Amylin-mediated regulation of LRP1 by miR-103/107 induces cerebral microvascular dysfunction and impairs β-amyloid efflux

Background : The cerebral small blood vessels of individuals with Alzheimer’s Disease (AD) often have deposits of amylin, an amyloid-forming protein secreted in the blood by pancreatic β cells. To determine whether systemic pancreatic amylin dyshomeostasis impairs amyloid β (A β ) efflux across the blood-brain barrier (BBB), we studied cerebral microvessels in humans and rats with pancreatic expression of amyloidogenic human amylin, and evaluated the effect of human amylin in an in vitro BBB model. Methods : Brain sections from AD and cognitively unimpaired individuals were co-stained with anti- Aβ and anti -amylin antibodies. In vivo analyses of Aβ efflux across the BBB were carried out in aged rats that express amyloid-forming human amylin in pancreatic β -cells and littermates expressing non-amyloidogenic rat amylin. We also used an in vitro BBB model of Aβ transcytosis in which the endothelial cell monolayer was exposed to amylin-mediated stress to determine whether amylin stress downregulates LRP1, the Aβ efflux transporter . This allowed us to use pharmacology to rescue the endothelial LRP1. Results : In human AD brains, Aβ accumulation within the perivascular space frequently co-localized with deposits of amylin in the vessel wall. In rats with pancreatic expression of amyloid-forming human amylin, the high blood levels of human amylin promoted amylin deposition in brain capillaries, increased brain Aβ level, lower ed plasma-to- brain Aβ ratio and suppressed expression of LRP1 protein. In vitro BBB model experiment revealed that amylin-induced stress downregulates LRP1 in endothelial cells through a miRNA-based translational repression mechanism. Conclusions : High blood human amylin levels cause cerebral microvascular dysfunction and interfere with A β efflux across the BBB through miRNA-mediated LRP1 downregulation. Lowering the blood amylin level in early AD could improve Aβ clearance from the brain. rescue endothelial LRP1 expression. Our results provide a basis for targeting amylin-mediated cellular pathways at the blood-brain interface to reduce AD pathology.

vessels [18]. We found that human amylin-expressing rats slowly accumulate aggregated amylin in the brain microvasculature with aging (> 12-month old rats) leading to microhemorrhages [18] and late-onset behavioral changes [18,28] that are similar to those in AD rat models. At the cellular and molecular levels, accumulation of aggregated amylin in brain capillaries is associated with astrocyte activation, neuroinflammation and oxidative stress [18,28].
Cell responses to stress conditions involve reprograming gene expression through noncoding RNAs such as microRNAs (miRNAs) [30]. They inhibit protein synthesis by suppressing the translation of protein coding genes or by degrading the mRNA [30]. Paralog miRNAs miR-103 and miR-107 have previously been shown to be dysregulated in AD [31]. These miRNAs also appear to mediate stress-suppressed translation of the low-density lipoprotein receptorrelated protein 1 (LRP1) [32], an apolipoprotein E (APOE) receptor that binds and internalizes soluble Aβ at the abluminal side of the BBB [33,34,35]. Because amylin deposition in the brain microvasculature affects vascular endothelial cells (ECs) [18], we hypothesized that amylin stress dysregulates miR-103/107 impairing LRP1 synthesis and Aβ efflux across the BBB, and that antagomirs against miR-103/107 modify the amylin-mediated stress effect on LRP1.
To determine whether amylin deposition in the brain microvasculature is associated with impaired Aβ efflux, we explored amylin-Aβ interaction in the human brain microvasculature and carried out in vivo analyses of Aβ efflux across the BBB in rats that express amyloid-forming human amylin in pancreatic β-cells versus littermates that express non-amyloidogenic rat amylin.
To further define the mechanism, we used an in vitro BBB model of Aβ transcytosis in which the EC monolayer was exposed to amylin-mediated stress; antisense microRNAs were used in an attempt to rescue endothelial LRP1 expression. Our results provide a basis for targeting amylinmediated cellular pathways at the blood-brain interface to reduce AD pathology.

