Exposure to Lead (Pb) in Drinking Water Causes Cognitive Impairment via an Alzheimer’s Disease Gene-Dependent Mechanism in Adult Mice

Abstract

Background:

Exposure to lead (Pb) is a major public health problem that could occur through contaminated soil, air, food, or water, either during the course of everyday life, or while working in hazardous occupations. Although Pb has long been known as a neurodevelopmental toxicant in children, a recent and growing body of epidemiological research indicates that cumulative, low-level Pb exposure likely drives age-related neurologic dysfunction in adults. Environmental Pb exposure in adulthood has been linked to risk of late-onset Alzheimer’s disease (AD) and dementia.

Objective:

Although the biological mechanism underlying this link is unknown, it has been proposed that Pb exposure may increase the risk of AD via altering the expression of AD-related genes and, possibly, by activating the molecular pathways underlying AD-related pathology.

Methods:

We investigated Pb exposure using a line of genetically modified mice with AD-causing knock-in mutations in the amyloid precursor protein and presenilin 1 (APP^ΔNL/ΔNL × PS1^P264L/P264L) that had been crossed with Lepr^db/db mice to impart vulnerability to vascular pathology.

Results:

Our data show that although Pb exposure in adult mice impairs cognitive function, this effect is not related to either an increase in amyloid pathology or to changes in the expression of common AD-related genes. Pb exposure also caused a significant increase in blood pressure, a well known effect of Pb. Interestingly, although the increase in blood pressure was unrelated to genotype, only mice that carried AD-related mutations developed cognitive dysfunction, in spite of showing no significant change in cerebrovascular pathology.

Conclusions:

These results raise the possibility that the increased risk of dementia associated with Pb exposure in adults may be tied to its subsequent interaction with either pre-existing or developing AD-related neuropathology.

INTRODUCTION

Lead (Pb) is a non-physiologic metal that has an extensive use history in numerous applications, including paint and fuel additives, glass, pigments, cosmetics, plumbing and, more recently, in electronic components and batteries. Environmental exposure to lead (Pb) has been linked to risk of Alzheimer’s Disease (AD) and related dementias, although the mechanism is unknown.^1–3 Pb is a versatile contaminant, with human exposure occurring via the inhalation of dust, or through its ingestion in soil, food, or drinking water. Although Pb has long been known as a neurodevelopmental toxicant in children, substantial toxicological and epidemiological research indicates that cumulative environmental Pb exposure in adulthood also has adverse effects on health, and may be a significant contributor to age-related neurologic dysfunction.⁴ Studies conducted on cohorts of workers who have been occupationally exposed to Pb showed that brain atrophy and neurobehavioral deficits were present years after Pb exposure.^3,5 For example, welders with a high blood lead level (BLL) show signs of early AD changes in brain.⁶ Pb exposure remains a problem in many American communities due to the persistence of Pb in the environment,⁷ and contaminated water continues to be a major public health issue.

Despite tremendous progress in reducing Pb exposure, substantial neighborhood disparities persist, with this burden falling disproportionately on lower income areas.⁸ Communities of color and residents in rural and urban disadvantaged areas are still exposed to high levels of environmental lead.⁹ Although Pb exposure has been in decline, there remains a persistent gap between the non-Hispanic Black and non-Hispanic white populations,¹⁰ and Black adults have higher Pb exposure than either white or Hispanic individuals.¹¹ Burdens of Pb exposure are unequally shared because of racial residential segregation, concentrated poverty, and housing market discrimination; what’s more, under resourced areas lack the means to effectively eliminate hazards such as Pb from their environment.^12,13

The biological mechanism underlying the link between Pb exposure and dementia remains unknown. Some epidemiologic studies have shown an association between BLL and stroke ¹⁴ and other forms of cardiovascular and cerebrovascular disease (CVD).^15–18 However, a major consequence of chronic Pb exposure is hypertension (reviewed extensively in ¹⁹), and chronic hypertension is a well-known, strong risk factor for dementia.²⁰ Although this could indicate that Pb exposure contributes to vascular complications of cognitive impairment and dementia (VCID), a causal link has yet to be demonstrated. It has been proposed, based on a limited number of animal studies that Pb exposure causes epigenetic changes in the methylation state of DNA,^21–24 and that this or related epigenetic changes might relate to the link to AD and related dementias (ADRD).^25,26 These studies concern early life effects of Pb exposure; there is no evidence that this mechanism contributes substantially to a link between ADRD and exposure in adulthood.

