]

The sedimentation assay data of all the proteins at pH 7.4 demonstrated a well-defined nucleation phase followed by the elongation phase (characterized by exponential decrease in the concentration of monomers) and a relatively short stationary phase. A conspicuous fourth phase (dissociation) at the end of the stationary phase was observed for all the proteins, regardless of the extent of ThT fluorescence enhancement. In contrast, our previous study at pH 2.0 demonstrated a long and stable stationary phase and did not show any dissociation of aggregates [11]. In addition, only ~70–80% of monomers aggregated in all the AL proteins studied at pH 7.4 by sedimentation assay as opposed ~95– 98% monomer aggregation observed at pH 2.0 [11]. These differences emphasize the importance of inherent conformational fluctuation originating from low pH in fibril assembly and stability, making the pH 2.0 aggregation reactions more favourable for these proteins. Although none of the aggregation phases have a well-defined demarcation, the approximate time limits of each phase are reported in Table ST2.

Elution properties of all proteins from SEC demonstrated a monodisperse peak with V_e ranging from 17.3–17.6 mL (at the beginning of the experiment). The apparent molecular weight calculated from the protein standards (Figure S2) within this range computes to 12.9–13.6 kDa − suggesting a predominance of monomeric protein [11]. The exception was AL-09 which presented a monodisperse peak with V_e = 18.9 mL (apparent molecular weight of 5.3 kDa). This drastic deviation in the apparent molecular weight observed for monomeric AL-09 was previously observed and reported, and is possibly due to a strong interaction between the protein and the Superdex 200 column matrix (supplementary information within references [9, 11]).

TEM images (Figure S3 – S8) of all proteins demonstrated two major aggregating species during the course of aggregation − NFM intermediates and fibrils. Normally the size and population of low molecular weight oligomers formed during the aggregation pathway can be evaluated by mass spectrometry, chromatography, gel electrophoresis, sedimentation (analytical ultracentrifugation), fluorescence correlation spectroscopy and light scattering techniques [25, 26, 27, 28]. However it was very difficult to quantify or assess the physical characteristics of the NFM intermediates owing to their large size and extreme morphological heterogeneity. Detailed biophysical analysis of NFM intermediates is beyond the scope of the present investigation.

Aggregation of κI O18/O8 (germline control) and κI O18/O8 Y87H (reciprocal mutant)

κI O18/O8 aggregated slightly faster than κI O18/O8 Y87H by sedimentation assay (Figure 1A and B; Table 1). The intrinsic tryptophan fluorescence assay at the beginning of the aggregation reaction showed a larger wavelength at which maximum emission occurs (${\lambda }_{max}^{em})$ in κI O18/O8 Y87H (337 nm) compared to κI O18/O8 (${\lambda }_{max}^{em}$ − 334 nm) indicating Trp 35 is slightly more exposed and accessible to the solvent in κI O18/O8 Y87H. Both proteins exhibited significant changes in the spectral properties during the course of aggregation (Figure 1C and D; S9A and B). κI O18/O8 Y87H demonstrated a significant red shift of ~ 8 nm prior to elongation phase, suggesting a change in the local hydrophobic environment for Trp 35. Considering the altered dimer interface (κI O18/O8 Y87H dimer is rotated at 180° with respect to the canonical dimer) [16, 29] and backbone deviations triggered by the Y87H mutation, the fluorescence data is consistent with an exposed Trp 35 as the aggregation reaction progresses. In contrast, κI O18/O8 exhibited a progressive quenching of Trp35 fluorescence without a shift in ${\lambda }_{max}^{em}$, suggesting a fine interplay between accessibility of Trp 35 to effective quenchers (disulfide bond, lysine, glutamine, asparagine, and tyrosine [30, 31]).

No significant changes in the secondary structural characteristics (Figure 1E and F; S9C and D) or in the elution properties (Figure S9E and F) of these proteins (using CD and SEC) were observed during nucleation phase. However the far UV-CD spectra in both proteins were distorted by the increasing amount of insoluble material present in our samples during the elongation phase. The apparent molecular weight calculated from V_e values for both proteins indicate a predominant population of monomers. TEM analysis of κI O18/O8 showed NFM intermediates during the different phases of aggregation at pH 7.4 (Figure S3A–D). In addition to the NFM intermediates, κI O18/O8 Y87H (Figure S4A–C) showed a small population of short fibrillar forms in the dissociation phase (Figure S4D) in agreement with the findings from ThT assay.

Aggregation of AL-09 (mutant) and AL-09 H87Y (restorative mutant)

Aggregation kinetics of AL-09 and AL-09 H87Y show amyloid fibril formation at very similar rates (Figure 2A and B; Table 1). AL-09 presented unique behaviour compared to AL-09 H87Y (and all the proteins in this study).

