Patulin suppresses α1-adrenergic receptor expression in HEK293 cells

Patulin (PAT) is a common mycotoxin contaminant of apple products linked to impaired metabolic and kidney function. Adenosine monophosphate activated protein kinase (AMPK), abundantly expressed in the kidney, intercedes metabolic changes and renal injury. The alpha-1-adrenergic receptors (α1-AR) facilitate Epinephrine (Epi)-mediated AMPK activation, linking metabolism and kidney function. Preliminary molecular docking experiments examined potential interactions and AMPK-gamma subunit 3 (PRKAG3). The effect of PAT exposure (0.2–2.5 µM; 24 h) on the AMPK pathway and α1-AR was then investigated in HEK293 human kidney cells. AMPK agonist Epi determined direct effects on the α1-AR, metformin was used as an activator for AMPK, while buthionine sulphoximine (BSO) and N-acetyl cysteine (NAC) assessed GSH inhibition and supplementation respectively. ADRA1A and ADRA1D expression was determined by qPCR. α1-AR, ERK1/2/MAPK and PI3K/Akt protein expression was assessed using western blotting. PAT (1 µM) decreased α1-AR protein and mRNA and altered downstream signalling. This was consistent in cells stimulated with Epi and metformin. BSO potentiated the observed effect on α1-AR while NAC ameliorated these effects. Molecular docking studies performed on Human ADRA1A and PRKAG3 indicated direct interactions with PAT. This study is the first to show PAT modulates the AMPK pathway and α1-AR, supporting a mechanism of kidney injury.


Results
PAT binds AMP-activated protein kinase gamma subunit 3 (PRKAG3) with high affinity via molecular docking studies. While it is known that PAT increases oxidative stress and impairs mitochondrial function (Supplementary Data S2), little is known about the effects of PAT on AMPK. Preliminary docking studies of PAT were performed using a theoretically modeled structure of the PRKAG3 protein as a template. The structure of the selective receptor protein active binding site was chosen based on the active domain reported in previous studies. The identified active site amino acid based binding pocket grid was created for further analysis. The results revealed molecular interactions which account for the observed affinity within 4 Å distance (Fig. 1). The hydroxyl groups of the myricetin interacted through hydrogen bonding with the side chain residues of ARG307, ARG454, and HIE453 on the PRKAG3 receptor (Fig. 1). The binding efficiency of PAT to PRKAG3 was relatively strong with an estimated affinity of − 5.734 kcal/mol. Both docked complexes were examined with an emphasis on visual rather than numerical appraisal, so XP were used for further results presented.
Environmental PAT exposure alters α-adrenergic receptor expression in HEK293 cells. α 1 -AR expression was determined in HEK293 cells exposed to PAT (0.2 µM; 0.5 µM, 1.0 µM) concentrations relevant to environmental exposure; including safety levels and concentrations recorded in incidence, evaluation and consumer studies 3,35 . As shown in Fig. 3; significant changes to ADRA1A (p = 0.0043; Fig. 3a) and ADRA1D (p = 0.0165; Fig. 3a) mRNA levels were observed after 24 h exposure. Measures of α 1 -AR protein expression (which was inclusive of all ADRA1 subtypes) was determined to have decreased consistently across treatments, most notably at 1 µM (p = 0.0037; Fig. 3b). An observed threefold increase in ADRA1D was noted at 0.5 μM. However, the relatively high SD prevented statistical significance.
PAT decreases α 1 -adrenergic receptor protein expression and opposes Epi and AMPK action. The effects of PAT on α 1 -AR were assessed using pathway agonist, Epi. In Epi-stimulated cells, α 1 -AR protein expression was elevated relative to standard culture conditions (p = 0.0265). Following PAT addition however, protein expression decreased significantly across all concentrations, most potently at 0.2 µM and 1 µM (p = 0.0145; Fig. 4a).
