1 Introduction

The gut microbiota is a complex ecosystem of microorganisms that reside in the gastrointestinal tract. It plays a crucial role in human health, including digestion, immune system regulation, and metabolism. The gut microbiota also produces metabolites including vitamins, secondary bile acids, and neurotransmitters that may contribute to the development of neurodegenerative diseases (Luan et al. 2019; Schoeler and Caesar 2019; Wang et al. 2023). Bile acids are produced in the liver and play a crucial role in lipid digestion and absorption. Bile acids also act as signaling molecules that regulate energy metabolism, glucose homeostasis, and inflammation (Chiang 2013; Jia et al. 2018). The gut microbiota further modifies the primary bile acids by deconjugation, dehydroxylation, and epimerization, resulting in the formation of secondary bile acids. These secondary bile acids have been shown to modulate various physiological processes, including inflammation, glucose metabolism, and gut permeability (Guzior and Quinn 2021).

Recent works show dysregulation of bile acids has been observed in the cerebrospinal fluid of Alzheimer's disease (AD) patients (Nho et al. 2019). Increased levels of microbiota-derived bile acids (MBA) are associated with AD and poor cognitive performance. MBA may influence AD pathogenesis by modulating the blood–brain barrier (BBB). The BBB is a critical barrier that separates the brain from the circulatory system and regulates the entry of substances into the brain. The study found that MBA could modulate BBB permeability and promote the entry of neurotoxic substances into the brain, leading to neuroinflammation and neuronal damage (Lalic-Popovic et al. 2013). MBA may modulate AD pathogenesis by influencing amyloid beta aggregation by altering the gut microbiota composition. Amyloid beta is a peptide that aggregates to form amyloid plaques, a hallmark of AD pathology (Dionisio et al. 2015).

Multiple proteomic technologies have been developed for the comprehensive analysis of entire proteomes, revealing hitherto unknown interactions between MBA and proteins (Backus et al. 2016; Mitchell et al. 2023; Prabhu et al. 2020). The traditional molecular biotechnologies have identified the farnesoid X receptor (FXR) and Takeda G protein-coupled receptor 5 (TGR5) as two well-characterized bile acid receptors. However, little attention has been paid to the protein targets of MBA connected with neurological diseases. Thermal proteome profiling (TPP) is a relatively new and powerful tool in the field of proteomics that allows for the identification of protein–ligand interactions in cell lysates, tissues, or living cells (Savitski et al. 2014; Zhong et al. 2021). The technique involves exposing cells to different temperatures, which can cause proteins to undergo conformational changes or denaturation. These changes can be detected using quantitative mass spectrometry, allowing us to identify and quantify proteins that interact with MBA under different thermal conditions (Mateus et al. 2020). In light of this, we have performed comprehensive thermal proteomic analyses to unravel and comprehend the protein targets affected by microbiota-derived bile acids, intending to understand the complex interplay between MBA and neurodegenerative diseases. We found 65 proteins in SH-SY5Y neuronal cells that are potential targets of deoxycholic acid (DCA) and revealed that the interaction between DCA and the Nicastrin subunit of gamma-secretase plays an important role in promoting amyloid beta formation.

2 Materials and methods

2.1 Reagents and cell culture

Deoxycholic acid (DCA, Cat.No. 30960), Dimethyl sulfoxide (DMSO, Cat.No. 276855), Dithiothreitol (DTT, Cat.No. 43815), Iodoacetamide (IAA, Cat.No. V900335) were purchased from Sigma-Aldrich with purity for analytical standards. Isopropyl alcohol (IPA), acetonitrile (ACN), Trypsin enzymes and iTRAQ® Reagent Application Kit (Cat.No. 4381663) were purchased from Sigma-Aldrich. A bicinchoninic acid (BCA) protein assay kit (Cat.No. 23227) were purchased from Thermo Scientific. All solvents used in this study were high-performance liquid chromatography grade or higher. SH-SY5Y cell line were obtained from the America Type Culture Collection (ATCC® CRL-2266™).