There are a limited number of studies examining the consequences of adult Pb exposure in AD-relevant mouse models.^27,28 One study, conducted on APP^V717F mice, was an acute exposure and lead to a small increase in amyloid-β (Aβ) in the brain ~24 hours later.²⁸ Another study,²⁷ in TgSwDI mice, extended this work to chronic Pb exposure, and resulted in both an increase in Aβ and a Morris water maze (MWM) impairment. The authors hypothesized that this effect could be due to an effect of Pb on amyloid fibrilization, although their data only show this occurring at mM concentrations of Pb – far beyond any level that might be reasonably expected in human brain. Although the authors showed an increase in Aβ in TgSwDI mice, they did not examine cerebrovascular pathology, even though it is a defining feature of this model.²⁹ Hence, in spite of its established epidemiologic links to cerebrovascular disease (CVD) and likely connection to ADRD, it remains unknown if Pb exposure will result in VCID.

Although this may point towards Pb as a driver of vascular complications of cognitive impairment and dementia (VCID), the limited amount of prior research in this area has been focused near exclusively on the traditional AD pathologic hallmarks of amyloid plaques and neurofibrillary tangles (NFTs). This raises the intriguing possibility that Pb exposure is connected to dementia by worsening CVD on a background of either pre-existing or developing AD neuropathology. Alternatively, Pb exposure could lead to cognitive impairment through AD-related pathways that are not directly connected to these traditional AD pathologic indices. In order to test this hypothesis, we chose a mouse line (an APPxPS1 double mutant knock-in, with an additional Lepr^db/db mutation) prone to develop both Aβ pathology, as well as CVD pathology and profound cognitive impairment.³⁰

METHODS

Animals and Treatments

For this study, we used a mouse model that we developed to study the complex interactions between AD and vascular co-morbidities in the brain.³⁰ This mouse line was made by crossing APP × PS1 double mutant knock-in (KI) mice (APP^ΔNLh/ΔNLh × PS1^P264L/P264L; ‘AD’),^31–33 with Lepr^db/db mice (‘db’),³⁴ resulting in the ‘db/AD’ lineage. The APP KI mouse contains both the Swedish ΔNL mutation at the β-secretase cleavage site, as well as a humanized Aβ sequence, and this was subsequently crossed with the gene targetted PS1 mutant line.^35,36 Both the AD and db/AD mice have significant cerebrovascular disease (CVD), as well as amyloid ³⁰ and tau ³⁷ pathology. Although substantial parenchymal Aβ pathology is present in these mice, cerebral amyloid angiopathy (CAA) is not a major feature of this model. The presence of the Lepr^db/db genotype does not lead to an overall increase in amyloid load (detectable from ~3 months of age), although it did lead to increased CVD (by ~10–11 months of age) and a parallel cognitive impairment as measured in the Morris water maze (MWM) as compared to wild type (‘WT’) mice. The db/AD mice are also normotensive.³⁰ Our findings in the APP × PS1 KI mice (AD), and in their db/AD variant, lead us to the idea that they might be useful for studying factors that might be involved in the occurrence of vascular co-morbidities that could impact the development of age-related dementia. One of the strengths of this mouse line is that they are not an overexpressing transgenic line, which could be an asset in the study of Pb exposure where one possible mechanism might be an effect on AD-related gene expression.

Tail snips were collected prior to weaning to determine genotypes and to make group assignments. Some Lepr^db/db and APP genotyping was performed by Transnetyx (Cordova, TN), although most was performed locally, as was all PS1 genotyping. Tissue was processed using the Wizard Genomic DNA kit (Promega; Madison, WI). APP and PS1 genotyping were performed by PCR as described,³⁸ except in this case using Platinum^® PCR High Fidelity Supermix (Invitrogen). Lepr^db/db genotyping was performed using a Taqman^® SNP genotyping kit (Applied Biosystems; Grand Island, NY). Mice of the appropriate genotypes were assigned randomly to the different experimental groups, while also maintaining a balanced distribution of males and females in each condition.

We exposed adult mice to Pb concentrations that had previously been shown ^23,39 to affect gene expression and neuropathology. Exposure via drinking water (0.2% Pb Acetate; Sigma-Aldrich) began at ~3 months of age and continued for ~3 months; control animals received normal, uncontaminated water. This timeframe covers the initial age of onset of primary amyloid pathology in this line. This concentration of Pb was chosen to attain a BLL approximately equivalent to what would be considered as cause for concern in adult humans (Occupational Safety & Health Administrations Lead. 1910.1025 and 1926.62.; Washington, DC: OSHA); according to 1926.62(j)(2)(i)(B): “[this applies to] each employee covered under paragraphs (j)(1)(i) or (ii) of this section whose last blood sampling and analysis indicated a blood lead level at or above 40 μg/dL, at least every two months”.

