Gene Deletion of Pregnancy-associated Plasma Protein-A (PAPP-A) Improves Pathology and Cognition in an Alzheimer’s Disease Mouse Model

Abstract

Alzheimer’s disease (AD) is a progressive neurodegenerative disease of age with no effective preventative or treatment approaches. Deeper understanding of the mechanisms underlying the accumulation of toxic β-amyloid oligopeptides and the formation of amyloid plaque in AD has the potential to identify new therapeutic targets. Prior research links the insulin-like growth factor (IGF) system to pathologic mechanisms underlying AD. Suppression of local IGF-I receptor (IGF-IR) signaling in AD mice has been shown to reduce plaque formation in the brain and delay neurodegeneration and behavioral changes. However, direct inhibitors of IGF-IR signaling are not a viable treatment option for AD due to the essentiality of the IGF-IR in physiological growth and metabolism. We have previously demonstrated a more selective means to reduce local IGF-IR signaling through inhibition of PAPP-A, a novel zinc metalloprotease that regulates local IGF-I bioavailability through cleavage of inhibitory IGF binding proteins. Here we tested if deletion of PAPP-A in a mouse model of AD provides protection against pathology and behavioral changes. We show that compared to AD mice, AD/PAPP-A KO mice had significantly less plaque burden, reduced astrocytic activation, decreased IGF-IR activity, and improved cognition. Human senile AD plaques showed specific immunostaining for PAPP-A. Thus, inhibition of PAPP-A expression or activity may represent a novel treatment strategy for AD.

INTRODUCTION

Alzheimer’s disease.

Alzheimer’s disease (AD) is a progressive and irreversible neurodegenerative disease of age that is associated with accumulation of extracellular amyloid plaques, neuroinflammation, formation of intracellular tau-containing neurofibrillary tangles and destruction of neurons in the brain, leading to severe memory loss and behavioral changes. This devastating disease takes a physical, mental, emotional, and financial toll on the patient as well as on their family, friends and caretakers. To date it is estimated that over six million Americans are living with AD, with medical expenses exceeding those for cancer and cardiovascular disease combined (www.alz.org). Unfortunately, the number of patients with AD is predicted to increase dramatically in the upcoming decades. Despite intensive research, there is no effective preventative or treatment strategy for AD. A better understanding of the mechanisms underlying the accumulation of toxic β-amyloid oligomers and formation of amyloid plaque in AD is likely to identify potential new therapeutic targets ^(1–5).

The IGF system and AD.

Prior research links the insulin-like growth factor (IGF) system to the pathologic mechanism underlying AD. There are numerous studies indicating that experimental AD in mice responds positively to reduction of local IGF signaling in the brain through genetic means. Thus, deletion or knock-down of IGF-I receptors (IGF-IR) or intracellular mediators of IGF-IR-activated signal transduction in various mouse models of AD resulted in a significant attenuation of pathological β-amyloid, neuro-inflammation, neuro-degeneration and subsequent behavior changes ^(6–8). Neuron-specific knock-out of IGF-IR correlated with sequestration of soluble Aβ peptides into dense aggregates of lowered toxicity, decreased neuro-inflammation and improved spatial memory ^{(6, 7, 9)}. Genetically ablating IGF-IR in neurons of the aging brain was protective against the neuro-inflammation, amyloid-β proteotoxicity, and memory impairment induced by intracerebroventricular injection of amyloidogenic Aβ oligomers ⁽¹⁰⁾. Furthermore, IGF-I impairs autophagy, and reduced IGF signaling helped to maintain autophagy in neurons with age and promote the clearance of toxic β-amyloid oligopeptides ^{(6, 7, 9)}. Thus, suppression of local IGF-IR signaling in AD mice can reduce plaque formation in the brain and prevent/delay neurodegeneration and associated behavioral changes. However, IGF-IRs are ubiquitous and serve fundamental roles in physiological growth and metabolism, therefore direct inhibitors of IGF-IR signaling are not a viable option for AD. A more selective means to reduce local IGF-IR signaling is through inhibition of pregnancy-associated plasma protein-A (PAPP-A).

PAPP-A.

