Age-Dependent Changes in the Plasma and Brain Pharmacokinetics of Amyloid-β Peptides and Insulin

Abstract

Background:

Age is the most common risk factor for Alzheimer’s disease (AD), a neurodegenerative disorder characterized by the hallmarks of toxic amyloid-β (Aβ) plaques and hyperphosphorylated tau tangles. Moreover, sub-physiological brain insulin levels have emerged as a pathological manifestation of AD.

Objective:

Identify age-related changes in the plasma disposition and blood-brain barrier (BBB) trafficking of Aβ peptides and insulin in mice.

Methods:

Upon systemic injection of ¹²⁵I-Aβ₄₀, ¹²⁵I-Aβ₄₂, or ¹²⁵I-insulin, the plasma pharmacokinetics and brain influx were assessed in wild-type (WT) or AD transgenic (APP/PS1) mice at various ages. Additionally, publicly available single-cell RNA-Seq data [GSE129788] was employed to investigate pathways regulating BBB transport in WT mice at different ages.

Results:

The brain influx of ¹²⁵I-Aβ₄₀, estimated as the permeability-surface area product, decreased with age, accompanied by an increase in plasma AUC. In contrast, the brain influx of ¹²⁵I-Aβ₄₂ increased with age, accompanied by a decrease in plasma AUC. The age-dependent changes observed in WT mice were accelerated in APP/PS1 mice. As seen with ¹²⁵I-Aβ₄₀, the brain influx of ¹²⁵I-insulin decreased with age in WT mice, accompanied by an increase in plasma AUC. This finding was further supported by dynamic single-photon emission computed tomography (SPECT/CT) imaging studies. RAGE and PI3K/AKT signaling pathways at the BBB, which are implicated in Aβ and insulin transcytosis, respectively, were upregulated with age in WT mice, indicating BBB insulin resistance.

Conclusion:

Aging differentially affects the plasma pharmacokinetics and brain influx of Aβ isoforms and insulin in a manner that could potentially augment AD risk.

INTRODUCTION

The blood-brain barrier (BBB) is responsible for maintaining the dynamic equilibrium between endogenous solute levels in the blood and the brain. The BBB endothelium serves dual functions, acting as both a gatekeeper to restrict the uptake of toxic substances circulating in the blood, and also as a trafficking portal to mediate the highly-selective delivery of essential nutrients to the brain. Additionally, the BBB plays a crucial role in clearing metabolic waste products from the brain. Disruptions to these spatially coordinated trafficking events have been implicated in the etiology of several neurodegenerative disorders, including sporadic Alzheimer’s disease (AD). Progression of physiological age is widely recognized as the greatest risk factor for AD [1]. While normal aging is associated with gradual loss of BBB function [2] and appearance of neuropathological changes, these are accelerated in AD brain and trigger catastrophic neurocognitive changes [3].

One major consequence of age-related BBB dysfunction in AD brain is the increased accumulation of amyloid-β (Aβ) peptides in the cerebral vasculature as amyloid deposits and in the brain parenchyma as senile plaques [4]. Of the two major Aβ isoforms that predominate in AD brain, Aβ₄₀ is the major isoform present in cerebrovascular amyloid deposits, whereas Aβ₄₂ is the major isoform present in parenchymal plaques. Notably, Aβ₄₂ is considered to be more neurotoxic and amyloidogenic than Aβ₄₀ [5]. Interestingly, Aβ₄₀ is also suggested to be protective against Aβ₄₂-induced neurotoxicity [6, 7]. However, changes in how the BBB handles each Aβ isoform during aging and AD remain unclear. Brain Aβ deposition precedes cognitive decline and continues to increase substantially over the course of AD progression. It has been estimated that 20–30% of all cognitively-normal older adults display significant Aβ deposition in the brain [4]. It is well established that decreased Aβ clearance from the brain promotes Aβ accumulation and plaque formation during aging and AD [8]. In contrast, very little is known about how alterations in Aβ handling by the periphery contribute to brain Aβ deposition in AD. Although the Aβ concentrations in plasma are typically ~ 6-fold lower than the concentrations in the brain interstitial fluid, the absolute amount of Aβ in plasma is estimated to be ~ 10-fold greater than the amount of soluble Aβ in the brain [9]. In aged nonhuman primates, it was shown that systemically injected Aβ efficiently crossed the BBB and became incorporated into existing parenchymal plaques [10]. Further, increased brain Aβ deposition in aged rats was associated with increased BBB expression of the receptor for advanced glycation end products (RAGE), which handles Aβ influx to the brain [11]. Elevated serum Aβ levels are also known to increase the risk of AD in elderly patients [12]. Therefore, it is likely that systemic Aβ can directly, by trafficking Aβ into the brain, or indirectly, by reducing the ability of BBB to clear brain Aβ, contribute to the brain Aβ load.

BBB dysfunction in aging and AD is associated with decreased influx of essential nutrients to the brain [13]. In particular, glucose and insulin are produced almost exclusively in peripheral tissues and must be delivered to the brain across the BBB. Decreased insulin levels in the cerebrospinal fluid are associated with progression of cognitive decline in early-stage AD patients [14]. In another study, insulin concentrations and insulin receptor (IR) density in postmortem brain decreased with age in subjects with or without AD [15]. We and others have demonstrated that Aβ peptide exposure interferes with the neurobiological functions mediated by insulin and IR [16–19]. Further, we reported that insulin modulates Aβ peptide transport at the BBB in mice, as well as the membrane expression of Aβ receptors (RAGE, LRP-1) in a BBB cell culture model [20]. Thus, the effects of insulin and Aβ on the BBB are strongly interconnected, and disruptions in this relationship are expected to promote BBB dysfunction in AD.