A total N of 136 mice were used in the overall experiment (66 Pb exposed, 70 control), broken down as follows [sex is also indicated for each group, female (F) or male (M)]: Lepr^+/+ × APP^+/+ × PS1^+/+ (WT), 13 control (8F, 5M), 14 Pb exposed (9F, 5M); Lepr^+/+ × APP^ΔNLh/ΔNLh × PS1^P264L/P264L (AD), 22 control (9F, 13 M), 25 Pb exposed (11F, 14M); Lepr^db/db × APP^+/+ × PS1^+/+ (db), 16 control (8F, 8M), 16 Pb exposed (8F, 8M); Lepr^db/db × APP^ΔNLh/ΔNLh × PS1^P264L/P264L (db/AD), 15 control (8F, 7M), 15 Pb exposed (7F, 8M). The animal numbers required for the current study were estimated based on the results of our earlier study.³⁰ Pb exposure had no significant effect on mortality or morbidity: over the course of the study, 4 AD (1 control, 3 Pb exposed) and 4 db/AD (3 control, 1 Pb exposed) animals died. Two additional male AD mice (both Pb exposed) were euthanized for excessive aggression. As Lepr^db/db mice are also overweight and have metabolic issues, we regularly weighed the mice (approximately once/week) and performed routine glucose tolerance tests.³⁰ Briefly, after fasting for ~6 hours, mice were injected with 2 mg of glucose / g of body weight, and their blood glucose levels monitored (using a commercial glucose meter) for 120 minutes, via sampling a drop of blood from the tail. We did not observe any differences in water consumption between mice drinking the Pb contaminated water versus uncontaminated water (not shown).

At the end of the study, all animals were euthanized by intraperitoneal sodium pentobarbital injection (150 mg/kg, Beuthanasia-D). Blood was collected into BD vacutainer trace element tubes (lead free) for ICP-MS analysis. One hemibrain was frozen on dry ice and stored at −80° C, while the other was fixed in 10% buffered formalin and stored at 4° C, until analyzed. All animal work was approved by the University of Kentucky Institutional Animal Care and Use Committee (IACUC), and was fully compliant with environmental health and safety and public health service guidelines.

Longitudinal, Physiological, and Behavioral Measures

Blood Pressure (BP) was measured using the Kent CODA 8 (Kent Scientific; Torrington, CT). Animals were acclimatized to the tail cuff for five minutes on a warming platform, while restrained in a clear Plexglas^® tube, before starting recording session. BP was measured for a minimum of 3 consecutive days. Data were collected over twenty cycles (diastolic / systolic), with a 20 second inter-cycle interval. The apparatus was carefully cleaned between animals, and measures were performed in the morning during the second to last week of exposure.

At approximately the same time as BP measures, a subgroup of animals (N = 20) underwent magnetic resonance imaging, as previously described.³⁰ These were all of the db/AD genotype, as these are the most vulnerable to developing vascular pathology; 8 males (5 Pb exposed, 3 controls) and 12 females (6 Pb exposed, 6 controls) were imaged. Briefly, animals were anesthetized under constant oxygenated desflurane (~3–5%), and imaged using a horizontal bore Bruker Clinscan (7.0T, 30 cm, 300 MHz: Billerica, MA) imager equipped with a triple-axis gradient (630 mT/m and 6300 T/m/s) and a helium-cooled 14K quadrature head cryo-coil, cooled to 20° K. Physiological parameters were continuously monitored, including body temperature and rate of respiration. T2*-weighted images were acquired with a 2D GRE sequence with at 34 μm × 34 μm × 400 μm resolution, 15mm FOV, 25 degree flip angle, 10 averages, TR 165ms, and TE 15.3ms. At least ten equally-spaced images were taken of each mouse brain. Images were evaluated by individuals blinded to all treatment conditions, and were scored for vascular events as previously described.³⁰

During the final week of exposure, the mice were tested for spatial learning and memory capabilities using the Morris water maze (a circular pool, 135 cm in diameter, filled with room temperature water, ~25° C). A 11 cm circular platform was submerged (~1 cm) in the northeast (NE) quadrant; nontoxic tempura paint was used to make the water opaque in order to render the platform invisible. The pool was surrounded by a ring of dark curtains with a variety of suspended, external maze cues, and dim lighting. Mice were given 4 acquisition trials / day, for 5 consecutive days. Animals were released from one of the 4 cardinal points (N, S, E, or W) in a pseudorandom order to avoid sequence effects, such that each animal was released from each point once / day. Maximum trial length was 60 seconds; if an animal failed to find the platform during the trial, it was placed on the platform by the investigator. In either case, the mouse was allowed to remain on the platform for 20 seconds before being returned to its cage to await the next trial. Cages were situated on warming pads to ensure that the animals did not become hypothermic. After a 30 minute rest on the fifth day, a probe trial was performed where the platform was removed and the animal was allowed to swim freely for one minute while its behavior was recorded for analysis. Following an additional rest period, mice were tested for visual acuity by moving the platform to a different quadrant (SW) and affixing a prominent striped flag to it in order to mark its location. The mice were given two trials, each starting from the E location, in which to swim to the visibly cued location. Data were collected using EthoVision XT software (Noldus Information Technology; Leesburg, VA).

