Pharmacodynamic and stage‐dependent therapeutic efficacy of SFRP1 neutralization in a mouse model of Alzheimer's disease

Abstract

INTRODUCTION

Alzheimer's disease (AD) is characterized by early synaptic dysfunction followed by progressive amyloid beta (Aβ) accumulation, neuroinflammation, and cognitive decline. We previously identified secreted frizzled‐related protein 1 (SFRP1) as a multifactorial contributor to AD pathogenesis and provided initial evidence that its neutralization ameliorates pathological AD‐like traits in mice. Here we evaluate the pharmacodynamics, biodistribution, and therapeutic window of an anti‐SFRP1 monoclonal antibody (α‐SFRP1) in double transgenic amyloid precursor protein (APP) and presenilin‐1 (PS1) mice (APP/PS1).

METHODS

Pharmacokinetics and target engagement of α‐SFRP1 were assessed in groups of both male and female APP/PS1 mice using biotinylated or Zirconium‐89 labeled (⁸⁹Zr) antibodies, with tissue distribution and α‐SFRP1 levels quantified by in‐house enzyme‐linked immunosorbent assay (ELISA) or positron emission tomography/computed tomography (PET/CT). Therapeutic efficacy was evaluated by administering α‐SFRP1 or the SFRP1 inhibitor WAY‐316606 at different stages of disease progression via retro‐orbital injection, followed by analysis of AD‐like pathology using ELISA and quantitative immunofluorescence assays and statistical analysis.

RESULTS

Using ⁸⁹Zr‐labeled antibodies, we show that intravenously administered α‐SFRP1 engages its target systemically and reaches the brain, although at substantially lower levels and with a rapid 24‐h clearance. Treatment with α‐SFRP1 had no apparent systemic side effects or sex‐dependent differences, but its therapeutic efficacy against AD‐like brain pathology was strongly dependent on disease stage. Although early administration reduced amyloid pathology in previous studies, treatment initiated at intermediate or advanced stages showed minimal benefit at standard doses. Higher antibody doses reduced amyloid burden and dystrophic neurites but were associated with increased mortality. Pharmacological inhibition of SFRP1 using a small‐molecule inhibitor similarly failed to ameliorate pathology at intermediate stages.

DISCUSSION

Together, these findings demonstrate that SFRP1 remains a relevant therapeutic target in AD, but its effective modulation is constrained by limited brain exposure and a narrow therapeutic window, underscoring the importance of early intervention and prompting the search for improved brain‐targeted delivery strategies.

Highlights

• SFRP1 is a novel therapeutic target in Alzheimer's disease

• Antibody therapy shows stage‐dependent efficacy in vivo

• Limited brain exposure constrains therapeutic benefit

• Increased dosing improves pathology but impacts survival

• SFRP1 targeting complements current anti‐amyloid therapies

1. BACKGROUND

Alzheimer's disease (AD) is a multifactorial, female‐prevalent neurodegenerative disorder (ND) and a leading cause of dementia. Its pathogenesis involves synaptic loss, extracellular and vascular amyloid beta (Aβ) plaques, intracellular hyperphosphorylated tau (p‐tau) tangles, chronic glia‐driven neuroinflammation, and vascular and neuronal damage leading to cognitive decline. ¹ Despite extensive research, ² , ³ , ⁴ the triggering mechanisms of AD remain unclear. Early synaptic dysfunction and chronic neuroinflammation are considered central pathogenic events, ⁵ , ⁶ whereas the accumulation of Aβ aggregates remains a defining hallmark of AD and the basis of the amyloid cascade hypothesis. ³

Aβ peptides originate from the proteolytic processing of the amyloid precursor protein (APP), heterogeneously expressed throughout the body with high levels in brain neurons. ⁷ APP can be processed via the non‐amyloidogenic pathway, in which ADAM10 cleaves APP within the Aβ domain, preventing Aβ formation upon presenilins 1 and 2 (PSEN1/2) cleavage, ⁸ or via the amyloidogenic pathway, mediated by β‐site APP cleaving enzyme 1 (BACE1) followed by PSEN1/2. If the balance between these pathways favors Aβ peptide generation, amyloid plaques build up in the brain constituting a diagnostic AD hallmark. Widespread recognition of this hallmark has driven the development of monoclonal antibodies (mAbs) against Aβ as the first disease‐modifying therapies for AD. ⁹ Nevertheless, only two (donanemab and lecanemab) have been commercialized in a limited number of countries, ¹⁰ highlighting the need for new therapeutic targets.