To investigate the contributions of aging to BBB dysfunction, we examined the plasma distribution and brain influx kinetics of ¹²⁵I radiolabeled Aβ₄₀, Aβ₄₂, and insulin in female wild-type (WT) and/or AD transgenic (APP/PS1) mice at various ages. Using single-cell RNA-Seq data previously published by Ximerakis et al. [21], we further assessed the activity of various transport/signaling pathways in brain endothelial cells obtained from WT mice at different ages. Our findings reveal an intricate switch in the brain influx of Aβ isoforms and insulin with age, accompanied by disruptions to transport/signaling pathways at the BBB. These changes are expected to contribute to pathological alterations in Aβ and insulin levels in the brain that are intricately linked with neuropathological changes observed in AD patients.

MATERIALS AND METHODS

Preparation of Aβ peptides

Aβ₄₀ and Aβ₄₂ peptides were custom synthesized on a fluorenylmethyloxycarbonyl (FMOC) column by the Mayo Clinic proteomics core facility (Rochester, MN). Solutions of soluble Aβ were prepared using the procedure developed by Klein et al. [22]. Briefly, solid Aβ was dissolved in ice-coldhexafluoro-2-propanol (HFIP) (MP Biomedicals, Santa Ana, CA) and then incubated at room temperature for 1h. The solvent was evaporated overnight, then further dried under vacuum, and the resultant Aβ films were stored at −20°C prior to use. Before labeling with ¹²⁵I, the Aβ films were dissolved in DMSO, diluted in Ham’s F-12 media (Mediatech, Manassas, VA), and centrifuged at 18,000 rpm to remove any insoluble Aβ aggregates.

Radioiodination of Aβ peptides and insulin

Carrier-free Na¹²⁵I and Na¹³¹I radionuclides were obtained from PerkinElmer Life and Analytical Sciences(Boston, MA). The following peptides/proteins were labeled with ¹²⁵I or ¹³¹I using the chloramine-T procedure as described previously [23, 24]: Aβ₄₀, Aβ₄₂, bovine serum albumin (BSA) (Sigma-Aldrich, St. Louis, MO), and human insulin (Sigma-Aldrich). Free radioactive iodine was separated from the radiolabeled peptide/protein by dialysis against 0.01M phosphate-buffered saline (PBS) at pH 7.4 (Sigma-Aldrich). Purity of the radiolabeled peptides/proteins was assessed by trichloroacetic acid (TCA) precipitation. The preparation was deemed acceptable if the precipitable radioactivity counts were greater than 95% of the total counts.

Animals

Animal studies were carried out in compliance with the U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The protocols were approved by the Institutional Animal Care and Use Committee at Mayo Clinic, Rochester, MN (Mayo IACUC #A00006176–21). The studies were documented according to the ARRIVE reporting guidelines. Wild-type (C57BL/6J) were purchased from The Jackson Laboratory (Bar Harbor, ME). The APP/PS1 mice were generated by mating hemizygous transgenic mice (mouse strain C57B6/SJL; i.d. no. Tg2576) expressing mutant human amyloid precursor protein (APP695) with a second strain of hemizygous transgenic mice (mouse strain Swiss-Webster/B6D2; i.d. no. M146L6.2) expressing mutant human presenilin 1 (PS1). APP/PS1 mice exhibit Aβ overexpression, accelerated brain Aβ deposition, and cognitive decline. Mice were housed in a virus-free barrier facility with 12-h light and dark cycles and were provided with pellet food and purified water ad libitum. All studies were performed using female mice. Prior to the experiments, the mice were confirmed by visual inspection to be in the diestrus phase of the estrous cycle [25], during which hormonal fluctuations are reported to be minimal [26].

Plasma pharmacokinetics and brain permeability of ¹²⁵I-Aβ₄₂ and ¹²⁵I-Aβ₄₀

These experiments were carried out as described in our earlier publications [24, 27]. The age groups were selected based on our previous findings which demonstrated that in APP/PS1 mice, brain Aβ plaques were undetectable at 12 weeks of age, with initiation of plaque formation at 24 weeks, and full-scale plaque burden at 52 weeks [28]. Briefly, wild-type (WT) or APP/PS1 transgenic mice at various ages (8, 24, or 52 weeks) weighing 25–35g were catheterized at the femoral vein and femoral artery while under general anesthesia (1.5% isoflurane; 4L/min oxygen). A single dose (100μCi) of ¹²⁵I-Aβ₄₀ or ¹²⁵I-Aβ₄₂ in PBS (100μL) was bolus injected into the femoral vein. Blood was sampled serially (20μL) from the femoral artery at 0.25-,1-,3-,5-,10-, and 15-min post-injection. Immediately after the 15-min sampling event, ¹³¹I-BSA (100μCi) was bolus injected into the femoral vein to serve as a measure of the residual plasma volume (V_p). One minute after the ¹³¹I-BSA injection, a final blood sample was collected, and the animal was euthanized. Blood samples were diluted in PBS and centrifuged to separate the plasma. The plasma was subjected to TCA precipitation and then centrifuged. The total ¹²⁵Iand ¹³¹I activity counts in the pellet, corresponding to intact radiolabeled peptide/protein, were assayed using a two-channel gamma counter (Cobra II; PerkinElmer Life and Analytical Sciences, Boston, MA).

The ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ plasma concentration versus time data was fitted with the following bi-exponential equation using Phoenix WinNonlin^® 6.4 (Certara, St. Louis, MO):

$$C\left(t\right)=A{e}^{-\alpha t}+B{e}^{-\beta t}$$ (1)

where $C\left(t\right)$ is the concentration (μCi/mL) in plasma at time $t$ (min), $A$ and $B$ are the intercepts of the distribution and elimination phases, respectively, and $\alpha$ and $\beta$ are the distribution and elimination macro-rate constants (min⁻¹), respectively. The secondary plasma pharmacokinetic parameters are reported in Table 1.

