Association of β-amyloid peptide fragments with neuronal nitric oxide synthase: Implications in the etiology of Alzheimers disease

Abstract

Neuronal nitric oxide synthase (nNOS) was purified on DEAE-Sepharose anion-exchange in a 38% yield, with 3-fold recovery and specific activity of 5 µmol.min⁻¹.mg⁻¹. The enzyme was a heterogeneous dimer of molecular mass 225 kDa having a temperature and pH optima of 40°C and 6.5, K_m and V_max of 2.6 μM and 996 nmol.min⁻¹.ml⁻¹, respectively and was relatively stable at the optimum conditions (t_½ = 3 h). β-Amyloid peptide fragments Aβ_17–28 was the better inhibitor for nNOS (K_i = 0.81 µM). After extended incubation of nNOS (96 h) with each of the peptide fragments, Congo Red, turbidity and thioflavin-T assays detected the presence of soluble and insoluble fibrils that had formed at a rate of 5 nM.min⁻¹. A hydrophobic fragment Aβ_17–21 [Leu₁₇ – Val₁₈ – Phe₁₉ – Phe₂₀ – Ala₂₁] and glycine zipper motifs within the peptide fragment Aβ_17–35 were critical in binding and in fibrillogenesis confirming that nNOS was amyloidogenic catalyst.

Keywords::

• Neural nitric oxide synthase
• Alzheimer disease; inhibition
• amyloid peptides

Introduction

Though the deposition of aggregated β-amyloid senile plaques and neurofibrillary tangles in the human brain are classic observations in the neuropathology of Alzheimer disease^1–3 an understanding of the mechanism of their formation remains elusive. Furthermore it also remains uncertain whether the elevated level of arginine in the brains and cerebrospinal fluid of Alzheimer patients is as a result, or a cause, of the disorder³. Any drug and/or metabolite that can inhibit the progression of the neurological disorder requires an understanding of the molecular causes underlying the neurodegenerative processes.

Since it has been reported⁴ that the astrocytes in the diseased brain not only function to store arginine but are surrounded by the insoluble amyloid plaques it would be logical, in the etiology and pathogenesis of Alzheimers disease, to investigate arginine-metabolising enzymes and their intimate association with amyloid peptides. Neuronal nitric oxide synthase (nNOS) [EC. 1.14.13.39] uses arginine as a substrate producing citrulline and nitric oxide (NO) (Scheme 1). This reaction, localized within the astrocytes in the brain, induces proteins in the cellular environment to unfold and trigger the aggregation of susceptible peptides⁵ such as amyloid peptides. An understanding of the function and action of nNOS, with respect to amyloid peptide aggregation and subsequent formation of senile plaques, may facilitate an understanding of neurodegeneration in Alzheimer disease.

Scheme 1.  Schematic representation of enzyme reaction for neuronal nitric oxide synthase.

Earlier reports from our group^6,7 have shown that certain amyloid peptide fragments, not only initially inhibit the arginine-metabolising enzyme—peptidylarginine deiminase—but also act as catalysts in the enzyme induced process of fibrillogenesis. It is thought that a decrease in Aβ catabolism is directly responsible for the accumulation of this peptide in the brain and its subsequent aggregation and plaque formation⁸. The higher the concentration of Aβ in vivo, the more likely they are to aggregate and form insoluble plaques. Amyloid fibrils are generally formed through a process of assembly of amyloid proteins either self-induced or by a proteolytic mechanism. It is relatively well known^8,9 that there are a number of amyloidogenic proteins and short peptide fragments, including the Aβ_25–35 fragment itself, that are capable of producing amyloid fibrils¹⁰ and neurotoxic oligomers¹¹.