BSO and NAC mimic and oppose PAT actions on GSH respectively. BSO was used to determine whether PAT-related thiol depletion had altered the α 1 -AR expression. In these treatments α 1 -AR protein expression was decreased compared to controls (p = 0.0286). The decrease was further exacerbated following PAT treatments  Fig. 4b). NAC, a GSH precursor was used to assess whether these effects could be reversed. Findings showed no change to α 1 -AR protein expression in PAT treatments (p = 0.0941; Fig. 4c) in cells pre-exposed to NAC. Metformin, a mitochondrial inhibitor and AMPK activator was used to assess the effects of PAT on α 1 -AR via the AMPK pathway. It was determined metformin caused a significant increase in α 1 -AR protein expression (p = 0.0294; Fig. 4d). PAT addition however significantly decreased α 1 -AR expression (p = 0.001; Fig. 4d), most significantly at 1 µM-above the established safety level.  www.nature.com/scientificreports/ These results allude to an inhibitory role for PAT in Epi-mediated α 1 -AR signalling and AMPK signalling with GSH depletion as a possible mitigating factor (Fig. 4).
PAT binds ADRA1 with strong affinity via computer docking studies. Computational methods were used to check the affinity binding between PAT and ADRA1, and Epi and ADRA1. Molecular docking predicts the binding modes and affinities of ligands and their receptors. An analysis of the docked complex of ADRA1 revealed highly significant interactions between PAT (ligand) and the ADRA1 receptor. Both 3D and 2D images were generated using Modeler software to visualise the interaction between PAT and ADRA1 ( Fig. 5a,b). The total free energy of binding was estimated to be − 6.4 kcal/mol, suggestive of a favourable reaction. PAT readily bound and formed close interactions with residues of ADRA1 through a variety of interactions including hydrogen bonding, pi-pi bonding, pi-alky bonding (Fig. 5c,d). The binding efficiency of PAT to ADRA1 was relatively strong with an estimated affinity of − 5.1 kcal/mol. Further, PAT bonded with ADRA1 though interactions with non-thiol containing amino acids side chains of serine (ser), tyrosine (tyr), phenylalanine (phe), glutamic acid (glu) and lysine (lys) (Fig. 5e), indicative of a novel binding. Interestingly, Epi was predicted to bind at a different location. It was determined that Epi had significant interactions with the side chains of glycine (gly), leucine (leu), aspartic acid (asp), proline (pro), valine (val), phenylalanine (phe) and glutamic acid (glu) (Fig. 5f). The binding affinity of Epi to ADRA1 was stronger than PAT with an estimated affinity of − 6.5 kcal/mol (Fig. 5g,h).

Discussion
The α 1 -AR mediates the effects of catecholamines throughout the human body 36 . These receptors are functionally linked to renal tone, metabolism, tubule function and sodium (Na + ) reabsorption in the kidney-where persistent suppression of α 1 -AR is associated with renal fibrosis 15,16,19,21,37,38 . This study showed that PAT altered the transcription and translation of α 1 -AR in HEK293 cells (Figs. 2,3,4). This finding was consistent across toxic, safety, environmentally and physiologically relevant concentrations, presenting insight into PAT-induced renal toxicity (Figs. 2, 3, 4). Western blotting determined (a) α 1 -AR expression decreased significantly following PAT administration in cells pre-exposed to α 1-AR agonist Epi (p = 0.0145). GSH depletion mimicked by (b) BSO (p = 0.0121) decreased expression significantly; while supplementation by (c) NAC (p = 0.0941) showed no changes before and after PAT exposure however, (d) metformin-exposed cells showed a significant increase in ADRA1 (p = 0.0294) and a significant decrease (p = 0.001) following PAT exposure ( # p < 0.05 relative to untreated control; *p < 0.05 relative to respective pretreated control) The original western blots are presented in Supplementary Data S7.
Scientific Reports | (2020) 10:20115 | https://doi.org/10.1038/s41598-020-77157-0 www.nature.com/scientificreports/ PAT, often found in mould-contaminated apples is a strongly electrophilic molecule and exerts toxicity by binding thiol groups. GSH is a prominent cellular thiol-containing compound involved in redox homeostasis. PAT has been linked extensively to GSH depletion and oxidative stress in the kidney 8,9,35 . PAT-induced suppression of α 1 -AR was potentiated in the presence of GSH inhibitor, BSO (Fig. 4b). GSH supplementation with NAC ameliorated this effect, alluding to thiol group depletion as a possible mechanism (Fig. 4c). Interestingly, molecular modelling challenged this finding (Fig. 5). These results showed PAT interacted directly with ADRA1 though hydrogen bonding, pi bonding and alkyl bonding with the side chains of Ser, Tyr, Phe, Glu and Lys (Fig. 5). None of the amino acids indicated contain a thiol side chain, revealing a novel mode of PAT affinity and binding. This finding offers an additional explanation for the low ADRA1 protein expression in PAT treatments (Figs. 2, 3, 4) and offers new insights into PAT induced toxicity beyond GSH inhibition and ROS generation. Most notably, PAT opposed the actions of α 1 -AR pathway agonist Epi and AMPK activator, metformin. A significant decrease in α 1 -AR was observed following PAT exposure in Epi (Fig. 4a) and metformin (Fig. 4d) pre-treatments.