Methods
The aim of this study was to determine whether elevated blood levels of human amylin impairs Aβ efflux across the BBB. We studied cerebral microvessels in humans and in rats with pancreatic expression of amyloidogenic human amylin versus littermates expressing nonamyloidogenic rat amylin for histological evidence of amylin-Aβ interaction at the BBB, and evaluated the effect of human amylin on Aβ efflux in an in vitro BBB model.

Human samples
The protocol concerning the use of autopsy tissues from patients was approved by the University of Kentucky Institutional Review Board (IRB) and informed consent was obtained prospectively.
We used paraffin embedded human brain tissues (n = 9) provided by the Alzheimer's Disease Center biobank at the University of Kentucky, to explore amylin-Aβ interaction in the brain microvasculature. Formalin fixed dorsolateral frontal cortex (Brodmann area 9) tissue was used from six autopsied individuals >80 years of age at death and three age-matched cognitively unaffected (CU) individuals. The disease group included patients with AD without diabetes (n = 3) and AD with diabetes (n = 3). The absence/presence of diabetes was determined during life (at longitudinal clinical visits) by patient or caregiver self-report and the use of diabetic medications.
Neuropathological information, neuritic amyloid plaques (Consortium to Establish a Registry for Alzheimer's Disease; CERAD), Braak NFT stage and CAA severity, along with age and sex of each individual included in the present study are summarized in the Supplementary Table 1.

Experimental animals
This investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996) and was approved by the Institutional Animal Care and Use Committee at the University of Kentucky.
To study the impact of systemic pancreatic amylin dyshomeostasis on the Aβ efflux from the brain, we used rats that express human amylin in pancreatic β-cells (i.e., HIP rats) [36]. The HIP rats (non-AD rats) develop systemic amylin dyshomeostasis by ~ 10-12 months of age, which is characterized by amylin deposition in the pancreas [36] and extra-pancreatic tissues [18,25,26,27,28], including the brain microvasculature [18]. Breeding pairs were purchased from Charles River Laboratory. Wild type (WT) littermates expressing non-amyloidogenic rat amylin served as controls.

In vitro BBB model of Aβ 42 transcytosis with amylin-induced stress at the blood side
We used an in vitro BBB model described previously [37] with modifications related to amylin deposition on the EC monolayer and consequent effects on the Aβ transport across the EC monolayer. Briefly, primary rat brain microvascular endothelial cells (Cell Applications Inc) were plated on 24-well Transwell-Clear inserts with 0.4 μm pore polycarbonate membrane (Costar, Corning, NY, USA) and primary rat brain astrocytes (Sigma) were cultured at the bottom wells.
Barrier integrity was measured from the trans-endothelial electrical resistance (TEER) as described previously [37,38] using the EVOM2 meter with STX-3 electrodes (World Precision Instruments).
In the Aβ42 transcytosis experiments, the EC monolayer was treated with a medium containing human amylin (10 μM)
After 12-hours, antagomir-treated cell groups were further treated with 10 uM human amylin for 24-hours. After 36-hours of transfection, cells were harvested for Western blot analysis.

MTS cytotoxicity assay
CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS) (Promega) was used to assess cytotoxicity of aggregated amylin on the EC monolayer.
In immunofluorescence experiments, we used formalin-fixed, paraffin-embedded human brain tissue processed as previously described [18,18,39], brain capillaries from HIP and WT rats isolated as previously described [18]. and cultured ECs Fluor 405 antibody was added after staining with human amylin and collagen IV; DAPI free mounting media was used. For Thioflavin S staining, after secondary antibody incubation, brain slides were incubated in 0.5% Thioflavin S for 30 minutes at room temperature. Slides were then incubated for 3 minutes in 70% ethanol, 5 minutes in 0.2% Sudan black before washing and mounting. Immunocytochemistry was performed as described previously [39,40].