Mass Spectrometry

Pb concentrations in blood and brains were analyzed by ICP-MS (Agilent 7500cx; Santa Clara, CA, USA) after microwave digestion. Briefly, ~100 mg of brain or ~100 μL of blood were accurately weighed and digested in a 3:1 HNO₃:H₂O₂ mixture in a microwave at 100 °C for 10 minutes. Following digestion, the sample was diluted to 15 mL for ICP-MS analysis. The method detection limit for blood was (MDL) was 0.39 μg Pb/dL. For tissue, the method detection limit varied by sample mass and ranged from 0.001–0.002 μg Pb/g fresh weight. Quality controls included analysis of procedural blanks, intial and ongoing calibration verification, spike recovery, analysis of duplicates and analysis of a certified reference material (Tort-2, National Research Council of Canada for Brain, and toxic elements in caprine blood, SRM 955c, National Institute of Standards and Technology, Gaithersburg, MD, USA). Values below the method detection limit (MDL) were assigned a value of half the MDLs for data analysis.

Immunoassays and Western Blots

Frozen samples were first homogenized in radioimmuniprecipitation assay buffer (RIPA; 50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholate, 0.1% SDS, pH = 8.0; 1.0 ml/150 mg of tissue) containing Halt^™ complete protease inhibitor cocktail (ThermoFisher), using a PowerMax AHS200 homogenizer. The homogenate was centrifuged at 14,000 × g for 30 minutes at 4°C, and the supernatant collected and stored frozen at −80 °C until analyzed. Protein concentration was determined using a bicinchoninic acid assay (Pierce).

Aβ was measured by sandwich ELISA as previously described.^30–32 Briefly, 384-well plates (Immulon 4HBX; Thermo Scientific, Waltham, MA) were coated with 0.5 μg / well of Ab42.5 (against Aβ_1–16) in PBS (pH = 7.4) overnight at 4 °C, and blocked with Synblock (Serotec; Raleigh, NC) for two hours. RIPA extracted samples were diluted 1:10 in antigen capture buffer (AC; 0.2 M sodium phosphate (pH7), 0.4 M NaCl, 2 mM EDTA, 0.4% Block Ace (Serotec; Raleigh, NC), 0.4% BSA, 0.05% CHAPS, 0.05% NaN₃) for analysis. A standard curve was prepared from human Aβ₄₀ or Aβ₄₂, where appropriate, diluted in AC/RIPA buffer. Samples and standards were incubated overnight at 4 °C. After multiple washes with PBS + 0.1% Tween20, wells were probed with 0.5 μg/mL of either biotinylated 4G8 (against Aβ_17–24; BioLegend, San Diego, CA), Ab13.1.1 (end-specific for Aβ₄₀; [40]), or Ab12F4 (end-specific for Aβ₄₂; BioLegend, San Diego, CA), in detection buffer (DB; PBS, 2 mM EDTA, 1% bovine serum albumin, and 0.00002% thimerosal). Following a 2 hour incubation at room temperature, plates were again washed multiple times with PBS + 0.1% Tween20, and probed with 0.1 μg/mL neutravidin-HRP (Pierce Technologies; Rockford, IL). Following a one hour incubation at room temperature, plates were again washed multiple times with PBS + 0.1% Tween20, and developed with 3′,3′,5′,5′-tetramethylbenzidine (TMB; Kirkeguard and Perry Laboratories; Gaithersburg, MD). After approximately 10 minutes, the reaction was stopped with 6% o-phosphoric acid. The absorbance at 450 nm was measured with a BioTek multiwell plate reader. Standards and samples were measured in triplicate.

Proteins levels of APP, BACE1, PS1, phosphorylated and total tau were detected by western blot. The samples were heated at 95°C for 5 min, then proteins were separated in MOPS or MES running buffer using 4–12% acrylamide Bis-Tris Criterion^™ XT precast gels (BioRad Laboratories; Hercules, CA). Samples were electrically transferred to 0.45 or 0.20 μm nitrocellulose membranes, and membranes blocked at 4°C overnight in 5% nonfat dried milk in TBS + 0.1 % Tween20. Primary antibodies used (~1 μg / mL) were AbCT20 (APP ⁴¹); 22C11 (APP, ThermoFisher); 10E5 (BACE1, Cell Signaling); Tau46 (Total tau, Cell Signaling); AT8 (phospho-tau, Sigma-Aldrich); AC15 (βActin, Sigma-Aldrich); anti-PS1 Loop (PS1; EMD Millipore); Ab1 (BACE2, EMD Biosciences); Nic (Nicastrin, Abcam); anti-GAPDH (ThermoFisher). Immunoreactive bands were visualized with Super Signal West Dura chemiluminescence HRP substrate (Pierce) after incubation with HRP-conjugated secondary antibodies and exposed to film. Alternatively, anti-rabbit antibody and reagents (LI-COR) were used along with the Odyssey imaging system. Densitometric analyses were performed using Image J software. Expression was standardized to βActin or GAPDH in the same lane.