Our earlier studies identified SFRP1 as a novel multifactorial player in AD pathogenesis. ¹¹ SFRP1 is a small (35 kDa) soluble protein that modulates Wnt signaling ¹² and ADAM10 enzymatic activity. ¹³ We showed that SFRP1 abnormally accumulates in the brain of AD patients and APP/PS1 mice, ¹¹ a finding supported by several proteomic studies confirming SFRP1 upregulation in the brain of AD patients ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ and related mouse models. ¹⁷ , ²¹ Furthermore, a recent large‐scale proteomic analysis stratified AD into three molecular subtypes, all correlating with SFRP1 levels and positioning it at the intersection of core disease mechanisms. ²² Functional studies show that SFRP1 not only favors Aβ formation and accumulation by hindering ADAM10‐mediated APP processing, ¹¹ but also sustains chronic neuroinflammation. ²³ In addition, chronic exposure to SFRP1 alone induces synaptic deficits and cognitive impairments resembling AD progression. ²⁴ Notably, its genetic or mAb‐mediated neutralization reduces amyloid plaque burden and prevents long‐term potentiation (LTP) loss in APP/PS1 mice. ¹¹

Together with the current need for a broader molecular and mechanistic understanding of AD, ²⁵ these findings support the therapeutic potential of SFRP1 in AD. Here we have further explored the pharmacodynamics and therapeutic efficacy of the already described α‐SFRP1.10.5.6 mAb (α‐SFRP1). ¹¹ We report that intravenous (IV) α‐SFRP1 administration enables target engagement in the brain, although at much lower levels than in other organs. In addition, the therapeutic window for mAb administration appears limited to early AD stages, while further research is necessary to assess potential off‐target effects. Our study provides the first in vivo pharmacodynamic and therapeutic characterization of SFRP1 targeting, establishing it as a functionally validated and actionable target in AD.

2. METHODS

Male and female APP/PS1 transgenic mice (4–12 months) were used under approved ethical protocols. α‐SFRP1 monoclonal antibodies were generated as described and biotinylated or ⁸⁹Zr‐labeled for biodistribution studies. Mice received α‐SFRP1, control immunoglobulin G1 (IgG1), or WAY‐316606 via systemic administration. Antibody distribution was assessed by positron emission tomography/computed tomography (PET/CT) imaging and quantified as standardized uptake values (SUVs). Blood and tissues were collected for biochemical, enzyme‐linked immunosorbent assay (ELISA), and histological analyses, including immunofluorescence (Table S1). Statistical significance was determined using standard parametric or non‐parametric tests (p < 0.05). A detailed version of the Methods is provided in the Supplementary Material.

3. RESULTS

3.1. Retro‐orbital vs intraperitoneal mAb administration

We previously showed that α‐SFRP1 (IgG1 backbone) efficiently neutralizes SFRP1‐mediated ADAM10 inhibition and reaches the brain parenchyma of APP/PS1 mice 24 h after retro‐orbital (RO) administration. ¹¹ However, RO delivery requires anesthesia, increasing animal stress, ²⁶ and is more technically demanding and time‐consuming than intraperitoneal (IP) injection. To improve animal welfare and facilitate experimental procedures, we compared the kinetics and half‐life of biotinylated α‐SFRP1 (α‐SFRP1^biot; 100 µg) following RO or IP administration in 4‐ to 8‐month‐old homozygous APP/PS1 mice (APP/PS1^Hom), a stage at which amyloid plaques are already present. ¹¹ α‐SFRP1^biot levels were quantified in serum and brain extracts using a home‐made ELISA. RO administration produced ∼3‐fold higher serum mAb levels than IP injection during the first 6 h post‐injection (Figure S1A), with a half‐life of 9.17 h. By 24 h, however, both routes yielded comparable α‐SFRP1^biot levels in serum and brain (Figure S1A,B).

These results suggest that RO administration enables a faster and more efficient distribution of the mAb, at least in the first few hours after inoculation, making it the route of choice for the following experiments.

RESEARCH IN CONTEXT

1. Systematic review: We reviewed the literature using PubMed and recent reports on Alzheimer's disease (AD) therapeutics, focusing on amyloid‐targeting strategies, ADAM10 regulation, and Wnt pathway modulators. Although ADAM10‐mediated non‐amyloidogenic processing is neuroprotective, its inhibition in AD is poorly understood. Converging proteomic and functional studies identify secreted frizzled‐related protein 1 (SFRP1) as consistently upregulated in AD and a negative regulator of ADAM10, yet its in vivo pharmacodynamics and therapeutic potential remain largely unexplored.

2. Interpretation: We provide the first comprehensive in vivo characterization of SFRP1 targeting, demonstrating that antibody‐mediated neutralization engages the brain but is constrained by limited penetration and a narrow, stage‐dependent therapeutic window. These findings position SFRP1 as a mechanistically relevant and druggable target, while highlighting key limitations shared with current disease‐modifying therapies.

3. Future directions: Advancing SFRP1‐based therapies will require improved brain delivery, biomarker‐guided early intervention, and rational combination with anti‐amyloid immunotherapies to achieve durable clinical benefit.