Immunoprecipitation
To immunoprecipitate rat Aβ from brain homogenates and plasma, a previously published protocol (16) was used. Briefly, 1000 μg of protein was incubated with anti-rat and human Aβ (2 µg; CST2454; Cell Signaling Technology) overnight with end-over-end rotation, at 4°C. All of the elution was used for Western blot analysis.

Western blot and enzyme-linked immunosorbent assay (ELISA)
Western blot analysis was performed on isolated brain capillaries, brain tissue homogenate and plasma from rats. Tissues were processed as described previously [16,18,28,39]. RIPA buffer with 2% SDS was used to retrieve Aβ monomers from frozen brain samples [41]. The lysate was centrifuged at 17,000 xG for 30-minnutes. The supernatant was separated from pellet after centrifugation and was then used for Western blotting. Total protein levels were estimated using a were loaded on 8% SDS-PAGE gel. Aggregated Aβ from brain homogenates were resolved in native-PAGE (non-reducing; non-denatured). Monomeric Aβ peptides were resolved in acidic Bis-Tris gel with 8M urea [35]. To enhance signal for monomeric Aβ, membranes were boiled for 3 minutes in PBS before the blocking step. LRP1 in cell and brain capillary lysates was resolved using 4-12% Bis-Tris gel under non-reducing condition. HRP-conjugated anti-rabbit or anti-mouse were secondary antibodies. Equal loading in Western blot experiments was verified by re-probing with a monoclonal anti-β actin antibody (raised in mouse, clone BA3R, Thermo Scientific; 1:2000). Protein levels were compared by densitometric analysis using ImageJ software. Levels of amylin in the rat plasma and red blood cells were measured using amylin ELISA kits (EZHA-52K, Millipore), according to the manufacturer's protocol.

Lipid peroxidation and reactive oxygen species (ROS)
Lipid peroxidation and ROS were measured in cultured rat brain microvascular ECs using previously published protocols [26,39].

Statistical analysis
Parametric comparison of two groups was done using two-tailed unpaired t-test. Welch's correction was used with t-test to account for unequal variance from unequal sample sizes, if necessary. Parametric comparisons between three groups or more were performed using one-way or two-way ANOVA with Dunnett's post hoc or Tukey's post hoc tests. Data are presented as mean ± S.E.M. Difference between groups was considered significant when P < 0.05. All analyses were performed using GraphPad Prism 8.1 software.

Aβ deposition in perivascular spaces co-occurs with amylin accumulation in vessel wall.
Co-staining of brain sections from the AD and CU individual groups with anti-amylin and anti-Aβ antibodies identified amylin and Aβ deposits in both groups, consistent with previous reports [16][17][18][19][20][21]. In AD brains, amylin immunoreactivity (brown) on the luminal side of small arterioles co-occurred frequently with patchy areas of Aβ immunoreactivity (green) within Virchow-Robin spaces ( Fig. 1a-1c). Confocal microscopy analysis of AD brain sections confirmed amylin deposition (green) on the luminal side and Aβ deposits (red) in perivascular spaces (Fig. 1d). The analysis also indicated amylin-Aβ tangled within the vessel wall (Fig. 1e). To confirm amylin accumulation within the vessel wall, we triple-stained AD brain sections for amylin, collagen and smooth muscle actin. Confocal microscopy analysis ( Fig. 1f and 1g) shows a propensity for amylin deposition at the vascular luminal side.
The results of our exploratory study in human brains revealed histological evidence of interaction between amylin secreted from the pancreas and Aβ at the blood-brain interface, in AD. Tangled amylin-Aβ deposits across cerebral blood vessel walls were identified in AD brains from individuals with AD independent of comorbid type-2 diabetes, consistent with previous studies [16]. The presence of amylin deposition at the luminal side of small blood vessels and Aβ in perivascular spaces suggest that systemic amylin dyshomeostasis may contribute to impaired Aβ efflux from the brain into the bloodstream in individuals with AD.