The ADRA1 protein results however, should be interpreted with caution. The amino acid sequence corresponding to the epitope for the α 1 -AR antibody used in this study is conserved between all three ADRA1 receptor subtypes-presenting a limitation due to potentially low specificity. In addition, overall protein expression may not correspond directly with plasma membrane localization or mRNA levels. Together, this may account for the increase in α 1 receptor subtype-specific mRNA levels in some PAT treatments (Fig. 3) despite consistently observed decreases in overall ADRA1 protein expression. As such, while the PAT molecular docking studies support overall protein findings, the radio ligand binding assay, generation of crystal structures and/or NMR is required to conclusively prove a mechanism of inhibition, (e.g., antagonism, reverse agonism or shift from active to inactive state)-presenting an interesting avenue for future studies [39][40][41][42][43] . In addition, potential PAT-induced changes to transcriptional, translational and post-translational processes pertaining to α 1 -AR expression could be explored. Nevertheless, the findings in this study suggest a novel role for PAT in AMPK and Epi-mediated www.nature.com/scientificreports/ signalling and metabolism with previously unknown molecular targets. This finding may be mechanistically linked to altered renal and cellular function discussed in previous studies on PAT 34,[44][45][46][47][48][49][50] . PAT exposure has been associated with impaired kidney function characterised by degeneration of glomeruli, haemorrhage in the tubules and cortical regions, tubular atrophy and diminished clearance abilities-hallmarks of kidney injury and disease 10,49-51 . AMPK has a unique regulatory role in the kidney at the junction of energy metabolism, ion transport, inflammation and stress. Molecular docking studies indicated PAT interacted directly with Arg and Hie residues on PRKAG3 through hydrogen bonding (Fig. 1). This was supported by findings in the Human AMPK Signalling PCR Array (Supplementary Data S3) and western blotting which showed PAT exposure opposed effects of AMPK activator metformin on ADRA1 protein expression (Fig. 3). α 1 -AR mediated AMPK activation; suppressed in the early stages of kidney injury and disease 16 -was also suppressed following PAT exposure in this study (Figs. 2, 3, 4)-supporting a novel mechanism for PAT-induced kidney injury. This is associated with reduced tubule function, decreased kidney mass and fibrosis. Hence this pathway has a potential role to play in modulation of PAT-induced kidney injury and progression of acute and chronic kidney disease 18 .
Studies on α 1 -AR ligand-binding report that occupancy of the receptor can trigger direct or parallel activation of-MAPK/ERK1/2. This provides evidence that α 1 -AR can activate ERK with or without the canonical GPCR pathway 42 . This is an important consideration given the different predictive modeling binding sites of PAT and Epi (Figs. 5, 6). The MAPKs play a central role in cellular signal transduction between the cell surface and nucleus. This study showed ERK1/2 signalling was increased following PAT exposure (Fig. 6), a finding corroborated by literature 1 . This trend was consistent following Epi stimulation and neutralised by metformin and NAC, no changes were noted in BSO treatments however, which may be related to direct effects of BSO on this redox sensitive pathway (Fig. 5) 52 . While the observed Epi trend was consistent with other findings, the ERK1/2 pathway integrates diverse signalling pathways and stimuli including growth factors, tyrosine kinase or GPCR mediated activation. Epi effects on the ERK1/2 pathway are also mediated through α 2 -and β-adrenergic receptors, which are potentially confounding factors in relation to this study parameter 15 . Future studies using receptor blockers are required to fully understand the effects observed on the pathway. With respect to α 1 -AR signalling-PKC and Ca 2+ can activate the ERK pathway directly which phosphorylates and activates other protein kinases and transcription factors involved in survival, proliferation and apoptosis-all of which have been linked to PAT-toxicity [53][54][55] . An investigation into potential changes in Ca 2+ flux resulting from PAT exposure could reveal further connections between these signaling networks and associated toxic outcomes.