3.2. α‐SFRP1 is widely distributed across the mouse body, including the brain

We next analyzed the biodistribution of α‐SFRP1^biot and a biotinylated isotypic control antibody (IgG1^biot) following administration to 4‐ to 8‐month‐old APP/PS1^Hom mice. Antibody levels were measured 24 h later in serum, liver, kidney, and brain. Both antibodies showed similar circulating levels at 6 and 24 h post‐injection (Figure S2A). However, α‐SFRP1^biot appeared to accumulate in the kidney, where SFRP1 is expressed, ²⁷ whereas comparable levels of both antibodies were detected in the brain and liver (Figure S2B‐D).

To obtain a more accurate estimate of α‐SFRP1 biodistribution, we inoculated Zirconium‐89 (⁸⁹Zr) radiolabeled antibodies (α‐SFRP1^89Zr and IgG1^89Zr; 100 µg) and used PET/CT for longitudinal and whole‐body tracing (2, 10, 24, and 168 h) of the antibodies in a group of five animals per treatment (Figure 1A). IgG1^89Zr concentrated in the bladder at earlier time points, suggesting faster clearance from the vascular system as compared to α‐SFRP1^89Zr (Figure 1G). In fact, α‐SFRP1^89Zr specifically accumulated in the brain, lungs, and heart up to 10 h post‐injection, reaching IgG1^89Zr comparable levels 24 h after the inoculation (Figure 1B,E,F). Surprisingly, we observed no significant differences in mAb concentrations in the liver (Figure 1C) or kidneys (Figure 1D), in contrast to the results we obtained with α‐SFRP1^biot and IgG1^biot.

FIGURE 1.

Whole‐body PET‐CT reveals transient differences in systemic biodistribution of ⁸⁹Zr‐IgG1 and ⁸⁹Zr‐α‐SFRP1 after intravenous administration. (A) Representative transverse whole‐body positron emission tomography/computed tomography (PET‐CT) maximum‐intensity projection images acquired at 2, 10, 24, and 168 h after tail‐vein administration of Zirconium‐89 (⁸⁹Zr)‐IgG1 (left) or ⁸⁹Zr‐α‐SFRP1 (right). (B–G) In vivo quantification of ⁸⁹Zr‐IgG1 and ⁸⁹Zr‐α‐SFRP1 over time in brain (B), liver (C), kidneys (D), lungs (E), heart (F), and bladder (G). Data are presented as standardized uptake values (SUVs) and shown as staggered individual replicates with a dotted line representing the mean of each experimental group. Males are represented by open circles and females by filled circles (⁸⁹Zr‐IgG1 n = 5 (2/3) and ⁸⁹Zr‐α‐SFRP1 n = 5 (2/3), male/female respectively). Statistical analyses were performed using two‐way ANOVA followed by Sidak's post hoc correction for multiple comparisons.

α‐SFRP1^89Zr accumulated throughout the brain when compared to IgG1^89Zr, with higher concentrations in the olfactory bulbs and entorhinal cortex (Figure 2A), where SFRP1 is more abundantly distributed. ²⁷ In APP/PS1 mice, amyloid‐ and SFRP1‐positive plaques primarily accumulate in the cortex and hippocampus. ¹¹ , ²⁸ Both regions were specifically targeted by α‐SFRP1^89Zr, although antibody concentrations reached IgG1^89Zr comparable levels within the first 24 h (Figure 2B,C), mirroring whole brain observations.

FIGURE 2.

Regional brain PET‐CTanalysis reveals transient differences in ⁸⁹Zr‐IgG1 and ⁸⁹Zr‐α‐SFRP1 uptake after intravenous administration. (A) Representative transversal brain positrion emission tomography/computed tomography (PET‐CT) maximum‐intensity projection images acquired at 2, 10, 24, and 168 h after intravenous tail administration of Zirconium‐89 (⁸⁹Zr)‐IgG1 (left) or ⁸⁹Zr‐α‐SFRP1 (right.). The leftmost image corresponds to the mouse brain template used for the analysis. Volumes of interest include the olfactory bulb (purple), cerebral cortex (green), striatum (orange), hippocampus (blue), thalamus (red), and cerebellum (yellow). (B, C) In vivo quantification of ⁸⁹Zr‐IgG1 and ⁸⁹Zr‐α‐SFRP1 over time in the brain regions of primary interest: cortex (B) and hippocampus (C). Data are presented as standardized uptake values (SUVs) and shown as staggered individual replicates with a dotted line representing the mean of each experimental group. Males are represented by open circles and females by filled circles (⁸⁹Zr‐IgG1 n = 5 (2/3) and ⁸⁹Zr‐α‐SFRP1 n = 5 (2/3), male/female respectively). Note that two ⁸⁹Zr‐α‐SFRP1 and one ⁸⁹Zr‐IgG1 male mice died 24 h after the first recording due to fights in the cage during the night. Statistical analyses were performed using two‐way ANOVA followed by Sidak's post hoc correction for multiple comparisons.

