PSEN1, but not PSEN2, mutation in familial form of Alzheimer’s disease is associated with impaired barrier phenotype in a stem cell-based model of the blood-barrier

Background: Alzheimer’s disease (AD) is the most common form of neurodegenerative disease. It is an irreversible condition marked by irreversible cognitive loss, commonly attributed by the loss of hippocampal neurons due to the formation of senile plaques and neurofibrillary tangles. Although the sporadic form is the most prevalent, the presence of familial form (involving several genes such as APP, PSEN1 and PSEN2) of the disease is commonly used as a model for understanding the pathophysiology of the disease. The aim of this study is to investigate the effect of mutation on PSEN1 and PSEN2 genes on the BBB function using induced pluripotent stem cells. Methods: iPSC lines from patients harboring mutations in PSEN1 or PSEN2 were used in this study and compared to a control iPSC line. Cells were differentiated into BMECs following existing protocols. Barrier function was assessed by measuring TEER and fluorescein permeability, drug transporters activity was assessed by uptake assay, glucose uptake and metabolism assessed by cell flux analyzer, mitochondrial potential by JC-1 and lysosomal acidification by acridine orange. Results: PSEN1-BMECs, but not PSEN2-BMECs, showed impaired barrier function compared to control group. Such impaired barrier function correlated with poor tight junction complexes and reduced drug efflux pump activity. In addition, both PSEN1 and PSEN2 displayed reduced glucose uptake and glycolysis, as well as impaired mitochondrial membrane potential and lysosomal acidification. Conclusion: Our study reports evidence that PSEN1 and PSEN2 mutations, two genes commonly associated with familial form of Alzheimer’s disease, can impair the development and the maintenance of the BBB, both by an impairment of the barrier function, vesicle trafficking and bioenergetics. Therefore, assessing the contribution of genetic mutations associated with Alzheimer’s disease will allow us to better understand detailed analysis of the glycolytic stress assay showed a decrease in glycolysis and non-glycolytic activity only in PSEN1-BMECs, whereas a decreased glycolytic capacity and glycolytic reserve were observed in both PSEN-BMECs.

3 the contribution of the BBB in dementia, but also in other neurodegenerative diseases.

Background
Alzheimer's disease (AD) is the most common neurodegenerative disorder accounting as the 6th cause of death in the United States. It is a fatal form of dementia that is progressive, irreversible, and a growing public health concern. It is estimated that over 10% of the US senior population is diagonosed with AD (1). It is characterized by the formation of senile plaques (2)(3)(4)(5).
Despite the important effort aimed to find a cure for such disease, there is still no cure for the disase (6), with the translation from pre-clinical models to clinically relevant candidates being an important pitfall. Therefore, a shift of paradigm from a neuron-centric to non-neuronal components is necessary to identify novel therapeutic targets.
Yet, the use of post-mortem tissue from AD patients strictly limits the ability to document dysfunctional signaling pathways at the BBB, whereas current in vitro studies using Aβ to assess its neurotoxicity at the BBB requires the use of amount not reflective of cocnentrations observed in vivo. Thus, assessing the contribution of genes associated with AD could provide a better understanding of AD pathophysiology at the neurovascular unit. The aim of this study is to document the effect of mutations in PSEN1 and PSEN2 genes on the barrier function using iPSCs derived from patients suffering from FAD (28)(29)(30). Using the differentiation protocol initially developed by Shusta and colleagues (31,32), this study investigated the effect of mutations on PSEN1 and PSEN2 genes on the barrier phenotype in iPSC-derived brain microvascular cells (BMECs).

Materials And Methods
Cell culture and iPSC differentiation Control (CS06iCTR), PSEN1 (CS40iFAD) and PSEN2 (CS08iFAD) iPSC lines used in this study were acquired from the Cedars-Sinai iPSC core (Los Angeles, CA). The PSEN1 iPSC line was isolated from a 56-year old Caucasian male diagnosed with memory impairment and harbors a Ala246Glu mutation. The PSEN2 iPSC line was isolated from a 81-year old Caucasian female diagnosed with progressive dementia and harbors a Asn141Ile mutation.

