Xenobiotic compounds modulate cytotoxicity of Aβ amyloids and interact with neuroprotective chaperone L-PGDS

A positive association of the exposure to different classes of xenobiotics such as commonly prescribed drugs and polycyclic aromatic hydrocarbons (PAH) typically those found in air pollution-related particulate matter with Alzheimer’s disease (AD) may point to direct physical interaction of those compounds with the amyloid formation and clearance processes. In this study, for the first time, we provide evidence of such interactions for three representative compounds from prescription drugs and air pollution, e.g. anticholinergic drugs Chlorpheniramine, a common antihistamine, and Trazodone, an antidepressant as well as 9,10-PQ, a common PAH anthraquinone abundantly present in diesel exhaust and associated with AD. We demonstrate that these three compounds bind to the lipophilic compound carrier and neuroprotective amyloid beta (Aβ) chaperone lipocalin-type prostaglandin D synthase (L-PGDS) with high affinity attenuating its neuroprotective chaperone function with Chlorpheniramine exhibiting markedly stronger inhibitory effects. We also show that these compounds directly interact with Aβ(1-40) increasing the fibril’s yield with altered fibril morphology and increased the cytotoxicity of the resulting fibrils. We propose that exposure to some xenobiotics in the peripheral tissues such as gut and lungs might result in the accumulation of these compounds in the brain facilitated by the carrier function of L-PGDS. This might lead to attenuation of its neuroprotective function and direct modification of Aβ amyloid morphology and cytotoxicity. This hypothesis might provide a mechanistic link between exposure to xenobiotic compounds and the increased risk of Alzheimer’s disease. Graphical Abstract

proteins typically associated with plaques [11]. L-PGDS is the second most abundant protein in human CSF (after albumin) with approximate concentrations of 26 mg/l [13] [14]. It is also abundantly expressed in peripheral tissues [15]. This protein exhibits a strong capacity to bind to a variety of lipophilic ligands [16] and was used as a drug delivery vehicle from gut to brain [17]. The 3D structure of human L-PGDS exhibits a single eight-stranded β-barrel with a deep calyx capable of simultaneous binding of several lipophilic or amphiphilic ligands [10] and Aβ peptides with nanomolar affinity [12] [11]. This function of L-PGDS might facilitate the accumulation of some xenobiotics in the brain followed by peripheral exposure in gut, nasal mucosa and lungs potentially interfering with the homeostasis of Aβ peptides associated with AD.
Association studies of prolonged exposures to some active xenobiotics available such as prescription drugs [18,19] [20] [21] or inhaled aerosol particulate matter [22][23][24][25] have shown that some compounds might significantly increase the risk of AD-associated dementia. However, the molecular mechanisms of these associations are not known. Given the fact that L-PGDS may contribute to controlling of Aβ aggregation [12] [11] and its ability to bind and transport a variety of drug-like small molecular ligands [26] [17], we hypothesize that these compounds might interfere with the Aβ homeostasis via their interaction with L-PGDS. Furthermore, some xenobiotics transported by L-PGDS may be capable of directly interacting with the Aβ peptides and modifying their morphology and cytotoxicity [27]. The morphology of amyloid fibrils found in different AD patients is evidently different from each other [28].
This morphological difference among patients may correlate with the origin and course of the disease and cytotoxicity of the amyloids [28] [29]. We propose that these xenobiotic compounds with high affinity to L-PGDS might be transported to Aβ aggregation sites and remodel the structure of the fibrils formed in the presence of these compounds. The altered fibrils might result in higher cytotoxicity towards neurons thus exacerbating the risk of AD.
In our study, we have selected two prescription drugs from the anticholinergic family, e.g.
Chlorpheniramine maleate (CPM) (Fig. S1B) is a common over-the-counter, first-generation antihistamine drug. CPM tablets are used clinically for the treatment of allergic diseases and symptoms of a cold [30].
Trazodone (TRD) (Fig. S1C) is among the top 5 most prescribed antidepressant drugs in the United States (US) [31]. Even though TRD was approved by the FDA as an antidepressant drug, it is one of the most widely prescribed off-label sleep aids in the US [32]. We have also selected 9,10-PQ (Fig. S1D), a common molecule found in the PM2.5 category of air pollutants and one of the main components of diesel exhaust. 9,10-PQ is a type of anthraquinone that can undergo redox cycling to produce reactive oxygen species (ROS) [33]. We report the effects of these xenobiotic compounds on the neuroprotective function of L-PGDS and the direct interaction between these compounds and Aβ peptides resulting in altered amyloid morphology and increased cytotoxicity. In this study, tryptophan fluorescence quenching assays and Isothermal Titration Calorimetry (ITC) are carried out to characterize binding of the selected compounds to human L-PGDS. Thioflavin T (ThT) assays are performed to demonstrate inhibition of nucleation and fibril disaggregase functions of L-PGDS. Binding sites of the ligands to L-PGDS are mapped using NMR titrations. The potential alteration of the fibril morphology and biological activity induced by the compounds are studied by using ThT assays, Transmission Electron Microscopy (TEM) and MTT cellbased assays.