Together, these results show that α‐SFRP1 engages its target in different organs, penetrating different brain regions of APP/PS1^Hom mice where it could mitigate AD progression. Notably, α‐SFRP1 seems to be cleared from the brain within the first 24 h after injection, allowing only a small time‐window for its therapeutic potential.

3.3. α‐SFRP1 administration has a reduced efficacy at later stages of the pathology

Aβ plaques start to appear as early as 3 months in the cortex and hippocampus of APP/PS1^Hom mice, ¹¹ followed by a marked astro‐ and microgliosis, neuronal loss, and cognitive impairment. ²⁸ , ²⁹ We have previously treated APP/PS1^Hom mice with α‐SFRP1 before the onset of the disease (2 months), observing a significant reduction of amyloid burden and related pathological hallmarks. ¹¹ However, diagnosis of dementia‐associated diseases occurs when patients’ brains have undergone major morphological alterations, ³⁰ hindering the effects of the attempted treatments. ³¹ , ³²

Thus, we asked whether α‐SFRP1 could prove beneficial at later stages of the disease. To answer this question, we performed injections of α‐SFRP1 or IgG1 (100 µg, RO) in 9‐month‐old APP/PS1^Hom mice (Figure 3A), once the AD‐like phenotype is fully developed, including cognitive deficits. ³³ Although the calculated half‐life of α‐SFRP1 was ≈9 h, we used weekly mAb administration, as reported previously, ¹¹ , ³⁴ because more frequent injections were excessively stressful for the animals. This dosing regimen likely underestimates the therapeutic potential of sustained SFRP1 neutralization. After 3 months of treatment, animals were sacrificed (12‐month‐old) and their brains analyzed. First, we employed a previously established SFRP1‐specific and highly sensitive ELISA ¹¹ to determine the levels of SFRP1 present in brain radioimmunoprecipitation assay buffer (RIPA)‐soluble fractions, finding no significant changes in SFRP1 concentration between control and α‐SFRP1 treated animals of either sex (Figure 3B). We next investigated the effects of the treatment on amyloid burden, neuronal damage, and gliotic response by immunostaining cortical and hippocampal slices with α‐Aβ (H31L21), α‐Glial fibrillary acidic protein (GFAP), and α‐Ionized calcium‐binding adapter molecule 1 (IBA1). We observed a non‐significant 33% decrease in the number of Aβ⁺ plaques, and no changes in the average Aβ⁺ plaque size, in the cortex of α‐SFRP1‐treated animals. Notably, α‐SFRP1 had no apparent effects on Aβ plaque deposition in the hippocampus (Figure 3D). The levels of astrocytic and microglial activation, measured as GFAP and IBA1 fluorescence intensity, remained unaltered between the two experimental groups (Figure 3E,F). We also observed no evident changes in the percentage of different blood cell populations, suggesting no apparent adverse peripheral immune effects under these treatment conditions (Figure S3).

FIGURE 3.

α‐SFRP1 treatment does not significantly modify AD‐related pathology in aged APP/PS1 mice. (A) Schematic representation of the experimental design: 9‐month‐old APP/PS1 homozygous mice received a weekly RO injection of IgG1 or α‐SFRP1 (100 µg) for 3 months; analysis in panels (B–F) were performed at 12 months of age. The time diagram indicates the expected age at which these mice initiate amyloid beta (Aβ) plaque accumulation. Illustration from NIAID NIH BioArt Source (https://bioart.niaid.nih.gov/). (B) ELISA quantification of SFRP1 levels in brain RIPA‐soluble homogenates. (C) Representative confocal images of cortices (left) and hippocampi (right) from IgG1‐treated (top) and α‐SFRP1‐treated (bottom) mice stained for Aβ42 as a marker of amyloid plaques. Scale bar: 100 µm. (D) Quantification of Aβ42⁺ plaques number (top) and mean plaque area (bottom) in the cortex (left) and hippocampus (right). (E–F) Quantification of GFAP (E) and IBA1 (F) fluorescence intensity in the cortex (left) and hippocampus (right), expressed as fold‐change relative to the mean value of the IgG1 group. Quantification analyses were performed for individual mice, averaging the results of three to eight slices per mouse. Data are presented as mean ± standard error of the mean (SEM), with males represented by open circles and females by filled circles (IgG1 n = 7 (3/4) and α‐SFRP1 n = 7 (3/4), male/female respectively). Note that some graphs show fewer than seven animals per group as immunohistochemical sections presenting disproportinate background staining were excluded from analyses. Statistical analyses were performed using Student's t‐test.

Similar results were obtained when we treated 4‐month‐old APP/PS1^Hom mice with weekly mAb administration (100 µg, RO) for 5 months (Figure 4A). No significant changes were observed in SFRP1 levels present in the RIPA‐soluble brain homogenates (Figure 4B) or in the amyloid burden (Figure 4D), dystrophic neuron load (Figure 4E), or astro‐ and microgliosis (Figure 4F,G) of the cortex and hippocampus of these mice. In addition, these animals showed no evident defects in the overall structure of their liver and kidneys (Figure S4).