qRT-PCR

RNA was extracted from frozen tissue (~100 mg, whole hemibrain) using 1 mL of TRIzol^® (Invitrogen) followed by a phenol/chloroform extraction and ethanol precipitation; in some cases, samples were further purified using RNeasy columns (Qiagen; Valencia, CA). Expression of AD-relevant target genes was determined by two step qRT-PCR. First, strand complementary DNA (cDNA) was synthesized from the isolated RNA using iScript^™ Reverse Transcription Supermix (BioRad; Hercules, CA). This was followed by qPCR with PerfeCTa^® SYBR^® Green FastMix^® Low ROX^™ (Quanta BioSciences, Inc; Gaithersburg, MD). RT-qPCR was performed on a ViiA 7 Real-Time PCR System (Applied Biosystems; Foster City, CA). The geometric mean of the C_T values for RPL30, cyclophilin, and RNA polymerase IIJ was used as an internal control to calculate and compare relative expression (2^−ΔΔC_T). Gene specific primer sets ^30,31 were obtained from IDT (Coralville, IA).

Histology

Fixed hemibrains were sectioned on a vibratome (at 50 μm) and mounted on glass slides. Perl’s Prussian blue staining of hemosiderin was performed as described.⁴² Sections were digitally imaged using the Aperio ScanScope XT (Leica). Microhemorrhages were counted by two individuals blinded to group and genotype.

Data Analysis and Statistics

Data were analyzed with SPSS (v29, IBM Corp; Armonk, NY, USA) by ANOVA, using the general linear model (GLM) module with the independent variables genotype, Pb exposure, and sex (ages were evaluated as a covariate). For Morris water maze data, trials / days were treated as a repeated measure, and adjusted for deviations from sphericity where appropriate. Levene’s test for equality of error variances, and Kolmogorov-Smirnov and Shapiro-Wilk tests for testing normality, were used to check assumptions. Post-hoc multiple comparisons were conducted using Tukey’s test, Dunnett’s test, or similar. When only a subgroup of the total number of mice was used for a specific measure, the numbers are indicated in the figure legend.

RESULTS

Lead was readily detectable in mice blood after one month of exposure in drinking water, as measured by ICP-MS (Fig. 1A). Although we also detected Pb in the brain, it should be noted that we did not perfuse the mice, and it is likely that a significant portion of the detected Pb in the brain was due to the presence of residual blood in the cerebral vasculature. Exposure to Pb resulted in a marked increase in both systolic and diastolic blood pressure (Fig. 1B). We did not detect any difference in general health in mice consuming the Pb contaminated drinking water as compared to controls, and animal mortality was minimal in both experimental groups. As approximately half of the mice were Lepr^db/db mutation carriers, we also assessed the potential for Pb to alter the physiological response to glucose. Pb had no impact on the outcome of a glucose tolerance test in the mice (Fig. 1C). In this experiment, we also did not detect a difference between db+ animals and db− animals in this test, although we have in the past,^30,37 indicating that at least in this study the animals were not yet diabetic or pre-diabetic. It should be noted that the Lepr^db/db mice were noticeably obese (weight, in grams, ± s.e.m.: Lepr^db/db, 63.4 ± 0.9; Lepr^+/+, 37.5 ± 0.8; p<0.0001), indicating that metabolic issues would likely have occurred if the study had continued for a sufficient amount of time. Overall, Pb exposure did not affect body weight (Fig. 1D).

Figure 1. Pb Exposure Resulted in Increased Blood Pressure.

(A) [Pb] by ICP-MS (n=6 / group; data shown are for 3 Lepr^+/+, 6 Lepr^db/+, 9 Lepr^db/db, ½ M / F, divided equally between exposure groups; in this study, all mice were APP^+/+ × PS1^+/+, but we have not noted any genotype effects in this or other studies); as mice were not perfused, Pb detected in the brain may be due to residual blood (**). (B) ~3 months of Pb exposure caused an increase in blood pressure (Pb Treated/Control: 11/12; sex = 13F / 10M; genotype = 15 Lepr^+/+, 8 Lepr^db/db, 3 APP^+/+ × PS1^+/+, 20 APP^ΔNLh/ΔNLh × PS1^P264L/P264L); we did not see differences by genotype, but we did see some evidence that the effect might be larger in males (sex*treatment interaction, p=0.02), possibly since the males had a lower baseline BP than the females (p<0.01; not shown). (C) Pb did not affect either fasting glucose levels, or affect changes in blood glucose following a bolus injection of dextrose (N = 44, ½ M / F; 23 Lepr^+/+, 21 Lepr^db/db; 17 APP^+/+ × PS1^+/+, 27 APP^ΔNLh/ΔNLh × PS1^P264L/P264L). (D) Pb had no effect on body weight (note the slight decrease during the water maze testing period; baseline weights / group (g): control, 46.7 ± 0.8; Pb exposed, 45.1 ± 0.8).