PAPP-A was originally discovered as a protein found at high levels in the plasma of pregnant women, hence the name. However, PAPP-A was subsequently found to function outside of pregnancy as well ⁽¹¹⁾. PAPP-A is a zinc metalloprotease that can increase local IGF-I bioavailability through cleavage of inhibitory IGF binding proteins, in particular IGFBP-4 ^{(11, 12)}. PAPP-A is a secreted protein that tethers to the surface of cells through heparan sulfate-like proteoglycan moieties in an autocrine/paracrine fashion. IGF bound to IGFBP-4 is unable to activate receptors. However, upon cleavage of IGFBP-4 by PAPP-A, IGF is liberated from the complex in the pericellular environment and IGF signaling is enhanced. Conversely, inhibition of PAPP-A expression or its proteolytic activity represents an alternative approach to decreasing local IGF availability resulting in moderate restraint of IGF-IR signaling. Inhibition of PAPP-A through gene deletion in mice has many beneficial effects, including a remarkable extension of lifespan by 30–40% and delayed onset and reduced severity of several age-related degenerative diseases in both male and female mice ⁽¹³⁾. Under basal conditions, these PAPP-A knock-out (KO) mice were not significantly different from their wild-type littermates in central regulation of coordination, learning and memory, and metabolism ^{(14, 15)}. Although PAPP-A KO mice are smaller, they exhibit no secondary endocrine abnormalities and have normal circulating levels of IGF-I ⁽¹⁶⁾. Furthermore, longevity in these mice is not determined by body size ^{(17, 18)}.

It has been suggested that approaches extending longevity may be protective in neurodegenerative diseases and provide a novel treatment strategy for AD ^{(19, 20)}. We propose PAPP-A inhibition and consequentially reduced local IGF signaling as such an approach ⁽²¹⁾. Interestingly, Zullo et al. ⁽²²⁾ recently reported that lower overall neuronal excitation is a major determinant of lifespan expansion, and that it is tied to reduced brain IGF signaling. There is also evidence linking neural over-excitation to AD ^{(23, 24)}. Therefore, it is not unreasonable to speculate that specific inhibition of PAPP-A expression or its proteolytic activity could be a novel treatment strategy for AD.

In this study we tested the hypothesis that deletion of the PAPP-A gene protects against AD-like pathology and behavior in a mouse model of AD.

MATERIALS AND METHODS

Mice.

All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Mayo Clinic, Rochester, MN. The role of IACUC is to ensure adherence to the Animal Welfare Program and The Guide for the Care and Use of Laboratory Animals.

APPswe/PS1dE9 transgenic mice ⁽²⁵⁾ (AD) were purchased from The Jackson Laboratory (Bar Harbor, ME; MMRRC stock number 34829-JAX). These bi-transgenic mice express chimeric mouse/human mutant amyloid precursor protein (APP) and mutant human presenilin (PS1dE9) controlled by a mouse prion protein promoter and directed to CNS neurons. The co-integrated transgenes breed as a single allele and are associated with early-onset amyloid plaque. The swe mutation increases APP and its processing through the β-secretase pathway, and the PS1dE9 mutation shifts further APP processing by γ-secretase to produce longer, more pathogenic Aβ oligomers, i.e., Aβ_1–42 and Aβ_1–40 ^{(6, 25, 26)}. These mice do not exhibit nuerofibrillary tangles ⁽²⁷⁾. The PAPP-A KO mice used in this study were derived from heterogeneous breedings ⁽¹⁶⁾. The hemizygous AD mice were crossed with PAPP-A KO mice to get mice heterozygous for PAPP-A with or without AD transgenes. Mice were then crossed to get four genotypes for the study: wild-type (WT), AD, PAPP-A KO, and AD/PAPP-A KO mice. Mice were housed in a pathogen-free mouse facility with up to 5 males or 5 females per cage, without separation by genotype. They were maintained ad libitum on standard chow diet (PicoLab^® Rodent Diet 20) and water. DNA was isolated from tail snips at weaning to identify genotype and again at harvest to confirm genotype, using previous primer sets for PAPP-A ^{(13, 16, 17)} and recommended primers for APP and PS1dE9 (Jackson Labs). We entered 30 or more male and female mice per genotype into the study and aged for 12 months. Losses before 12 months were ~10% per group, except for female mice without AD where there were no losses.