2.2 Sample preparation for TPP

The SH-SY5Y human neuroblastoma cell line was chosen for this study due to its established relevance in neurobiological research(Kovalevich and Langford 2013). Previous literature has extensively characterized SH-SY5Y cells as a suitable in vitro model for investigating various aspects of neurological diseases (de Medeiros et al. 2019; Xicoy et al. 2017). SH-SY5Y cells cultured in the 10 cm dishes were treated with 0, 10 and 40 μM DCA in a 5% CO2 incubator at 37°C for 24h respectively. DCA was dissolved in DMSO and final concentration of DMSO was 0.1%. Then cells were washed with 3 mL cold PBS twice and scraped down with another 1 mL cold PBS supplemented with EDTA-free protease inhibitor cocktail. After centrifugation at 300 g for 5 min at 4°C, the cell pellets were resuspended in 200 µL PBS and transfered equally to new PCR tubes. Subsequently the samples were heated at 37 or 53°C for 3 min and followed by incubation at room temperature for 3 min. Afterwards, cells were lysed by repeated freezing and thawing and centrifuged at 20,000 × g for 20 min at 4°C. Then supernatant was collected and BCA assay was performed to determine the protein concentration.

2.3 Tryptic digestion, ITRAQ labeling

The same amount of proteins from each sample were reduced by 10 mM DTT for 30 min at 55°C and then alkylated with 30 mM IAA for 30 min at room temperature. Excess IAA was quenched by 10 mM of DTT. Proteins were then purified by acetone precipitation and protein precipitates were dissolved in 50 µL Dissolution buffer from the iTRAQ reagent kit (0.5M TEAB, PH8.5). Proteins were digested overnight by trypsin at a ratio of 20:1 (protein:trypsin). After digestion, each peptide sample was incubated with iTRAQ reagent at room temperature for 2 h. HPLC grade water was added to quench the reaction. Then the combined iTRAQ-labeled peptides were desalted and fractionated on a SCX and C18 column using stepwise elution with 3%, 6%, 9%, 15% and 80% (v/v) ACN in 0.1% ammonia. Each fraction was lyophilized and resuspended in 0.1% (v/v) formic acid for MS analysis.

2.4 LC–MS/MS analysis

LC–MS/MS analysis was performed on an Easy-nLC 1000 (Thermo Fisher Scientific) chromatography system coupled to an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific). Peptide samples were separated on an analytical column (75 μm i.d. × 15 cm, packed with 3 μm C18 particles) using a gradient of 50 min from 5 to 18% phase B (0.1% FA in acetonitrile), followed by 10 min from 18 to 28% at a flow rate of 250 nL/min. Survey full scan MS spectra (m/z 350–1550) were acquired in the Orbitrap with a resolution of 120,000, AGC target value of 2 × 106, and maximum injection time of 40 ms. Data-dependent acquisition of MS/MS spectra was performed using a top speed of 3 s and dynamic exclusion of 30 s. Precursors were selected using an isolation window of 1 Da and then fragmented by HCD using collision energy of 38%. Fragment ions were recorded by the orbitrap mass analyzer with first mass of 100 m/z, orbitrap resolution of 30,000, AGC target value of 5 × 104, and a maximum injection time of 60 ms.

2.5 Protein identification and quantification

Raw data were searched through UniProt Homo sapiens proteome database via Proteome Discoverer 2.4 software (Thermo Scientific). The search results were filtered to 1% false discovery rate to reduce the probability of false peptide identification. iTRAQ quantitation was performed according to previously reported rules. The functional annotation and pathway enrichment analysis were integrated and analyzed in Ingenuity Pathway Analysis (IPA, QIAGEN INC.).

2.6 Molecular docking

DCA and protein structure files were downloaded from the PUBCHEM database and Protein Data Bank, respectively, as described for each docking software program. Autodock vina v1.1.2 was used to visualize ligand and receptor docking, and the final results were visualized by PYMOL and Discovery studio software.