Table 1.

Plasma pharmacokinetics of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ in WT and APP/PS1 transgenic mice at 8, 24, and 52 weeks

Wild-type
Age (weeks)	125I-Aβ40			125I-Aβ42
	Cmax (μCi/mL)	AUC (μCi.min/mL)	Clearance (mL/min)	Cmax (μCi/mL)	AUC (μCi.min/mL)	Clearance (mL/min)
8	23.2 ± 0.6	63.5 ± 6.5	1.6 ± 0.2	11.3 ± 0.5###	78.9 ± 11.8	1.3 ± 0.2
24	16.4 ± 0.4	43.5 ± 3.1	2.3 ± 0.2	5.2 ± 0.3###	40.1 ± 9.4	2.5 ± 0.3*
52	12.8 ± 0.1	373 ± 63***	0.26 ± 0.04**	6.0 ± 0.3###	31.2 ± 3.5####	3.2 ± 0.3***
APP/PS1
8	17.7 ± 1.8	49.7 ± 12.4	2.1 ± 0.5	3.4 ± 0.3###	27.6 ± 9.4	3.6 ± 1.2
24	12.5 ± 0.2	103.0 ± 34.5	1.0 ± 0.3	7.1 ± 0.2##	72.2 ± 19.7	1.4 ± 0.3
52	13.7 ± 0.2	86.8 ± 22.8	1.2 ± 0.3	5.1 ± 1.0###	30.9 ± 7.8	3.2 ± 0.8

Data represent mean±SE, n = 3–5. Significance was assessed over age (*p<0.05, **p<0.01, and ***p<0.001) and ¹²⁵I-Aβ₄₀ versus ¹²⁵I-Aβ₄₂ (##p<0.01, ###p<0.001, and ###p<0.0001; two-way ANOVA with Bonferroni post-tests).

At the end of the experiment, the brain was removed from the cranial cavity and dissected into anatomical regions (cortex, caudate putamen, hippocampus, thalamus, brain stem, and cerebellum). Individual brain regions were assayed for ¹²⁵I and ¹³¹I activity using a two-channel gamma counter. The residual plasma volume $\left(V_p\right)$ at each brain region (mL per gram tissue) was estimated using the equation:

$$V_p=\frac{q_p}{C_{v}W}$$ (2)

where $q_p$ is the total ¹³¹I-BSA activity in the brain region (μCi), $C_v$ is the ¹³¹I-BSA concentration in plasma (μCi/mL), and $W$ is the weight of the brain region (g). Given the total ¹²⁵I-Aβ₄₀ or ¹²⁵I-Aβ₄₂ activity in the brain region $\left(q_{\tau };\mathrm{\mu }\mathrm{C}\mathrm{i}\right)$, the amount that permeated into the brain extravascular space ($q;\mathrm{\mu }\mathrm{C}\mathrm{i}$ per gram tissue) was estimated using the equation:

$$q=\frac{q_{\tau }}{W}-V_p{C}_a$$ (3)

where $C_a$ is the final ¹²⁵I-Aβ concentration in plasma (μCi/mL). The permeability-surface area (PS) product (mL/s per gram tissue) for ¹²⁵I-Aβ₄₀ or ¹²⁵I-Aβ₄₂ at each brain region was estimated using the equation:

$$PS=\frac{q}{60{\int }_0^t{C}_{p}dt}$$ (4)

where ${\int }_0^t{C}_{p}dt$ is the area under the observed plasma concentration versus time profile (AUC) from 0–15 min (min∗μCi/mL) obtained using the linear-trapezoidal method. The PS product for the whole brain was estimated based on the sum of $q$ in all six dissected brain regions. The % of the injected dose that permeated into the whole brain per gram tissue (% ID/g) was estimated using the equation:

$$\mathrm{\%}ID/g=\frac{q}{\mathit{Dose}\left(100\mu Ci\right)}$$ (5)

Plasma pharmacokinetics and brain permeability of ¹²⁵I-insulin

These experiments were performed as described above for ¹²⁵I-Aβ. Briefly, WT mice at various ages (8, 24, or 52 weeks) were bolus injected with a single dose (100μCi) of ¹²⁵I-insulin in PBS (100μL) into the femoral vein. Blood was sampled serially (20μL) from the femoral artery at 0.25-, 1-, 3-, 5-, 10-, and 14-min post-injection. Immediately after the 14-min sampling event, ¹³¹I-BSA (100μCi) was bolus injected into the femoral vein to serve as a measure of ${\mathrm{V}}_{\mathrm{p}}$. One minute after the ¹³¹I-BSAinjection,a final blood sample was collected, and the animal was euthanized. The plasma pharmacokinetics and brain uptake of ¹²⁵I-insulin were assessed as described above for ¹²⁵I-Aβ. The secondary plasma pharmacokinetic parameters for ¹²⁵I-insulin are reported in Table 2.

Table 2.

Plasma pharmacokinetics of ¹²⁵I-insulin in WT mice at 8, 24, and 52 weeks

125I-insulin
Age (weeks)	Cmax (μCi/mL)	AUC (μCi.min/mL)	Clearance (mL/min)
8	19.6 ± 0.2	83.7 ± 7.9	1.19 ± 0.11
24	23.5 ± 0.5****	170 ± 15	0.61 ± 0.06***
52	28.0 ± 0.4****	357 ± 68**	0.28 ± 0.05****

Data represent mean±SE, n = 3–5. Significance was assessed over age (**p <0.01, ***p <0.001, ****p <0.0001; one-way ANOVA with Bonferroni post-tests).