Tangled amylin-Aβ across the BBB impairs the Aβ efflux from the brain, in rats.
To study in vivo how systemic pancreatic amylin dyshomeostasis impairs Aβ transcytosis across the BBB, we used HIP rats that express amyloid-forming human amylin in pancreatic β-cells [36] and accumulate amylin in brain capillaries [18]. The average circulating level of amylin in 16-month old HIP rats was ~ 2-fold higher compared to that in wild type (WT) littermates ( Supplementary Fig. S1a). This was associated with amylin accumulation in small cerebral arterioles ( Supplementary Fig. S1b) and presumable capillaries (Supplementary Fig. S1c; arrows), which was not detected in WT rat brains ( Supplementary Fig. S1d).
Co-staining of brain slices from HIP and WT rat brains with anti-Aβ (green) and antiamylin (brown) antibodies showed vascular amylin-Aβ interaction in HIP, but not WT rats (  S2a). In HIP rat brains, amylin immunoreactivity was detected solely on the luminal side of the blood vessel, whereas those of Aβ were seen within perivascular, Virchow-Robin spaces (Fig.   2a) and within the blood vessel wall (Fig. 2b). In addition, brain slices from HIP rats had sporadic amylin-Aβ deposits ( Fig. 2c; circle) that were seen in association with capillaries positive for luminal amylin accumulation and for Aβ deposition within the surrounding tissue ( Fig. 2c; arrows), consistent with our findings in human AD brains (Fig. 1a-c). Scattered Aβ immunoreactivity was also detected in HIP rat brains ( Supplementary Fig. S2b), but not in brains of WT littermates ( Supplementary Fig. S2c).
Western blot analysis of HIP rat brain homogenates shows accumulation of Aβ in the brain ( Fig. 2f and Supplementary Fig. S2d). Rat Aβ40 peptide and brain homogenate from a 12month old APP/PS1 rat were the positive controls for Aβ accumulation. To test whether both soluble and insoluble aggregated Aβ accumulated in HIP rat brains, we performed native-PAGE followed by Western blot (Fig. 2g). The levels of soluble and insoluble Aβ aggregates were higher in HIP rat brains compared to those in WT littermates.
AD model rats are genetically determined to develop brain Aβ pathology, whereas rats expressing human amylin in the pancreatic islets may accumulate Aβ in the brain due to changes associated with chronically elevated blood levels of human amylin. To test whether the Aβ efflux from the brain to bloodstream is altered in HIP vs. WT rats, we used immunoprecipitation to enrich Aβ in plasma samples and brain homogenates from age-matched rats in the two groups followed by Western blot analysis of Aβ ( Fig. 3a and 3b). The ratio of plasma-to-brain Aβ levels was lower in HIP compared to WT rats (Fig. 3c), which suggests that Aβ efflux across the BBB is impaired in HIP rats.

High blood human amylin suppresses the Aβ efflux transporter expression.
Amylin deposition in the brain microvasculature may induce stress in ECs and decline of the Aβ efflux transporter LRP1 expression. To test this hypothesis, we analyzed LRP1 protein expression in brain capillary lysates from aged HIP rats vs. WT littermates and EC lysates from EC monolayers that were subjected to amylin-induced stress.
Vascular amylin-induced LRP1 downregulation in the brain endothelium. Brain capillaries were isolated from HIP and WT rats and tested for the presence of amylin deposition and LRP1 protein expression by immunofluorescence and Western blot. Confocal microscopy analysis of isolated brain capillaries (Fig. 3d) showed that amylin deposition (green) co-localized with caveolin-1 (red), a protein that is abundant in ECs and further confirmed amylin deposition in HIP brain capillaries (Fig. 3e). Staining for LRP1 revealed lower LRP1 immunoreactivity signal in brain capillaries from HIP rats compared to WT littermates ( Fig. 3f-g). Consistent with this result, Western blot analysis showed reduced LRP1 protein levels in brain capillary lysates from HIP rats compared to those in WT littermates ( Fig. 3h and 3i). In contrast to protein expression, LRP1 mRNA levels were increased in HIP vs. WT rat brain capillaries (Fig. 3j), suggesting that the decrease in LRP1 protein in HIP rat capillary ECs occurs at a posttranscriptional level.