2.7 Statistical analyses

The proteins with fold changes greater than 1.2 and p-value less than 0.05 in either 10 µM or 40 µM DCA-treated group under 53°C conditions while they remained unchanged at 37°C were screened as DCA’s protein targets. Comparisons between two groups were performed using Welch’s t-test. The relative protein ratio was plotted versus DCA concentration and fit to a standard four-parameter logistic model to calculate EC50. Heatmap and volcano plot were generated in R package “pheatmap”and “statTarget”(Luan et al. 2018), respectively.

3 Result and discussion

3.1 TPP for identification of protein targets of DCA

TPP has been widely used to measure protein thermal stability and abundance for unbiased identification of direct and indirect drug targets (Savitski et al. 2014). It offers a number of advantages over traditional proteomics approaches, including high throughput, sensitivity, precision, accuracy, and the ability to provide insights into complex biological systems (Becher et al. 2016). We analyzed the proteome changes after incubation of live cells with DCA. Theoretically, the heat stability of the target proteins can be enhanced upon ligand binding in solution than the unbound proteins (Martinez Molina et al. 2013). Therefore, these DCA-bound proteins would be more abundant in the supernatant after DCA treatment than in the control after thermal challenge and centrifugation. We evaluated the melting curve of SH-SY5Y lysate after the incubation with DCA at conditions ranging from 37 to 67°C and found that the median protein thermal melting temperature (Tm) was 53°C, which was subsequently selected as the temperature for screening DCA target proteins (Fig. S1). Figure 1A shows the detailed workflow of TPP applied to screen target proteins for DCA in living cells. Briefly, cells were treated with DCA and DMSO and heated at 37 and 53 °C for 3 min. Subsequently, proteins were extracted, digested, and further labeled with iTRAQ stable isotope reagents for quantitation analysis.

Fig. 1
figure 1

TPP for identification of protein targets of DCA. A Workflow of TPP experiments. B Abundance rank plot of identified 4076 proteins. C Venn plot showing stabilized protein targets shared by different groups (n = 3). D and E Volcano plot showing protein targets stabilized in cells following administration of DCA (10 μM or 40μM). Stabilized proteins are indicated in red (p < 0.05 and fold change > 1.2). Statistical significance was determined using Welch’s t-test

Full size image

As shown in Fig. 1B, A total of 4076 protein groups were annotated from 22,679 peptide groups. After normalization of the quantitative data with the vehicle, proteins with significant enrichment fold change more than 1.2 and adjusted p-value less than 0.05 in the 53 °C heated DCA treatment groups were selected as the potential targets. Overall, we have identified 65 unique proteins as the potential targets of DCA in the 10 μM and 40 μM treatment groups under 53 °C (Fig. 1C, D, and E). Among them, Nicastrin (NCSTN) and Casein kinase 1 epsilon (CSNK1E) were identified with high confidence in the DCA treatment groups, compared with the vehicle group (Fig. 1D and E; Table S1).

Although the use of chemo-proteomic probes has led to the identification of hundreds of bile acid-interacting proteins, encompassing recognized receptors, transporters, and biosynthetic enzymes of bile acids (Zhuang et al. 2017), It is worth noting that the NCSTN and CSNK1E are novel bile acid–protein interactions discovered in our study. NCSTN is a highly glycosylated type I transmembrane protein and is widely distributed in all cell types of humans or mice as one of the important component proteins of the γ-secretase complex. It is not only closely related to the assembly and maturation of γ-secretase but also plays an important role in regulating the production and degradation of γ-secretase activity and amyloid beta protein in AD (Xie et al. 2014). Another MBA-interacting protein, CSNK1E, is part of the CK1 family of ubiquitous serine/threonine-specific protein kinases, which has been suggested to have a role in AD pathology. CSNK1E is involved in regulating the cleavage of the γ-secretase of amyloid precursor protein, leading to increased formation of amyloid beta peptides (Chen et al. 2017; Flajolet et al. 2007). Our study identified NCSTN and CSNK1E as MBA-interacting proteins and may provide a novel link for MBA in the direct regulation of amyloid beta peptide formation.