During the final week of the exposure period, mice were tested for spatial learning and memory ability using the Morris water maze task. We have used this task to successfully measure cognitive dysfunction in rats and mice in multiple studies, including this mouse line.³⁰ Pb exposure had no effect on swim speed, nor did it affect performance on the cued version of the platform task, which tests whether mice can use visual cues to navigate (Fig. 2A). This outcome indicated that Pb had no impact on sensorimotor functions that might affect the ability of a mouse to perform the hidden platform version of the task. As previously reported,³⁰ genotype also had minimal impact on this test. Genotype did, however, affect the ability of the mice to learn the location of the hidden platform, with the db/AD mice performing worse than the 3 other genotypes (p<0.001; Fig. 2B), similar to what was previously reported.³⁰ However, Pb-exposed mice were impaired (p<0.003) at learning the location of the hidden platform (Fig. 2C). When groups were broken out by genotype, only the mice with the APP and PS1 mutations were vulnerable to the effects of Pb exposure. These effects were also seen in the probe trial, following removal of the platform (Fig. 2D). As expected, the mice with the APP^ΔNLh/ΔNLh × PS1^P264L/P264L genotype (those carrying AD-related mutations) performed worse compared to those that carried the wild type alleles (p<0.001); animals that also harbored the Lepr^db/db mutation were further impaired (db × AD interaction, p<0.03; AD vs. db/AD mice, p<0.05). The Pb effect was significant for both genotypes.

Figure 2. Pb Exposure Impaired Cognitive Function, but Only in Mutant Mice.

(A) Pb had no effect on either swim speed (all genotypes combined) or on the ability to locate a flagged platform (separated by genotype) in the Morris water maze; N = 135 mice: WT (Lepr^+/+ × APP^+/+ x PS1^+/+), n=27; db (Lepr^db/db × APP^+/+ x PS1^+/+), n=32; AD (Lepr^+/+ × APP^ΔNLh/ΔNLh × PS1^P264L/P264L), n=46; db/AD (Lepr^db/db × APP^ΔNLh/ΔNLh × PS1^P264L/P264L), n=30; Pb treated, n = 69 / untreated, n = 66; sex balance = 69F / 66M. Note again the lack of genotype differences on the visual cued task. (B) Mice were given 4 trials / day over 5 days in the MWM; consistent with our earlier results in this line, the db/AD mice were the worst at this task (shown is the overall main effect of genotype, regardless of Pb exposure). (C) Pb exposed mice were impaired at learning the location of the hidden platform (p<0.003); however, this effect was driven almost entirely by mice with APP and PS1 mutations (p<0.005), as wild type animals were unaffected. The Pb effect was particularly pronounced on the last day of training (N.B.: days 2 and 3 have been left off of the genotype figure due to space constraints). (D) APP × PS1 mice also showed a deficit on a probe test of memory for the hidden platform location and, as expected, the db/AD mice performed worse than AD mice; in both cases, Pb exposure made the deficit worse. We did not detect a difference on training trial escape latency between Lepr^+/+ and Lepr^db/db mice. (E) Representative mouse swim patterns (shown are both male, Lepr^+/+ × APP^ΔNLh/ΔNLh × PS1^P264L/P264L). * = p<0.05, ** = p<0.01; relative to comparison group.

Next, we sought to determine if the cognitive impairment was due to changes in standard indices of AD related neuropathology. Given prior reports indicating that a possible mechanism of Pb exposure might be changes in gene expression, we examined a panel of AD related genes by RT-PCR (Fig. 3A). We did not detect a significant effect of Pb exposure on any of these mRNAs after adjustment for multiple comparisons. Although we did detect a small decrease in expression in some genes when looking only at the unadjusted p-values, these were not consistent between experiments, nor did they align with the negative immunoblot results. If should also be noted that if the connection between Pb exposure and ADRD risk was through this mechanism, we would expect an upregulation of expression, rather than a downregulation (for example, in the case of APP this would be protective). This panel covered the substrate for Aβ (APP), as well as the critical enzymes involved in its production (β-secretase: BACE1 and BACE2; γ-secretase: PS1, PS2, APH1, PEN2, and NIC) and clearance (IDE, NEP, ECE1, ECE2). An examination of the amount of total extractable Aβ by ELISA showed that there was no difference between Pb exposed and control mice; Aβ₄₀ and Aβ₄₂ levels were similarly unchanged (Fig. 3F). Since mRNA levels do not necessarily predict the amount of translated protein, we also examined several targets via Western blot. We did not detect any difference in expression at the protein level (Fig. 4), and this was confirmed by densitometry (not shown).

Figure 3. Pb Exposure Did Not Affect AD-related mRNA Expression or the Amount of Amyloid In the Brain.