Amylin-induced LRP1 downregulation in endothelial cells, in vitro.
To further evaluate the relationship between vascular amylin deposition and LRP1 protein expression, we incubated rat brain microvascular ECs with various concentrations of human amylin for 24 hours followed by analysis of LRP1 protein expression by Western blot (Fig. 4a). LRP1 protein levels decreased with increasing concentrations of human amylin; LRP1 expression was reduced by more than 50% in ECs incubated with 10 µM human amylin. This result was further confirmed by immunofluorescence measurements in ECs incubated for 24-hours with 10 µM human amylin ( Fig. 4b). In contrast, non-amyloidogenic rat amylin (10 µM; 24-hour incubation time) had no effect on LRP1 protein levels ( Fig. 4b-d), as indicated by analyses of immunoreactivity by confocal microscopy (Fig. 4b) and Western blot (Fig. 4c). Viability of the ECs was not affected by incubation with human amylin (Supplementary Fig. S3), indicating that decreased LRP1 protein expression is not due to cell death. Consistent with this result, the capacity of ECs to induce transcript expression of LRP1 was not affected by amylin stress; consistent with the findings in HIP rat cerebral capillaries, LRP1 mRNA levels were greatly elevated in ECs incubated with human amylin vs. control cells and ECs incubated with rat amylin (Fig. 4d).

Impaired endothelial Aβ transcytosis by amylin in a 3-dimensional BBB model.
To determine whether the amylin stress-induced LRP1 downregulation affects Aβ transcytosis, we employed a well-established model of BBB [37] in which the EC monolayer was exposed to human amylin on the luminal side (as shown in Fig. 4e). The BBB model was tested for monolayer formation and tightness by measuring TEER (Fig. 4f). All experiments were done with a fully formed EC monolayer characterized by the maximum TEER = 110 ± 5 Ω/cm 2 .
First, we tested the effect of human amylin on EC monolayer structural integrity. We then measured Aβ transcytosis across the BBB using FAM tagged Aβ42 (Aβ42-FAM) and FITC-dextran as a paracellular diffusion marker. TEER, cell morphology within the EC monolayer and permeability to FITC-dextran (4 kDa) were measured following the incubation of the ECs for 24hours with 10 μM human amylin or similar concentrations of rat amylin or vehicle ( Supplementary Fig. S4a-S4e). Although TEER decreased following the treatment with human amylin (Supplementary Fig. S4b), neither the intercellular space, as assessed from FITC-dextran permeability ( Supplementary Fig. S4c-d), nor EC monolayer morphology (Supplementary Fig.   S4E) were significantly altered by the treatment with human amylin. To assess the impact of amyloid-forming amylin on Aβ transcytosis across the BBB, human amylin or vehicle were applied at the luminal (blood) side for 24-hours followed by washing of the EC monolayers with PBS and application of FITC-Dextran or Aβ42-FAM at abluminal (brain) side of the BBB for 1hour. The amounts of Aβ42-FAM and FITC-Dextran that cross the monolayer were estimated from the fluorescence intensity in the medium samples collected from luminal side and used to calculate the Aβ transcytosis quotient [38]. Amylin-pretreatment reduced the Aβ transcytosis quotient by 20 ± 5% (P < 0.05) (Fig. 4g), which indicates that amylin blocks transcytosis of Aβ42 across the in vitro BBB.
From these results (Fig. 4), we infer that amylin accumulation at the luminal side of the BBB impairs Aβ transcytosis across the ECs via LRP1 downregulation. As the LRP1 transcript was upregulated in response to amylin effects on ECs both in vivo (Fig. 3j) and ex vivo (Fig. 4d), this suggests that LRP1 protein downregulation may involve altered protein translation.