This present study uses a series of readily available different sized Aβ peptides [Aβ_17–28, Aβ_22–35 and Aβ_32–35] along with the neurotoxic elements Aβ_25–35 and Aβ_1–40, in order to investigate which sequence would, primarily, be involved in inhibition of nNOS and which sequence would be involved in the formation of amyloid fibrils induced by nNOS. Kinetic analysis was used to measure the kinetic parameters (V_max, K_m] and affinity constants (K_i) while turbidity at 400 nm¹², Congo Red assay¹³ and Thioflavin-T (Th-T) staining fluorescence^14–16 were used to investigate the formation of insoluble fibrils and aggregation kinetics of nNOS induced fibril formation.

Materials and methods

Materials

Bovine brain was kindly donated by Rosedale abattoir (Grahamstown, South Africa). All amyloid peptides [Aβ_17–28, Aβ_22–35, Aβ_32–35, Aβ_25–35 and Aβ_1–40], acrylamide; bis-acrylamide, bromophenol blue, N, N, N’, N’- tetramethylethylenediamine, ammonium persulphate, β-mercaptoethanol, Coomassie Blue commercial stain, Nα-benzoyl-l-arginine ethyl ester hydrochloride, DL-citrulline, calcium chloride, tetrahydro-l-biopterin dichloride, β-nicotinamide adenine dinucleotide phosphate reduced tetrasodium salt, thiosemicarbazide, 2,3 butanedione monoxime, DEAE-Sepharose^®, bovine serum albumin, Bradford reagent, ethylenediaminetetracetic acid and 4-(2-hydroxyethyl) piperazine-N’-(2-ethane-sulphonic acid monosodium salt [Hepes] were obtained from Sigma-Aldrich (Johannesburg, South Africa). Commercial nNOS obtained from rat brain, dithiothreitol (DTT), sodium dodecyl sulphate (SDS), glycerol (50%), Tris (hydroxymethyl) aminomethane, phenylmethylsulphonylfluoride (PMSF), ferric chloride hexahydrate (FeCl₃), sodium chloride (NaCl), perchloric acid, sulphuric acid, ortho-phosphoric acid, glacial acetic acid and hydrochloric acid were obtained from Merck Chemicals (Statsoft Inc., Sandton, South Africa). Molecular weight markers were obtained from PEQGOLD (Erlangen, Germany). In all the experiments, reagents were dissolved in milli-Q water and all the experimental conditions throughout the purification were performed at 0–4°C except where indicated. Snake skin dialysis tubing was obtained from Pierce (Rockford, IL). UV analyses were performed on a PowerWave XTM microplate reader (Bio-Tek Instruments, Winooski, VT).

nNOS assay

This assay was conducted, with slight modifications, according to the protocol previously described¹⁷. The reaction mixture contained benzoyl-l-arginine ethyl ester (5 mM, 10 μl), CaCl₂ (5 mM, 10 μl), DTT (2 mM, 10 μl) in Tris-HCl buffer (50 mM, pH 7.6, 940 μl). The reaction was started by the addition of brain extract (10 μl) in NADPH (1.0 mM, 10 μl) and allowed to incubate at 40°C for 2 min, before being stopped with perchloric acid (5 M, 10 μl). This mixture (250 μl) was treated with chromogenic reagent (250 μl) and cooled to 22°C for 2 min. The reduction of benzoyl-l-arginine ethyl ester was then determined spectrophotometrically at 530 nm. Chromogenic reagent was made from thiosemicarbazide (18 mM) in 2.3 butanedione monoxime solution (500 mg in 100 ml distilled water) and sulphuric:orthophosphoric acid (1:1) in the presence of FeCl₃ (0.25 g). The citrulline extinction coefficient (ϵ₅₃₀) was 27.3 ml.μmol⁻¹¹⁸. One unit of activity was that amount of nNOS that produced 1 μmol of citrulline per min per ml reaction mixture.

Protein determination

The protein concentration for all experiments was routinely determined according to the method of Bradford¹⁹. The assay was performed in triplicate in a 96-well microplate. Enzyme extract (5 μl) was incubated (22°C, 10 min) with Bradford reagent (245 μl), the absorbance measured at 595 nm and the concentration determined from a BSA standard curve.

