Alzheimer’s disease and cigarette smoke components: effects of nicotine, PAHs, and Cd(II), Cr(III), Pb(II), Pb(IV) ions on amyloid-β peptide aggregation

Cigarette smoking is a significant risk factor for Alzheimer’s disease (AD), which is associated with extracellular brain deposits of amyloid plaques containing aggregated amyloid-β (Aβ) peptides. Aβ aggregation occurs via multiple pathways that can be influenced by various compounds. Here, we used AFM imaging and NMR, fluorescence, and mass spectrometry to monitor in vitro how Aβ aggregation is affected by the cigarette-related compounds nicotine, polycyclic aromatic hydrocarbons (PAHs) with one to five aromatic rings, and the metal ions Cd(II), Cr(III), Pb(II), and Pb(IV). All PAHs and metal ions modulated the Aβ aggregation process. Cd(II), Cr(III), and Pb(II) ions displayed general electrostatic interactions with Aβ, whereas Pb(IV) ions showed specific transient binding coordination to the N-terminal Aβ segment. Thus, Pb(IV) ions are especially prone to interact with Aβ and affect its aggregation. While Pb(IV) ions affected mainly Aβ dimer and trimer formation, hydrophobic toluene mainly affected formation of larger aggregates such as tetramers. The uncharged and hydrophilic nicotine molecule showed no direct interactions with Aβ, nor did it affect Aβ aggregation. Our Aβ interaction results suggest a molecular rationale for the higher AD prevalence among smokers, and indicate that certain forms of lead in particular may constitute an environmental risk factor for AD.

increased fluorescence intensity when bound to amyloid aggregates 15 . In addition to amyloid plaques, AD brain tissue typically exhibits a second type of lesion in the form of intracellular neurofibrillary tangles consisting of aggregated hyperphosphorylated tau proteins. Exactly how Aβ or tau aggregation can induce the neuronal death associated with AD remains a point of contention. Different mechanisms for development of cell toxicity have been proposed that need not be mutually exclusive 16 . Although only around five percent of all AD cases are caused by inherited genetic conditions (i.e., the familial form), the higher incidence of AD among patients with aggregation-enhancing Aβ mutations provides an irrefutable link between AD and Aβ aggregation 17 .
In addition to nicotine, tobacco contains high levels of various metals. In fact, the tobacco plant is so effective at extracting metals from the ground that it is sometimes used for phytoremediation of metal-contaminated soil and groundwater 47,48 . The metal content in tobacco leaves is further increased by the use of metal-containing fertilizers, herbicides, pesticides, and insecticides such as lead arsenate when growing commercial tobacco, and by the metal boxes used for growing, drying, and curing tobacco 49 . With additional contributions from other sources such as brightening agents in rolling paper, cigarettes end up containing non-negligible concentrations of metals such as Al, As, Cd, Co, Cr, Cu, Hg, Mn, Ni, Pb, Se, Tl, V, and Zn 47,50-53 , many of which (i.e., Al, As, Cd, Cu, Cu, Hg, Mn, Ni, Pb, Tl and V) are considered neurotoxic 54 . Numerous studies link AD to the essential transition metals Cu, Fe, and Zn, as they are elevated in phosphorylated tau tangles 55 and in AD plaques compared to the surrounding nerve tissue (i.e., AD neuropil) [56][57][58] . Cu(II), Fe(II), and Zn(II) ions display specific binding to the Aβ peptide and modulate its aggregation pathways [59][60][61][62][63][64] , and Cu exposure increases Aβ levels in mice 65,66 . Cu(II) and Fe(III) ions furthermore generate harmful reactive oxygen species (ROS) when bound to the Aβ peptide 64 , and such ROS damage likely contributes to the neuroinflammatory condition associated with AD 67 . The previously suggested connections between AD and metals such as Al 68,69 and Hg 70 remain unclear but possibly valid. Although metal ion interactions and general metal dyshomeostasis appears to be an integral part of AD pathology 59,71 , studies on non-essential metals in AD at a molecular level are rare [72][73][74] .