After behavioral testings, mice under deep anesthesia were transcardially perfused with ice-cold phosphate-buffered saline (PBS). Brains were removed with the left hemisphere being immersion-fixed in a large volume (20X that of tissue volume) of neutral phosphate-buffered 10% formalin for subsequent histology and immunohistochemistry. The other hemisphere was flash frozen for Aβ-oligomers.

In a second set of aged mice, blood was collected by cardiac puncture from WT, AD, PAPP-A KO, and AD/PAPP-A KO mice for measurement of IGF-I serum levels by ELISA (R&D Systems, MG 100). Brains were rapidly removed, put on ice, rinsed with ice-cold PBS and cortical/hippocampal sections isolated and flash frozen for gene analyses.

Histology and immunohistochemistry.

Fixed brain samples were paraffin-embedded and sectioned (coronal, 5 μm) at mouse brain ‘Level 3’ of the Brain Nonneoplastic Lesion Atlas, which includes cortex and hippocampus ⁽²⁸⁾, using Standard Operating Procedures at the Small Animal Histology Core at Mayo Clinic, Scottsdale. Sections were H&E stained for histology and immune-labeled using specific antibodies for Aβ-amyloid plaque (82E1; RRID AB_10707424; Immuno Biological Laboratories #10323; 1:100 dilution). Visualization by ImmPACT Novo Red (Vector laboratories). Histology slides were reviewed by a board-certified anatomic veterinary pathologist (NMG). Analyses were done on one tissue per mouse, with 3–4 fields per tissue depending on the size of the specimen. Fields were adjacent on the tissue slide. Plaques were manually counted and stratified by location, i.e., cortex and hippocampus. Reactive astrocytes were identified in the cortex using immunohistochemical staining for glial fibrillary acid protein (GFAP; RRID AB_2924325; Abcam ab207165; 1:400 dilution). Visualization by DAB (3,3’-diaminobenzidine, Abcam). Stained slides were scanned at 20X using Motic Easy Scanner. Measurements were made by manually tracing structures or regions of interest using mean grayscale intensity measurements. Area and density were then analyzed by Fiji ImageJ software program, version 2.140. Histograms of plaque area were constructed as visual representations of the quantitative area data. The range of values were binned into a series of intervals and then counted as to how many values fell into each interval.

ELISAs.

Mouse Aβ_1–40 and Aβ_1–42 levels were measured in duplicate by specific ELISA kits (Invitrogen Aβ_1–40 KMB3481, Aβ_1–42 KMB3441) following manufacturers’ instructions. Briefly, brain tissue was homogenized in 5 M guanidine-HCl/50 mM Tris (8 volumes per grams of tissue). The two ELISAs had the same basic steps except for different detector antibodies and standard curves provided by the manufacturer: 8–500 pg/ml for Aβ_1–40, dilution 1:50. 3–200 pg/ml for Aβ_1–42, dilution 1:100. Reactions were read at an absorbance of 450 nm.

RT-qPCR.

Total RNA was extracted as previously described ^{(29, 30)}. Briefly, tissue sections were treated with Trizol (Ambion Life Technologies, Carlsbad, CA) and further processed as per manufacturer’s instruction. RNA (1 μg) was reversed transcribed with the SuperScript III First-Strand Synthesis System (Life Technologies) and evaluated in duplicate by quantitative real-time PCR using the CFX Connect Real-Time System with iTAQ Universal SYBR Green Supermix (Bio-Rad, Hercules, CA). Amplification plots were analyzed using CFX Maestro Software version 4.1 (Bio-Rad). Amplified PCR products were purified through QIAquick Gel Extraction Kit (Qiagen Hilden, Germany), quantified and serial diluted from 10⁸ to 10³ molecules. Relative quantification and fold changes were based on the standard curve for each gene, i.e. IGFBP-5/TBP. TBP was chosen for the reference gene because the efficiencies and relative expression were similar to the target gene. Primers for mouse IGFBP-5 and TBP reference genes: Mus musculus insulin-like growth factor binding protein 5 (Igfbp5), NM_010518.2, Forward: gaacactgcccaccccagag, Reverse: ccacgggagggcttacactg. Mus musculus TATA box binding protein (m.TBP), NM_013684.3, Forward: ctcagttacaggtggcagca, Reverse: cagcacagagcaagcaactc.

Novel Odor Recognition.