PSEN1-BMECs but not PSEN2-BMECs showed impaired barrier function
Assessment of the barrier function in these iPSC-derived BMECs monolayers was the first outcome investigated in this study (Fig. 1). PSEN1-BMECs displayed a lower immunoreactivity to claudin-5 and occludin (Fig. 1A), such lower immunoreactivity was further confirmed by quantification of protein expression by fluorescence intensity (Fig. 1B). PSEN1-BMECs displayed overall lower expression of these two tight junction (TJ) proteins, whereas PSEN2-BMECs showed no differences to the control iPSC-line.
Next, changes in the barrier function was assessed by measuring changes in TEER and fluorescein permeability (Fig. 1C&D). As expected, PSEN1-BMECs showed impaired barrier function compared to the two other iPSC lines, as a significantly lower TEER ( ̴ 150 Ω.cm 2 ) and higher fluorescein permeability.
Such impaired phenotype appeared not limited to BMECs, as iPSC-derived neurons originated from PSEN1 iPSCs displayed an impaired formation of maturing neurons as represented by formation of neurites compared to control and PSEN2-neurons ( Supplementary Fig. 1). Taken together, PSEN1 but not PSEN2, seems to impair the formation of tight BMECs monolayers.

PSEN1-BMECs have impaired drug efflux pumps activity
To further investigate the effect of mutations on PSEN1 and PSEN2 genes, changes in drug efflux pump transporters expression and activity were assessed in iPSC-derived BMECs (Fig. 2). No significant differences were observed in the expression of major drug efflux transporters were observed, although P-gp and MRP1 expression appeared lower in PSEN1-BMECs. Such decreased expression correlated with a decreased activity of both pumps. PSEN1-BMECs showed higher drug uptake levels for rhodamine-123 (Fig. 2B, a P-gp substrate) and DCFDA (Fig. 2C, an MRPs substrate), and confirmed by the relative absence of change in cellular uptake following treatment with cyclosporine A (CsA, a P-gp inhibitor) or MK571 (a pan-MRP inhibitor). In contrast, no significant differences were observed in regards of BCRP activity. In conclusion, mutation in the PSEN1 gene may impair the activity of certain drug efflux transporters. PSEN1 and PSEN2-BMECs display impaired glucose uptake and metabolism Next, changes in glucose uptake and metabolism between iPSC lines was assessed ( Fig. 3). No significant changes in glucose transporter isoforms (GLUT1, GLUT3, GLUT4) at the BBB were observed (Fig. 3A). However, PSEN1-derived BMECs showed a lower glucose uptake (Fig. 3B) compared to controls and PSEN2-BMECs. Although PSEN2-BMECs showed no significantly lowe glucose uptake than controls, these cells still showed a slight decrease compared to control. In addition, both PSEN1-BMECs and PSEN2-BMECs failed to show inhibition of glucose uptake following treatment with glucose transporter inhibitor II (GTI). Notably, similar pattern was observed with iPSC-astrocytes ( Supplementary Fig. 2), as PSEN1-astrocytes showed lower glucose uptake than control-astrocytes. However, all three groups showed significant decrease in glucose uptake following treatment with GTI.
To investigate the impact of such impaired glucose uptake on the cell metabolism, we investigated changes in glycolysis in iPSC-derived BMECs using a cell flux analyzer ( Fig. 3D-F). Both PSEN1-BMECs and PSEN2-BMECs showed a basal extracellular acidification rate (ECAR) compared to control-BMECs. Both PSEN-BMECs showed a metabolic phenotype considered "quiescent", compared to a "glycolytic" phenotype observed with control-BMECs. A detailed analysis of the glycolytic stress assay showed a decrease in glycolysis and non-glycolytic activity only in PSEN1-BMECs, whereas a decreased glycolytic capacity and glycolytic reserve were observed in both PSEN-BMECs.
In conclusion, mutations in PSEN genes maybe impairing glucose uptake and metabolism at the BBB.