Expression and purification of wild type, human unlabeled L-PGDS
Glycerol stock of Rosetta 2 DE3, E.coli cells (Novagen) with pNIC-CH vectors was prepared via transformation. The rest of the protocol was followed from the paper [10]. The concentrated protein was injected into the AKTA purifier Fast Performance Liquid Chromatography (FPLC) (GE Healthcare, USA) and further purified using Superdex 75 column in 50 mM sodium phosphate buffer (pH 7.0). Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) was run to check the purity of the different fractions obtained after the FPLC run.

Expression and purification of wild type, human 15 N labeled L-PGDS
The glycerol stock of the E.coli cells used for the 15 N labeled L-PGDS was the same as used for unlabeled L-PGDS. Starter culture was prepared in TB broth with kanamycin (50mg/ml) and chloramphenicol (25mg/ml) at 1:1000 dilutions and inoculated with the glycerol stock containing the transformed cells. Expression, extraction and purification of the 15 N labeled L-PGDS was done in the same manner as the unlabeled L-PGDS.

ThT inhibition assay
ThT dye (Sigma-Aldrich, Co) was dissolved in Mili Q water and filtered through a 0. transferred from previously assigned spectra [10] and chemical shifts of the cross peaks were analyzed using Computer aided resonance assignment (CARA) (www.nmr.ch) [43]. The chemical shift perturbation of the affected residues was calculated using the formula: And the resulting was mapped onto the previously published L-PGDS crystal structure (PDB code: 4imn) to characterize binding site of the drug -CPM onto the human WT L-PGDS.

Transmission Electron Microscopy (TEM)
To visualize the direct interaction of CPM, TRD and 9, 10-PQ on the Aβ(1-40) fibrils formation, Aβ  monomer was incubated at 37°C for 72 h in 50 mM sodium phosphate buffer (pH 7.0). Aβ  fibrils samples treated with and without compounds were applied on copper-rhodium 400 mesh grids with 15 nm carbon coating (thickness) (prepared in-house) followed by negative staining with 2% uranyl acetate and then air dried. The samples were then viewed under FEI T12. 120 kV Transmission electron microscope equipped with a 4K CCD camera (FEI) between 48000X to 68000x magnification under low dose conditions.

MTT Metabolic Assay
The toxicity effects of all the amyloid fibrils were tested on SH-SY5Y human neuroblastoma cell line using in each well containing 100 µL culture media. Before the fibrils were added to the cells, they underwent a washing protocol to remove excess/unbound compounds as described previously [44]. The samples were added into the cell assay and incubated for 20 hours to test their toxic effects. After incubation, 10 µL of the stock MTT reagent (5mg/ml) was added to each well. After 4 hours of conversion into formazan product, DMSO was added to dissolve the purple crystals left in the dark for 1 hour before measurement of absorbance at 570 nm with 630 nm as a subtracted reference wavelength.