Purification of nNOS

Isolation

Fresh bovine brain (374 g) was homogenized by sonication (10W, 30 s intervals, 4 min) in Hepes buffer (50 mM, pH 7.6, 600 ml) that contained EDTA (1.0 mM), NADPH (1.0 mM), DTT (0.5 mM) and PMSF (0.43 mM). The cell debris was removed by centrifugation (10,000g, 4°C, 30 min) and the crude cell-free extract assayed for nNOS activity and protein concentration then stored as 20 ml aliquots at −70°C until required.

Anion- exchange on DEAE-Sepharose

The crude extract (20 ml), obtained after centrifugation and sonication was applied to a DEAE-Sepharose anion-exchange resin, previously equilibrated with Tris-HCl buffer (50 mM, pH 7.6) and washed with the same buffer until A₂₈₀ was at the baseline. The adsorbed proteins were then eluted with NaCl (0–1 M) in the same buffer at a flow rate of 2 ml.min⁻¹ and fractions (5 ml) collected, assayed for nNOS and protein and active fractions pooled and dialysed overnight against Tris-HCl buffer (50 mM, pH 7.6).

Characterisation of nNOS

pH, temperature, stability and kinetic profiles

To determine the pH optimum, the enzyme extract (10 µl) was assayed in different buffers [sodium acetate (pH 3–5.5, 50 mM); Hepes (pH 6–7, 50 mM); Tris-HCl buffer (pH 7.5–9, 50 mM)] under the standard assay conditions. The temperature optimum of the partially purified enzyme was determined in Tris-HCl buffer (pH 7.6, 50 mM) over a range of 20–70°C. The reaction was started by addition of enzyme suspension (10 µl) at the different temperatures. The temperature stability of nNOS was determined at the optimum temperature and pH. nNOS activity at time zero was considered to be 100% relative activity and aliquots (10 µl) were removed at 1 h intervals and analyzed for nNOS activity for a maximum period of 12 h. The kinetic properties (K_m and V_max) were determined by varying the substrate (benzoyl-l-arginine ethyl ester, 3 ml) between 0 and 10 µM. nNOS activity was determined at each substrate concentration.

SDS and native PAGE analyses

The effectiveness of the purification process was determined by SDS-polyacrylamide gel electrophoresis (PAGE) after the method of Laemmli²⁰ on samples exhibiting synthase activity. Samples from each purification step (20 µl) and a standard molecular weight marker (10–170 kDa), were electrophoresed on 12% SDS-PAGE at 200 V. The gels were stained with Coomassie Brilliant Blue R-250, then destained in methanol: acetic acid: water (1:1:8 v/v/v). The molecular weight of the partially purified nNOS was determined using a standard curve of log molecular weight versus distance migrated. The procedure for native PAGE was performed with 12% separating gel at the same constant voltage of 200 V. After electrophoresis, the protein band was sliced out, re-suspended in Tris-HCl buffer (20 mM, pH 6.5, 1.0 ml) and incubated with benzoyl-l-arginine ester (0.1 g). The activity was assayed as previously described.

Interaction of nNOS with amyloid peptides

Kinetic analysis

Partially purified nNOS (5 μl) in Tris HCl buffer (50 mM, pH 7.6) was treated with amyloid peptides Aβ_1–40, Aβ_22–35, Aβ_17–28, Aβ_32–35 and Aβ _25–35 at 2 µM in a total volume of 1.0 ml and in the presence of benzoyl arginine ethyl ester (0–10 µM). The nNOS activity was assayed, at each substrate concentration.