Dynamic SPECT/CT imaging of ¹²⁵I-insulin influx to the brain

WT mice at 12 or 104 weeks of age weighing 25–35g were catheterized at the femoral vein and femoral artery while under general anesthesia (1.5% isoflurane; 4L/min oxygen). A single dose (500μCi) of ¹²⁵I-Insulin diluted in PBS was bolus injected into the femoral vein. Immediately following which, the entire animal was placed inside the single-photon emission computed tomography (SPECT/CT) scanner (Gamma Medica, Northridge, CA), and the biodistribution of radioactive signal was imaged at 1-min intervals over the next 40 min, followed by a 5-min CT scan to locate regions of interest (ROIs). The brain influx of ¹²⁵I-insulin was assessed by Gjedde-Patlak graphical analysis [23], which involved plotting:

$$\frac{X_b\left(t\right)}{C_p\left(t\right)}vs\cdot \frac{{\int }_0^t{C}_{p}dt}{C_p\left(t\right)}$$ (6)

where $X_b\left(t\right)$ is the measured amount (μCi) of radioactive signal in the brain ROI at time $t$ (min), $C_p\left(t\right)$ is the plasma concentration (μCi/mL) at time $t$, and ${\int }_0^t{C}_{p}dt$ is the plasma AUC (min∗μCi/mL) from $0-t$. Both $C_p\left(t\right)$ and ${\int }_0^t{C}_{p}dt$ were predicted based on the primary plasma pharmacokinetic parameters obtained in a separate experiment. The slope obtained after linear regression corresponds to the brain influx clearance, ${\mathrm{K}}_{\mathrm{i}}$ (mL/min) of ¹²⁵I-insulin.

Transcriptomic analysis of brain endothelial cells from WT mice

A comprehensive, transcriptomic analysis of gene expression profiles of brain endothelial cells obtained from WT mice at 12 and 104 weeks of age was recently published by Ximerakis et al. [21]. In these studies, the transcriptomes of 50,212 single cells obtained from the brains of young or aged mice were assessed by single-cell RNA-Seq. Following which, an established clustering method was used to group transcriptionally similar cells based on the expression of multiple cell-type-specific marker genes, such as those specific for brain endothelial cells (CD31, tight junction proteins, etc.).

Using the single-cell RNA-Seq data publicly available at http://shiny.baderlab.org/AgingMouseBrain/, we applied a non-parametric Gene Set Variation Analysis (GSVA) method [29] to assess the pathway-level activity in young and aged cohorts at an individual sample level. Curated gene sets defining specific pathways were obtained from MSigDB. The GSVA method condenses a gene-level expression matrix into pathway enrichment scores. The directionality of pathway scores was determined from the mean difference between the cohorts. The significance was assessed by unpaired, two-tailed t-tests. The analysis was performed using the GSVA package in R studio (R software; Boston, MA).

We also performed Gene Set Enrichment Analysis (GSEA) on the brain endothelial cell expression data published by Ximerakis et al. [21]. Similar to GSVA, the GSEA method can also be used to assess differences in pathway-level activity when comparing cohorts [3]. The same pathways (gene sets) assessed by GSVA were assessed by GSEA. Pathway enrichment scores were calculated based on the list of genes ranked according to their differential expression. The significance was assessed by a permutation test, which determines the probability of obtaining an enrichment score that is as strong as or stronger than that observed under the permutation-generated null distribution. The analysis was performed using the fgsea package in R studio.

Statistical analysis

Statistical tests were performed using GraphPad Prism 8.4 (GraphPad software; La Jolla, CA). A p-value of less than 0.05 was regarded as statistically significant. Two-way ANOVA followed by Bonferroni post-hoc tests were used to compare the plasma pharmacokinetic parameters or PS products of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ at 8, 24, and 52 weeks of age in WT mice, as well as in APP/PS1 mice. One-way ANOVA followed by Bonferroni post-hoc tests were used to compare the plasma pharmacokinetic parameters or PS products of ¹²⁵I-insulin in WT mice at 8, 24, and 52 weeks of age. Unpaired, two-tailed t-tests were used to compare the slopes obtained from Gjedde-Patlak plots for ¹²⁵I-insulin in WT mice at 12 and 104 weeks of age.

RESULTS

Age-dependent changes in plasma pharmacokinetics of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ in WT and APP/PS1 mice (Fig. 1, Table 1)

Fig. 1.

Plasma pharmacokinetics of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ in WT and APP/PS1 mice at 8, 24, and 52 weeks. WT or APP/PS1 transgenic mice were bolus injected with 100Ci of ¹²⁵I-Aβ₄₀ (A, C) or ¹²⁵I-Aβ₄₂ (B, D) into the femoral vein. Plasma was sampled periodically between 0–15min, and radioactivity in the intact ¹²⁵I-Aβ fraction was measured. The plasma concentration versus time data was fit to a bi-exponential equation. Shown are the observed values (mean±SD, n=3–5) overlaid on the predicted curves.

