The role of histidines in amyloid β fibril assembly

Low pH has a strong stabilising effect on the fibrillar assembly of amyloid β, which is associated with Alzheimer's disease. The stabilising effect is already pronounced at pH 6.0, suggesting that protonation of histidines might mediate this effect. Through the systematic substitution of the three native histidines in Aβ for alanines, we have evaluated their role in fibril stability. Using surface plasmon resonance, we show that at neutral pH the fibrillar forms of all His‐Ala variants are destabilised by a factor of 4–12 compared to wild‐type Aβ. However, none of the His‐Ala Aβ variants impair the stabilising effect of the fibril at low pH.

The self-assembly of amyloid b (Ab) into amyloid plaques is strongly linked to the development of Alzheimer's disease, which is the most common form of dementia and afflicts around 35 million people worldwide [1]. Ab is a 39-42 residue long peptide where the two forms Ab 1-40 and Ab 1-42 dominate. Ab is derived from the proteolytic cleavage of the single transmembrane Ab precursor protein by band c-secretase activity [2,3]. Depositions of amyloid fibrils seen as plaques as well as intracellular neurofibrillary tangles and neuronal loss are clinical hallmarks of Alzheimer's disease [4].
A delicate balance controls the difference between a native monomeric Ab peptide and its pathological aggregated state, and just a small shift in this equilibrium can have detrimental effects [5][6][7][8][9][10]. Elucidation of the mechanisms that mediate the pathological self-assembly is consequently of interest for developing therapeutic interventions against amyloid diseases.
The formation of Ab fibrils requires an initial nucleation event, and fibril elongation involves the addition of monomers to an existing fibril. In analogy to all polymerisation processes, Ab fibril formation is concentration dependent, and at a concentration of free monomers below the dissociation constant (K D ), also known as the critical concentration, fibril formation cannot occur [11,12]. We have previously determined that the critical concentration for Ab at neutral pH is around 100 nM for Ab  and slightly lower for Ab   [13]. A similar value was also previously found by Hasegawa and co-workers [14]. However, the extracellular concentration of Ab in vivo is around 1 nM for Ab  and even lower for the Ab 1-42 variant [15]. The low concentration in vivo in combination with the significantly higher K D has puzzled the field because fibrillar formation under these conditions should, in theory, be prohibited.
It has recently been shown that Ab self-assembly is highly dependent on pH and that a lowering of the pH significantly enhances the stability of the fibrils [11]. The K D for Ab 1-40 can be as much as 120 times lower at pH 4.5 than pH 7.4. This has several implications because it shows that Ab assembly can indeed form at physiological Ab concentrations and also strongly suggests that it occurs within intracellular compartments having a lower pH, such as endosomes or lysosomes [11]. It is therefore of interest to elucidate the stabilising factors associated with fibril stability and the potentiating effect of low pH.
The strong pH dependency is likely caused by a change in charge through the protonation of certain amino acid side chains. The stabilising effect is pronounced already at pH 6.0, and within this range only the histidine side chains, having a pKa around 6.0 are significantly affected. It has also previously been shown that truncating the first 16 amino acids of the N-terminus of the peptide results in a loss of the stabilising effect at low pH [11], which indicates that the N-terminal part controls this mechanism. The Ab sequence contains three histidines; all located in the N-terminal part of the peptide at positions 6, 13 and 14. In the present work, we have evaluated how the stabilising effect of lowering the pH is affected in the Ab 1-40 variant by the systematic substitution of the three histidines for alanines. Using surface plasmon resonance (SPR), we studied the single mutation variants Ab 1-40 H6A , Ab 1-40 H13A and Ab 1-40 H14A as well as a variant where all three histidines have been exchanged (Ab 1-40 H6A, H13A, H14A ).
Amyloid polymerisation follows a template-dependent polymerisation process where monomers are incorporated onto the fibrillar ends. The stability of an amyloid fibril (in analogy to all binding interactions) is determined by the rate of association (k (on) ) and the rate of dissociation (k (off) ), which reflects how fast a monomer is incorporated onto the fibrillar end and how long it remains bound.
Using SPR, it is possible to monitor both the k (on) and the k (off) in a polymerisation reaction, such as the formation of amyloid. The method is based on the immobilisation of preformed fibrils onto a surface followed by the subsequent probing of free monomers. The Ab monomers then incorporate onto the fibrillar ends and a continuous increase in the mass on the SPR surface can be monitored. Upon stopping the injection, while maintaining a flow of the running buffer, fibril dissociation into monomers can selectively be observed. Through monitoring both the association and the dissociation, the strength of the interaction, K D , reflecting the fibril stability, can be evaluated [13,14], further discussed below.
Using SPR, we show that substituting any of the histidines results in significantly lower stability of the fibrils, at neutral pH, compared to the wild-type peptide (Ab 1-40 wt ). However, no single site mutant impaired the stabilising effect of low pH and, similar to Ab 1-40 wt , they all acquired more than a 200-fold increase in fibril stability upon shifting the pH from 7.4 to 5.5. This suggests that histidine residues contribute to fibril stability at neutral pH but are not involved in the strong potentiating effect on fibril stability induced by low pH.