Congo Red assay

Congo Red assay was performed according to that reported in the literature¹³ with slight modifications. Partially purified nNOS (5 μl) in Tris HCl buffer (100 mM, pH 7.6) was mixed with amyloid peptides Aβ_1–40, Aβ_22–35, Aβ_17–28, Aβ_25–35 and Aβ_32–35 (10 nM). Aliquots (20 µl) were periodically removed over 96 h and incubated (30 min) with a Congo Red solution (12.5 µM), prepared in Tris-HCl buffer (pH 7.6, 100 mM) containing NaCl (150 mM) in a final volume of 1.0 ml. Absorbance was read at 480 and 540 nm at 24 h intervals and the degree of fibrillogenesis estimated from the amount of Congo Red bound to the fibrils (equation 1) where the values 25.955 and 46.306 are the extinction coefficients of Congo Red-fibril complex and Congo Red, in units of ml.μmol⁻¹, respectively.

1

Th-T assay

Th-T was used to indicate the presence of Aβ fibrils using a modified version of that previously reported^14–16. nNOS solution (5 μl) was added to a reaction mixture containing Th-T (2 μl, 3.14 mM) and NaOH (200 μl, 10 mM) in Tris HCl buffer (pH 7.6, 100 mM) and NaCl (150 mM) in a final volume of 1.0 ml. Increasing concentrations of each Aβ peptide were added and incubated (22°C, 30 min) and the shift in fluorescence monitored at an excitation wavelength of 440 nm and an emission wavelength of 482 nm.

Turbidity assay

The turbidity assay was performed as previously described¹² with a slight modification. Aliquots (5 µl) of each Aβ peptide dissolved in dimethylsulphoxide were added to Tris-HCl buffer (10 mM, pH 7.6) and partially purified nNOS (5 µl) to give a final concentration of 5 nM. Aggregation of fibrils was measured as turbidity at 400 nm.

Transmission electron microscopy (TEM)

A 10 µl sample of either amyloid peptide alone or those that had been challenged with nNOS was placed onto a carbon-coated copper grid, excess liquid removed by carefully blotting with filter paper and then the sample negatively stained with uranyl acetate (2%, 30 s), washed twice in deionised water, blotted dry with filter paper and viewed using a JEOL JEM-1210 TEM at an acceleration voltage of 100 kV.

Statistical analyses

All experiments were carried out in triplicate. Mean and standard deviation calculations and comparison of data using analysis of variance was performed to 5% level of significance (p < 0.05) using Statistica for Windows, version 8 (Statsoft Inc.) and Microsoft Excel 2007.

Results and discussion

The enzyme proved difficult to obtain in a pure form and so a partially purified product was used. Preliminary attempts to purify nNOS by ammonium sulphate fractionation and/or polyethylene glycol led to about 80–85% loss in activity and so were abandoned in favour of ion-exchange chromatography on DEAE-Sepharose. Eight active fractions (39–46) were eluted with ~0.75 M NaCl in Tris-HCl buffer (50 mM, pH 7.6) (Figure 1) to afford partially purified nNOS in 38% yield, 3-fold recovery and a specific activity of 5 µmol.min⁻¹.mg⁻¹. SDS-PAGE analysis revealed that the partially purified nNOS from the DEAE-Sepharose column appeared to be a dimer of ~75 kDa and 150 kDa (Supplementary material, Figure 1A). This indicated an overall molecular weight of the total enzyme of 225 kDa which was similar to values of 230 kDa reported^21,22. The source of the commercial sample of nNOS (Supplementary material, Figure 1A; Lane C) was not from the bovine brain but from the rat brain and under the conditions of the SDS-PAGE did not show any dimers whatsoever. Native-PAGE revealed a single band for the partially purified enzyme estimated at 225 kDa (Supplementary material, Figure 1B).

Figure 1.  Elution profile of DEAE-Sepharose ion-exchange chromatography. Proteins eluted with linear NaCl gradient (0–1 M) in Tris-HCl buffer (50 mM, pH 7.6); Flow rate, 2 ml/min (5 ml per tube).

