PERFLUOROOCTANESULFONIC ACID (PFOS) AS A POTENTIAL RISK FACTOR FOR LATE-ONSET ALZHEIMERâ•ŽS DISEASE

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that accounts for about 60 to 80% of dementia cases worldwide. AD is characterized by two important physiological features: extracellular senile plaques consisting of β-amyloid (Aβ) peptide that is cleaved from a larger protein called the amyloid precursor protein (APP), and intracellular neurofibrillary tangles (NFTs) containing hyperphosphorylated tau protein. The pathogenesis of sporadic or late-onset AD (LOAD) which represents 95% of the cases appears to not be heritable and may be related to epigenetic programming and environmental factors acquired earlier in life. Apolipoprotein allele 4 (ApoE4) protein is encoded by the ApoE gene and is considered the only established genetic risk factor for LOAD. It is suggested that ApoE4 affects the pathogenesis of AD by increasing tau hyperphosphorylation, inhibiting the clearance of amyloid-β (Aβ), and promoting Aβ aggregation. Perfluorooctanesulfonic acid (PFOS) is a persistent organic pollutant known to cause neurotoxicity and poses a major threat to the ecosystem and human health. The aim of this study was to investigate PFOS as a potential risk factor for LOAD by assessing the potential mechanism and impact of PFOS on three major AD-associated pathways: amyloidogenesis, tau pathology, and ApoE using in vivo and in vitro models. Our findings demonstrated that SH-SY5Y neuroblastoma cells exposed to low concentrations of PFOS showed increased protein levels of APP, tau, and phosphorylated tau (p-tau) at Ser-404 and Thr-181 sites. PFOS exposure also caused an elevation in the kinase GSK3β, ApoE4 protein levels in vitro; however, only GSK3β and ApoE were modulated in vivo. These data suggest that ApoE4 is inducible by