Intrinsic fluorescence shows that Trp 35 is slightly less exposed in AL-09 (t=0: ${\lambda }_{max}^{em}$ − 332 nm) as compared to AL-09 H87Y (t=0: ${\lambda }_{max}^{em}$ − 336 nm) (Figure 2C and D). No significant shift in ${\lambda }_{max}^{em}$ was observed over the course of the reaction. AL-09 presents a progressive increase in the fluorescence intensity (~nucleation and elongation phases) followed by a sequential decrease in stationary phase (Figure S10A). Such an oscillation in fluorescence intensity might result from the moderate restructuring of the protein, leading to increased solvent accessibility to Trp 35. The oscillation in fluorescence intensity is followed by a significant quenching of Trp 35 possibly due to the close proximity to effective quenchers [30, 31]. This quenching is likely due to the formation of higher order aggregates in the elongation and stationary phase. AL-09 H87Y presents distortions in the emission intensity observed in the late elongation phase that may result from Raman scattering (Figure 2D and S10B).

We observe progressive loss of secondary structure in far UV-CD spectra in both proteins during the nucleation phase (Figure 2E and F; S10C and D). AL-09 presented a unique stepwise obliteration of the minima followed by the complete disappearance of the negative peak at ~ 235 nm and a significant shift of the negative peak at ~ 217 nm.

AL-09 forms tetrameric and decameric intermediates during the nucleation phase as shown with SEC (Figure S10E). This is in contrast with the predominant monomeric population found for AL-09 H87Y (Figure S10F) and the rest of the proteins in this study. TEM image analysis of both AL-09 (Figure S5A–D) and AL-09 H87Y (Figure S6A–D) showed a small population of NFM intermediates during the nucleation phase followed by detection of codominant population of both NFM intermediates and bundles of short fibrils in the elongation phase. In the late stages of aggregation – stationary and dissociation phases – we observed confluent masses of short fibrillar forms co-existing with a small population of NFM intermediates.

Aggregation of AL-12 (mutant) and AL-12 R65S (restorative mutant)

AL-12 aggregated slightly faster than AL-12 R65S as determined by sedimentation assay. The ThT fluorescence intensity signal was very stochastic for both proteins (Figure 3A and B; Table 1). We observed a larger ThT fluorescence enhancement for AL-12 R65S. The ThT fluorescence data of AL-12 (Figure 3A) represents data from a single aggregation experiment among at least three independent replicates. Although AL-12 did not present ThT fluorescence enhancement, we did observe significant increase in ThT fluorescence intensity in one of the biological replicates towards the end of the reaction. We further corroborated the presence of amyloid in AL-12 by TEM in that biological replicate. Intrinsic fluorescence of both proteins demonstrated that the Trp 35 is slightly more exposed in AL-12 [t=0: ${\lambda }_{max}^{em}$− 335 nm] as compared to AL-12 R65S [t=0: ${\lambda }_{max}^{em}$− 328 nm] (Figure 3C and D). AL-12 presents a small blue shift in Trp fluorescence during the nucleation phase (Figure 3C), while AL-12 R65S presented a significant red shift (~8 nm) (Figure 3D). No significant changes in the emission intensity of these proteins were observed (Figure S11A and B). This suggests that AL-12 and AL-12 R65S undergo different conformational changes prior to aggregation involving the Trp 35 microenvironment.

The secondary structural characteristics (Figure 3E and F; S11C and D) and elution properties (Figure S11E and F) of both proteins remain unaltered during the nucleation phase compared to their native monomeric state. We observed a significant distortion in the far UV-CD spectra in both proteins in the elongation phase, similar to what we observed for κI O18/O8, possibly due to the high concentration of insoluble material. TEM images of aggregating species in both proteins (AL-12: Figure S7A–D and AL12 R65S: Figure S8A–D) shows a dominant population of NFM intermediates throughout the aggregation reaction followed by a small population of short fibrillar species in the dissociation phase, similar to what was observed with κI O18/O8 Y87H.

Discussion

Our study demonstrates for the first time the dissociation of aggregates in AL proteins having broad and significant relevance to the pathophysiology of AL amyloidosis. We observed an inverse correlation between amyloid yield and the dissociation observed in different proteins. We established that at pH 7.4, all AL proteins studied here aggregate via the formation of NFM intermediates, suggesting a role of these intermediates in AL protein fibril formation.

Amyloidogenicity of proteins and its correlation with dissociation

We observed the presence of short amyloid fibrils of varying morphologies using TEM at the end of the aggregation reactions in all proteins except κI O18/O8. Previous reports from our laboratory identified AL-09 H87Y and AL-12 R65S as non amyloidogenic [9, 17], in contrast with our observations in this study. This discrepancy could be explained by the differential sensitivity of AL proteins to its surface environment (Eppendorf tubes compared to polystyrene plates) and the total volume of the reaction mixture as we reported before [11].

The sedimentation data, in combination with TEM, demonstrated that the decrease in the concentration of monomers observed during the elongation phase of κI O18/O8, κI O18/O8 Y87H, AL-12, and AL-12 R65S was exclusively due to the formation of condensed masses of sedimentable NFM intermediates. In contrast, AL-09 and AL-09 H87Y present codominant populations of NFM intermediates and fibrillar forms during the elongation phase.