FIGURE 4.

Long‐term α‐SFRP1 treatment does not significantly modify AD pathophysiology in adult APP/PS1 mice. (A) Schematic of the experimental design. Four‐month‐old homozygous APP/PS1 mice received weekly retro‐orbital injections of IgG1 or α‐SFRP1 (100 µg) for 5 months; analyses in panels (B–G) were performed at 9 months of age. The time diagram indicates the expected age at which these mice initiate amyloid beta (Aβ) plaque accumulation. Illustration from NIAID NIH BioArt Source (https://bioart.niaid.nih.gov/). (B) ELISA quantification of SFRP1 levels in brain RIPA‐soluble homogenates. (C) Representative confocal images of cortices (left) and hippocampi (right) from IgG1‐treated (top) and α‐SFRP1‐treated (bottom) mice stained with ThioS to label amyloid plaque cores. Scale bar: 100 µm. (D,E) Quantification of the number (top) and mean plaque area (bottom) of ThioS+ plaques (D) and LAMP1+ puncta (E) in the cortex (left) and hippocampus (right). (F,G) Quantification of GFAP (F) and IBA1 (G) fluorescence intensity in the cortex (left) and hippocampus (right), expressed as fold‐change relative to the mean value of the IgG1 group. Quantification analyses were performed for individual mice, averaging the results of three to eight slices per mouse. Data are presented as mean ± standard error of the mean (SEM), with males represented by open circles and females by filled circles (IgG1 n = 6 (2/4) and α‐SFRP1 n = 5 (2/3), male/female respectively). Note that some graphs show fewer than seven animals per group as immunohistological sections presenting disproportionate background staining were excluded from analyses. Statistical analyses were performed using Student's t‐test.

These results indicate that α‐SFRP1, at the applied concentration, has little and no sex‐dependent effects once pathological alterations are well established in the brain of APP/PS1^Hom mice. We thus tested whether a higher dose of α‐SFRP1 would be necessary to show beneficial effects in APP/PS1^Hom mice at this age. To this end, we used 4‐month‐old APP/PS1^Hom mice and inoculated them weekly with α‐SFRP1 or IgG1 (200 µg, RO) (Figure 5A). The experiment was terminated after 2 months due to the low survival rate (43%) in the α‐SFRP1‐treated group (Figure 5B). Substantial evidence indicates that brain‐targeted immunotherapies highly increase the patients’ risk of suffering edema or hemorrhages, known as amyloid‐related imaging abnormalities (ARIA). ³⁴ , ³⁵ , ³⁶ To test whether similar vascular alterations might underlie the premature death of these mice, we performed CD31 immunostaining on brain sections from surviving animals. However, no evident vascular abnormalities were detected (Figure S5). Notably, surviving α‐SFRP1‐treated animals showed a 50% decrease in SFRP1 brain concentration as compared with IgG1‐injected mice (Figure 5C) and their number of ThioS⁺ plaques was significantly reduced in both the cortex and hippocampus (Figure 5E). The amelioration of Aβ load was paralleled by a substantial reduction in the number of LAMP1⁺ spots surrounding the plaques in both brain regions (Figure 5G). Although no changes were observed in overall amyloid plaque size, detailed analysis of their distribution across different size ranges revealed a reduction in plaque number across all size categories in α‐SFRP1‐treated brains, with the strongest effect on small deposits (Figure 5E,F). We also observed non‐significant trends, particularly in microglial activation, with reduced IBA1⁺ signal intensity in the cortex (35%) and hippocampus (49%) of surviving α‐SFRP1‐treated mice (Figure S6B). In contrast, astrogliosis was unchanged in the cortex and showed a slight increase in hippocampal GFAP intensity (Figure S6A).

FIGURE 5.