Aβ efflux transporter expression is suppressed by amylin-induced endothelial cell stress.
Paralog miRNAs miR-103 and miR-107 are upregulated by oxidative stress [42] and repress LRP1 translation in several cell lines [32]. Thus, we investigated if these miRNAs are involved in amylin-induced LRP1 downregulation in the BBB.
Amylin accumulation in brain capillaries induced oxidative stress in ECs by forming deposits with biochemical properties of amyloid (Fig. 5a), which was shown to alter structural stability of the cellular membranes [25,26,39]. This is evidenced by accumulation of the lipid peroxidation marker 4-hydroxynonenal (4-HNE) (Fig. 5b). Oxidative stress also occurred in rat brain microvascular ECs incubated with human amylin (10 μM amylin; 24-hour incubation time), as indicated by lipid peroxidation of the EC membranes (Fig. 5c-e) and increased generation of ROS ( Fig. 5f and 5g). Pancreas from a HIP rat was the positive control ( Supplementary Fig. S5a) and pancreas tissue from an AKO rat was the negative control ( Supplementary Fig. S5b) for amylin amyloid. The amylin stress on ECs was associated with elevated levels of miR-103 and miR-107 (Fig. 5h). Brain capillary lysates from HIP rats also had elevated miR-103 and miR-107 levels compared to those in WT littermates (Fig. 5i).
The results support the hypothesis that miRNA is upregulatied by amylin-mediated endothelial stress and suggest that additional pathways may compromise the capability of ECs to express LRP1, which were not linked to peroxidative membrane injury.

Antisense microRNAs rescue amylin-induced suppression of Aβ efflux transporter.
TargetScan predicts that miR-103 and miR-107 bind directly to LRP1, with the biding site located at the 3'UTR region of rat LRP1 (Fig 6a). To further determine whether high miR-103 and miR-107 levels suppress LRP1 translation, we co-transfected miR-103 and miR-107 mimics (100 nM) into rat brain microvascular ECs. Cell lysates were tested after 24-hours for LRP1 protein expression by Western blot. The average LRP1 expression level was lower in ECs cotransfected with miR-103 and miR-107 mimics compared to miR-control (Fig. 6b).
These results indicate that: 1, endothelial LRP1 downregulation associated with amylin stress is a miRNA-based translational repression mechanism; and 2, LRP1 downregulation by amylin-induced stress on ECs can be reversed by modulating miR-103 and miR-107.