The plasma pharmacokinetics of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ were assessed in WT and APP/PS1 transgenic mice at 8, 24, and 52 weeks of age. Female mice were used for the studies, given that AD disproportionately affects women, and that female APP/PS1 transgenic mice display higher brain Aβ levels and greater occurrence of histopathological hallmarks compared to their male littermates [18, 30]. Following a bolus injection into the femoral vein, both peptides exhibited bi-exponential decline in their plasma concentrations with time (Fig. 1), which is consistent with our earlier publications [20, 24]. The plasma exposure to ¹²⁵I-Aβ₄₀ in WT mice, estimated by the area under the concentration versus time curve (AUC), was unaltered between 8 and 24 weeks, but increased by ~ 7-fold at 52 weeks; a concomitant decrease in plasma clearance was observed (Fig. 1A; Table 1). In contrast, the plasma AUC of ¹²⁵I-Aβ₄₂ reduced by ~ 2-fold in WT mice around 24 or 52 weeks of age compared to 8 weeks; a concomitant increase in plasma clearance was observed (Fig. 1B; Table 1). Further, at 52 weeks, the plasma AUC of ¹²⁵I-Aβ₄₀ in WT mice was ~ 12-fold higher than that of ¹²⁵I-Aβ₄₂, whereas no significant differences were observed between the two peptides at 8 or 24 weeks (Table 1).

In APP/PS1 mice, the plasma AUC of ¹²⁵I-Aβ₄₀ was not significantly altered across all three age groups, although an increasing trend was observed at 24 and 52 weeks compared to 8 weeks (Fig. 1C; Table 1). The plasma AUC of ¹²⁵I-Aβ₄₂ was also not significantly different across all three age groups of APP/PS1 mice, but an increasing trend was observed at 24 weeks compared to 8 or 52 weeks (Fig. 1D; Table 1). In APP/PS1 mice, no significant differences were observed between the plasma AUC of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ at any of the three age groups, although an increasing trend was observed for ¹²⁵I-Aβ₄₀ compared to ¹²⁵I-Aβ₄₂ at 52 weeks (Table 1). Interestingly, in both WT and APP/PS1 mice, the maximum observed plasma concentration (C_max) was higher for ¹²⁵I-Aβ₄₀ compared to ¹²⁵I-Aβ₄₂ at all three age groups.

Brain uptake of ¹²⁵I-Aβ₄₀ decreases and that of ¹²⁵I-Aβ₄₂ increases with age in WT mice, which is disrupted in APP/PS1 mice (Fig. 2)

Fig. 2.

Brain uptake of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ in WT and APP/PS1 mice at 8, 24, and 52 weeks. A) Experiment scheme. WT or APP/PS1 transgenic mice were bolus injected with 100Ci of ¹²⁵I-Aβ₄₀ or ¹²⁵I-Aβ₄₂ into the femoral vein. At the end of the experiment, 100μCi of ¹³¹I-albumin was injected to serve as a marker of V_p. The brain regions were dissected and assayed for radioactivity. Shown are the PS value estimates for ¹²⁵I-Aβ₄₀ (B, D) and ¹²⁵I-Aβ₄₂ (C, E) uptake at various brain regions. F) The PS values are shown for the whole brain. G) The overall brain accumulation was assessed as % ID/g. Data represent mean±SD, n = 3–5. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001; two-way ANOVA with Bonferroni post-tests).

Following a bolus injection of ¹²⁵I-Aβ₄₀ or ¹²⁵I-Aβ₄₂ into the femoral vein of WT or APP/PS1 transgenic mice at 8, 24, and 52 weeks of age, the brain uptake was assessed as the permeability-surface area (PS) product. In WT mice, the PS values of ¹²⁵I-Aβ₄₀ in various brain regions were unaltered at 8 and 24 weeks but decreased by ~ 4-fold at 52 weeks (Fig. 2B). In contrast, the PS values of ¹²⁵I-Aβ₄₂ were unaltered at 8 and 24 weeks but increased by ~ 1.5-fold at 52 weeks in the WT mice (Fig. 2C). In APP/PS1mice,thePSvaluesof ¹²⁵I-Aβ₄₀ were unaltered at 8 and 24 weeks but decreased by ~ 1.5-fold at 52 weeks (Fig. 2D). Interestingly, the PS values of ¹²⁵I-Aβ₄₂ at various brain regions in APP/PS1 mice were unaltered at 8 and 52 weeks but decreased by ~ 1.5-fold at 24 weeks (Fig. 2E).

When comparing age-matched WT and APP/PS1 mice, the PS products of ¹²⁵I-Aβ₄₀ in the whole brain were ~ 1.5-fold lower in APP/PS1 mice compared to WT mice, at both 8 and 24 weeks. However, by 52 weeks, at which point the PS products of ¹²⁵I-Aβ₄₀ had decreased substantially in WT mice, no differences were observed between WT and APP/PS1 mice (Fig. 2F). Interestingly, for ¹²⁵I-Aβ₄₂, no significant differences between WT and APP/PS1 mice were observed at any of the age groups (Fig. 2F). When comparing Aβ₄₀ versus Aβ₄₂ in WT mice, the PS products in the whole brain were ~ 1.5-fold higher for ¹²⁵I-Aβ₄₀ compared to ¹²⁵I-Aβ₄₂ at both 8 and 24 weeks; however, by 52 weeks, a dramatic shift was observed in the PS product of ¹²⁵I-Aβ₄₂, which increased by ~ 4-fold compared to ¹²⁵I-Aβ₄₀ (Fig. 2F). Conversely, in APP/PS1 mice, the PS products in the whole brain were similar for ¹²⁵I-Aβ₄₀ compared to ¹²⁵I-Aβ₄₂ at both 8 and 24 weeks; however, by 52 weeks, again a shift was observed in the PS product of ¹²⁵I-Aβ₄₂,whichincreasedby~3-fold compared to ¹²⁵I-Aβ₄₀ (Fig. 2F). Similar trends were observed in the overall brain accumulation, which was expressed as the % of the injected dose (ID) accumulated in the brain per gram of tissue (% ID/g) (Fig. 2G).

Age-dependent changes in plasma pharmacokinetics of ¹²⁵I-insulin in WT mice (Fig. 3, Table 2)

Fig. 3.