INTRODUCTION
Alzheimer's disease (AD) is a progressive neurodegenerative disorder that accounts for about 60 to 80% of dementia cases worldwide 1,2 . Several pathologies have been implicated with AD including neuroinflammation, microglia activation, and oxidative stress 3 . The major AD hallmarks are the presence of extracellular senile plaque consisting of β-amyloid (Aβ) peptide that is cleaved from a larger protein called the amyloid precursor protein (APP), and intracellular neurofibrillary tangles (NFTs) containing hyperphosphorylated tau protein 4 . Tau is the major neuronal intracellular microtubule binding protein, it is essential for stabilization of microtubules, and axonal transport in neurons 5 . Hyperphosphorylated tau which occurs in pathological conditions, reduces tau affinity to microtubules and impairs axonal transport 6,7 . Several protein kinases are involved in the phosphorylation of tau protein such as the proline-directed protein kinases: cyclin-dependent kinase-5 (CDK5) and glycogen synthase kinase-3 beta (GSK3β) 8 .
The etiology of Late-onset Alzheimer's Disease (LOAD) which represents 95% of the cases is sporadic in nature and the only established gene associated with LOAD is Apolipoprotein (ApoE) 9 . Mounting evidence has indicated that the human ApoE gene is the strongest genetic risk factor for the development of LOAD 10,11 . The presence of one copy of ApoE4 allele increases the risk of acquiring LOAD by about 3 times while the risk is increased by 15-fold if 2 copies of ApoE4 allele are present 9,11 . In the central nervous system (CNS), ApoE is involved with the repair process after injury 10 , it functions as a lipid transporter and/ or signaling molecule 12,13 . There are 3 different isoforms of ApoE in humans, however, ApoE4 is the one mainly associated with increased prevalence of AD 9,10,14 . Previous studies have reported that fragments of ApoE4 accumulated in NFTs in 3 brains of AD patients 15 , and are detected in neurons of ApoE4 transgenic mice 16 . ApoE4 appears to result in behavioral deficits and neurodegeneration through disruption of the cytoskeleton, stimulation of tau phosphorylation, generation of NFTs 17,16,18 as well as by interfering in Aβ clearance thereby facilitating the aggregation of Aβ and formation of senile plaque 9,10,11,17,19,20,21 .
The random nature of LOAD suggests that environmental factors may contribute to the onset of LOAD, particularly environmental exposures acquired earlier in life 22 . Polyand perfluoroalkyl substances (PFASs) are a class of emerging persistent organic pollutants (POPs) with neurotoxic properties. They are manmade fluorinated chemicals that have been widely used in numerous industrial and consumer products such as textile products, firefighting foams, and oil-resistant coatings for paper products. PFASs persist in the environment and bioaccumulate in the food chain 23 . One of the most widely used PFASs is the 8-carbon chain (C8) Perfluorooctanesulfonic acid (PFOS) which has been detected in adult human serum 24 and in children with a serum concentration equal or even higher than adults 25 . PFOS can cross the placenta, the blood brain barrier (BBB) 26,27,28 and was found in breast milk 29 . PFOS strongly binds to plasma albumin 30 and it has an average serum half-life of 5.4 years in humans 31 , and 36.9 days for CD-1 mice 32 .
Even though levels of PFOS exposure have decreased during the last decade due to industrial pashing out of PFOS, it persists in the environment due to its resistance to photolysis, hydrolysis, and biodegradation 23 . Developmental studies using rodents showed that exposure to PFOS can cause neural damage manifested by chronic glial activation, release of inflammatory factors 33 , altered susceptibility of the cholinergic system, resulting in behavioral deficits, reduced spatial learning, memory and cognitive impairment which 4 can persist into adulthood 34,35,36,37 . Neonatal exposure of mice to a single PFOS dose 21 mol/kg body weight (11.3 mg/kg body weight) on PND 10 increased cerebral tau protein expression and caused behavioral abnormalities in adult mice 34,37,38 . Moreover, PFOS resulted in an increase in β-amyloid aggregation, increased tau levels, and tau hyperphosphorylation in adult rats 39 suggesting an association between developmental PFOS exposure and LOAD. Therefore, the main aim of this study was to investigate the potential of PFOS exposure as a risk factor for LOAD by assessing its impact on biomarkers associated with three major AD-related pathways in an in vitro model using the human SH-SY5Y neuroblastoma cell line. Furthermore, the same pathways were examined following developmental PFOS exposure in timed-pregnant CD-1mice.

Animal model and PFOS exposure
Twelve timed-pregnant CD-1 mice were purchased from Charles River For dose calculation, the dams body weights were recorded every 3-4 days and the last dose given was on PND 19. On PND 1, neonates were randomly assigned to dams within the same treatment group with 10 pups assigned to each dam in order to ensure equal lactational PFOS exposure, excess pups were culled. On PND 20, dams and pups were fasted for 4-6 hours and euthanized via cardiac puncture with subsequent decapitation. The brain tissues were collected and stored at -80°C for biomarker assessment. During this study, the animals were under the supervision of the URI CBRC staff. The animal protocols and the study procedures were reviewed and approved by the University of Rhode Island Institutional (URI) Animal Care and Use Committee (IACUC). Cells were then sub-cultured at a density of 10 5 cells/mL into a six-well plate and upon 80% confluency, cells were differentiated for 6 days using 10 µM all-trans retinoic acid (Sigma Aldrich, St. Louis, MO) dissolved in DMEM/F12 medium containing 1% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin in the dark. Cells were observed for neurite outgrowth with media being changed every 48 hours as described previously 40 . Furthermore, differentiated cells were exposed to PFOS dissolved in DMSO for 24, 48, and 72 hours at 37°C with the following concentrations: 0, 0.001, 0.1, 5, 25, 50, 100 µM. The concentration of DMSO in the cell culture media was maintained at 0.1% for control and across all treatment groups.