While the ERK1/2/MAPK pathway has been linked directly to α 1 -AR signalling; debate surrounds the distinct activation and function of α 1 -AR activated PI3K signalling 20,25,27,28,56 . This study showed PAT suppressed PI3K/ Akt signalling, even when stimulated with Epi and metformin (Fig. 6). α 1 -AR stimulated PI3K activation was www.nature.com/scientificreports/ also shown in murine keratinocytes 56 . Evidence suggests this has a functional role in glycogen regulation and receptor desensitization; which may be supported by findings in this study (Supplementary Data S3) and others showing PAT alters PI3K/Akt signalling in the same cells 20,28,31 . Several other studies however have found this pathway is primarily activated via co-ordinated cross talk between ERK1/2/MAPK pathways, Epi-stimulated signalling and receptor tyrosine kinases (RTK) 20,27,57 ; providing further substantiation for PAT-induced suppression of the pathway and Epi-mediated action (Fig. 6). These pathways have been linked to DNA synthesis, cell cycle progression, proliferation and metabolic regulation-often associated with carcinogenesis. Our findings indicated PAT slowed the kidney metabolic machinery and proliferation pathways-as a possible energy conservation strategy, evidenced by decreased PI3K/Akt activation (Fig. 7). The potential role for the mitochondria and AMPK in PAT-mediated toxicity is supported by significantly reduced PI3K/Akt activation in PAT-exposed metformin pre-treatments (Fig. 7). Suppression of this pathway is also linked to apoptosis and cell death via mitochondrial dysfunction-a key feature of PAT toxicity 29,30 . This is shown in literature confirming PAT impairs ATP and mitochondrial function, causes DNA damage, cell cycle arrest and cell death 11,34,53,55,58,59 .
Stress response mechanisms determining cell death and survival are central modulators in recovery and progression of renal injury. The apoptotic response to stress in the kidney; closely associated with PAT exposure in vitro and in vivo-is a critical event in the loss of tubular epithelial cells observed in kidney injury. AMPK Figure 7. PAT alters PI3K/Akt signalling in HEK293 cells. Western blotting revealed PAT significantly deceased PI3K/Akt signalling. Similar trends were observed in Epi-stimulated and NAC treated cells, while cells preexposed to BSO showed no statistically significant changes. Metformin treatments however showed a significant in decrease in pAkt following PAT exposure (*p < 0.05 relative to respective control). The original western blots are included in Supplementary Data S9 and S10. www.nature.com/scientificreports/ signal transduction via α 1 -AR mediates cell survival, repair and mitochondrial biogenesis. This study shows the pathway was suppressed by PAT-directly and through thiol depletion-providing a mechanistic explanation for previous cell death, mitochondrial impairment and nephrotoxicity studies on PAT 1,32,34,48,49 . Literature indicates PAT is nephrotoxic and has been associated with oxidative stress, cell cycle changes, apoptosis, diminished clearance abilities and changes in blood flow 34,[44][45][46]60 . This is the first study to show PAT alters α 1 -AR signalling and Epi-mediated action on the pathway, associated with changes in downstream signalling. This provides renewed insight to previous physiological and descriptive data on PAT-mediated toxicity. The distribution of these receptors in the brain, heart, vasculature and liver together with the functional relevance of Epi-mediated signalling and PAT target organs, warrants further investigation into PAT-induced changes in this pathway.
In conclusion, PAT-induced decreases in transcription and translation of α 1 -AR was associated with changes in downstream PI3K and MAPK signalling-most significantly at 1 µM, above safety level concentrations, supporting existing regulation and the need for food monitoring. This effect was consistent in cells stimulated with pathway agonist Epi and AMPK activator metformin-suggestive of a suppressive role for PAT in Epiand AMPK-mediated signalling. The effects of this parameter on downstream signalling was inconclusive and requires further investigation. Future studies including both pre-treatments and post-treatments, radio ligand binding assays, α 1 -AR and β-AR antagonists could elucidate mechanistic data and improve understanding of the downstream signalling observed. This study provides significant insights to previous studies on PAT and yields potential for further investigation. A study by Dailey et al. in both male and female rats showed 36% PAT was recovered in the urine 7 days post-administration 64 . This study found PAT accumulated in blood rich organs specifically the kidney, liver, erythrocytes and spleen. While current data on the absorption, distribution, metabolism, and excretion of PAT are limited, other studies have confirmed the bioaccumulation of PAT in the kidney. An in vivo study in albino mice found 1 μM oral administration of PAT led to glomerular haemorrhage, damage to the cortical regions and tubules of the kidney 48 . Another study (in 2012) using 6-22 μM PAT in mice also reported kidney damage and established a link between these effects and GSH depletion 50 . A recent in vitro study on HEK293 (kidney) cells by Jin and colleagues (2016) selected a range of 0-9 μM PAT over 24 h to determine a role for p53 in PAT mediated kidney damage 3,48,59 . The highest concentration used in our study (2.5 μM) is below the upper range cited in previous PAT exposure investigations-but within range of concentrations reported in incidence studies. Previously, a range of PAT concentrations (0-100 μM) was tested on HEK293 cell viability (using the MTT assay) following 24 h exposure 33 . The MTT assay indicated 100 µM PAT reduced HEK293 cell viability to 2% following 24 h exposure and an IC 50 of 2.5 μM was determined.