Cigarette smoke contains thousands of organic compounds that are either present in the tobacco plant, added during cigarette manufacture, or produced by pyrolysis during smoking. The latter group includes polycyclic aromatic hydrocarbons (PAHs), which have been associated with a variety of adverse health effects in humans and wildlife 75,76 . Among the known harmful PAHs, toluene is a neurotoxicant used in glue 77 , benzo[a]pyrene (B[a]P) is both a neurotoxicant 78 and a potent carcinogen 79,80 , while naphthalene can cause hemolytic anemia and is the active substance in mothballs 81 . Several of the 16 PAHs identified by the US Environmental Protection Agency as priority pollutants are present in cigarette smoke, such as B[a]P, pyrene, and phenanthrene 82 , but possible relations between PAH exposure and AD remain largely unexplored 72 .
In this study, we monitored in vitro how a variety of substances in cigarette smoke affect Aβ aggregation and fibrillation, at a molecular level, using the Aβ(1-40) variant as a model peptide (from now on Aβ 40 ). Biophysical techniques including nuclear magnetic resonance (NMR) spectroscopy, atomic force microscopy (AFM) imaging, ThT fluorescence assays, and mass spectrometry (MS) were used to investigate interactions between Aβ and i) Figure 1. Simplified overview of the aggregation pathway for the amyloid-β peptide. Thioflavin T (ThT) fluorescence spectroscopy can be used to monitor the self-assembly of Aβ peptides from soluble monomers and oligomers into amyloid fibrils via various intermediate states such as protofibrils. The sigmoidal aggregation curve consists of a lag phase, or nucleation phase, and the elongation phase, or growth phase, before the saturation phase is reached. Monomeric and oligomeric conformational changes and nuclei formation processes occurs in the lag phase, followed by a rapid elongation process into partly insoluble fibrils. Aggregation modulation effects can be studied by monitoring the ThT fluorescence when Aβ is incubated with various compounds.
the alkaloid nicotine, ii) hydrocarbons with one to five aromatic rings (Fig. 2), i.e., toluene, naphthalene, phenanthrene, pyrene, and B[a]P, and iii) the metal ions Pb(II), Pb(IV), Cd(II) and Cr(III). Many of these compounds are associated with adverse biological effects on survival, growth, development, reproduction, metabolism, and tumor formation 75,[79][80][81] , and neurotoxic effects in particular have been associated with exposures to toluene 77 , B[a]P 78 , and several of the metals found in cigarette smoke 54,[83][84][85][86] . The results presented below show that some of these substances may also contribute to AD pathogenesis and progression, by modulating the Aβ fibril formation process or/and by inducing ROS damage.

Results
NMR spectroscopy. NMR experiments were conducted to investigate possible molecular interactions between the monomeric Aβ 40 peptide and the studied substances (Figs 3 and 4 and S1-S2). The finger-print region of a 1 H, 15 N-HSQC spectrum of 84 µM (550 μl) monomeric unstructured 15 N-labeled Aβ 40 peptide is shown in Fig. 3, before and after addition of 504 µM Pb(IV) acetate (Fig. 3A), 840 µM nicotine (Fig. 3B), or 20 µL neat toluene (Fig. 3C). All compounds were titrated to the Aβ solution in small steps, starting at sub-stoichiometric ratios, but only spectra for larger additions are shown where the interaction effects (or lack thereof) are most pronounced. Pb(IV) ions clearly induce loss of signal intensity in the NMR amide crosspeaks corresponding to certain N-terminal Aβ 40 residues, indicating specific binding to this part of the peptide (Fig. 3A). Similar effects were seen in the 1 H, 13 C-HSQC spectra of 13 C, 15 N-labeled Aβ 40 , where addition of Pb(IV) ions reduces the intensity of the aromatic and Cα-H crosspeaks for Y10 and the three histidines H6, H13, and H14 (Fig. 4). The loss of signal intensity is particularly strong for the Y10 crosspeaks (Fig. 4A), suggesting that Y10 or/and possibly E11 (Fig. 3A) is involved in Pb(IV) coordination. Although the present NMR observations cannot be directly interpreted in terms of binding geometry, the observed loss of NMR signal for specific residues is arguably caused by intermediate chemical exchange on the NMR time-scale between a free and a metal-bound state of the Aβ 40 peptide. The residues most affected by the Pb(IV) ions are likely the most specific and strongly binding ligands 62 .