The enzyme indicated a pH optimum at pH 6.5 as there was a 40% decrease in activity between pH 4.5 and 6.5 and 6.5 and 8 but rapidly lost its activity beyond these extremes (Supplementary material, Figure 2A). Similar values of 6.8 were reported by Riveros-Moreno et al.²³ Based on the temperature profile (Supplementary material, Figure 2B), the enzyme showed an optimum of 40°C as there was a 40% decrease in activity between 25 and 40°C and 80% decrease between 40 and 60°C. This optimum temperature was comparable to a reported value of 37°C for nNOS²⁴. The nNOS had a half life of ~3 h (Supplementary material, Figure 2C) when incubated under optimal conditions of temperature and pH.

Figure 2.  A: Michaelis–Menten and B: Hanes–Woolf plots for Aβ_17–28 interaction with nNOS in the presence of different concentrations of benzoyl arginine ethyl ester.

A study into the effect of substrate concentration on the kinetic activity of nNOS was investigated by measuring enzyme activity over a range of benzoyl arginine ethyl ester concentrations (0–10 µM). A typical increase in substrate concentration resulted in a proportional increase in activity as shown by the Michaelis–Menten and Hanes–Woolf plots (Figure 2) and for benzoyl arginine hydrolysis, the V_max and K_m were 996 nmol.min⁻¹ and 2.6 μM, respectively. The turnover number (k_cat) and the catalytic efficiency, calculated using equations 2 and 3 and were found to be 50.8 min⁻¹ and 19.54 min⁻¹.μM⁻¹, respectively. The total amount of enzyme [Et] was 19.6 nmol.

2 3

The kinetic constant K_m (2.6 μM) indicated a similar affinity for benzoyl arginine compared with other reported values (11 μM; 2 μM; 3.2 μM; 1.5 μM) for nNOS^22–25 and the V_max of 996 nmol.min⁻¹ was only slightly lower than that of 1 μmol.min⁻¹ reported elsewhere²⁵.

The affinity of the nitric oxide synthase for amyloid peptides was determined by including the peptides to a final concentration of 2 µM with the substrate benzyl arginine ethyl ester (0–10 µM, 1 ml). The results indicated that with Aβ_17–28 the V_max decreased from 996 to 286 nmol.min⁻¹ ml⁻¹ (Figure 2) while the K_m value remained unchanged at 2.6 µM indicating that the nNOS was inhibited, in a non-competitive fashion, by the amyloid peptide. Furthermore each of the other amyloid peptides (Aβ_1–40, Aβ_22–35, Aβ_32–35 and Aβ_25–35) at a concentration of 2 µM, also inhibited nNOS. The inhibitor constant (K_i) calculated from equation 4, indicated the binding affinity of enzyme for inhibitor (Table 1).

4

where Aβ = concentration of amyloid peptide; V_max^app = apparent maximum velocity in the presence of amyloid peptides. Results showed that Aβ_17–28 inhibited nNOS the most as reflected by its low K_i value of 0.81 µM compared to the other K_i values for Aβ_1–40, Aβ_22–35, Aβ_25–35 and Aβ_32–35 (14.4 μM; 7.4 μM; 10 μM and 8.8 μM) (Table 1).

Table 1.  Inhibitor binding constants (K_i) for amyloid peptides on nNOS.

Aβ	1–40	22–35	17–28	32–35	25–35
Ki (µM)	14.4	7.4	0.81	8.8	10

When each of the amyloid peptides (at 2 µM) was incubated with the enzyme for an extended period (20–30 min) interesting results were forthcoming in that full enzyme activity was restored (Table 2). This time-dependent observation suggested an association—dissociation between nNOS and the amyloid peptide fragment of which the strength and time of effect was dependent on the type and concentration of the fragment. Furthermore it pointed to the fact that the enzyme converted the amyloid peptide into a form that was no longer capable of binding. In our opinion (and as we will indicate later) there was a strong probability that, first soluble then insoluble, amyloid fibrils had been generated by a process called fibrillogenesis. Since the enzyme activity was restored to its original value after about 20 min it ruled out any possibility that the fibrils blocked the enzyme active site. Though the data in Table 2 does not infer any quantifying evaluation for the interaction of the amyloid peptides with the enzyme it does support the fact that Aβ_17–28 peptide caused rapid inhibition (95% within 2 min). All of the other peptides had slightly less influence producing about 90% inhibition within 4–6 min.