Plasma pharmacokinetics of ¹²⁵I-insulin in WT mice at 8, 24, and 52 weeks. Mice were bolus injected with 100μCi of ¹²⁵I-insulin into the femoral vein. Plasma was sampled periodically between 0–14 min, and radioactivity in the intact ¹²⁵I-insulin fraction was measured. The plasma concentration versus time data was fit to a bi-exponential equation. Shown are the observed values (mean±SD, n = 3–5) overlaid on the predicted curves.

The plasma pharmacokinetics of ¹²⁵I-Insulin were assessed in WT mice at 8, 24, and 52 weeks of age. Following a bolus injection into the femoral vein, a bi-exponential decline in ¹²⁵I-insulin plasma concentrations was observed with time (Fig. 3), which is consistent with earlier publications [31]. The plasma AUC of ¹²⁵I-insulin increased by ~ 3-fold at 52 weeks compared to 8 or 24 weeks; a concomitant decrease in clearance was observed. The C_max was also increased at 52 weeks compared to 8 or 24 weeks (Table 2).

Brain uptake of ¹²⁵I-insulin decreases with age in WT mice (Figs. 4 and 5)

Fig. 4.

Brain uptake of ¹²⁵I-insulin in WT mice at 8, 24, and 52 weeks. A) Experimental scheme. WT mice were bolus injected with 100μCi of ¹²⁵I-insulin into the femoral vein. At the end of the experiment, 100μCi of ¹³¹I-albumin was injected to serve as a marker of V_p. The brain regions were dissected and assayed for radioactivity. B) The PS value estimates for ¹²⁵I-insulin uptake in various brain regions are shown. Data represent mean±SD, n = 3–5. *p<0.05, ***p<0.001, ****p<0.0001; two-way ANOVA with Bonferroni post-tests. C) The overall brain accumulation was assessed as % ID/g. **p<0.01; one-way ANOVA with Bonferroni post-tests.

Fig. 5.

Dynamic SPECT/CT imaging of ¹²⁵I-insulin uptake to the brain. A) Experimental scheme. WT mice were bolus injected with 500μCi of ¹²⁵I-insulin into the femoral vein and the accumulation of ¹²⁵I-insulin in the brain was monitored between 0–40 minutes post-injection by dynamic SPECT/CT imaging. B) The brain influx clearance (K_i) of ¹²⁵I-insulin was estimated by the slope obtained from Gjedde-Patlak graphical analysis. In the graph, Amount_brain(t) is the measured radioactivity in the brain ROI (Ci) at each time point (data are mean±SD, n=4), whereas AUC_plasma(0-t) (min.μCi/mL) and Conc._plasma(t) (μCi/mL) were predicted using the plasma pharmacokinetic parameters. The K_i predictions presented in the table are mean±SE, n=4. ***p<0.001; unpaired two-tailed t-test.

Following a bolus injection of ¹²⁵I-insulin into the femoral vein of WT mice at 8, 24, and 52 weeks of age, the brain influx was assessed as the PS product (Fig. 4A). The PS products of ¹²⁵I-insulin at various brain regions decreased by ~ 5-fold at 24 weeks compared to 8 weeks but did not further decrease at 52 weeks (Fig. 4B). Similarly, the % ID/g of ¹²⁵I-insulin in the whole brain decreased by ~ 3-fold at 24 and 52 weeks compared to 8 weeks. To study the effects of advanced aging, the brain influx of ¹²⁵I-insulin was further assessed in WT mice at 12 and 104 weeks of age by dynamic SPECT/CT imaging. After femoral injection, the accumulation of ¹²⁵I-insulin in the brain was imaged from 0–40min (Fig. 5A). Gjedde-Patlak plots were constructed to estimate the brain influx clearance [32], which corresponds to the slope (K_i; mL/min) obtained after linear regression (Fig. 5B). The brain influx clearance of ¹²⁵I-insulin decreased by ~ 2-fold at 104 weeks compared to 12 weeks (Fig. 5 inset).

Age-dependent changes in transporter/signaling pathways at the BBB in WT mice (Fig. 6)

Using the publicly available single-cell RNA-Seq data published by Ximerakis et al. [21], we performed Gene Set Variation Analysis (GSVA) and also Gene Set Enrichment Analysis (GSEA) to assess the activity of various cellular pathways that are implicated in Aβ or insulin trafficking at the BBB. Pathways relating to RAGE signaling, insulin/Akt signaling, and T2DM pathology were upregulated in brain endothelial cells obtained from WT mice at 104 weeks compared to 12 weeks (Table 3). In contrast, ABC transporter and tight junction networks were downregulated in the brain endothelial cells obtained at 104 weeks compared to 12 weeks (Table 3).

Table 3.

Enrichment of various transporter/signaling pathways in brain endothelial cells from WT mice at 12 and 104 weeks

Pathway	GSVA		GSEA
	Directionality	p	Directionality	p
KEGG AGE RAGE Signaling Pathway	Up	0.088	Up	0.25
PID_Insulin Pathway		0.035		0.11
PID.PI3KCI AKT Pathway		1.5E-08		0.0043
KEGG_Type II Diabetes Mellitus		0.012		0.12
KEGG.ABC Transporters	Down	0.0058	Down	0.086
REACTOME_Tight Junction Interactions		2.9E-05		0.13

The GSVA and GSEA methods were used to compare the enrichment of various transporter/signaling pathways in brain endothelial cells obtained from WT mice at different ages, using the single-cell RNA-Seq data published by Ximerakis et al. [21]. The directionality indicates upregulation or downregulation of the pathway in the aged compared to young mice.