Addition of nicotine does not selectively affect the intensity or position of any particular Aβ 1 H, 15 N-HSQC crosspeak, indicating that there are no specific interactions between nicotine and the Aβ monomer (Fig. 3B). Similarly, the metal ions Pb(II), Cd(II), and Cr(III) do not induce any specific changes in the Aβ 1 H, 15 N-HSQC spectrum, indicating they have no strong specific binding to monomeric Aβ ( Supplementary Fig. S1). Additions of naphthalene, phenanthrene, pyrene, and B[a]P induced small changes in crosspeak intensities and chemical shifts (Fig. S2). All these hydrocarbons were however dissolved in DMSO, and the small crosspeak effects observed are virtually identical to the spectral effects induced by small amounts of pure DMSO additions onto Aβ ( Supplementary Fig. S1). Toluene, which was added neat, induced minor non-specific chemical shift changes in the crosspeak positions (Fig. 3C). We conclude that none of the studied hydrocarbons display any specific molecular interactions with monomeric Aβ.
ThT fluorescence kinetics. Figures 5 and S3 show ThT fluorescence intensity curves for Aβ 40 aggregation kinetics that monitor the formation of amyloid in the presence of the studied substances. These kinetic curves have a general sigmoidal shape, and the kinetic parameters τ ½ and r max obtained from curve-fitting to Eq. 1 (in Materials and Methods) are shown in Table 1. The hydrocarbons toluene and naphthalene have no significant effect on the aggregation kinetics, while the larger phenanthrene, pyrene, and B[a]P molecules all increase the Aβ aggregation rate. The metal ions Cr(III) and Pb(II) significantly slow down the Aβ aggregation kinetics, which however is promoted in the presence of Pb(IV) ions. Cd(II) ions as well as nicotine appear to induce slightly slower aggregation, but the differences are too small to be significant ( Table 1). The Aβ aggregation in the presence of both Cr(III) ions and naphthalene, or both Cr(III) ions and phenanthrene, is faster than with Cr(III) ions alone, yet slower than with naphthalene or phenanthrene alone (Table 1). Thus, the aggregation-promoting effects of the hydrocarbons -especially phenanthrene -seem to counteract the aggregation-retarding effect of the Cr(III) ions. The ThT fluorescence intensity at the endpoint plateau phase arguably corresponds to the amount of ThT-active (amyloid) aggregates being present. Following this assumption, it seems that the organic compounds have no significant effect on the amount of amyloid formed, while Cd(II), Cr(III), and Pb(IV) ions appear to induce a lower amount of ThT-active amyloid material at the end of the reaction (Table 1) Mass spectrometry. Using protocols for soft ionization developed during the last decade, the recorded MS spectra provide information on non-covalent aggregated states of the Aβ 40 peptide (i.e., from monomers up to dodecamers under favourable conditions) [87][88][89] . Under the experimental conditions used, we observe that toluene and Pb(IV) ions induce clear effects on the Aβ oligomer distribution up to tetramers (Figs 7 and S4). The Aβ sample freshly prepared in ammonium acetate solution is seen to be in equilibrium between the dominant monomeric form and smaller fractions of soluble dimers, trimers, and tetramers. The sample prepared in presence of toluene (1:1 Aβ:toluene ratio) shows a lower relative amount of tetramers (Fig. 7). The sample prepared in the presence of Pb(IV) ions (1:1 Aβ:Pb ratio) shows decreased relative amounts of dimers and trimers, but not tetramers. The total MS signal was lower for the sample containing Pb(IV) ions, which could indicate that addition of Pb(IV) ions immediately shifts the equilibrium towards higher molecular weight species. The presence of free metal ions in the sample should however also result in lower ionization efficiency for the Aβ peptide, which would reduce the signal intensity.