Materials
In the current study 2.5 μM (previous IC 50 ) represents the highest concentration of PAT used. Interestingly, this concentration corresponds to approximately 36% less than the upper concentration of 4 μM PAT (from literature)-accounting for clearance and recovery; while the lowest concentration used was 0.2 μM-below the established safety level (0.3 μM), accounting for lower range incidences.
Hence, the concentration range selected in our study (0.2 μM; 0.5 μM; 1 μM; 2.5 μM), in general, relates to the incidence of PAT found in food and beverages and accounts for clearance, recovery and retention levels as reported in previous studies.
Treatment conditions. Preliminary experiments used toxic exposure (2.5 µM PAT) determined from cytotoxic assays in previous studies using the same model 33,34 . Concentrations relevant to environmental and safety levels (0.2 µM; 0.5 µM; 1 µM PAT)-determined from incidence, monitor studies and literature were used to validate findings and molecular mechanisms in subsequent assays 2,3 .
Experiments examining the effects of PAT on ADRA1 protein signalling were exposed to PAT as described above. Cells were preincubated with α 1 -AR agonist Epi (Sigma, St Louis, USA) 10 µM for 30 min (min) prior to PAT exposure and further incubated for 24 h 36 . Buthionine Sulphoximine (BSO) 5 mM (Sigma, St Louis, USA) and N-acetylcysteine (NAC) 2 mM were preincubated for 1 h followed by the PAT exposure, to simulate GSH depletion and supplementation, respectively (Supplementary Data S1). This was included to determine whether observed effects were related to previously established mechanisms of PAT toxicity 65  www.nature.com/scientificreports/ using cells pre-exposed to AMPK activator metformin (5 mM) for 30 min-5 mM concentration was selected from literature 66 .
Molecular docking. Molecular docking was performed to determine a mechanism of substrate/inhibitor selectivity and to visualize the ligand orientation and the active site cavity of the protein 68 . The protein sequence of human ADRA1A and PRKAG3 was retrieved from PubMed. The 3D model of ADRA1 and PRKAG3 was generated using Homology modeling and Modeler software 69 . The chemical structure of PAT was drawn using Chem draw software. Molecular docking is a technique to predict the preferential orientation between two molecules to Scientific Reports | (2020) 10:20115 | https://doi.org/10.1038/s41598-020-77157-0 www.nature.com/scientificreports/ form a stable complex. Docking calculations are done to investigate the binding affinity between selected small molecules and target proteins. Glide docking requires a receptor grid and a set of ligand structures for flexible docking using the Monte Carlo based simulated algorithm. This technique was carried out using Glide v12.1 in which Glide SP and XP were applied. SP docking was done to screen ligands, which were large in number, and then XP docking was done which is more powerful as its run time is longer than SP. XP docking uses Extra precision and write XP descriptor information generates favorable ligand poses which were further screened through filters to examine the spatial fit of the ligands in the active site. Ligand poses which pass through initial screening are subjected to the evaluation and minimization of grid approximation. In the grid-based docking technique, the receptor is basically rigid. XP mode is tolerant than SP mode because it can screen out the false positive. XP is designed to place the active ligands that bind to the receptor in particular conformation. The best pose of each ligand was ranked based on Glide XP Glide score 70 .
Statistical analysis. Results are represented as mean ± standard deviation (SD) relative to normalized control. This is representative of three independent experiments completed in triplicate and error bars showing standard deviation. Statistical significance was assessed using t tests, one-way and two-way ANOVA with appropriate post hoc comparisons on GraphPad Version 5.0 Software. p values less than 0.05 were considered significant.