MTS assay
The viability of differentiated SH-SY5Y human neuroblastoma cells was assessed after exposure to PFOS using the MTS assay kit (Promega, Madison, WI). The cells were seeded in 96-well plate (2 x 10 4 cells/well) and treated with different concentrations of PFOS ranging from 0-100 µM for 24, 48, and 72 hours. After the completion of the incubation period, the media containing PFOS was aspirated out and fresh media with no PFOS was added in each well followed by treatment with 20 µl of MTS solution. The plates were incubated at 37°C for 2 hours and after incubation, the absorbance was measured at 490nm by using the Spectramax M2 Multi-mode microplate reader (Molecular Devices, San Jose, CA).

Protein Extraction and Western Blotting
The brain cortex samples and cells were lysed using radio-immunoprecipitation assay (RIPA) lysis buffer (Sigma Aldrich, MO) supplemented with phosphatase and protease inhibitor tablet: one tablet was dissolved in 10 ml of RIPA buffer (Pierce Protease and Phosphatase Inhibitor Mini Tablets, EDTA-Free, Thermo Scientific, Waltham, MA).
The samples were incubated on ice for 30 min, then vortexed and sonicated for 5 minutes.
The samples were subsequently centrifuged at 10,000 g for 20 min and the supernatants were collected and stored at -80C for protein quantification and western blot analysis. The  Anti-APP (Cell Signaling Technology, MA) at a dilution of 1:1000 in TBST. On the following day, the membranes were washed three times with TBST for 5 min each and exposed at room temperature for 1 hour to either goat anti-mouse or goat anti-rabbit IRDye ® 680 or 800 LT infrared dye (LI-COR Biotechnology, Lincoln, NE) at a 1: 10,000 dilution factor in TBST then the membranes were again washed three times with TBST.
Also, the specific target protein/GAPDH ratio was obtained by re-probing the membranes for 2 hours at room temperature with primary antibody for the protein Glyceraldehyde 3phosphate dehydrogenase (Rabbit Anti-GAPDH) at 1:5,000 dilution as a control, followed by washing and re-exposure to goat anti-rabbit IRDye ® 680 or 800 LT infrared dye (LI-COR Biotechnology, Lincoln, NE) at a 1: 10,000 dilution factor in TBST (Cell Signaling Technology, MA). Finally, images were captured using the Odyssey ® Infrared Imaging System (Li-Cor, NE), and Band intensities were measured using NIH ImageJ software and normalized against GAPDH.

Total RNA isolation, synthesis of cDNA, and RT-q PCR
Total RNA was isolated from cerebral cortex tissues and the SH-SY5Y cells using The plates were maintained at room temperature for 2 hours to allow the binding of the immobilized (capture) antibody and the Aβ antigen. Then, the plates were washed four times by using the washing solution provided in the kit. Subsequently, a solution of rabbit antibody specific for the C-terminus of either Aβ1-42 or Aβ1-40 was added to the wells (100 µl in each well) except for the blanks. The plates were then covered and incubated again for 1 hour at room temperature. After incubation, the wells were washed 4X using the washing solution and 100 µl of horseradish-peroxidase-labeled anti-rabbit antibody was added to each well except for the blanks. Then, the plate was covered and incubated at room temperature for 30 minutes followed by four times washing. Then, 100 µl of the stabilized chromogen solution was added to each well and the plate was incubated at room temperature for 20-30 minutes in the dark. Once, the reaction mixtures turned blue, 100 µl per well of the stop solution was added to each well to stop the reaction and the mixture turned yellow. Finally, the absorbance was detected at 450 nm using the Spectramax M2 Multi-mode microplate reader (Molecular Devices, San Jose, CA)

Statistical analysis
The results of various exposed groups versus the control were expressed as the mean ± the standard error of the mean (SEM). The data were analyzed by using the statistical tests: unpaired two-tailed Student's t-test for comparison between two groups and the one-way analysis of variance (ANOVA) followed by the Dunnett or Tukey test for multigroup comparisons. The GraphPad Prism 8.0 computer software (La Jolla, CA, USA) was used for the analysis, and p-value < 0.05 was considered to indicate a statistically significant difference.