The Novel Odor Recognition (NOR) tests for changes in non-spatial short-term memory in an open field and takes advantage of a mouse’s innate preference for novelty by assessing the ability to discriminate between familiar and new ^{(31, 32)}. It was performed as previously described ⁽³³⁾, except 100% pure essential oils (Lemon or Eucalyptus & Rosemary) were used randomly as the novel scent. In brief, mice from each group were acclimatized to a 50 cm × 50 cm testing environment for a period of two minutes. After acclimatization, the mice were removed, the testing area cleaned with hydrogen peroxide, and two identical scents were placed in either corner of the testing area approximately 5 cm from either wall. Mice were reintroduced and the number of visits and the time spent at each scent were recorded for a period of 5 minutes (training). Recording was performed from above by live tracking cameras (Panasonic WV-CP294) and analyzed with Top-Scan Version 3.00 (Clever Sys). For testing, the mice were removed, the testing area cleaned with hydrogen peroxide, and one scent was replaced with a novel scent #2. The mice were reintroduced after a 90 minute interval and the number of visits (frequency) and total time spent at each scent (duration) were recorded as for training. Data are expressed as a Discrimination Index (DI), i.e., frequency and time at novel scent #2 divided by frequency and time at training scent #1.

Stone T-maze.

A standard protocol by Pistell and Ingram for the Stone T-maze (STM) was used to assess spatial learning and long-term memory over seven days ⁽³⁴⁾. The STM has black acrylic sides and a clear acrylic ceiling. It is constructed so that mice need to wade through water 1.8 cm deep and 22–24°C to reach a dry dark box ⁽³⁴⁾. Thus, the STM exploits a primary motivation of mice -- escape to a safe place – in this case a location dry and dark. As previously described ⁽¹⁴⁾, mice first undergo a straight run training to establish the concept that moving forward would allow them to escape from the water and light into the goal box. Any mice that were unable to reach the goal box in 15 s or less on 6 trials were excluded from further testing. Acquisition trials in the maze were performed the next day and consisted of 6 trials to learn the correct sequence of left and right turns to reach the goal box. The primary measures of learning were the time taken to reach the goal box (latency) and the number of errors committed before reaching the goal box. An error was noted with the complete entry of a mouse’s head into an incorrect path. During acquisition, if a mouse failed 3 times to reach the goal box within 3 min then the trial was terminated and the mouse removed from study. Memory retention was evaluated one week following acquisition. As prospectively designed, group failures >20% of total number of mice were excluded from further analyses (supplemental Table 1).

Human brain tissue.

Four slides of human brain AD tissue were obtained from Dr. Ronald Petersen, PI of the Mayo Clinic Study of Aging (U01 AG006786) and Mayo Clinic Alzheimer’s Disease center (P30 AG062677). These were obtained from AD patients with documented amyloid plaque pathology of 5 (on a scale of 1–5). Slides of cortex and hippocampus were stained for human PAPP-A, as previously described ^{(35, 36)}.

Biostatistics.

In general, primary analyses were comparisons between AD and AD/PAPP-A KO mice. F testing indicated normal distribution. Student’s t-test was used for statistical comparisons between AD and AD/PAPP-A KO mice. Results from WT and PAPP-A KO mice were included to control for any non-AD effects. Non-parametric Mann Whitney testing was used in NOR to compare the two groups. For STM, differences between AD and AD/PAPP-A KO mice were evaluated by Student’s t-test. ANOVA with Tukey was used for multiple comparisons. Time-related data were analyzed by repeated measures ANOVA. Significance was set at P < 0.05.

RESULTS

Phenotype of WT, PAPP-A KO, AD and AD/PAPP-A KO mice.

Table 1 presents the body and brain weights of WT, PAPP-A KO, AD, and AD/PAPP-A KO mice at harvest. The PAPP-A KO growth genotype in both male and female mice clearly dominated with 30–40% reduction in body weights of PAPP-A KO and AD/PAPP-A KO mice compared to WT and AD mice (P < 0.0001). PAPP-A KO mice are born as proportional dwarfs ~60% the size of WT littermates and retain this reduced size difference through adulthood ⁽¹⁶⁾. The AD transgenes had little or no effect on the growth phenotype. Brain weights of AD mice were reduced by 15–20% of non-AD (P < 0.0001), as expected ⁽³⁷⁾. Serum IGF levels did not vary significantly among the different groups (supplemental Table 2).