We furthermore observed that the presence of Pb(IV), a known oxidative agent 90,91 , induced significant oxidation of the M35 residue: 6% of the Aβ 40 monomers were oxidized in the Pb(IV) sample compared to 2% in the control and toluene samples ( Supplementary Fig. S4).

Discussion
Nicotine. Nicotine is the most abundant alkaloid in tobacco leaves and one of the most addictive substances known 92 . At high doses nicotine is toxic and even lethal 93 . Smoking a cigarette yields about 1-2 mg of absorbed nicotine, which readily is transported to the brain where it acts as an agonist on nicotinic acetylcholine receptors in the central nervous system 94 . A few hours after exposure, the absorbed nicotine is metabolized into various forms of cotinine 94 .
Our current results show that even at a 10:1 ratio, nicotine does not interact with the Aβ 40 monomer; does not affect Aβ aggregation; and does not alter Aβ fibril morphology. We conclude that nicotine has no significant effect on Aβ and its aggregation pathway, which is in line with certain previous observations 95,96 . As AD is considered to be strongly related to amyloid aggregation 12,97,98 , a substance may affect an individual's Aβ amyloid burden in any of three principal ways: by modulating the Aβ aggregation process, by binding to Aβ aggregates and thereby altering their biological effects, or by affecting Aβ production, degradation, or/and localization. For nicotine, our current results rule out the first two mechanisms, but effects on AD via the third mechanism remains a possibility that should be further explored. In addition, nicotine has been suggested to affect tau phosphorylation 99 , and to attenuate Aβ neurotoxicity by regulating metal homeostasis 100 .  a family of metal ions displaying specific and relatively strong binding interactions with the Aβ peptide. Addition of Pb(IV) ions to Aβ 40 strongly affects the NMR signals of H6, Y10, E11, H13, and H14, indicating these residues as likely binding ligands. Previous studies on Aβ interactions with e.g. Cu(II) and Zn(II) ions indicate that multiple binding conformations likely co-exist 59 , and that the Aβ binding properties change at lower pH when the histidines become protonated 59,61,63 . Here, Y10 is strongly affected by Pb(IV), suggesting it is a major binding ligand (Figs 3A and 4). Pb(IV) ions are therefore likely to have a different binding coordination to Aβ than e.g. Cu(II) and Zn(II) ions, which are coordinated mainly by the three N-terminal histidines (H6, H13, and H14) together with the D1 residue 103 . This finding shows that Aβ metal-binding may be more complex and varied than previously thought, and should be further investigated.
While normal Aβ fibrils form in the presence of Cr(III) ions, our AFM images show formation of amorphous Aβ aggregates in the presence of Cd(II), Pb(II), and Pb(IV) ions. Thus, even though their binding is non-specific and weak, Cd(II) and Pb(II) ions are capable of altering the Aβ aggregation pathway in vitro. This is in line with previous work showing that Ca(II) ions promote Aβ fibrillation, even though Ca(II) ions show no specific interaction with the Aβ monomer 104 . However, in vivo mainly metal ions with relatively strong and specific binding to Aβ, and which are present in the brain in reasonable concentrations, are believed to modulate the Aβ aggregation process. Typical examples are Cu(II) and Zn(II) ions, which both have strong affinities for the Aβ monomer, and which both are released in high local concentrations from neuronal synapses 105 . As Cd(II), Cr(III), and Pb(II) ions display weak binding to Aβ and are present only as trace contaminants in human fluids, it appears unlikely that these metal ions would have a strong effect on Aβ aggregation in vivo, at least on their own. Pb(IV) ions, on the other hand, display