Discussion
By investigating brain tissues from humans with AD and from non-AD rats with chronically elevated blood levels of human amylin, we found that Aβ clustering in the cerebral capillary bed consistently co-localized with luminal accumulation of amylin amyloid secreted by pancreatic βcells. We then used aged rats with pancreatic expression of amyloid-forming human amylin vs.
littermates expressing non-amyloidogenic rat amylin and found that chronically elevated blood levels of amyloid-forming human amylin results in tangled amylin-Aβ across the BBB and endothelial transcytosis dysfunction (see Fig. 6d for the proposed mechanism). These results suggest that LRP1-mediated Aβ efflux could be affected (or restored) through cellular pathways modulated from the luminal BBB side.
Although APOE/LRP1-regulated metabolic pathways have a well-established role in Aβ clearance [44], a large fraction of Aβ fragments is eliminated from the central nervous system through transcytosis across the BBB, as shown by kinetic studies in humans [45] and in mice [33,35]. This pathway of Aβ clearance from the brain is reduced early in the onset of AD causing Aβ deposition in capillary beds [46]. In contrast to rats expressing endogenous nonamyloidogenic rat amylin that had unimpaired Aβ efflux across the BBB, human amylinexpressing rats showed amylin amyloid deposition in the brain microvasculature and lower Aβ efflux across the BBB. This was caused by miRNA-based translational repression of LRP1 and was reversed by antisense microRNA. Our results suggest that miRNA-mediated suppression of endothelial LRP1 may be a mechanism by which amylin accumulation in the brain microvasculature reprogram ECs toward an AD phenotype.
Because amylin is co-secreted with insulin by pancreatic β-cells [22], conditions associated with hyperinsulinemia and prediabetic insulin resistance always coincide with hypersecretion of amylin. Patients with type-2 diabetes accumulate amylin amyloid in pancreatic islets that causes oxidative stress and inflammation leading to cell apoptosis and depletion of βcell mass [22]. We reported the presence of aggregated amylin with similar molecular weights in matched plasma and brain homogenate from humans with AD [16], and suggested the hypothesis that toxic aggregated amylin circulates in the blood in humans with AD. Soluble aggregated amylin is endocytosed into the cells through macropinocytosis or caveolae-mediated pathways and degraded in the acidic compartment of lysosomes [47]. In human amylin-expressing rats, we found amylin deposition in the brain microvasculature, consistent with amylin deposition in brains of humans with AD [16,17,18,19,20]. From these results, we infer that chronically elevated blood amylin overwhelms the EC capacity to clear aggregated amylin through endocytosis letting the BBB vulnerable to aggregated amylin toxicity.
MiR-103 and miR-107 are upregulated by oxidative stress [42] and repress LRP1 translation in vitro [32]. Several human diseases [30,31,48,49], including AD [31] and type-2 diabetes [49] are associated with altered miR-103/107. Our results show that, although amylin deposition in the brain microvasculature in HIP rats causes stress on the brain endothelium, the brain capillary ECs can induce transcript expression of LRP1; however, a miRNA feedback loop prevents the translation of LRP1 transcripts. In support to this proposed mechanism (Fig. 6d), we found evidence of dose-dependent regulatory mechanism of LRP1 expression that occurred at the post-transcriptional level in ECs incubated with amyloid-forming human amylin (but not non-amyloidogenic rat amylin) and by miR-103 and 107 transfections in ECs. In future studies, it would be interesting to determine whether miRNA-mediated suppression of LRP1 show an inverse expression pattern with lowering blood amylin concentration.
Finding Aβ accumulation in the brains of non-AD rats that express human amylin in pancreatic islets was unanticipated. AD model rats are genetically determined to develop brain Aβ pathology, whereas rats expressing human amylin in the pancreas accumulate Aβ in the brain in association with amylin deposition in the brain microvasculature and impairment of LRP1medated Aβ efflux across the BBB. Our results suggest that rats "humanized" for amylin expression may serve as a model for the amylin contribution to late-onset AD pathology. LRP1 downregulation in brain capillaries is common in aging [50] and further accelerated in AD [33,34,35]. We showed that changing blood components in rats by pancreatic expression of human amylin provoked miRNA-based LRP1 downregulation in brain capillary endothelium, which suggests that high circulating human amylin is a specific stressor on the luminal side of the BBB. Thus, lowering blood amylin levels or/and inhibiting miRNA-mediated LRP1 downregulation caused by amylin stress on the BBB could improve Aβ clearance from the brain and reduce AD pathology. Finding that endothelial LRP1 expression/ function can be restored via cellular pathways that are modulated from the luminal side of the BBB has the potential to change the current paradigms in the treatment of AD.

Conclusions
While the association between dysregulation of circulating amylin and AD pathology is not yet as widely studied as other aspects of AD biology, accumulating evidence from multiple research teams describe amylin-Aβ cross-seeding in brains of individuals with sporadic AD [16][17][18][19][20][21]. Here, we show that high blood human amylin levels cause cerebral microvascular dysfunction and interfere with Aβ efflux across the BBB through miRNA-mediated LRP1 downregulation. This study constitutes a basis for additional hypothesis driven research and evidence-based validation studies to help understanding what triggers amylin deposition in the cerebral microvasculature and whether lowering the blood amylin level in early AD could improve Aβ clearance from the brain and may reduce the progression of AD pathology.

Ethics approval and consent to participate
The protocol concerning the use of autopsy tissues from patients was approved by the University of Kentucky Institutional Review Board (IRB) and informed consent was obtained prospectively.