Interactions of 9,10-PQ, TRD and CPM with L-PGDS
In order to assess binding affinity of the xenobiotic compounds to L-PGDS, we carried out tryptophan quenching and calorimetric analysis on 9,10-PQ/L-PGDS, TRD/L-PGDS and CPM/L-PGDS complexes. There are three tryptophan residues in human L-PGDS: Trp43 is located at the bottom of the L-PGDS cavity while Trp54 and Trp112 are located on the H2 helix and EF loop respectively (Fig. S1) [34].
Concentration-dependent quenching of the intrinsic tryptophan fluorescence of L-PGDS was monitored as a function of the concentration of 9,10-PQ, TRD and CPM (Fig. 1A, 1B and 1C) and the extracted binding parameters are summarized in Table 1. Due to insufficient quenching of intrinsic fluorescence (Fig. 1C), binding stoichiometry of CPM to L-PGDS was not estimated from the quenching data. Even though it is unclear as to why we observe a great difference in the tryptophan quenching efficiency of the three compounds -9,10-PQ, TRD and CPM, we suspect different binding sites in L-PGDS resulting in variable proximity of the ligand and the tryptophan residues in the complex [34].
To further confirm binding of the three compounds to L-PGDS, thermodynamic analysis between L-PGDS and 9,10-PQ/TRD/CPM at pH 7.0 and 25°C were examined using ITC. From the negative peaks of the titration curves obtained from ITC, binding of L-PGDS to all compounds was found to be an exothermic reaction, as indicated by the favorable enthalpy changes (Fig. 1D, 1E and 1F upper panel). with Kd obtained from the tryptophan quenching data (Table 1).

CPM inhibits the chaperone and disaggregase function of L-PGDS by occupying Aβ(1-40) peptide binding site
We used the Thioflavin T (ThT) fluorescence assay to monitor spontaneous aggregation of Aβ  in the presence and absence of L-PGDS in complex with the selected compounds. Aβ(1-40) aggregation exhibits a characteristic sigmoidal curve indicative of amyloid formation via primary nucleation, fibril elongation and secondary nucleation [11] (Fig. 2A). The elongation phase of the control 50 μM Aβ  peptide starts at ca 10 h after the lag phase reaching the steady phase at ca 30 h which is typical for fibril formation [35]. The presence of L-PGDS at 5 μM extends the elongation phase and decreases the final amount of Aβ(1-40) peptide aggregates to 40% when compared to the control ( Fig. 2A) using HADDOCK, an online protein-ligand docking platform [36]. Our docking model revealed that CPM molecules bind near the β-calyx entrance of L-PGDS as shown in Fig. 3C. We discovered that the entrance of the calyx of L-PGDS can accommodate up to three CPM molecules, thus supporting the stoichiometric binding ratio estimated from our ITC measurements.

Xenobiotic compounds alter the Aβ(1-40) fibril morphology and increase its cytotoxicity
To further investigate the direct interaction between the selected compounds and the Aβ(1-40) we performed the ThT assay (Fig. 4A). At 1:1 stoichiometry, all three compounds induce faster fibril formation with the overall higher fibrillar content as compared to Aβ(1-40) control (Fig. 4A). The elongation phase of Aβ(1-40) fibrils start at ~20 h, ~25 h and ~15 h when incubated with 9,10-PQ, TRD and CPM respectively, as compared to ~30 h for Aβ(1-40) control. TRD, CPM and 9,10-PQ shorten the lag phase while significantly increasing the total fibril content (Fig. 4A). The increase in the fibril content was also visualized in terms of increased A fibril numbers as observed by TEM (Fig. 4B-E).
Upon further increase in the concentration of CPM, TRD and 9,10-PQ to 1:5 (Aβ: compound), changes in the morphology of the Aβ(1-40) peptide fibrils were observed. The electron micrographs obtained from the TEM were analyzed using Image J software to quantify the number and size of both modulated and control fibrils of Aβ(1-40) [11]. For CPM, in addition to the increase in the total fibril content ( Fig. 5B and 5F), the fibrils also became shorter and thicker in terms of width ( Fig. 5B and 5E) when compared to the control fibrils. For 9,10-PQ, the total fibrillary content increases significantly ( Fig.   5C and 5F) with increasing concentration of the compound along with a significant increase in the fibril length ( Fig. 5C and 5D) resembling a 'ribbon-like' morphology (Fig. 5C). The fibrils grown in the presence of TRD do not show visible differences in number and morphology when compared to the control (Fig.   5D). Since the modified fibrils have different morphology compared to the Aβ(1-40) control, we speculate that the difference in morphology may show different biological activities of the resulting AB fibrils as it was demonstrated in the subsequent cell viability assays.
In addition to changes in morphology, the altered Aβ