High‐dose α‐SFRP1 treatment reduces amyloid burden in older APP/PS1 mice. (A) Schematic of the experimental design. Four‐month‐old homozygous APP/PS1 mice received weekly retro‐orbital injections of IgG1 or α‐SFRP1 (200 µg) for 2 months; analyses in panels (B–G) were performed at 6 months of age. The time diagram indicates the expected age at which these mice initiate amyloid beta (Aβ) plaque accumulation. Illustration from NIAID NIH BioArt Source (https://bioart.niaid.nih.gov/). (B) Kaplan–Meier curves showing mouse survival during the treatment period; each vertical step represents a death event. (C) ELISA quantification of SFRP1 levels in brain RIPA‐soluble homogenates. (D) Representative confocal images of cortices (left) and hippocampi (right) from IgG1‐treated (top) and α‐SFRP1‐treated (bottom) mice stained with ThioS to label amyloid plaque cores. Scale bar: 100 µm. (E) Quantification of the number (top) and mean plaque area (bottom) of ThioS+ plaques in the cortex (left) and hippocampus (right). (F) Quantification of plaque number across 100 µm² size bins in the cortex (left) and hippocampus (right). The inset plots for both regions show plaque numbers across 20 µm² size bins for plaques between 10 and 380 µm². This size range represents ≈90% of all plaques. (G) Quantification of the number (left) and mean plaque area (right) of LAMP1+ puncta in the cortex and hippocampus. Quantification analyses were performed for individual mice, averaging the results of three to eight slices per mouse. Data are presented as mean ± standard error of the mean (SEM), with males represented by open circles and females by filled circles (IgG1 n = 5 (1/4) and α‐SFRP1 n = 3 (1/2), male/female respectively). Note that males and females have been pooled together in and are shown together in (F) for better visualization. Statistical analyses were performed using Student's t‐test (C,E,G) or mixed‐effect analysis followed by Sidak's post hoc correction for multiple comparisons (F).

Taken together, these results indicate that higher doses of α‐SFRP1 are required to alleviate AD traits in mice when brain morphological alterations are well established. This dose is, however, associated with an increased mortality, the causes of which remained unidentified.

3.4. Chemical inhibition of SFRP1 does not alleviate AD at intermediate stages of the disease

The increased mortality observed at high‐dose α‐SFRP1 suggests that this concentration has a narrow therapeutic safety margin. With the view of circumventing this problem, we tested WAY‐316606 (WAY), a specific SFRP1 inhibitor, for its potential efficacy. WAY has been shown previously to decrease Aβ production in AD patient–derived organoids ³⁷ and to improve experimentally induced osteoporosis in vivo, ³⁸ with no apparent toxic side effects.

To test this compound, we employed heterozygous APP/PS1 mice (APP/PS1^Het), which show a slower progression of the AD‐like phenotype, starting to show Aβ plaque accumulation at 6 months of age. ²⁸ We treated 8‐month‐old APP/PS1^Het mice with RO weekly injections of either 1 µM WAY or 2% dimethyl sulfoxide (DMSO), as control, for 2 months (Figure 6A). We observed a non‐significant 34% reduction in the levels of SFRP1 in WAY‐injected mice brains (Figure 6B). However, WAY treatment did not affect canonical Wnt signaling in the brain of treated mice, as shown by unchanged levels of the pathway targets AXIN2 and Cyclin D1 in Western blot analyses of brain extracts (Figure S7). In addition, no apparent differences were found in the number or size of ThioS⁺ amyloid plaques or LAMP1⁺ dystrophic neuron spots (Figure 6D,E).

FIGURE 6.

Pharmacological inhibition of SFRP1 does not significantly modify AD‐related pathology in adult APP/PS1 mice. (A) Schematic of the experimental design. Eight‐month‐old heterozygous APP/PS1 mice received weekly retro‐orbital injections of WAY‐316606 (1 µM) or vehicle (2% DMSO) for 2 months; analyses in panels (B–E) were performed at 10 months of age. The time diagram indicates the expected age at which these mice initiate amyloid beta (Aβ) plaque accumulation. Illustration from NIAID NIH BioArt Source (https://bioart.niaid.nih.gov/). (B) ELISA quantification of SFRP1 levels in brain RIPA‐soluble homogenates. (C) Representative confocal images of cortices (left) and hippocampi (right) from vehicle‐treated (top) and WAY‐316606‐treated (bottom) mice. Scale bar: 100 µm. (D,E) Quantification of number (top) and mean plaque area (bottom) of ThioS+ plaque (D) and LAMP1+ puncta (E) in the cortex (left) and hippocampus (right). Quantification analyses were performed for individual mice, averaging the results of three to eight slices per mouse. Data are presented as mean ± standard error of the mean (SEM), with males represented by open circles and females by filled circles (DMSO, n = 6 (1/5) and WAY, n = 5 (2/3), male/female respectively). Statistical analyses were performed using Student's t‐test.

Taken together, these data indicate that the α‐SFRP1 mAb may represent a better tool to target SFRP1 in AD conditions, although optimization of its dose and delivery to the brain is needed.

4. DISCUSSION

AD is characterized by a prolonged preclinical phase during which progressive molecular and cellular alterations impair synaptic integrity and neuronal plasticity, and ultimately promote amyloid‐driven neurodegeneration. By the time cognitive symptoms emerge, pathological changes are already well established, greatly limiting the efficacy of disease‐modifying therapies. This temporal disconnection between disease onset and clinical diagnosis remains a major challenge for successful AD treatment. ³⁰ , ³² , ³⁹ Together with the complexity of AD pathogenesis, this gap underscores the need for therapeutic targets that act early, have pleiotropic functions, and remain effective beyond the prodromal stage. We propose SFRP1 as such a candidate. Proteomic studies have confirmed its accumulation at prodromal AD stages, ¹⁵ whereas functional studies have implicated it in several processes preceding overt neurodegeneration. ²³ , ²⁴ , ³⁷ , ⁴⁰ These findings support the idea that SFRP1 neutralization may provide multiple benefits against AD progression. The present study extends this evidence, demonstrating the therapeutic potential of mAb‐mediated SFRP1 neutralization even after pathology is established, while revealing important limitations related to brain delivery, dose requirements, and increased mortality at higher doses.