DISCUSSION

BBB dysfunction is observed in the aging brain [3] and is thought to be among the earliest changes driving AD pathogenesis [33]. The BBB dysfunction is customarily studied from the perspective of loss of tight junction integrity [34] and reduced cerebral blood flow [35]. However, age-related changes in cellular and molecular mechanisms that drive these changes in BBB functions are only partially understood. Even less understood are age-related changes in diverse BBB functions that coordinate the delivery of essential nutrients like glucose and insulin to the brain [13], as well as the clearance of toxic metabolites like Aβ peptides from the brain [36].

Our findings demonstrate that plasma pharmacokinetics of Aβ isoforms and insulin are differentially affected by aging in WT mice. The liver and kidneys are the major organs responsible for clearing Aβ from the periphery [37], but it remains poorly understood how systemic Aβ clearance is affected during normal aging and in AD. We showed that in WT mice, the plasma AUC of ¹²⁵I-Aβ₄₀ increased with age, but that of ¹²⁵I-Aβ₄₂ decreased. This demonstrates an age-dependent switch in the plasma pharmacokinetics of the two major Aβ isoforms. Consistent with these findings, the plasma AUC of Aβ₄₀ was reported to increase with age in squirrel monkeys [38]. The observed age-dependent changes in Aβ₄₀ plasma pharmacokinetics in WT mice were evident at an even earlier age in APP/PS1 transgenic mice, likely due to the accelerated AD pathology in these animals. Similarly, it was reported that Aβ₄₀ plasma concentrations increased with age in healthy patients, and substantially elevated in AD patients [39]. This is expected to promote a decrease in the Aβ₄₂/Aβ₄₀ ratio in plasma with age, which has been strongly linked to amyloid plaque deposition and AD risk in patients [40, 41]. The increase in ¹²⁵I-Aβ₄₀ plasma AUC with age predicts increased exposure of this isoform to the luminal surface of the BBB. Increased plasma Aβ₄₀ levels are associated with cerebrovascular disease, which is prevalent in approximately 90% of all AD patients [42].

The current study demonstrates that plasma-to-brain influx of Aβ isoforms and insulin are progressively and differentially affected by aging in WT mice. Using the publicly available single-cell RNA-Seq data of BBB endothelial cells in WT mice [21], we identified age-dependent changes to various trafficking/signaling pathways implicated in Aβ and insulin delivery to the brain. Changes in Aβ trafficking are evident in AD transgenic mice at a much younger age, suggesting that age-related pathophysiological changes are accelerated in AD. The brain influx of ¹²⁵I-Aβ₄₀ is lower in APP/PS1 mice than in WT mice at both 8 and 24 weeks of age; but no significant differences in ¹²⁵I-Aβ₄₂ influx were observed between WT and APP/PS1 mice in these age groups. At 52 weeks, the differences between WT and APP/PS1 mice were entirely lost, with Aβ₄₂ influx dominating over Aβ₄₀ influx. Previously, we reported that in APP/PS1 mice, initiation of Aβ plaque formation occurs at 24 weeks of age, with full-scale plaque burden observed by 52 weeks [28]. It was also reported that cognitive deficits first appear in APP/PS1 mice at 24 weeks, and become more severe by 52 weeks [43]. Thus, we speculate that the shift in the preferential brain influx of Aβ₄₂ versus Aβ₄₀ observed at 52 weeks is associated with enhanced amyloid deposition and cognitive decline in the APP/PS1 mice.

Normal aging, AD, and T2DM have all been associated with hyperinsulinemia and insulin resistance, which manifests as increased insulin plasma levels and impaired insulin sensitivity in the brain [44, 45]. We showed that the plasma AUC of ¹²⁵I-insulin increased from 8 to 24 weeks, and then further increased at 52 weeks in WT mice. This could be due to impaired renal clearance of insulin, which was claimed to be a major driver of hyperinsulinemia [46], and/or decreased distribution from plasma to peripheral tissues and the brain. In the AD brain, increased levels of toxic Aβ₄₂ peptides are paralleled by decreased levels of the essential growth factor insulin [14]. The PS products of ¹²⁵I-insulin in WT mice declined steeply from 8 to 24 weeks, then remained unchanged from 24 to 52 weeks. In contrast, the PS products of ¹²⁵I-Aβ₄₀ and ¹²⁵I-Aβ₄₂ remained unchanged from 8 to 24 weeks, then decreased and increased, respectively, at 52 weeks. It was also shown that WT mice demonstrate learning/memory deficits at 52 weeks compared to 24 weeks [47]. Taken together, these findings suggest that age-related pathological changes in insulin transport at the BBB precede the changes in Aβ transport at the BBB and the onset of cognitive decline in WT mice.

The brain uptake of insulin is extremely rapid relative to other serum proteins, indicating a highly specialized and efficient trafficking system for delivering insulin across the BBB [23]. To capture the initial rate of insulin uptake, dynamic SPECT/CT imaging studies were performed to measure ¹²⁵I-insulin uptake to the brain after systemic injection in WT mice at 12 and 104 weeks. These age groups were selected based on a previous study that established age-dependent changes in insulin sensitivity in WT mice [48]. Based on Gjedde-Patlak graphical analysis, the ¹²⁵I-insulin brain influx clearance decreased at 104 weeks compared to 12 weeks. Thus, we showed that compared to young WT mice, insulin uptake to the brain was decreased in both middle-aged (52 weeks) and advanced-aged (104 weeks) mice. Furthermore, we recently showed that ¹²⁵I-insulin uptake to the brain is decreased in APP/PS1 mice compared to age-matched WT mice [19]. Together, these findings suggest that reduced insulin transport at the BBB during aging in WT mice may decrease further in APP/PS1 mice and reduce brain insulin levels.