Discussion
L-PGDS exhibits multiple functions in the organism. It can act as a PGD2-synthesising enzyme and an extracellular transporter for lipophilic compounds [16]. Due to its ability to bind and carry lipophilic compounds, it was proposed as an efficient drug delivery vehicle to the brain dramatically increasing in-situ drugs' availability [17]. Exposure of the peripheral tissues such as gut, nasal mucosa and lungs to the lipophilic xenobiotics with potentially strong biological activities might result in binding of these compounds to the abundant lipophilic carriers with subsequent transfer and accumulation of those compounds to the remote tissues. L-PGDS is standing out from the other lipophilic carriers by its abundance and unique ability to release its cargo into the blood [17]. For example, studies show that some air-pollution related compounds can accumulate in the brain at a concentration of 0.1-30 M [33]. The accumulation of the biologically active xenobiotics at these elevated concentrations might interfere with a variety of processes including the neuroprotective function of L-PGDS itself and direct interference with the amyloidogenesis in the context of AD [27]. In this study, we selected three representative compounds and explored their abilities to bind to L-PGDS, interfere with its amyloid nucleation and propagation inhibition as well as its amyloid disaggregase functions. Also, we show that at elevated concentrations, which are still within the physiological range found in the brain tissues, these compounds are capable of altering the morphology of the resulting A fibrils typically increasing their resistance to disaggregation and cytotoxicity.
Recently, it has been reported that the morphology of amyloid fibrils found in different AD patients are evidently different from each other [28]. It has been suggested that the morphological difference among patients correlate with the difference in disease progression [28] [29]. However, the exact reason for the different fibril morphology observed remains unknown. Here, from our study, we suggest that the modulation of fibril morphology by xenobiotic compounds could be one of the possible causes of the morphology differences observed among different patients. Given our constant exposure to these xenobiotic compounds in terms of air pollutants and common drugs, it seems reasonable to suggest that the Aβ fibrils found in AD patients can be remodeled by these compounds. Hence, exposure to these compounds might result in the different morphology of fibrils observed among different AD patients [28]. The modulated fibrils have different fibrillary morphologies and are more cytotoxic than the unmodified fibrils, indicating that certain amyloid morphologies are more toxic than others. We believe that future therapeutic interventions must take into consideration this possible morphological alteration in amyloid fibrils potentially induced by drugs. Reported here is our pilot study pointing to the direct mechanistic link between exposure to xenobiotics and change in the homeostasis of Ab peptides typically associated with AD.
We addressed the lipophilic carrier function of L-PGDS by investigating direct binding of the selected compounds to the protein. We selected compounds which have been associated with AD and belong to commonly prescribed drugs group and air-pollution-related compounds. The intrinsic tryptophan fluorescence quenching data and the ITC data establish binding of the three compounds 9,10-PQ, TRD and CPM to human L-PGDS with binding affinities of Kd ~ 10 M and ~5 M respectively (Table 1). Interestingly, we observed that CPM is the only compound that interferes with the interaction between the Aβ(1-40) peptide and L-PGDS ( Fig. 2A and S2) most likely inhibiting binding of Aβ  monomers to L-PGDS. While interacting with A (1-40), the chemical shifts of residues D37 and R144 of L-PGDS were significantly shifted and chemical shifts of A49 and G140 were completely attenuated [11].