Preparation of monomeric Ab peptides
Recombinant variants of Ab 1-40 , Ab 1-40 H6A , Ab 1-40 H13A , Ab 1-40 H14A and Ab 1-40 H6A, H13A, H14A were obtained from AlexoTech (Ume a, Sweden). All lyophilised peptides were solubilised in 20 mM NaOH and separated by size-exclusion chromatography (Superdex 10/300, GE Life Science, Uppsala, Sweden) in PBS with 2 mM EDTA, 5 mM NaOH and 150 mM NaCl. The pH of all purified peptide solutions was adjusted using a stock solution of phosphate or acetate buffer to give a final buffer concentration corresponding to 20 mM phosphate (pH 7.4) or acetate (pH 5.5) and 150 mM NaCl.

Surface plasmon resonance (SPR)
The interaction study between the fibrillar form of Ab and its monomeric form was performed using a BIA core Ò 3000 biosensor (GE HealthcareÒ, Uppsala, Sweden) equipped with a CM5 sensor chip (GE HealthcareÒ). First, the fibrillar form of Ab was immobilised at a density of 3500 response units (RUs) on the dextran layer of the chip using standard amino coupling reagents at pH 4.0. All SPR experiments were performed at 25°C at a flow rate of 20 lLÁmin À1 in either a 20 mM phosphate buffer (pH 7.4) or 20 mM acetate buffer (pH 5.5). All experiments were performed in the presence of 150 mM NaCl, which represents a physiological ion-strength.
Due to the polymeric nature of amyloid fibrils, the covalent immobilisation on the sensor chip, likely only occurs in minor fraction of the accessible amino groups. This treatment is therefore not anticipated to influence the overall incorporation of monomers in a significant manner. However, the influence from using immobilised fibrils or the dextran layer cannot be fully excluded and should therefore be disclosed.
All sensograms were corrected for nonspecific interactions using a reference surface according to standard procedures [16].

Transmission electron microscopy
A total volume of 4 lL of the different Ab variants was absorbed for 2 min onto glow-discharged carbon-coated copper grids. Samples were washed in water and immediately stained in 50 lL of 1.5% uranyl acetate solution for 30 s. Negatively stained samples were examined on a JEM1230 transmission electron microscope (JEOL) at 80 kV. Transmission electron micrographs were recorded with a Gatan UltraScan 1000 2k 9 2k pixel CCD camera using DIGITALMICROPGRAPH TM software (Gatan, Pleasanton, CA, USA.).