PSEN1 mutation impairs mitochondrial and lysosomal acidification
Finally, we investigated changes in cell metabolic activity and investigated the effect of PSEN on mitochondrial and lysosomal function (Fig. 4). Firstly, changes in cell metabolic activity was assessed in iPSC-BMECs monolayers using an MTS assay (Fig. 4A).
Interestingly, PSEN1-BMECs showed a higher cell metabolic activity than control and PSEN2-BMECs. Similar observation was done in iPSC-derived neurons ( Supplementary   Fig. 2B), whereas iPSC-derived astrocytes showed no differences in cell metabolic activity.
To confirm if such differences in cell metabolic activity were due to changes in mitochondrial potential, changes in JC-1 fluorescence (Fig. 4B). Under resting condition, control-BMECs showed a majority of cell events in high intensity red fluorescence. investigating the contribution of genetic mutations associated with AD (e.g. APP, PSEN1, PSEN2…) on the BBB dysfunction remains anectodal. In this study, we investigated the effect of mutations in PSEN1 and PSEN2 on the BBB function, using iPSCs obtained from patients suffering from FAD. This study identified PSEN1, rather than PSEN2, as a possible gene important in the formation and maintenance of the BBB, as PSEN1-BMECs showed an overall worsened outcome than PSEN2-BMECs including formation of the BBB phenotype, glucose metabolism and mitochondrial function. Interestingly, such results correlated with previous findings reported by Searson and colleagues using an iPSC line sharing the same mutation in PSEN1 (35), as we both observed that PSEN1-BMECs BMECs showed poor barrier function (as measured by TEER and permeability with a paracellular marker), as well as a compromised activity. Furthermore, our group showed that MRP-mediated efflux in PSEN1-BMECs was also affected, as well as glucose metabolism (glucose uptake, glycolysis), as well as glucose metabolism, mitochondrial function and lysosomal acidification. Our study also suggests that this effect maybe restricted to PSEN1, as the phenotype observed with PSEN2 mutant was milder, although signs of impaired glucose metabolism, mitochondrial function and lysosomal acidification were reported by our group. A limitation of our approach is the limited number of iPSC lines available from patients with FAD at the time of publication. Hence, future studies are aimed to investigate and confirm our observation by the inclusion of additional cell lines from patients with mutations in APP, PSEN1 and PSEN2 respectively.
A particular feature observed in our study was the lower glucose uptake in both PSEN1 and PSEN2 iPSC lines compared to control. Such lower uptake was accompanied by a lack of response to GLUT inhibition by GTI, and by a much lower ECAR values and glycolytic capacity compared to controls. Although the overall expression of GLUT1 appeared unchanged, we cannot exclude a possible impaired GLUT1 activity due to instrisic factor. GLUT1 has been documented to have a particular interactions with Aβ, as a recent study by Zlokovic and colleagues reported a worsened outcome in AD transgenic mice crossed with Slc2a1+/-deficient mice (22). Hence, our future direction will be to further investigate the relationships and interactions between Aβ peptides and GLUT1.
The effect of PSEN1 and PSEN2 on the BBB maturation and maintenance is intruiging. Both proteins are known to be part of the γ-secretase complex, which ultimately drives the formation of Aβ peptides. In addition, a survey of the literature also identified γ-secretase as an important modulator of the WNT signaling pathway (36,37). WNT signaling is an important pathway involved in the development and maintenance of the BBB (38,39). At this point, we cannot restrict and determine if the impairment of the BBB by PSEN1 is driven by an increase in Aβ production, or by an impairment of the endogenous WNT signaling. A limitation of our study is the absence of documentation of Aβ1-40 and Aβ1-42 production by our BMECs monolayers. Assessing differential secretion of these Aβ peptides between PSEN mutants and control iPSC lines could help us better understand the contribution of each of these pathways on the BBB. Finally, we have reported an impaired mitochondrial function (as seen by JC-1 staining) and lysosomal acidification. These are two components playing essential roles in the maintenance of energy homeostasis as well as vesicular trafficking. These two features remains largely undocumented at the BBB despite their important contribution in neurological diseases. Thus, a better understanding on how PSENs impact these pathways may increase interests in understanding the contribution of these pathways on the BBB dysfunction during neurological diseases.

Conclusion
In conclusion, this study suggests the importance of PSEN1 on the BBB development and maturation, as mutation in PSEN1 appears to have detrimental effect on the BBB function.
Such study raises the importance to investigate the contribution of genetic disorders at the BBB, and the possible inclusion of a dysfunctional BBB in the pathophysiology of the disease.

Declarations
Availability of data and materials: Data are available upon reques to the corresponding author.

Supplementary Figures
Supplementary Figure 1: Phenotype of iPSC-derived neurons differentiated from the iPSC lines.
Cells were differentiated into neurons following existing protocol (33,40). Neurons were stained against nestin (red), bIII-tubulin. DAPI was used as nuclear counterstaining.   Representative ECAR diagram following treatment with various inhibitor. Cells were incubated for 2 hours in glucose-free medium prior onset of experiment. Cells were maintained in medium with L-glutamine, and subsequently given 10mM D-glucose, followed by incubation with 1µM oligomycin (mitochondria respiratory chain inhibitor) and 2deoxyglucose (100mM). (D) Energy consumption profile of iPSC-BMECs. OCR denotes oxygen consumption rate, ECAR denotes extracellular acidification rate. Note the shift of metabolic activity from "glycolytic" to "quiescent" phenotype. (E) Glycolytic parameters extrapolated.

Supplementary
Noted the lower glycolytic capacity and reserve in PSEN1 and PSEN2 iPSC-BMECss compared to control iPSC-BMECs, whereas PSEN1 showed a lower glycolysis and non-glycolytic acidification rate. N=3/group, * denotes P<0.05 versus control group, # denotes P<0.05 versus non-inhibited group.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download. FBCNS_FAD_BBB_R0_Suppl_Figures.pdf