Table 1.

Body and brain weights of ~ 12-month-old mice at harvest

	Body wt (g)	Brain wt (g)
Males
WT	36.4 ± 1.02	0.30 ± 0.005
PAPP-A KO	24.2 ± 0.72	0.25 ± 0.005
AD	42.3 ± 0.88	0.31 ± 0.005
AD/PAPP-A KO	24.6 ± 1.22	0.27 ± 0.005
Females
WT	35.2 ± 1.48	0.30 ± 0.006
PAPP-A KO	21.7 ± 1.18	0.26 ± 0.004
AD	35.0 ± 1.61	0.30 ± 0.006
AD/PAPP-A KO	23.8 ± 1.43	0.25 ± 0.005

Results are mean ± SEM, n = 10–15 per group

Consequences of PAPP-A gene deletion on brain pathology in AD mice.

Immunohistochemistry.

Brains were removed with the left hemisphere immersion-fixed in neutral phosphate-buffered 10% formalin for plaque, as well as for reactive astrocytes that accumulate around amyloid plaque and are associated with neuro-inflammation and toxicity ^{(38, 39)}. Results for Aβ plaque staining are shown in Fig. 1. There were no detectable plaques in brains of WT or PAPP-A KO mice. In general, visible plaque deposits were more prevalent in the cortex than the hippocampus of AD mice (supplemental Table 3). In male AD mice, plaque number was 27 ± 2 (mean ± SEM). In male AD/PAPP-A KO mice, plaque number was 9 ± 2. The 67% difference in number of plaques was highly significant (P = 0.004). Female AD mice had twice the number of plaques than male AD mice. Female AD/PAPP-A KO mice had reduced number of plaques compared to AD mice (~40%), but the difference was not statistically significant. As shown in the histograms, plaques were of varying sizes but there were more very large area plaques (>2000 pixels²) in male AD compared to AD/PAPP-A KO brains when normalized by total plaque number (18% vs 6%). Percentage of small plaque sizes (0–500 pixels²) did not differ between the two male groups (Fig. 2A). In female AD and AD/PAPP-A KO mice (Fig. 2B), there was no difference in the large plaques, but the percentage of small plaques was reduced in AD/PAPP-A KO compared to AD mice (6% vs. 13%). In both males and females, the distribution was skewed toward the left. Plaque intensity values were widely varied in both male and female AD and AD/PAPP-A KO brains (data not shown). Examples of plaque images are presented in supplemental Fig. 1.

Figure 1.

Total number of Aβ plaques in brains of AD (orange) and AD/PAPP-A KO (blue) mice

Fixed brain samples from mice ~12-months of age were paraffin-embedded and coronal sections were immunostained using a specific antibody for Aβ-amyloid plaque (82E1). Data are mean ± SEM, n = 8. Student’s t-test

Figure 2.

Histograms of plaque area (pixels²) in brains of male (A) and female (B) AD (orange) and AD/PAPP-A (blue) KO mice. Frequency of plaques in the different bin sizes were assessed.

n = 8 per group for males and n = 5 per group for females

Reactive astrocytes are found close to amyloid plaques and play a major role in neuroinflammation in AD ^{(38, 40)}. Utilizing GFAP antibodies that recognize reactive astrocytosis, staining of stellate-like cells in the cortex was significantly reduced in AD/PAPP-A KO brains compared to AD in both males and females (P = 0.025 and P = 0.038, respectively). WT and PAPP-A KO brains showed minimal GFAP staining. Representative stainings are shown in Fig. 3.

Figure 3.

Immunohistochemistry of mouse brain for astrocytosis

Fixed brain samples from mice ~12-months of age were paraffin-embedded and coronal sections were immunostained using a specific antibody for reactive astrocytes (GFAP). Representative images of GFAP at 10X.

Aβ oligomers.

Table 2 presents Aβ oligomer levels in AD and AD/PAPP-A KO brains. In males, Aβ_1–40 levels were reduced by 33% in AD/PAPP-A KO brain compared to AD (P = 0.002). Aβ_1–42 trended toward lower levels in male AD/PAPP-A KO brain than in AD, but the difference was not statistically significant. Females showed no difference in Aβ_1–40 levels between groups but showed a significant increase (33%) in AD/PAPP-A KO brain compared to AD.