PFOS effects on APP mRNA and protein levels in differentiated SH-SY5Y cells. (amyloidogenic pathway)
Further, we investigated the effects of PFOS on AD biomarkers related to the amyloidogenic pathway in human neuroblastoma SH-SY5Y cells. APP protein level was significantly elevated as compared to the control after 24 h exposure to 0.001 µM concentration of PFOS, p= 0.0152 ( Figure 3A, B). We did not observe any effects on APP mRNA levels ( Figure 3C).

PFOS effects on Tau mRNA expression, Total Tau, and P-tau protein levels in differentiated SH-SY5Y cells (tau Pathway)
SH-SY5Y cells showed a significant increase in total tau, P-tau (Ser-404) and Ptau (Thr-181) protein levels as compared to controls after 24 h exposure to 0.001 µM of PFOS, P= 0.0307, P= 0.0062, P=0.036 respectively ( Figure 4A, B, D-G). Also, P-tau (Ser-404) was significantly elevated as compared to controls in SH-SY5Y cells after 24 h exposure to 0.1 µM of PFOS, P= 0.0058 ( Figure 4D, E). Based on one-way ANOVA, there was no effect of PFOS exposure on total tau mRNA expression ( Figure 4C). Next, we investigated the effect of PFOS on kinase pathways in differentiated SH-SY5Y cells.
GSK3β protein levels were significantly increased as compared to controls following 24 h exposure of SH-SY5Y cells to PFOS at 0.001 µM concentration, p=0.012 analyzed by the one-way ANOVA ( Figure 5A, B). We did not observe any significant changes in CDK5 protein levels and mRNA expression levels of GSK3β and CDK5 ( Figure 5C-F), based on a one-way ANOVA followed by Dunnett's for multiple comparisons.

PFOS effects on ApoE and ApoE4 protein levels in differentiated SH-SY5Y cells (Non-neural Pathway)
We further analyzed the impact of PFOS exposure on non-neural pathways associated with AD. SH-SY5Y cells exposed to PFOS at 0.001 µM concentration for 24 h, showed a significant increase in ApoE4 protein levels as compared to controls, P= 0.026 ( Figure 6C, D) analyzed by the one-way ANOVA followed by Dunnett's for multiple comparisons. Interestingly, PFOS had no effect on total ApoE protein levels in SH-SY5Y exposed cells at 0.001 and 0.1 µM concentrations following 24 h of exposure ( Figure 6A, B).

Effects of developmental exposure of PFOS on APP mRNA expression, APP protein levels, Aβ1-42 and Aβ1-40 accumulation (amyloidogenic pathway)
We observed an increasing trend in the effect of PFOS exposure on APP mRNA expression and APP protein levels as compared to controls ( Figure 7A-C). PFOS exposure caused an increased trend in Aβ42 and the ratio of Aβ42/Aβ40 proteins levels, however, these findings were not statistically significant. The Aβ40 protein levels were not affected ( Figure   7D-F).

Effects of developmental PFOS exposure on Tau mRNA expression, Total Tau and P-tau protein levels (tau Pathway)
Tau mRNA expression in the cortex of PFOS exposed group showed no significant difference as compared to controls ( Figure 8C). Western blot analysis showed no significant difference between the PFOS exposed group and the control group in Total tau and P-tau (Ser-404) protein levels ( Figure 8A, B, D, E). The protein level of P-tau (Thr-181) was trending higher in the PFOS exposed group compared to controls ( Figure 8F, G). The PFOS exposed group showed a significant increase in GSK3β protein levels as compared to control group, p= 0.05 determined by two-tailed unpaired t-test ( Figure 9A, B). The GSK3β mRNA expression was trending higher in the PFOS exposed group compared to controls ( Figure 9C). We did not observe any significant effects on CDK5 mRNA expression and protein levels in the PFOS treated group compared with the control group ( Figure 9D-F).