Kinetic analysis of fibril formation using SPR
In analogy to all molecular interactions, the stability of the binding between an Ab monomer and the fibrillar end is dependent on both the rate of assembly k (on) and the rate of dissociation k (off) .
Probing free monomeric Ab against immobilised fibrils using SPR is a frequently used technique for measuring the intrinsic kinetic properties of Ab polymerisation [11,14,[17][18][19]. The SPR sensograms can be separated into an association phase (k (on) ) and a dissociation phase ((k (off) ). Figure 1 shows the result when immobilised Ab 1-40 wt fibrils are probed with free Ab 1-40 wt monomers.
The end state of the association phase is indicated by arrow 1 in Fig. 1.
Note that the dissociation phase exposes a biphasic curvature. This is highly representative and frequently observed when performing these kinds of experiments. The suggested mechanism behind this phenomenon is further discussed below. However, the initial fast phase of dissociation (indicated by arrow 2, Fig. 1) is much weaker than the second phase (indicated by arrow 3, Fig. 1) and does not contribute significantly to the fibril stability. All analyses of decay rates within this work are therefore exclusively focused on the slower second phase of dissociation.
Through a systematic exchange of the three histidine residues of Ab, we will expose below their role within fibril formation with respect to association, dissociation and fibril stability.
Kinetic analysis of the association rate, k (on) Fibrils from recombinant variants of Ab 1-40 , Ab 1-40 H6A , Ab 1-40 H13A , Ab 1-40 H14A and Ab 1-40 H6A, H13A, H14A were prepared by prolonged incubation in either PBS (pH 7.4) or acetate buffer pH 5.5 containing 150 mM NaCl. Fibrils immobilised on a CM5 chip were then probed with their monomeric counterpart at different concentrations of Ab monomers and analysed by SPR. Samples were injected for exactly 60 s and the maximal value at the end state of each injection (indicated by arrow 1 in Fig. 1A) was plotted as a function of the monomeric concentration, Fig. 2. This generates an association rate at each point of measurement. A guideline is given in the figures to illustrate the linearity.
By comparing the rate between the different Ab variants, it is possible to determine relative differences in k (on) .
At pH 7.4, all histidine mutants of Ab had a k (on) slightly slower than Ab 1-40 wt . The results are shown in Fig. 2 Table 1.
Interestingly, the corresponding k (on) at pH 5.5 revealed that all Ab His-Ala variants, in analogy to Ab 1-40 wt , displayed a much faster rate of association as a function of a decrease in pH to 5.5. The degree of potentiation, indicated in Table 1, is similar to Ab 1-40 wt and a significant difference in RA was only noted regarding Ab 1-40 H6A and Ab 1-40 H6A, H13A, H14A , which displayed a 1.94 and 1.57 times higher RA when compared to Ab 1-40 wt .

Kinetic analysis of dissociation rate
In contrast to the association rate, k on , the rate of dissociation, k off , is independent of the concentration of monomers. Subsequent to probing, the immobilised fibrils with free monomers, the rate of dissociation can be seen as a decrease in the signal over time.  with previous results [11], is significantly stabilised by the decrease in pH. The results, however, also expose a significant stabilisation for all Ab His-Ala variants and a significantly slower phase of dissociation is observed upon lowering the pH to 5.5.
Because the latter part of the curve, representing the slower phase of dissociation (indicated by arrow 3 Fig. 1), strongly dominates the interaction strength, we have exclusively focused on this part, and no points before 100 s were included in the curve. The dissociation curve could then be fitted to a one exponential decay. The relative dissociation rate in relation to Ab 1-40 wt (RD) for the different mutants is given in Table 1.

Determination of the relative fibril stability
While a simple interaction between, for example, a receptor and ligand can easily be determined through saturation experiments, this cannot be done with polymers since they, per definition, cannot be saturated.
This problem can, however, be circumvented by defining the concentration of free monomers where the fibrils are at steady state, i.e. the concentration of free monomers where neither association nor dissociation can be detected. We have previously shown that the K D for Ab 1-40 wt at neutral pH is around 100 nM and around 1 nM at pH 5.5 [11,13]. The K D at neutral pH has also been investigated by Hasegawa and co-workers [14] with similar results. All figures used here are therefore given as relative values to the K D for Ab 1-40 wt corresponding to 100 nM at neutral pH and 1 nM at pH 5.5.
Multiplying the relative RD/RA with the K D for Ab 1-40 wt will then give the K D for the histidine derivatives.
For all of the evaluated His-Ala Ab variants, the substitutions destabilised the fibrillar conformation at neutral pH as indicated by a more rapid decay compared to Ab 1-40 wt (Fig. 3A). The overall stability was decreased 4-12-fold in all of the His-Ala variants compared to Ab 1-40 wt . At low pH, all of the histidine to alanine substitutions displayed a pronounced stabilisation. The relative stabilisation was even stronger than observed in Ab 1-40 wt and the potentiation ranged between 238 and 358 times simply by shifting the pH from 7.4 to 5.5. The dissociation relative to Ab 1-40 wt and the overall potentiation factor (PnF) considering both the k on and k off is indicated in Table 1.

Transmission electron microscopy verifies a fibrillar morphology
To verify the fibrillar morphology of the assembled peptides, we evaluated the acquired aggregates using negative-stain transmission electron microscopy. Assemblies of Ab 1-40 wt and the different Ab His-Ala  variants, formed at pH 7.4 as well as 5.5, were analysed. The results showed that none of the variants were impaired in forming fibrils and all displayed a similar morphology as Ab 1-40 wt fibrils, Fig. 4A-J.