Table 2.

β-amyloid oligopeptides (ng/mL)

	Aβ1–40	Aβ1–42
Males
AD	9 ± 1.0	50 ± 3.2
AD/PAPP-A KO	6 ± 0.7*	41 ± 4.0
Females
AD	11 ± 0.5	63 ± 2.8
AD/PAPP-A KO	11 ± 0.8	84 ± 7.3*

Aβ_1–40 and Aβ_1–42 levels in brain were determined by specific ELISA kits.

Results are mean ± SEM, n = 9.

^*P < 0.05

IGF-I activity.

Expression of an IGF-responsive gene, such as Igfbp-5 ⁽⁴¹⁾, can be used as a surrogate indicator of local IGF signaling in vivo. Igfbp-5 is transcriptionally upregulated with IGF stimulation and downregulated with IGF inhibition in several cell types and tissues, including brain ⁽⁴²⁾. Decreased IGFBP-5 mRNA abundance has been used to indicate decreased IGF-I signaling associated with PAPP-A deficiency in kidney, aortic plaque, and neointima ^(43–45). In this study, Igfbp5 was significantly reduced in the cortex/hippocampus of male AD/PAPP-A KO brains compared to AD (P = 0.006). Igfbp5 expression levels in brain of AD/PAPP-A KO mice were not significantly different from WT and PAPP-A KO mice (Fig. 4A). In female mice (Fig. 4B), there was reduced Igfbp5 expression in AD/PAPP-A KO mice compared to AD mice (P = 0.023), but expression in AD/PAPP-A KO mice was greater than in mice negative for AD genes (P = 0.008).

Figure 4.

Igfbp-5 gene expression in (A) male brains and (B) female brains

Brain cortical/hippocampal sections were harvested and processed for mRNA analyses by RT-qPCR. TBP was used as an internal reference control. Results are mean ± SEM, n = 9–12 (AD, AD/PAPP-A KO mice) and 5–9 (WT, PAPP-A KO mice).

ANOVA with Tukey

Impact of PAPP-A gene deletion on cognitive function in AD mice.

In the NOR test, male AD/PAPP-A KO mice were more inquisitive toward the novel scent #2 than AD mice in both frequency and duration, P = 0.005 and P = 0.017 (Mann-Whitney), respectively (Fig. 5A). Analyzed another way, 60–70% of AD/PAPP-A KO mice showed improvement in frequency and duration (DI >1), whereas AD mice showed ~20–30% improvement by CHI-square testing. WT male mice behaved nearly identically to AD mice with DI >1 in frequency and duration, 20% and 10%, respectively (supplemental Fig. 2) suggesting that elimination of PAPP-A mitigated the short-term memory loss in male AD/PAPP-A KO mice. However, in the absence of the AD genes, loss of PAPP-A had little effect on memory. Notably the overall distance traveled during testing did not differ between the AD/PAPP-A and AD mice (data not shown). In contrast to the males, there was no significant difference by NOR testing in female AD/PAPP-A KO and AD mice (Fig. 5B).

Figure 5.

Novel Odor Recognition in (A) male and (B) female AD (orange) and AD/PAPP-A KO (blue) mice

Data are expressed as a Discrimination Index (DI), i.e., frequency and time at novel scent #2 divided by frequency and time at training scent #1. Each dot represents an individual mouse. Mann-Whitney

For the STM, we recorded number of errors and time to reach the safe box (latency). Groups were only analyzed if the group had < 20% failures (set a priori to the experiment). Number of failures per group are presented in supplemental Table 1. Only male AD and AD/PAPP-A KO mice fit the criteria for statistical analyses. As shown in supplemental Fig. 3, there was no significant difference between AD and AD/PAPP-A KO mice in errors in acquisition runs or (data not shown) in duration of the runs. For evaluation of memory retention one week later (Fig. 6), AD/PAPP-A KO mice had 40% fewer errors and a 30% shortened latency to reach the box than AD mice. However, these did not reach statistical significance (P = 0.06 and P = 0.13, respectively).

Figure 6.

One-week memory retention - Stone T-maze

Number of errors and latency to target in male AD (orange) and AD/PAPP-A KO (blue) mice, n = 16. Student’s t-test

PAPP-A and human amyloid plaque.