Table 2.  Influence of amyloid peptides on nNOS activity over time.

nNOS activity (µmol.min^−1.ml^−1
Aβ	Time (min)
	0	2	4	6	8	10	12	14	16	18	20	30
1–40	5.00	2.58	4.73	1.44	3.65	1.54	2.58	1.61	4.08	4.51	5.00	5.00
17–28	5.00	0.35	3.00	0.52	2.35	3.63	2.67	2.00	2.98	4.02	5.00	5.00
22–35	5.00	3.00	2.25	0.22	3.57	4.44	0.52	3.13	2.49	3.44	5.00	5.00
25–35	5.00	3.57	0.41	2.55	1.38	3.19	0.63	1.39	3.56	4.65	5.00	5.00
32–35	5.00	4.00	2.28	0.49	2.52	1.35	0.35	2.25	3.88	4.72	5.00	5.00

For all β-amyloid peptides studied, the concentration of fibrils was quantified through the Congo Red assay according to equation 1. From the results (Table 3) about 80% soluble fibrils were formed from all of the peptides within 0.5 h of incubation with nNOS. Moreover, it was interesting to note that up to 96 h the percentage rate of decrease in soluble fibrils, as detected by the Congo Red assay, was mirrored by the rate of formation of insoluble fibrils detected by the turbidity assay (Table 3). All of the various Aβ-peptides, in the absence of nNOS, showed no change in absorbance over the 96 h aggregation period, thus concluding that the peptide themselves did not form fibrils. With Aβ_1–40 there was an ~10–15% decrease in soluble fibril every 24 h up to 96 h while with Aβ_22–35 the amount of soluble fibrils remained at about 69% up to 48 h and then decreased by a further 18% up to 72 h with only 9.3% soluble fibrils remaining after 96 h. Aβ_17–28 showed a similar trend with 56% soluble fibrils remaining after 48 h followed by a surprising increase to 76.7% at 72 h and then 14% remaining at 96 h. With respect to Aβ_25–35 and Aβ_32–35 a steady rate in the decrease of soluble fibrils every 24 h up to 96 h was observed with the former leaving 28% and the latter 52.3% soluble fibrils after 96 h.

Table 3.  % Soluble and insoluble fibrils formed over time from incubation of amyloid peptides with nNOS.

Aβ	Time (h)
	0	0.5	24	48	72	96
% Soluble fibrils
1–40	nd	81.2	69.7	53.1	37.4	24.7
17–28	nd	83.8	70.1	56.2	76.7	14.2
22–35	nd	69.4	69.2	70.2	51.6	9.3
25–35	nd	81.7	70.0	56.3	36.0	27.9
32–35	nd	79.6	60.0	41.8	75.9	52.3
Aβ	Time (h)
	0	0.5	24	48	72	96
% Insoluble fibrils
1–40	0.00	28.6	39.1	61.4	90.0	100
17–28	0.00	29.7	40.2	60.7	89.9	100
22–35	0.00	40.0	60.0	79.4	90.2	100
25–35	0.00	19.8	39.2	60.0	94.1	100
32–35	0.00	8.4	39.9	55.7	92.7	100

The extent of aggregation of each peptide was determined with the formation of insoluble fibrils as determined by turbidity, at 400 nm, as a function of time (Table 3). All controls for the various Aβ peptides in the absence of nNOS showed little or no turbidity concluding that the amyloid peptides on their own did not produce insoluble fibrils correlating with the control results from the Congo Red assay. From the turbidimetric results, the various Aβ peptides indicated an amount of 8–40% of insoluble fibrils during the initial 0.5 h incubation which increased gradually to 100% by 96 h, and this correlated with soluble fibril decrease as obtained from the Congo Red results.