(A) There were no effects (all adjusted p-values were n.s.) of Pb on major AD-related genes by qRT-PCR (Amyloid Precursor Protein, APP; Anterior Pharynx-Defective 1, APH1; β-Amyloid Precursor Protein Cleaving Enzyme, BACE (1 or 2); Endothelin Converting Enzyme, ECE (1 or 2); Insulin Degrading Enzyme, IDE; Neprilysin, NEP; Nicastrin, NIC; Presenilin, PS (1 or 2); Presenilin Enhancer 2, PEN2). Of these genes, only APP (F[2,13] = 6.61, p=0.010) and PS1 (F[2,13] = 5.71, p=0.017) showed a significant difference in the unadjusted p-values. RNA were extracted from one hemibrain; we did not evaluate expression by region. Data shown are for 7 Lepr^+/+, 13 Lepr^db/+, 10 Lepr^db/db, 17 M / 13 F, divided between exposure groups as follows: Control (n=8), 0.004% Pb (n=13), 0.2% Pb (n = 9). All mice shown in this combined subgroup were APP^+/+ x PS1^+/+, and we have not observed any genotype differences. Note that this subgroup was exposed earlier, during weaning and pre-weaning period, in order to maximize chances to detect a difference in gene expression. (B) Aβ pathology was unchanged (human Aβ specific ELISA); WT (Lepr^+/+ × APP^+/+ x PS1^+/+) and db (Lepr^db/db × APP^+/+ x PS1^+/+) mice show minimal signal and are not shown.

Figure 4. Pb Exposure Did Not Affect Expression of AD-related Proteins.

(A) RIPA-extracted brains from Pb exposed mice were separated by SDS-PAGE, and blotted for multiple AD-related proteins. Shown are Tau (detected using Tau46; Cell Signaling), and APP (detected using AbCT20. Equal protein loading was verified by probing for βActin (AC15, Sigma).

A longer exposure was used to visualize CTFs (top); a lighter APP exposure is also shown (bottom) to better show intermouse variability. Note that KI mice can be clearly identified by an increased APP CTF signal, even though they do not over express APP. Representative blots are shown for a subset of mice; quantitation was performed using spot and western blots on all genotype, sex, and treatment combinations (n = 3 / subgroup; N = 48); we did not find any effects of Pb. (B) Pb exposure did not affect BACE1 (AD and WT are shown here, but other genotypes were also not different); equal protein loading shown through GAPDH levels.

Cerebrovascular pathology is well known to interact synergistically with AD-related pathology to cause dementia,⁴³ and the main hallmarks of AD (plaques and tangles) are present with CVD in up to 50% of dementia cases.⁴⁴ We originally developed this mouse model to study the complex interactions between AD and vascular co-morbidities in the brain.^30,37 The db/AD mice have a range of CVD pathologies – including spontaneous strokes and microhemorrhages, as well as amyloid pathology.³⁰ Although these mice show significant parenchymal Aβ pathology, they do not show evidence of substantial cerebral amyloid angiopathy (CAA), and have no appreciable Congo Red, Thioflavin S, or resorufin staining in cerebral blood vessels.³⁰ Even though these mice are prone to develop substantial microbleeds and other vascular pathologies with age, Pb exposure had no effect at this age (Fig. 5).

Figure 5. Pb Exposure Had No Effect on Vascular Events.

We did not detect any increase in the occurrence of microbleeds in the db/AD mice, either via MRI (top) or via examination by Prussian blue staining. We did detect some Prussian blue infarcts (an example is shown at higher magnification in the bottom left), but these were rare events. N = 36; 18 Lepr^+/+, 18 Lepr^db/db; 18 APP^+/+ x PS1^+/+, 18 APP^ΔNLh/ΔNLh × PS1^P264L/P264L; 12 M / 24 F; 12 Pb exposed, 24 control).

DISCUSSION

Collectively, our data suggest that although AD-related pathology (such as Aβ) may be necessary for the detrimental effects of Pb exposure in adult mice, the Pb exposure itself need not cause amyloid pathology to worsen or lead to changes in AD-related gene expression (at either the mRNA or protein levels). This was a somewhat surprising outcome, since it could mean that the impact of adulthood Pb exposure requires the presence of an AD-related vulnerability that we do not understand or were unable to detect at this younger age (~6 months). At this age, this mouse line does not show significant neuroinflammation, tau, or cerebrovasular pathology, and has only a relatively modest amount of deposited Aβ that is difficult to detect in the insoluble (FA) fraction. Although it is possible that these outcomes would change in response to Pb-exposure at older ages, they seem unlikely to be able to account for effects on cognitive function at this age. This outcome has important implications for our understanding of how low level, environmental exposure to Pb could affect the long-term development of neuropathology, and for our general understanding of ADRD pathogenesis.