Analyses of brain tissue from four patients with documented AD indicated strong immunostaining for PAPP-A in senile plaques. Figure 7 is a representative image of staining in the hippocampus of a 65-year-old male with advanced amyloid plaque pathology.

Figure 7.

Representative PAPP-A immunostaining of amyloid plaque in brain tissue from AD patients

Left panel: PAPP-A antibody

Right panel: IgG control

DISCUSSION

These studies are the first to generate and characterize a mouse model to study the role of PAPP-A, a novel zinc metalloprotease, in AD. We determined the effects of PAPP-A gene deletion on amyloid plaque burden, neuroinflammation, and memory, and found that AD/PAPP-A KO mice, compared to AD mice, had 1) Reduced amyloid plaque number in the cortex and hippocampus, along with reduced plaque size; 2) Reduced astrocytosis, indicative of neuroinflammation; 3) Reduced IGF-I activity in the brain; 4) Improved cognitive behavior. These findings were statistically significant in male mice, although females showed a similar trend. It is well established that AD pathology manifests differently among male and female humans ⁽⁴⁶⁾. In our study and others, plaque load in female AD mice exceeded that of male AD mice ^(47–49). Papazoglou et al. ⁽⁵⁰⁾ recently presented a sex-specific transcriptome profile using the same AD mouse model as in this study. The reason for this difference is poorly understood ⁽⁵¹⁾. It has been suggested that part of reason could be different neuroendocrine control of metabolism, sex hormones, and/or response of astrocytes to inflammatory challenges ⁽⁵²⁾. In regard to the latter, we found much greater astrocytosis in female vs. male mouse AD brains. It was interesting that male AD mice had more large-sized plaques than AD/PAPP-A KO mice but the same number of smaller plaques. On the other hand, female AD mice had the same number of large plaques but much fewer smaller plaques than AD/PAPP-A mice. The significance of these findings is unclear but may suggest that PAPP-A gene deletion influences plaque initiation in females and progression in males.

In males, brain Aβ-oligomers were reduced in AD/PAPP-A KO mice compared to AD mice, but in females Aβ_1–42 was significantly increased in AD/PAPP-A KO mice. This was unexpected, but we measured the levels at one point in time so we can only speculate on the dynamics of Aβ production, accumulation and clearance ^{(6, 53)}.

Neuroinflammation plays a critical role in AD pathology ⁽⁵⁴⁾. We have shown that deletion of the PAPP-A gene in AD mice reduces the burden of reactive astrocytes, as measured by immunostaining with GFAP. Thus, both amyloid plaque and associated reactive astrocytes that are known to contribute to neuroinflammation ^{(38, 40)} are significantly reduced in AD mice with PAPP-A gene deletion.

These differences between AD and AD/PAPP-A KO mice were also associated with decreased IGF-IR signaling in the cortex and hippocampus in brain. Igfbp-5 was used as a surrogate marker of IGF-IR activity in vivo ^(41–45) and was found to be significantly reduced in AD/PAPP-A compared to AD brain. It is important to emphasize the local effects of PAPP-A and IGF-IR signaling. There were no significant differences in circulating IGF-I levels in WT, AD, AD/PAPP-A KO, and PAPP-A KO mice. However, elevated circulating levels of IGF-I have been associated with beneficial effects on the aging brain ^{(55, 56)}. The underlying mechanism for this seeming discrepancy between local brain IGF and circulating IGF is not well understood. However, the findings underscore the importance of distinguishing between local and systemic regulation of IGF signaling in AD ^{(55, 57, 58)}. Furthermore, long-term blockade rather than enhanced IGF signaling supports neuronal function and neuroprotection ⁽¹⁰⁾.

We used NOR (non-spatial working memory) and STM (orientation and memory performance) to assess cognition in these mice. Male AD/PAPP-A KO mice showed significantly improved attention, working memory, and long-term spacial memory over AD mice. There were no significant differences in the female mice. Similar results for NOR testing was seen in another AD mouse model, i.e., there were differences in male but not female mice ⁽⁵⁹⁾. Male AD/PAPP-A KO mice also performed better than AD mice both in terms of latency and success in finding the escape box in STM. None of the female groups had <20% errors and, therefore according to our criteria, were not analyzed further. We suspect that the female mice were too far along in AD. Progressive deterioration of short-term memory occurs before losses in long-term reference memory in these AD mice ⁽⁶⁰⁾. Interestingly, the same seems to be true in humans.