a stronger and specific binding to monomeric Aβ 40 , and the oxidation of Aβ residue M35 by Pb(IV) ions show that these ions can act as oxidizing agents and thus produce harmful ROS when bound to the peptide. Although Cu, Fe, and Zn are the main metals found in the amyloid brain plaques in AD patients 21,[56][57][58][59]106 , lower concentrations of other metals including Pb have also been observed 58 . These lower Pb levels may not be surprising, as about 95% of the Pb that enters the adult body accumulates in the skeleton where Pb(II) replaces Ca(II) in the bone apatite 107 . Yet, the minor fraction of Pb that enters the brain appears to have biological impact, as Pb exposure has been correlated with a variety of adverse effects on neuronal formation, neurotransmission, and cognitive function 4,86,108 . Given our NMR results, it can be speculated that the Pb previously observed in AD brain plaques c Pb(IV). Future research will hopefully be able to shed more light on the oxidation states of the Pb that is distributed in different body tissues and fluids. It has previously been suggested that the Cu and Fe ions bound to Aβ plaques may generate damaging ROS via Fenton chemistry 59,64,109 , which would contribute to the neuroinflammation observed in AD patients 67 . In such scenarios the less reactive Zn(II) ion may protect nerve cells from radical damage by competing away harmful Cu and Fe ions from the Aβ metal binding sites 59 . As generation of oxygen radicals is one of the main mechanisms of Pb toxicity 90,91 , it appears very likely that the Pb bound to Aβ plaques will produce harmful ROS in the areas around these plaques. Given the local accumulation of Pb in these plaques, the resulting ROS damage likely affects AD pathology more than the aggregation-modulating properties of Pb(IV) ions, as we are not aware of any evidence for co-localization of elevated Pb and Aβ aggregation pathways. Nevertheless, together with previous observations that Pb exposure induces elevated Aβ levels in rats 110 , increased Aβ plaque formation in monkeys 111 , and enhanced tau production and phosphorylation in both mice and monkeys 112,113 , our current results that Pb(IV) ions display specific binding to the Aβ peptide, alter its aggregation process, and cause oxidative effects in the presence of Aβ, clearly show that Pb ions can modulate the Aβ amyloid cascade events that are associated with AD. Hydrocarbons. Cigarette smoke contains a wide range of aromatic hydrocarbons that deposit and accumulate in the lungs as cigarette tar. Some of this tar is absorbed by the body, and the hydrophobic hydrocarbons can permeate the lipophilic blood-brain barrier membrane, allowing transport into the brain 114 . Within a few days, absorbed aromatic hydrocarbons are typically metabolized in multistep reactions into e.g. epoxides and polar hydroxyl-derivatives, leading to end products that the body readily can excrete 115 . Our ThT fluorescence results show that addition of phenanthrene, pyrene, or B[a]P increase the Aβ 40 aggregation rate, which remains largely unaltered when toluene or naphthalene is added (Fig. 5; Table 1). The AFM  images show that toluene, naphthalene, and B[a]P induce formation of amorphous Aβ aggregates, while relatively unaltered fibrils are formed in the presence of phenanthrene and pyrene (Fig. 6). Thus, there is no clear correlation between the aggregation kinetics and the aggregation products formed. Instead, the kinetic monitoring and the AFM images provide complementary information. For example, even though pyrene promotes and Cr(III) ions retard Aβ aggregation ( Fig. 5; Table 1), proper fibrils are formed in the presence of both compounds (Fig. 6). Cd(II) ions also retard Aβ aggregation, but here amorphous aggregates are formed instead of fibrils (Figs 5-6; Table 1).