We previously showed that early SFRP1 neutralization with low α‐SFRP1 doses reduces plaque burden and neuroinflammation while preserving synaptic plasticity in APP/PS1 mice, ¹¹ although the mAb pharmacokinetics, biodistribution, and therapeutic window remained unexplored. Here we show that RO administration in mice results in significantly higher circulating α‐SFRP1 levels than IP delivery. Although IP injections are more practical and less stressful in preclinical mouse studies, IV administration is commonly used in human antibody therapies and therefore has greater translational relevance. ⁴¹

Using biotinylated and ⁸⁹Zr‐radiolabeled antibodies combined with PET/CT imaging, we further demonstrate that IV administered α‐SFRP1 engages its target systemically, accumulates in different organs, and reaches the brain, including AD affected regions such as the cortex and hippocampus. Nonetheless, brain exposure was markedly low relative to peripheral organs, and declined rapidly within 24 h. This observation aligns with extensive literature showing that under physiological conditions only a small fraction (≈0.1%) of circulating IgG crosses the intact blood–brain barrier (BBB). ⁴² BBB integrity is partially compromised in AD, particularly in regions affected by amyloid pathology. ⁴³ Therefore, the detectable hippocampal accumulation observed in our study is biologically plausible and comparable to what has been reported for other therapeutic antibodies targeting Aβ in mouse models. ³⁴ Nevertheless, we found 10‐fold less concentration in the brain than in the kidney or the liver, indicating that mAb penetration remains limited even in AD conditions and still represents a major obstacle for central nervous system (CNS) immunotherapies. It should be noted that IV inoculation of drugs was performed consistently through the RO sinus, with the exception of the ⁸⁹Zr‐radiolabeled antibodies, which were administered through the tail vein. Given that both of these routes seem to provide comparable absorption and pharmacokinetic activity for therapeutic antibodies, ⁴⁴ they were considered equivalent for the purpose of this study.

In addition to restricted brain availability, the effects of α‐SFRP1 proved highly dependent on disease stage. When treatment was initiated in mice with established plaque pathology (4–9 months), standard antibody dosing failed to significantly reduce SFRP1 concentration in the brain. This may reflect that SFRP1 accumulates in the plaques and cannot be neutralized, further becoming resistant to degradation, as shown in a recent study. ²⁰ This persistent SFRP1 presence may also explain why amyloid burden, neuritic dystrophy, or gliosis are unaffected by the mAb administration.

Only upon dose escalation did surviving animals show significant reductions in plaque number across different size ranges, along with fewer dystrophic neurites. This suggests that SFRP1 neutralization may disrupt plaque growth and favor clearance of established amyloid deposits, supporting the idea that SFRP1 remains mechanistically involved even at later stages of the disease, but cannot be effectively neutralized at conventional doses due to limited CNS exposure. This underscores a key challenge shared across neurodegenerative immunotherapies: sufficient target engagement in the brain often requires antibody concentrations that are difficult to achieve safely through systemic administration. In fact, similar dose‐limiting effects have been described for other antibody‐based therapies in AD models, such as aducanumab and lecanemab. ³⁴ , ⁴⁵

The increased mortality observed at higher antibody doses further highlights the narrow therapeutic margin for CNS‐directed immunotherapy in AD. The causes of death were not fully investigated in this study but CD31 staining did not reveal overt vascular alterations in the brain of surviving treated animals, as reported in the case of Aβ‐targeting antibodies such as aducanumab, lecanemab, and donanemab. ³⁴ , ³⁵ , ³⁶ High‐dose systemic α‐SFRP1 administration may nevertheless present safety limitations that warrant further investigation.

Notably, a study in 5xFAD mice, another AD mouse model, described sex‐related differences in SFRP1 upregulation. ²¹ In contrast, we never observed differences in SFPR1 expression between males and females or any evident sex‐dependent differences in response to α‐SFRP1 treatment in APP/PS1 mice. Our results are in line with a recent proteomic study analyzing sex‐specific differences in AD and other dementias. ⁴⁶

The commercial availability of small‐molecule SFRP1 inhibitors, such as WAY‐316606, prompted us to test it as an alternative to α‐SFRP1. However, this pharmacological modulation was insufficient to reverse pathology once amyloid deposition is established. Although WAY has been reported to reduce Aβ production in patient‐derived organoids, ³⁷ its administration in intermediate‐stage APP/PS1^Het mice produced only modest and non‐significant reductions in brain SFRP1 levels, without measurable effects on plaque burden or neuritic dystrophy. This negative result raises the possibility that the compound did not effectively reach the brain. Although this could not be directly assessed, the absence of changes in Wnt signaling read‐outs further supports this interpretation. However, previous work from our lab showed that SFRP1 induces a neurodegenerative‐like phenotype independently of canonical Wnt signaling, ²⁴ and that SFRP1 genetic inactivation reduces amyloid burden and gliosis in mice without altering AXIN 2 levels. ¹¹ Together, these findings support the idea that SFRP1 effects on AD progression may occur independently of canonical Wnt signaling and that WAY may have little effect on AD pathology.