Fluorescence of Th-T is used extensively for the identification, quantification and rate of formation of amyloid fibriuls; free Th-T showed weak, negligible fluorescence in solution. It is not the intention in this present manuscript to discuss the mechanism of the formation of the amyloid fibrils as this had been covered in detail elsewhere^26–28. Nevertheless it was pertinent to mention that since there is stoichiometry and saturation binding between Th-T and the fibrils any increased change in fluorescence can be equated directly with their rate of formation. When each of the amyloid peptides [Aβ_1–40, Aβ_22–35, Aβ_17–28, Aβ_32–35 and Aβ_25–35] at increasing concentrations was incubated for 30 min with nNOS in the presence of Th-T there was a gradual change in fluorescence up to a saturation maximum (Figure 3). This supported our findings with the Congo Red and turbidity assays that nNOS was catalytic towards fibril formation. Furthermore all controls for the various Aβ peptides in the absence of nNOS showed little or no fluorescent change concluding that the amyloid peptides on their own did not undergo fibrillogenesis. It was also consistent with the possibility that amyloid peptides were converted into a form that could no longer bind to the enzyme.

Figure 3.  Change in fluorescence of Th-T as an indicator to the formation of amyloid fibrils catalysed by nNOS on amyloid peptide fragments.

All of the peptides produced fibrils at a rate of 5 nM.min⁻¹ (Figure 3). However nNOS induced both Aβ_17–28 and Aβ_25–35 to afford fibrils with a yield 8-fold greater than from Aβ_32–35. The other peptides [Aβ_1–40 and Aβ_22–35] gave fibrils with only three to 4-fold yield more than Aβ_32–35 (Figure 3). The incubation time of nNOS with amyloid peptides and Th-T was 30 min and substantiated the results obtained for Congo Red and turbidity as fibrils began to form rapidly within 0.5 h, irrespective of whether the fibril was soluble or insoluble. From the data obtained by both Th-T staining fluorescence and Congo Red assay we were confident that the aggregates formed were in fact amyloid.

Aβ peptides interact with hydrophobic environments^29,30 and so it was more than likely that nNOS-Aβ complexes were hydrophobic-hydrophobic associations. Examining the amino acid sequence of the Aβ-peptides that were readily induced to form fibrils pointed us towards two hydrophobic patches with sequences Leu₁₇ – Val₁₈ – Phe₁₉ – Phe₂₀ – Ala₂₁ and Gly₂₅ – Ser₂₆ – Asn₂₇ – Lys₂₈ – Gly₂₉ – Ala₃₀ – Ile₃₁ – Ile₃₂ – Gly₃₃. This latter sequence looked especially promising in view of the glycine zipper [G-X-X-X-G-X-X-X-G] recognition motif³¹. We were also very aware of a third hydrophobic peptide glycine zipper motif that existed with a sequence Gly₃₃ – Leu₃₄ – Met₃₅ – Val₃₆ – Gly₃₇. These characteristic glycine zipper motifs have been identified in many proteins as a transmembrane helix-turn-helix and, in the case of Aβ, were completely conserved suggesting that such motifs played an important role in their normal function that could have serious implications in the etiology and pathophysiology for Alzheimers disease^32,33. There was evidence^34–36 that fibrillogenesis of amyloid peptides may be catalysed by several biomacromolecules including enzymes and/or proteins though the actual specific mechanism remains elusive. By correlating our present kinetic analysis, Th-T fluorescence data and the Congo Red and turbidity assays it was tempting to suggest that the hydrophobic patches and/or glycine zippers between residues 17 and 35 mediated the process of fibrillogenesis of the amyloid peptides. Though we have no definitive proof whether the original peptide fragments were random coils or α-helices there was evidence of a rapid conversion of the peptide, initiated by the enzyme, into a soluble β-conformation rendering it no longer able to bind with the enzyme. Then after ~30 min there was a further change, also initiated by nNOS, and the formation of insoluble fibril aggregates.