The NMR data reveal that there is no strong interaction between the monomeric Aβ 40 peptide and any of the studied hydrocarbons, suggesting that hydrocarbons are unlikely to initiate (seed) Aβ aggregation. Taken together, these results can be explained in terms of hydrophobicity. The amphiphilic Aβ monomers are not attractive binding partners for the hydrophobic hydrocarbons. Aβ aggregation is however driven by hydrophobic interactions, where the oligomers formed are considered to be micelle-like entities with a hydrophobic core 11 . Such a core would readily attract aromatic molecules, arguably leading to formation of micelle-like Aβ-PAH co-aggregates (depending on the concentrations and Aβ/PAH ratios involved), thereby inducing deviations from the fibril-forming aggregation pathway. Our mass spectrometry data support such a scenario, as hydrophobic toluene was found to affect larger Aβ tetramers but not the smaller dimers or trimers (Fig. 7). The Pb(IV) ions, on the other hand, affected the dimer and trimer populations more than the tetramers (Fig. 7). This suggests that electrostatic interactions may be more important for the first steps of Aβ aggregation, and hydrophobic effects more important for larger oligomer formation. Such a scenario is consistent with nicotine not affecting Aβ aggregation at all, as nicotine is a hydrophilic weak organic base that is only slightly positively charged at neutral pH (its pyrrolidino N has a pKa around 8 116 ). Elucidating the forces governing different stages of Aβ aggregation is important not only for understanding the amyloid formation process as such 117 , but may also help in designing interacting molecules (drugs) that can selectively target certain (toxic) aggregation states 118 . Thus, this difference in the interaction of Aβ with metal ions and PAHs respectively should be further explored, together with the question how promoting or retarding Aβ aggregation -into amyloid fibrils or amorphous material -affects AD progression. Future research on the molecular details of the likely toxic Aβ oligomers, where the structures remain poorly understood 10,11 , may also be able to explain why the more hydrophobic three-, four-, and five-ring molecules affect the Aβ aggregation kinetics more than the one-and two-ring compounds (Fig. 5, Table 1).

Combined effects.
Toxic or biological effects of chemical substances are often investigated one substance at a time, even though real-life exposure scenarios typically involve exposure to multiple chemicals. The thousands of different compounds in cigarette smoke is a case in point. For PAHs it is well known that synergistic effects can make them far more toxic in combination than alone 119 . For AD, a recent study showed that rats exposed to a mixture of Pb, Cd, and As produced greater increases in Aβ levels and cognitive impairment related to oxidative stress and inflammation than the sum of the individual metals 120 . This could be due to synergistic effects, or due to overloaded protective mechanisms. Our current measurements indicate that the retarding effects of Cr(III) ions and the promoting effects of naphthalene/phenanthrene on Aβ peptide aggregation kinetics counteract each other when both substances are present (Fig. 5C). Thus, given the multitude of compounds present in cigarette smoke, the overall effect of cigarette smoking on Aβ aggregation and AD pathology will be difficult to elucidate at a molecular level, at least until precise neurotoxic mechanisms underlying AD have been firmly established.
Due to the requirements of the analytical techniques used, the in vitro experiments were carried out with somewhat higher reagent concentrations than typically found in the human body. The Aβ 40 concentrations in our experiments were in the range 10-100 uM, while the Aβ 40 concentration in cerebrospinal fluid (CSF) is in the range 14-23 ng/L (3-5 pM) for AD patients and 10-18 ng/L (2-4 pM) for healthy controls 121 . Aggregation of Aβ in vivo is however likely to take place in e.g. membrane environments, where the local Aβ concentration is higher. The cigarette smoke compounds were also investigated at higher concentrations than what is found in vivo, although for the studied interactions the Aβ:compound ratio is likely more important than the actual compound concentration. The superstoichiometric ratios required for the studied compounds to have significant effects indicate that they are not likely to be major in vivo modulators of Aβ aggregation on their own, but could be important factors in combination with other compounds. As concentrations of e.g. Cu, Fe, Pb, and Zn are elevated in amyloid brain plaques, the combined ROS generated by Cu, Fe, and Pb ions arguably contribute to the neuronal damage observed in AD, and as metal dyshomeostasis appears to be involved in AD pathology 59,71 , future research should elucidate how different combinations of metal ions affect the Aβ amyloid cascade events.

Conclusions.
Smoking is an established risk factor for Alzheimer's disease and other neurodegenerative disorders. Here, five aromatic hydrocarbons and four metal ions present in cigarette smoke were found to affect the Aβ 40 peptide aggregation process. Metal ions such as Pb(IV) appear to mainly affect formation of Aβ dimers and trimers, while hydrocarbons such as toluene appear to mainly affect larger oligomeric and hydrophobic forms such as tetramers. Some metal ions and hydrocarbons counteract each other's overall effects. The uncharged and hydrophilic nicotine molecule has no direct effect on Aβ or its aggregation process. As Pb(IV) ions interacting with Aβ were found to act as oxidizing agents -likely harmful -the specific binding observed between Aβ and Pb(IV) ions warrant further investigation, particularly given that significant sources of Pb exposure remain a major problem worldwide and especially in developing countries 86 .