This study provides foundational data and clues for more directed queries into mechanism. However, there are several limitations of this study: First, the studies were done using the APP/PS1 transgenic mouse model ⁽²⁵⁾. We acknowledge that no single mouse AD model provides an all-encompassing view of human AD. However, AD models can be used as reductionist tools to describe complex systems in terms of simple or fundamental mechanisms ^{(61, 62)}. We carefully considered the pivotal role of Aβ accumulation in the pathogenesis of the APP/PS1 mouse and speculated that Aβ might be a potential target for PAPP-A intervention as we showed for PAPP-A in the development of atherosclerotic plaque ⁽⁴⁴⁾. Indeed, the pathogenesis of AD and atherosclerosis may involve some common mechanisms ⁽⁴⁷⁾. Studies with PAPP-A overexpression in human atherosclerosis ⁽⁶³⁾ and PAPP-A gene deletion in mice ^{(44, 64, 65)} suggest a similar mechanism in plaque formation and progression. Second, we chose to look at a single time point when we would be better able to assess cognitive decline. Further time course studies are needed to assess mechanisms of PAPP-A on Aβ oligomer production, aggregation, and clearance in AD ^{(6, 9, 53)}. This knowledge would help our basic understanding of the dynamics of plaque formation, and the seemingly paradoxical results on Aβ oligomers in this study. Nevertheless, our plaque findings are strong. Third, we cannot rule out peripheral effects of the constitutive PAPP-A gene KO that could impact the brain. Our attempt to generate brain-specific PAPP-A KO mice failed due to germ line recombination ⁽³⁷⁾.

Despite great efforts made to identify pharmacologic interventions for the treatment of AD, the disease remains the only top contributor to mortality without a disease modifying intervention, aside for a recent provisionally FDA-approved drug, lecanemab. A large part of the lack of progress has been the inability to translate findings in preclinical mouse models to humans. Herein, we present pathological and behavioral characterization of a novel AD/PAPP-A KO mouse model for valuation of PAPP-A and its effect on local IGF signaling in the brain, and demonstrate that loss of PAPP-A expression reduces pathological and cognitive changes in a mouse model of AD. Moreover, we present evidence for an association of PAPP-A and human plaque in AD. These data support development of novel neurodegenerative therapies based on targeted inhibition of PAPP-A to prevent/delay AD progression and possibly other neurological diseases.

Supplementary Material

1

Supplemental Figure 1. Fixed brain samples from mice ~12-months of age were paraffin-embedded and coronal sections were immunostained using a specific antibody for Aβ-amyloid plaque (82E1).

2

Supplemental Figure 2. Novel Odor Recognition in male and female WT (grey) and PAPP-A KO (green) mice.

Data are expressed as a Discrimination Index (DI), i.e., frequency and time at novel scent #2 divided by frequency and time at training scent #1.

3

Supplemental Figure 3. Acquisition - Stone T-maze

Number of errors in (A) male and (B) female AD (orange) and AD/PAPP-A (blue) KO mice during the acquisition phase.

4

Supplemental Table 1. Failures in Stone T-maze during acquisition trials

5

Supplemental Table 2. Serum IGF-I

6

Supplemental Table 3. Number of plaques in the cortex and hippocampus

HIGHLIGHTS.

• Generations of an AD mouse model of amyloid plaque accumulation with the PAPP-A gene knocked out.

• Compared to AD mice, AD/PAPP-A KO mice had reduced plaque number and size, reduced astrocytosis, and reduced brain IGF activity

• Compared to AD mice, AD/PAPP-A KO mice had improved cognitive behavior.

• Human amyloid plaque immunostained for PAPP-A

ABBREVIATIONS

PAPP-A

pregnancy-associated plasma protein-A

AD

Alzheimer’s disease

IGF

insulin-like growth factor

IGF-IR

IGF-I receptor

WT

wild-type

KO

knock-out

PBS

phosphate-buffered saline

GFAP

glial fibrillary acid protein

NOR

Novel Odor Recognition

DI

Discrimination Index

STM

Stone T-maze

Data Availability Statement

Data supporting the results in this paper will be peer-reviewed. No large datasets were generated during the study that would require archiving in a repository.