Collectively, our study defines critical pharmacodynamic and temporal constraints for targeting SFRP1 in AD. Our findings indicate that SFRP1 neutralization is most effective at early disease stages, before widespread plaque deposition, glial activation, and synaptic degeneration are established, although improved CNS delivery may extend its efficacy into symptomatic phases. Although APP/PS1 mice do not fully recapitulate the temporal complexity of sporadic human AD, the treatment paradigms used here likely reflect distinct biological stages of disease progression. In this context, the partial efficacy observed at later stages suggests that SFRP1 remains mechanistically relevant beyond early pathology, but requires stronger target engagement. These observations parallel the broader field of AD therapeutics, where the clinical benefit of anti‐Aβ antibodies is greatest in prodromal or early symptomatic populations. Together, our data support SFRP1‐directed therapies primarily as an early disease‐modifying strategy, potentially aided by improved brain‐targeted delivery approaches. The use of biomarkers for patient stratification, including, for example, cerebrospinal fluid/plasma Aβ and p‐tau, GFAP, soluble triggering receptor expressed on myeloid cells 2 (sTREM2), neurogranin, or vascular/barrier‐associated markers, ⁴⁷ may help sharpening the temporal windows in which SFRP1 neutralization is most effective. An additional important limitation of the present study is the absence of functional and cognitive assessments, as performed previously for early treatments. ¹¹ Future studies are therefore necessary to determine whether the observed pathological improvements translate into functional benefit.

Current AD therapies either aim to alleviate symptoms, including, for example, acetylcholinesterase inhibitors and N‐methyl‐d‐aspartate (NMDA) receptor antagonists, or are disease‐modifying compounds, such as lecanemab and donanemab. ¹⁰ Systematic reviews of the current literature and regulatory reports indicate that at least lecanemab and donanemab, when administered to patients at early stage of the disease, can slow disease progression by ≈30%, but are associated with ARIA side effects and a rather heavy follow‐up protocol. ³⁵ , ³⁹ Our preclinical analysis in mice provides a comprehensive evaluation of α‐SFRP1 pharmacodynamics and stage‐dependent efficacy in vivo, suggesting that this approach could be used as an alternative or in combination with current treatments. However, additional efforts are needed to optimize its dose and brain delivery. Strategies such as BBB shuttle engineering, Fc receptor‐mediated transcytosis, or encapsulation into brain‐targeted lipid vesicles are possible options to substantially expand the therapeutic window of α‐SFRP1. Combinatorial strategies, including SFRP1 neutralization with lecanemab or donanemab may provide synergistic benefit by simultaneously facilitating plaque clearance and restoring non‐amyloidogenic APP processing, ¹¹ while reducing neuroinflammation ²³ and preventing synaptic alterations. ²⁴

AUTHOR CONTRIBUTIONS

P.M. and M.M.B. performed all experiments with the help of M.J.M.B. P.M. and M.M.B. analyzed and interpreted the data with P.B. I.M. and M.D. generated, purified, and verified the activity of the mAb. P.M. drafted the manuscript and P.B. produced the final version. All authors read and approved the final manuscript. P.B. supervised the work and obtained funds.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.

ETHICS STATEMENT

All experiments were performed according to procedures approved by the ethical committees of the Centro de Biologia Molecular Severo Ochoa (CBM), the Bioethics sub‐committee of Consejo Superior de Investigaciones Científica (CSIC, Madrid, Spain) and the Comunidad Autónoma de Madrid under the protocol approval number PROEX 092.6/21; RD 53/2013 or the Ethical Committee of Centro de Investigación Cooperativa en Biomateriales (CIC biomaGUNE) and authorized by the local Diputación Foral de Gipuzkua under the protocol approval number PRO‐AE‐SS‐212 in case of radiolabeled antibodies administration.

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Miaja P, Martinez‐Baños M, Martin‐Bermejo MJ, Moreno I, Dominguez M, Bovolenta P. Pharmacodynamic and stage‐dependent therapeutic efficacy of SFRP1 neutralization in a mouse model of Alzheimer's disease. Alzheimer's Dement. 2026;12:e70301. 10.1002/trc2.70301

DATA AVAILABILITY STATEMENT

All data generated or analyzed during this study are included in this published article and its supplementary materials.