Interestingly, our results show dissociation of aggregates (dissociation defined as the increase in soluble proteins in the solution after a continuous decrease in the early phases of the reaction) is inversely correlated with amyloid yield. The proteins κI O18/O8, κI O18/O8 Y87H, AL-12 and AL-12 R65S (which form NFM intermediates during the stationary phase) dissociate extensively. However, the concentration of the soluble proteins after dissociation was observed to be extremely stochastic in the independent biological replicates. In contrast, the highly amyloidogenic AL-09 and AL-09 H87Y presented the lowest residual monomer concentration during the entire aggregation reaction, suggesting a very favourable amyloid formation reaction. This is compelling evidence that intuitively confirm NFM intermediates are unstable and rapidly dissociate to more soluble forms at physiological pH. Since spectroscopic evidence suggest presence of non-native monomers in these intermediates, it could be argued that the species originating from NFM intermediates dissociation are soluble proteins likely composed of non-native monomers which may be in different self-associated states. We compared the critical concentration (C_r, defined as the concentration of residual monomeric protein after the aggregation reaction reaches equilibrium and bears an inverse correlation with aggregate stability [32]) of all aggregation reactions at pH 7.4 with our previously reported C_r values at pH 2.0 [11]. We observed that the C_r values range from 4.7 − 7.7 μM at pH 7.4 compared to 1.2 − 3.0μM at pH 2.0, suggesting that the aggregates (including NFM intermediates) formed at pH 7.4 are less stable than the fibrils formed at pH 2.0.

Dissociation of aggregates in AL proteins has never been reported before to the best of our knowledge. Dissociation of fibrils and oligomer formation are not new concepts for other amyloid diseases [33, 34]. Dissociation is proposed to arise due to packing defects within the fibril core. Aggregate dissociation leading to formation of soluble species could play a role in the overall pathophysiology of AL amyloidosis since soluble proteins in AL amyloidosis have been reported to cause cell dysfunction and apoptosis [35, 36] and oxidative stress [37, 38]. Further studies needs to be conducted to fully understand the mechanism of dissociation, and structural and mechanistic properties of the dissociated species. This will be an important consideration for potential therapeutic strategies that aim at fibril binding and dissociation.

Structural variations and the evolution of intermediates

AL-09 presents a progressive loss in secondary structure concurrent with an increased solvent accessibility of Trp 35. We propose that these changes facilitate the formation of AL-09 oligomeric intermediates. However, we were unable to conclusively identify the role of these oligomeric intermediates in AL-09 fibrillation at pH 7.4 due to the concomitant presence of sedimentable non-fibrillar intermediates (as determined by TEM).

CD and fluorescence data, in conjunction with TEM show that NFM intermediates have undergone structural changes, suggesting that the protein associated with these intermediates is in a non-native conformation. We asked whether or not somatic mutations or thermodynamic stability may play a role in the stabilization or generation of these intermediates. We found no correlation between thermodynamic stability (T_m and ΔG_folding) and the formation of NFM intermediates or fibrillar forms either quantitatively or qualitatively (Table 1). These NFM intermediates were also observed previously in our studies conducted at pH 2.0, particularly during nucleation phase. However, at pH 7.4, varying population of NFM intermediates was consistently observed during all phases of aggregation in the different proteins. This suggests that although these species are present at both pH values, they are more transient at pH 2.0.

In summary, our data show that NFM intermediates, act as crucial intermediate species in the process of fibril formation in AL proteins. However it is unclear how the NFM intermediates might mature into fibrillar species or play a role in fibril propagation. Based on our TEM observation and in conjunction with sedimentation and ThT assay we have presented a schematic model of evolution of species during aggregation of κIO18/O8, AL-09 and AL-12 at pH 7.4 (Figure 4).

Figure 4.

Schematic representation of the aggregation mechanism of κIO18/O8, AL-09 and AL-12 at pH 7.4. During the early phases we observe the formation of NFM intermediates. κIO18/O8 was devoid of any fibrillar species throughout aggregation. AL-09 demonstrated a significant increase in amyloid yield concomitant with the decrease in the population of NFM intermediates. AL-12 presented very low amyloid yield during the course of aggregation.

The observed NFM intermediates have never been characterized for their possible cytotoxic properties in AL amyloidosis models. The characterization of these new species is critical to understand the molecular mechanism of overall pathogenicity in AL amyloidosis as it is well established that low molecular weight oligomeric intermediates have significant cytotoxic implications for different amyloid disease models [25, 26, 39, 40, 41]. Future studies will be conducted to identify the role of NFM intermediates in monomer recruitment, fibril propagation, heterologous amyloid recruitment and cytotoxicity to fully understand the complex pathophysiology of AL amyloidosis.

Abbreviations:

AL

amyloidogenic immunoglobulin light chain

CD

Circular Dichroism spectroscopy

SEC

Size exclusion chromatography

TEM

Transmission electron microscopy

t₅₀

time taken to complete 50% of fibril formation reaction

ThT

Thioflavin T

NFM intermediates

non-fibrillar macromolecular intermediates

PBS

phosphate buffered saline

HPLC

high performance liquid chromatography

T_m

melting temperature

Trp

tryptophan

V_e

elution volume