3.2 MBA-interacting proteins and pathway enrichment analysis

Screening of MBA-interacting proteins from cells treated with different drug concentrations and temperatures was employed to identify bile acid-interacting proteins. Overall, 65 out of 4076 proteins were identified as MBA-interacting proteins, which has their good thermal stability upon interactions with DCA (Fig. 2A). The 65 MBA-interacting proteins were analyzed through the use of Ingenuity Pathway Analysis (IPA, QIAGEN INC.) to identify enriched pathways and associated networks based on a calculated probability score of ≥ 2 and a p-value of < 0.05 using IPA software application. The top-ranked canonical pathways influenced by these 65 proteins were shown in Fig. 2B. The pathways associated with “Glutathione Biosynthesis “, and “Glutathione Redox Reaction II” were the most significantly enriched pathways, indicating MBA may be associated with the antioxidant defense mechanism and oxidative damage (Wang et al. 2018). Notably, two MBA-interacting proteins (NCSTN and CSNK1E) were highlighted in the amyloid processing pathway (Fig. 2B). From the data for canonical pathways and biological functions, IPA assessed networks which revealed the direct or indirect association of the 65 enriched MBA-interacting proteins with each other. The first and most relevant network is depicted in Fig. 2C. The network consisted of 26 out of the 65 MBA-interacting proteins analyzed. A central component of this network includes γ-secretase and NCSTN, which are associated with the processing of amyloid precursors in AD (Zhou et al. 2019).

Fig. 2
figure 2

DCA-interacting proteins and pathway enrichment analysis. A The levels of stabilized protein targets were visualized by the depicted heat map. B The top-ranked canonical pathways. C typical protein–protein interactions networks revealed the association of stabilized protein targets

Full size image

3.3 Binding affinity evaluation of DCA and protein targets

We generated isothermal (37 °C and 53 °C) dose–response curves and calculated EC50 values to evaluate the binding affinities between DCA and the involved MBA-interacting proteins. The cell lysates were treated with two different doses of DCA as well as one vehicle control. As shown in Fig. 3A, an increase in the concentration of the target proteins was observed as the ligand concentration increased to a level of protein binding saturation. The EC50 values of NCSTN and CSNK1E were calculated as 6.03 μM and 7.22 μM, respectively, indicating the binding affinity between DCA and target proteins (Kurzawa et al. 2020).

Fig. 3
figure 3

Binding affinity evaluation of DCA and protein targets. A Dose-dependent stabilization of NCSTN and CSNK1E by DCA at 37°C and 53°C, respectively. Stabilization of NCSTN and CSNK1E in response to DCA treatment was measured relative to vehicle-treated control and fitted with saturation curves to determine the concentrations leading to half-maximal stabilization, EC50 (− log10-transformed compound concentrations). B Evaluation of the binding affinity between DCA and the involved proteins (NCSTN at the top layer and CSNK1E at the bottom layer) by molecular docking analysis

Full size image

We further evaluated the binding affinity between DCA and the involved proteins (NCSTN and CSNK1E) by molecular docking analysis (Meng et al. 2011). Figure 3B shows that DCA could be embedded into the active pocket of NCSTN and CSNK1E with the binding energy of -7.9 and -7.8 kcal/mol, respectively. DCA is capable of hydrogen-bonding, salt-bridging, and hydrophobic interactions with the binding pocket of NCSTN protein. Subsequently, by analyzing the three-dimensional interactions, it was found that the DCA is capable of forming hydrogen-bonding interactions with proteins TYR283, THR387, ASP447, and ILE468; the carboxylic acid functional group of the compounds is negatively charged under physiological pH conditions, and the negatively charged functional group of the compounds forms salt-bridging interactions with the positively charged LYS470 of the protein. Similarly, DCA is also capable of hydrogen-bonding and hydrophobic interactions with the CSNK1E protein binding pocket and hydrogen-bonding interactions with proteins ALA56, THR57, and ASP336. The compound is capable of hydrophobic interactions with ASP143 and TYR173 of the protein. These interactions promote DCA binding to the active pocket of the protein to form a complex. These results indicated that NCSTN and CSNK1E are the targets of DCA, and the binding strength is positively correlated with the concentration of DCA.