Consent for publication
Not applicable

Competing interests
The authors declare that no conflict of interest exists.

Fig. 1: Amylin-Aβ interaction at the blood-brain interface in human AD brains. a, b
Representative immunohistochemical (IHC) micrographs of brain sections from patients with sporadic AD (a) and cognitively unimpaired (CU) (b) individuals that were co-stained with antiamylin (brown) and anti-Aβ (green) antibodies. c Analysis of the number of blood vessels with vascular mixed amylin-Aβ deposition (per mm 2 ) in brains from AD (n = 6) and CU (n = 3) (2 slides/brain) groups assessed from IHC. d, e Representative images of confocal microscopy analysis of brain sections from the same AD individuals (n = 6; 2 slides/brain) as in above showing amylin (green) and Aβ (red) deposits in small-type vessels in which amylin is present on the luminal side (d) and within the vessel wall (e). e-g Representative immunofluorescence images of AD brain sections triple stained with anti-amylin (green), anti-collagen IV (red) and anti-smooth muscle actin (blue) (n = 3; 2 slides/brain). Scale bars, 50 µm. Data are mean ± SEM. P ≤ 0.05 *; by two-tailed, unpaired t test (c).

Fig. 2: Pancreatic amylin accumulates in the brain microvasculature and impairs Aβ efflux
from the brain. a-d Representative IHC micrographs of brain sections from HIP (a-c) and WT (d) rats co-stained with anti-amylin (brown) and anti-Aβ (green) antibodies. e Analysis of the number of blood vessels with vascular amylin-Aβ deposition (per mm 2 ) in brains of WT and HIP rats (n = 5/group) (3 slides/brain), assessed from IHC. f Representative Western blot and densitometry quantification of Aβ in brain homogenates from HIP rats and WT littermates using acidic urea gel (n = 3/group) to resolve monomers. Rat Aβ40 peptide and APP/PS1 rat brain homogenate were used as positive controls. Aβ densitometry was normalized to loading control actin. Aβ densitometry quantification was calculated from two experiments (n = 7-8/group). g Representative Western blot and densitometry analyses of aggregated Aβ in brain homogenates from HIP rats and WT littermates (n = 7/group) using native-PAGE. Scale bars, 50 µm (a-d).

Fig. 3: High blood amylin levels downregulate the Aβ efflux transporter at the BBB. a-b
Immunoprecipitation and western blot analyses of Aβ in the plasma (a) and brain homogenates (b) from WT and HIP rats (n = 5/group). Age-matched APP/PS1 rat brain homogenate were used as a positive control. c Ratio of plasma Aβ-to-brain Aβ levels in HIP and WT rats assessed from Western blot analysis of Aβ enriched by immunoprecipitation from plasma (a) and brain (b) homogenates. d Confocal fluorescent micrographs of amylin (green) and endothelial cell marker caveolin-1 (red) in brain capillaries isolated from HIP rats and WT littermates. e Levels of amylin fluorescence intensity signals quantified from n = 21 and n = 47 isolated capillaries from WT rats and HIP littermates (n = 3 rats/group), respectively, and averaged for each rat. f Confocal immunofluorescent micrographs of LRP1 (green) and nuclei (blue) in isolated capillaries from WT rats and HIP littermates. g Levels of LRP1 fluorescence intensity signals quantified from WT rats (n = 4) and HIP littermates (n =4) and averaged for each rat. h Representative western blot and densitometry quantification (i) of LRP1 in brain capillary lysates isolated from HIP rats and WT littermates (n = 3 rats/group). j LRP1 mRNA levels (fold difference using 2 -ΔΔCt method) in brain capillary lysates isolated from the same HIP and WT rats as in (h) (n = 3 rats/group). Scale bars, 10 µm (d, e). Data are mean ± SEM. P ≤ 0.05 *, P ≤ 0.01 ** by two-tailed, unpaired t test (c, e, g, i, j).