Lead exposure is known to cause hypertension,^45–47 and elevated Pb in adults is associated with hypertension, diabetes, and cardiovascular disease-related mortality.¹¹ However – despite the widely known connections between hypertension, CVD, and ADRD – to the best of our knowledge, a link with Pb exposure has not been causally demonstrated. We should note here that the effect of Pb on blood pressure is both well documented ¹⁵ and multifactorial ⁴⁸, and the mechanism for this effect is beyond the scope of this study. We also cannot rule out the possibility that the relative increase in blood pressure might be lessened in comparison to equimolar sodium acetate in drinking water, although the overall effect is unlikely to change.^49,50 In this study, regardless of the increase in blood pressure in mice with and without AD-related mutations, only the mice carrying the APP and PS1 mutations displayed a Morris water maze deficit. Even though Pb exposure did not result in any changes in the amount of Aβ in the brain, our data imply that the presence of (or vulnerability to) amyloid pathology may be required for adult Pb exposure to impair cognitive function. Currently, no human neuropathology studies have explored the effects of Pb exposure in adulthood on ADRD, though the epidemiological links between BLL and cardiovascular and cerebrovascular disease suggest that the association with late-life dementia is probably more akin to vascular cognitive impairment, or VCID.^14–18 If this sequence of events is correct – that the long-term neurologic consequences of Pb exposure in adults is a form of VCID – then it is conceivable that using blood pressure lowering strategies might be a viable treatment strategy. This could be clinically significant, as chelation therapy to remove Pb has been ineffective at mitigating the long-term consequences of exposure.⁵¹

CAA is characterized by the deposition of Aβ within the walls of cerebral and leptomeningeal blood vessels, and is a common pathology in AD patients.⁵² CAA is well known to contribute to vascular dysfunction in the AD brain, and is a known cause of intracerebral hemorrhaging.^53–55 However, even in the absence of CAA, the presence of soluble Aβ in the brain can lead to impairment of vascular function.^56–60 The APP × PS1 mice that were used for this study do not exhibit CAA pathology.³⁰ However, other lines of genetically modified mice, such as TgSwDI (which express the human APP₇₇₀ isoform containing the Swedish ΔNL double mutant, as well as the Dutch E693Q and Iowa D694N mutations, under a Thy1 promoter ^61–64) or Tg2576 (which express the human APP₆₉₅ isoform containing the Swedish ΔNL double mutant, under a hamster PrP promoter ^60,65) have either dominant or mixed CAA pathology, respectively. Given that hypertension is a known risk factor for ADRD, it is possible that a mouse line with prominent CAA might eventually suffer from an increase in age-related CVD as a consequence of prolonged Pb exposure. Indeed, chemically induced hypertension causes increases in CVD, including microhemorrhages, in both TgSwDI and Tg2576 mice.^66–69 Although this possibility is beyond the scope of the current study, it represents a promising avenue for future research.

One of the most striking observations in this study was that although we saw increased blood pressure following Pb exposure, we only observed cognitive dysfunction in mice with Aβ pathology (either the AD or db/AD genotypes). Interestingly, when normal mice were subjected to the same Pb concentration (0.2%) and duration (3 months) as in this study, but exposure began prenatally, they displayed cognitive dysfunction in adulthood.⁷⁰ However, initiating Pb exposure in adult mice did not produce the same cognitive deficits, suggesting that Aβ pathology might be a prerequisite for Pb-induced cognitive dysfunction. A single study in TgSwDI mice showed an increase in Aβ and a Morris water maze impairment.²⁷ The authors hypothesize that this effect may be due to an effect of Pb on amyloid fibrilization, although their data only show this occurring at mM concentrations of Pb – far beyond the level that might be reasonably expected to be seen in the human brain. The role of trace metals ^71,72 in vivo is no longer believed to be a major factor in the development of AD-related pathology ⁷³. A possible way to address this discrepancy would be to remove Aβ from the brain, using an approach such as immunotherapy ^74–76, and determine if Pb exposure still causes cognitive dysfunction.

Recent clinical outcomes have provided some hope that targeting the Aβ peptide may be beneficial for treating AD,⁷⁷ indicating that animal models based on familial AD mutations are valuable for developing viable therapies. Interestingly, the findings in this study could also suggest that the ambiguous relationship between ADRD and Pb exposure might not involve Aβ at all. Instead, these observations could be due to an interaction with the APP mutation,⁷⁸ or perhaps involve a different γ-secretase substrate or signaling pathway beyond APP. It is possible that future studies might help to better elucidate this mechanism.

Highlights:

• Exposure to Pb in drinking water over several months does not affect expression of key AD-related genes in genetically modified mice, and does not affect key indices of AD-related neuropathology.

• Exposure to Pb in drinking water in adult mice causes an increase in blood pressure regardless of mouse genotype.

• Exposure to Pb in drinking water in adult mice causes cognitive impairment only in those mice that carry mutations in the AD-related genes APP and PS1, indicating that at least some predisposition to AD-related pathology is necessary.

DATA AVAILABILITY STATEMENT

The data supporting the findings of this study are available on request from the corresponding author.