Both insulin and Aβ peptides are known to exhibit saturable, receptor-mediated transcytosis at the BBB [49, 50]. The brain uptake of Aβ peptides is thought to be primarily mediated by RAGE, expressed on the luminal surface of the BBB [9]. Membrane RAGE is reported to localize in caveolae microdomains [51], while IR is reported to localize in clathrin-coated pits [52]. Although the role of IR at the BBB in insulin delivery to the brain is controversial [49, 53], insulin transcytosis across microvascular endothelial cells was reported to be clathrin-dependent [54]. We previously showed that Aβ₄₂ uptake in BBB endothelial cells is caveolae-dependent, whereas Aβ₄₀ uptake is clathrin-dependent [64]. This suggests that in addition to RAGE, Aβ₄₀ uptake at the BBB may be handled by other receptor(s) that involve clathrin-dependent endocytosis. In a recent study, aging in WT mice was shown to be associated with increased expression of caveolin-1 and decreased expression of clathrin heavy chain in the brain microvessels [55]. Therefore, our current findings that aging in WT mice is associated with increased brain uptake of Aβ₄₂ and decreased uptake of Aβ₄₀ and insulin could be resulting from the age-dependent switch from clathrin to caveolae-mediated endocytosis at the BBB.

In addition to serving as an endocytosis receptor, RAGE also functions as a signaling receptor to trigger the activation of pro-inflammatory pathways, which have been implicated in Aβ uptake at the BBB [56]. It was reported that RAGE expression at the BBB increased with age in rats [11]. By performing pathway-level analyses (GSVA and GSEA) on publicly available single-cell RNA-Seq data [21], we found that the RAGE signaling pathway was upregulated in brain endothelial cells obtained from WT miceat104weekscomparedto12weeks.Incontrast, ABC transporter proteins, which were claimed to mediate Aβ efflux from the brain [57] as well as tight junction proteins, were downregulated at 104 weeks. Thus, increased RAGE expression and signaling at the BBB likely contributed to the increased ¹²⁵I-Aβ₄₂ uptake observed in the aged mice. However, these trends cannot explain the concomitant decrease in ¹²⁵I-Aβ₄₀ uptake. We speculate these differences are engendered by other unidentified luminal receptors reliant on clathrin-dependent endocytosis that demonstrate higher selectivity for Aβ₄₀ over Aβ₄₂. During normal aging and AD progression, downregulation of such receptors could promote a decrease in the brain influx of Aβ₄₀. The differential handling of Aβ₄₀ versus Aβ₄₂ by cell-surface receptors and intracellular mechanisms governing their disposition require further investigation.

Insulin signaling networks in the brain are disrupted during normal aging and AD, which triggers brain insulin resistance and contributes to cognitive decline [58,59].ThePI3K-AKTsignalingpathwayis downstream of IR and has canonical functions in glucose metabolism. By performing GSVA and GSEA on the single-cell RNA-Seq data [21], we found that insulin signaling and PI3K-AKT pathways were upregulated at the BBB in WT mice at 104 weeks compared to 12 weeks. Previously, it was reported that sustained over-activation of PI3K-AKT signaling in the aging brain is associated with brain insulin resistance and Aβ pathology [60, 61]. Consistent with these findings, we observed upregulation of a set of proteins implicated in insulin resistance/T2DM in the 104-week-old mice. These age groups match with those used in the ¹²⁵I-insulin SPECT/CT imaging studies. Based on these reports and our current findings, we speculate that in aged WT mice, insulin signaling disruptions at the BBB triggered by hyperinsulinemia and peripheral insulin resistance inhibit insulin transcytosis at the BBB, resulting in decreased insulin uptake to the brain.

Insulin signaling networks have also been implicated in the regulation of Aβ trafficking receptors at the BBB [62]. We previously reported that stimulation with insulin modulates the plasma membrane expression of LRP-1 in hCMEC/D3 cell monolayers (BBB cell model) [20], while others demonstrated similar findings in hepatocytes [63]. Insulin is also suggested to modulate expression of other important Aβ receptors, specifically RAGE and P-glycoprotein, at the BBB [20, 64, 65]. Moreover, we have previously shown that insulin differentially regulates the trafficking of Aβ isoforms at the BBB. Specifically, systemic insulin injection in WT mice decreased the brain influx of ¹²⁵I-Aβ₄₂, but increased the brain influx of ¹²⁵I-Aβ₄₀, with concomitant changes in their plasma pharmacokinetics [20]. Therefore, we speculate that impaired insulin signaling in the BBB endothelium contributes to the Aβ trafficking perturbations observed in aged mice.

In summary, we demonstrated age-dependent changes in the plasma disposition and brain influx of Aβ isoforms and insulin in female WT mice. The brain influx of both ¹²⁵I-insulin and ¹²⁵I-Aβ₄₀ decreased and ¹²⁵I-Aβ₄₂ influx increased in WT mice with age. The increase in plasma AUC of ¹²⁵I-insulin with age is indicative of hyperinsulinemia, whereas the increase in ¹²⁵I-Aβ₄₀ plasma AUC and decrease in ¹²⁵I-Aβ₄₂ plasma AUC could reduce the Aβ₄₂/Aβ₄₀ ratio in plasma. Both of these changes are heavily implicated in AD risk. Thus, pathological changes during normal aging alter systemic clearance and brain influx of both insulin and Aβ peptides in a manner that could augment AD risk. Additionally, aging was associated with various transporter/signaling deficits in BBB endothelial cells, which is indicative of BBB dysfunction and insulin resistance. Further studies are needed to clarify the molecular mechanisms by which insulin and Aβ trafficking in the brain and periphery are impacted during normal aging and in AD.