3.4 DCA can interact directly with NCSTN and affect the formation of Aβ

NCSTN is the putative substrate recruitment component of the γ-secretase complex that is responsible for the production of A amyloid beta peptides (Zhou et al. 2019). We further evaluated whether DCA can affect the formation of amyloid beta. We have observed that 10 and 40μM DCA treatments have resulted in the enrichment of NCSTN and CSNK1E playing vital roles in the amyloid processing at 53 °C, indicating that the processing of amyloid precursor protein to amyloid beta might be an important target pathway of DCA (Figs. 3A and 4A). In contrast, there was no significant change in the concentration of these proteins after 37 °C heating in DCA-incubated samples, suggesting that the DCA does not affect their constitutive abundance (Fig. 4A and B). Furthermore, we incubated different concentrations of DCA with live cells. As shown in Fig. 4C and D, the concentration of Aβ40 in live cells was significantly increased under 10 and 40 μM DCA treatment. We also found that APP protein was significantly increased only under high concentrations of DCA treatment. These results suggest that DCA may promote the conversion of amyloid precursor protein to amyloid beta.

Fig. 4
figure 4

DCA interacts with NCSTN and affect the formation of Aβ. A and B The level changes of NCSTN and CSNK1E. The column marked as red and blue refers to the cells treated with DCA 10 μM at 37 and 53 °C, respectively (n = 3). The DMSO treatment groups (37 or 53 °C) were normalized as 1, while the fold changes of DCA treatment groups were calculated after normalization with the vehicle group. C and D bar plots showing increased αβ40 and increased Amyloid precursor protein (APP) levels upon DCA treatment (n = 3). Asterisk (* and **) indicates a statistically significant difference (p < 0.05 and p < 0.01). NS indicates not significant. Statistical significance was determined using Welch’s t-test.

Full size image

Our study shows that thermal proteomic analysis are promising for revealing the protein targets of MBA, providing new insights into their role in interacting with these MBA in neurodegenerative diseases. A limitation of our study is the lack of further investigation into the potential molecular mechanisms underlying the observed interactions between MBA and protein targets. These interactions can be confirmed by co-immunoprecipitation or in vitro assays (Lin and Lai 2017). DCA represents just one type of MBA, highlighting the need for broader exploration into various MBAs and their associated protein targets. Moreover, TPP technology faces certain limitations, such as the challenge of detecting low-abundance proteins and its relatively low analytical throughput.

4 Conclusion

The intricate interplay between gut microbiota and host metabolism has been the subject of numerous investigations (Luan et al. 2019). Of particular interest are microbial bioactive metabolites, such as bile acids, amino acids, and short-chain fatty acids, which have been implicated in neurodevelopmental processes and neurodegenerative disorders (Dalile et al. 2019; Luan et al. 2017; Winston and Theriot 2020). Despite this knowledge, the specific protein targets of MBA remain elusive. In our study, employing the TPP method, we successfully identified NSCTN and CSNK1E as direct protein targets of deoxycholic acid (DCA), an MBA associated with Alzheimer's disease. Furthermore, we confirmed the ability of DCA to promote the formation of Aβ, a hallmark of Alzheimer's pathology. Our findings shed light on the underlying mechanisms connecting MBA and neurodegenerative diseases, providing valuable insights into these complex pathological processes.