The MD simulation model obtained from the study showed that Aβ(1-40) peptide interacts with L-PGDS at the entrance of the L-PGDS calyx (Fig. S3). In this study, WT L-PGDS in complex with CPM showed large chemical shifts for similar residues (Fig. 3A)  for Aβ fibrils [38]. As Aβ fibrils grow via the addition of monomers at fibrils' end [39], it is likely that these compounds interact with the short axis of the fibril and cause the elongation of the fibrils. Hence, the fibrils grown in the presence of the compounds had higher fibril content and the length of the fibrils was significantly longer. With regards to the change in morphology of the fibrils at higher concentrations of the compounds (1:5; Aβ(1-40) peptide: compounds), we propose that the large number of compounds present at such concentrations is sufficient to interact with the long axis of the fibrils (long side interactions). This interaction between the fibrils and compounds would affect the fibril packing resulting in the change in morphology of the fibrils [40] as observed in TEM images (Fig. 5A-D).
From the results of the MTT (Fig. 6A) and ThT assays (Fig. 6B), we conclude that different morphologies of fibrils caused by the interaction of the compounds can exhibit different toxicities and biological activities. We suggest that when the morphology of the fibrils is changed by the compounds, different sets of amino acids would be buried in the core of the fibrils. Hence, causing an alteration of the interaction sites of the modulated fibrils, leading to the fibrils displaying different toxicity and biological activity as compared to the unmodulated fibrils. This hypothesis is further supported by Petkova et al where they also suggested that the reason for different morphology leading to different biological activities could be due to the exposure of different amino acid side chains on the fibril surface [41]. It is therefore worth exploring with future studies on the detailed structure of the modulated fibril to unravel the relationship between fibril morphology and its respective toxicity and biological activity.
Furthermore, it has been reported that the toxicity of Aβ fibrils could be related to the process in which the fibrils are grown during the initial nucleation process [42]. Since we have shown in our experiment that the compounds most likely interact with the Aβ monomers during the fibrillation process (Fig. 4A), the molecular mechanism on how these compounds interact and alter the structure of monomers and fibrils might also provide valuable insights for the different toxicity observed in this study.
Together, our studies show that xenobiotic compounds (9,10-PQ, CPM and TRD) might be linked to AD via transport with lipophilic carriers such as L-PGDS and disruption of neuroprotective function important for the Aβ homeostasis. We show that these compounds are capable of directly interacting with the amyloid fibrils and altering their morphology. This change in morphology resulted in various toxicity of the modulated fibrils and their respective resistance towards the disaggregation of L-PGDS. As a result, it is possible that these compounds indirectly disrupt the chaperone clearance of the fibrils. CPM is also capable of directly interfering with the chaperone network by inhibiting the chaperone and disaggregase function of L-PGDS. This resulted in a slower clearance of the aggregated Aβ load and further exacerbated the aggregation of the Aβ peptides. At present, we hope that our study would provide possible molecular mechanisms for the association between these xenobiotic compounds with AD.

Conclusion
In conclusion, we have shown that the xenobiotic compounds, 9,10-PQ, TRD and CPM, can be associated with AD via two different possible mechanisms. Firstly, they inhibit the chaperone protein such as L-PGDS, important for the Aβ aggregate clearance pathway by binding to the A site and interfering with the chaperone activity of the protein. Secondly, they directly interact with the Aβ peptides (monomers and/or oligomers) to increase the Aβ peptide aggregation as well as altering the morphology of the modified fibrils and/or affecting both mechanisms at the same time. From our study, we hope to generate possible insights for future therapeutic intervention for AD, increase awareness of the potential long-term effects of the xenobiotic compounds and trigger further interest in finding more about other mechanisms that might link other xenobiotics compounds with increased AD risk. The morphological and toxicity differences of the modulated fibrils induced by xenobiotic compounds may also serve as a useful model for investigating the relationship between amyloid fibril structure and the resulting biological activities observed in our study.

Accession numbers:
PDB ID referenced in this manuscript : 4IMN