Discussion
A very delicate balance controls the self-assembly of Ab monomers in vivo, and also a very small shift in the equilibrium can have detrimental effects [6][7][8][9]20]. Today, major efforts are being directed towards identifying methods that can prevent either Ab excision from APP [3,21,22] or impair Ab assembly [23][24][25][26].
It is therefore of interest to understand the factors controlling fibril formation, as well as the properties of the acquired assemblies.
It has recently been shown how low pH has a strong potentiating effect on fibril stability [11]. This finding also provides a possible explanation to how fibrils may form at the very low Ab concentrations found in the human brain, which had previously been considered to be far below the critical concentration for Ab fibril formation. This finding also strongly indicates that fibril formation occurs within the low pH compartments of the cell, such as endosomes and lysosomes.
From a mechanistic perspective, the effects of a change in pH on fibril formation are most likely due to protonation of amino acid side chains. Interestingly, a moderate lowering of the pH also has a strong effect where the potentiation factor already at pH 6.5 is around 20 times [11]. This suggests that some functional groups, having a pKa, within this pH range, likely mediate the effect of pH. The histidine side chains in Ab have a pKa corresponding to 6.0 [27] and their protonation may therefore possibly represent a part of the mechanism.
In this work, we show how a single exchange of any of the three histidines in Ab 1-40 decreases the fibril stability at neutral pH relative to Ab 1-40 wt . Substitution of either His13 or His14 results in an approximately seven-to eightfold lower stability of the fibril, while the exchange of His6 results in an approximately fourfold decrease in fibril stability. Exchange of all three histidines results in an overall 12 times less stable fibrillar assembly. In all cases, we show that the destabilising effect is dependent on both a decreased rate of association and an increased rate of dissociation.
Somewhat surprisingly, all His-Ala variants displayed a strong potentiation in binding strength upon lowering the pH from 7.4 to 5.5 and increased the fibrillar stability from 240-350 times. The result suggests that the histidine side chains do not mediate the potentiating effect. It should also be noted that the relatively higher potentiation regarding the Ab His-Ala variants is mostly an effect of the decreased stability at neutral pH and does not imply a stronger fibril than Ab 1-40 wt at low pH. At pH 5.5, all single-substitution variants had 1.42-2.22-fold lower stability compared to Ab 1-40 wt , while the fibrillar fold of the triple-substituted variant had a 5.26-fold lower stability.
Notably, the initial rapid phase of decay is also much more pronounced in the His-Ala substitution variants. A biphasic behaviour of the dissociation curve has been reported previously and is suggested to be an effect of a dock and lock mechanism [28][29][30]. This phenomenon implies that an initial poor association of the free monomer onto the fibrillar end first occurs, which is defined as the 'dock' phase. The peptide then adopts a more optimal orientation, with a stronger interaction between the newly incorporated monomer and the templating fibril, which is defined as the 'lock' phase.
A recent publication, however, suggests an alternative explanation where the initial low affinity interaction is due to a lateral assembly of monomers along already formed fibrils [31] and consequently does not reflect the interaction within the fibrillar end.
However, in terms of binding affinities, the initial rapid phase of dissociation only provides a small contribution to the overall binding. Upon analysis of the overall binding strength, the first phase has therefore been disregarded.
It is also of interest to discuss the role of histidines from a pathology perspective of AD. The role of fibrils as inert bystanders or active participants of the pathology is still a matter of debate and it is well known that prefibrillar assemblies, such as oligomers, exhibit a much higher cytotoxic effect than mature fibrils [32][33][34][35]. Factors that partly impair the assembly into fibrils may, instead, facilitate the formation of alternative assemblies. Ab also has metal-binding properties and the binding of both Zn 2+ and Cu 2+ has been shown [36][37][38][39] which are both potent inducers of Ab assembly [40,41]. However, the structure of the acquired aggregates differs significantly as compared to fibrils acquired in the absence of metals [36]. Interestingly, a dysregulation of Zn 2+ , in the AD brain has also been assigned a pathophysiological role within AD where the increased levels of Zn 2+ results in alternative Ab assemblies [42]. The binding of both Zn 2+ and Cu 2+ to Ab involve all three histidines [43,44] and, given present results showing that histidine residues affect the fibrillar stability, may suggest that metals may have the same effect.
Taken together, it can be concluded that the histidine residues of Ab contribute to fibrillar stability at neutral pH but that their protonation is not a key event that can explain the strong potentiation in Ab fibril stability that is induced as a function of low pH.