Pathway)
Western blot analysis showed that PFOS significantly increased total ApoE protein levels in brain cortex as compared to the control group, p= 0.0025 analyzed by two-tailed unpaired t-test ( Figure 10A, B).

DISCUSSION
As a persistent organic pollutant, PFOS has been identified as a developmental neurotoxicant 24,41,42 . Previous studies have reported the developmental effects of PFOS on neurological disorders, behavioral disorders, and cognitive function 35,43,44 . A few reports indicated that developmental exposure to PFOS induced tau protein levels 37 Figure 4A, B, D-G) providing more evidence that PFOS can promote AD pathogenesis through modulating the tau pathway. The upregulation of total tau and the abnormal hyperphosphorylation of tau (P-tau) proteins are responsible for the loss of microtubule stability and axonal transport in AD 49,50 . This is consistent with the study by  which reported that developmental exposure to PFOS induced β-amyloid aggregation, total tau, and phosphorylated tau proteins levels in adult rats at PND 90 45 . Also, Johansson et al (2009) reported that neonatal mice exposed to a single PFOS dose on PND 10 showed a significant elevation of cerebral tau protein levels 37 . This is consistent with our data suggesting the impact of PFOS on amyloidogenic and tau pathways related to AD.
Moreover, we investigated the effect of PFOS on the kinase pathways that are associated with AD. GSK3β is an active serine/threonine proline-directed kinase, it is involved in many processes including inflammatory responses, memory impairment, cholinergic deficits, increased the production of Aβ and hyperphosphorylation of tau 51 .
Also, the sites of tau phosphorylation found in AD brain are proline-directed (Thr-Pro or Ser-Pro). Thus, GSK3β plays an important role in the pathogenesis of AD 52 . In this study, we observed the ability of PFOS to promote an increase in the GSK3β protein levels at very low concentrations (0.001 µM; Figure 5A, B). There was no effect observed on CDK5 ( Figure 5D-F).
In addition to the most common neural pathways associated with AD, this study to our knowledge is the first study to investigate the effect of PFOS on non-neural pathways associated with AD, namely Apolipoprotein E (ApoE). It has been reported that ApoE4 can stimulate the phosphorylation of tau and the generation of NFTs 18 as well as facilitating the aggregation of Aβ and formation of senile plaque 9,10,11,17 . Also, a recent in vitro study indicated that GSK3β activity can be upregulated by ApoE4 through the inhibition of the wnt signaling pathway via Low-density lipoprotein-related protein-5 and -6 (LRP5 and LRP6) receptors 53 . The invitro work here showed that PFOS exposure at 0.001 µM for 24 h increased ApoE4 protein levels significantly ( Figure 6C, D). It is worth noting here that our in vivo findings did not demonstrate that PFOS exposure alters amyloidogenic and tau pathways in wild-type CD-1 mice on PND 20 (Figure 7 and 8). Similar to previous developmental behavioral study by Butenhoff et al 2009 indicated that rat offspring exposed from GD0 to PND 20 to 1 mg/kg/day PFOS didn't show any changes in learning and memory abilities on PND 20 54  Furthermore, our in vivo data showed that developmental exposure to PFOS caused a significant increase in GSK3β protein levels ( Figure 9A, B). PFOS resulted in a significant elevation of ApoE protein levels in the cerebral cortex of wild-type CD-1 mice ( Figure 10A, B). It is noteworthy here that humans express 3 isoforms of ApoE (h-ApoE) protein while mice express only one isoform (m-ApoE) 55

CONCLUSION
The results presented herein suggest that PFOS impacts amyloidogenic, tau, and kinase pathways related to AD particularly in human cells through the upregulation of ApoE4 protein levels via a mechanism associated with GSK3β. This study may contribute to understanding the mechanism by which PFOS triggers AD pathogenesis. However, further investigation is warranted such as examination of potential involvement of phosphatase enzymes, microRNA, cellular degradation pathways including insulin degrading enzymes in the mechanism of PFOS to further confirm these findings and to provide an insight into the neurodegenerative mechanisms induced by PFOS.