The rapid fibril formation, detected by Th-T, could be explained via the nucleation effect which was accelerated by seeding. An ordered nucleus was formed only after a lag phase and in a ‘supersaturated’ solution of fibrils that had formed from ‘seeds’ (initially present) which eventually exceeded the critical concentration of amyloid peptide^37,38. This effect was enhanced by micelles that were in equilibrium with soluble Aβ monomers that provided nuclei for new fibrils allowing further formation of fibrils with time. Based on this model, it was suggested that with increased amount of fibrils, there were more micelles that remained in solution until a point of saturation. It had been suggested that these soluble amyloid oligomers, rather than the mature aggregates represented the primary toxic mechanism of amyloid pathogenesis³⁹. Once the aggregation process had begun and a critical nucleus had formed, the soluble fibrils, that had been detected by the Congo Red assay, now formed insoluble fibrils and precipitated out of solution (Table 3).

The structure of the insoluble amyloid fibrils generated from the proteolytic action of nNOS with Aβ_17–28 was examined under TEM to investigate aggregate morphology. They revealed extensive formation of fibrils compared to the enzyme (nNOS) and amyloid peptide before induction (Supplementary material, Figure 3A and B). Within the first 30 min, fibrils appeared mainly as short, rod-like structures (Supplementary material, Figure 3C) and after 24 h, fibrillization of amyloid peptides appeared to have progressed from these short discrete fibrils to a mass of much longer fibrils in a branched network (Supplementary material, Figure 3D). This supported the morphology pattern of fibrils as illustrated from literature^30–41. These observations also substantiated Congo red, turbidity and Th-T results which showed that fibrils, initially formed within 30 min, remained soluble and then after 24 h aggregated to form a mass of slender, long fibrils in a typical branched network.

Conclusions

All Aβ peptides inhibited nNOS with Aβ_17–28 inhibiting the most (K_i = 0.81 μM) within 2 min. Extended incubation of the peptides with nNOS suggested the presence of an association—dissociation equilibrium and helped to explain the onset of fibrillogenesis and the role of nNOS as an amyloidogenic catalyst.

The Congo Red assay proved to be successful in the quantification of soluble fibrils while the turbidity assay and Th-T were simple and reliable methods in identifying and quantifying the process of aggregation driven by the formation of insoluble fibrils. Most importantly the enzyme was inhibited when the Aβ peptide initially was bound to nNOS. Then after several minutes the peptide dissociated and was converted into a soluble fibril which remained attached to the enzyme but did not inhibit its activity. Eventually the nNOS-Fib(sol) complex reached a saturation point in solution resulting in a rapid dissociation of fibrils to account for the increase in soluble fibrils with time as revealed by the Congo Red assay. Subsequent decrease in soluble fibrils over 96 h was mirrored by an increase in concentration in insoluble fibril as detected by turbidity. The Th-T assay showed that the rate of formation of insoluble fibrils was 5 nM.min⁻¹ with Aβ_17–28 and Aβ_25–35 the most prominent peptides in fibrillogenesis confirming that nNOS was an amyloidogenic catalyst. It also suggested that amyloid fibril formation required proteolysis and that different amyloid peptides produced different quantities of fibrils over a given time. We are confident that the hydrophobic pentapeptide patches and/or glycine zipper motifs between residues 17 and 37 were responsible for both initial binding and proteolytic catalysis into fibrils.

Acknowledgements

The financial assistance from The Medical Research Council (South Africa) toward this research is hereby acknowledged.

Declaration of interest

The authors report no conflict of interest. The authors alone are responsible for the content and writing of the paper.