Materials and Methods
Sample preparation. Unlabeled or uniformly 15 N-or 13 C, 15 N-labeled Aβ(1-40) peptides were bought lyophilized from AlexoTech AB (Umeå, Sweden). The peptide samples were freshly dissolved and prepared before the measurements, according to previously published protocols 61 . Liquid (-)-nicotine (Sigma-Aldrich) and acetate salts of Cd(II), Cr(III), Pb(II), and Pb(IV) (Sigma-Aldrich) were dissolved or diluted in 20 mM phosphate buffer. Naphthalene, phenanthrene, pyrene, and B[a]P were first dissolved in 50% DMSO and then diluted in 20 mM phosphate buffer. NaOH/HCl was used to adjust the pH of all stock solutions to 7.35. NMR spectroscopy. 2D heteronuclear single quantum coherence ( 1 H, 15 N-HSQC and 1 H, 13 C-HSQC) NMR experiments were performed on Bruker Avance 500 MHz and 700 MHz spectrometers equipped with cryoprobes. Spectra of monomeric 84 μM Aβ 40 peptides uniformly labeled with 13 C or/and 15 N isotopes were recorded at + 5 °C in 20 mM sodium phosphate buffer pH 7.35 with 10% D 2 O. The studied cigarette smoke substances were titrated to the Aβ sample in small steps to final Aβ:substance molar ratios above 1:10. The Aβ 40 HSQC crosspeak assignment is known from previous work 122 . All data was processed with the Topspin version 3.2 software and referenced to the 1 H signal of TSP. AFM imaging. Solid state AFM images were recorded using a Bruker's Scan Asyst (Bruker Corp., USA) unit operating in peak-force mode or tapping mode with a resolution of either 256 × 256 or 1024 × 1024 pixels. Solutions of 100 μM Aβ 40 peptide in 20 mM sodium phosphate buffer at pH 7.35 were incubated at +37 °C in Eppendorf tubes at 200 rpm for 6 hours, either in absence (control) or in presence of one or two of the cigarette smoke substances (1000 μM additions). The incubated samples were then diluted and applied on freshly cleaved mica substrates. After 20 minutes the mica substrates were three times washed with distilled water and left to air-dry.
ThT fluorescence. A 96-well FLUOstar Omega plate reader (BMG LABTECH, Germany) was used to record fluorescence spectra (excitation 440 nm; emission 480 nm) every three minutes for samples containing 10-20 μM Aβ 40 peptides, 40 μM ThT dye, and 100-200 μM cigarette smoke substances together in 20 mM sodium phosphate buffer at pH 7.35. The measurements were running real-time for up to 48 hrs at +37 °C under quiescent conditions. The Aβ aggregation kinetic parameters τ ½ and r max were calculated by fitting five or six replicates per condition to a sigmoidal curve according to Eq.(1) 123 : where F 0 is the fluorescence intensity baseline, A is the fluorescence amplitude, r max is the maximum growth rate, and τ ½ is the time when half the monomer population is depleted.
Mass spectrometry. Mass spectra of 20 μM Aβ 40 peptide dissolved in 20 mM ammonium acetate buffer, pH 7.4, with and without addition of toluene and Pb(IV) acetate at 1:1 ratios were recorded three times each on a Synapt G2-Si high definition mass spectrometer (Waters corporation) equipped with a conventional ESI source operating in positive ion mode. Flow rate was 20 μl/min, capillary voltage 2.5 kV, cone voltage 40 V. Analysis was done in high-resolution mode (average resolution of 30 000) in the 500-4000 m/z range. Data processing was done using the Proteowizard 124 , UniDec 125 , and mMass 126 softwares. Peaks were identified by comparison between raw experimental data and generated theoretical peak lists, and by analysis of isotopic patterns (Supp. Fig. S4). All data were normalized to the +4 charge state signal of the Aβ monomer, to account for small deviations in concentration and ionization efficacy across samples. Data availability. All data generated or analysed during this study are included in this published article (and its Supplementary Information file).