Meprin-β levels are increased in the brain and the cerebrospinal fluid of Alzheimer’s disease patients

Abstract

The aim of this study is to assess the levels of meprin-β, an alternative β-secretase in the brain and cerebrospinal fluid of subjects suffering Alzheimer’s disease. The full-length zymogen and the active species of meprin-β were characterized by immunoblotting and immunoprecipitation using ectodomain N-terminal and C-terminal antibodies. We examined meprin-β changes in extracts from Alzheimer’s frontal cortex (n = 13, Braak stages I–II; n = 12, Braak stages III–IV; n = 12, Braak stages V-VI) compared with controls (n = 14), as well as in cerebrospinal fluid samples from Alzheimer’s patients (n = 15) or non-Alzheimer’s controls (n = 15). Cerebrospinal fluid meprin-β levels were also determined in 19-month old TgF344-AD and wild-type rats. Brain and cerebrospinal fluid from a Mep1b-null mouse served to characterize the specificity of the immunoreactive bands for meprin-β. Human induced pluripotent stem cell (iPSC)-derived neuronal cultures were treated with 2 μM Aβ42 for 24 h. We found that meprin-β is present in human brain extracts and cerebrospinal fluid as several distinct species, attributed to the zymogen and the mature species. These species were absent in brain and cerebrospinal fluid from a Mep1b-null mouse. No fragments of meprin-β were detected in brain extract, either in cerebrospinal fluid. Only the mature forms of meprin-β are increased in frontal cortex extracts (from Braak V-VI), but not the zymogen. The MEP1-B mRNA levels are increased in Braaks III-IV and V-VI compared with controls. The meprin-β species are increased in human neuronal cultures treated with Aβ42. Finally, we demonstrated elevated levels of the mature meprin-β in the cerebrospinal fluid of patients with Alzheimer’s disease and in TgF344-AD rats. Our data support a role of meprin-β in Alzheimer’s pathology.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13195-026-02092-7.

Introduction

Alzheimer’s disease (AD), the most common form of dementia, is characterized by neuropathological observation of neuritic plaques and neurofibrillary tangles, which are known to be composed of β-amyloid (Aβ) and paired helical filament tau [42]. The most exclusive hallmark lesion of AD is the abnormal accumulation of Aβ in the brain. The Aβ peptide is generated by processing of a larger transmembrane spanning glycoprotein, the amyloid precursor protein (APP), through the successive action of proteolytic enzymes called secretases.

Sequential processing of APP begins with either the action of α-secretase or β-secretase for the ectodomain shedding (sAPPα or sAPPβ), leading to the non-amyloidogenic or amyloidogenic pathways, both followed by γ-secretase cleavage, releasing the APP intracellular domain (AICD) and the non-amyloidogenic p3 or the Αβ fragments. The primary neuronal β-secretase responsible for Aβ generation in the brain is an aspartyl protease type-I transmembrane protein, the β-site amyloid precursor protein cleaving enzyme 1 or BACE1 [17]. Proteases such as meprin-β, cathepsin B and ADAMTS4 (A Disintegrin and Metalloproteinase with Thrombospondin Motifs 4) have been identified as alternative β-secretases [1]. Among the several proteases previously proposed as alternative β-secretases, some, such as the β-site amyloid precursor protein cleaving enzyme 2 (BACE2), are now considered to act as non-amyloidogenic α-secretases, exerting neuroprotective effects (discussed in Sáez-Valero et al., [40]).

Clarifying how alternative APP-cleaving enzymes and fragments regulate Aβ generation is fundamental for evaluating the toxic cycle that drives pathology and for developing new therapeutic approaches. Meprin-β appears to be of particular interest as an alternative β-secretase [3]. In previous studies, the knockout of meprin-β in a mouse model of AD that overexpressed human APP resulted in the reduction of brain Aβ levels [32], while its overexpression in neurons of the cortex and hippocampus led to a marked increase in soluble Aβ levels [24], suggesting that meprin-β contributes to Aβ generation in vivo.

The metalloprotease meprin-β (EC: 3.4.24.63) is a multi-domain enzyme belonging to the astacin family of zinc endopeptidases (reviewed in [27]. The human meprin-β is a single-pass type I glycoprotein of 701 amino acids possessing a signal peptide (residues 1 to 21), a pro-domain (residues 22 to 61), and a mature chain (residues 62 to 701), containing the catalytic domain (residues 62 to 259) and MAM, MATH (conserved domains necessary for correct folding and transport, and involved in protein–protein interactions and dimerization) and epidermal growth factor (EGF)-like extracellular domains (residues 260 to 673), the transmembrane domain (residues 653 to 673), and the short C-terminal cytoplasmic domain (residues 674 to 701). The mouse meprin-β subunit is composed of 704 amino acids and a very similar sequence structure [16].

Expressed as a zymogen, pro-meprin-β reaches the cell surface and is activated by matriptase-2, cleaving off the pro-peptide between Arg61 and Asn62 and activating full-length meprin-β at the cell surface [19]. Moreover, pro-meprin-β can be shed as a zymogen by the α-secretases ADAM10 (A Disintegrin and Metalloproteinase domain-containing protein 10) or ADAM17 (A Disintegrin and Metalloproteinase domain-containing protein 17) [41]; next, this soluble form is activated by soluble serine proteases [35]. Thus, meprin-β can act as an APP β-secretase, but also be a substrate of the α-secretase. In this complex network, ADAM10 can be also activated by meprin-β through cleavage of the pro-peptide [21]. Highlighting the potential role in AD, meprin-β can compete with ADAM10/17 for the cleavage of Trem2 (triggering receptor expressed on myeloid cells 2) [4], a transmembrane protein expressed on myeloid cells including macrophages, dendritic cells and microglia with a role in response to amyloid plaque, and that displays rare variants associated with high AD risk [8]. In addition, meprin-β plays a role in regulating blood–brain barrier integrity; thus, deregulation can contribute to barrier breakdown [15].

Despite the potential role of meprin-β in Aβ generation and in the regulation of APP proteolytic processing, very few studies have addressed whether meprin-β levels are altered in the brain of AD patients. In a previous study, meprin-β mRNA levels were reported to be significantly higher in the brain of AD subjects versus controls, and protein levels were also increased. However, meprin-β was identified as a ~ 50 kDa species that does not correspond to the active species [33]. In another study, meprin-β protein levels were also shown to be increased in sporadic AD by immunohistochemical analysis, corroborated by western blotting, displaying proper molecular mass, in a reduced number of cases (3 control vs 3 AD cases) [32].

In the present study, we examined whether the mRNA and protein levels of meprin-β are altered in individuals with sporadic AD discriminating by Braak stages. We characterized and estimated meprin-β levels attributable to active species by electrophoresis followed by quantitative fluorescent western blotting. In addition, we also examined meprin-β changes in lumbar cerebrospinal fluid (CSF) samples designated as AD or non-AD controls according to CSF core biomarker levels, and in TgF344-AD, a rat transgenic model of AD. Brain and CSF samples from a Mep1b-null mouse served to characterize the specificity of the immunoreactive band for meprin-β protein in tissue extracts and body fluids. We report an increase in the mature species of meprin-β in brain and CSF of AD patients and in human induced pluripotent stem cell (iPSC)-derived neuronal cultures treated with Aβ42.

Methods

Human brain and CSF samples

This study was exploratory and not preregistered. No sample size was calculated a priori; the sample size was limited by sample availability and estimated according to our previous experience. This study was approved by the ethics committee of both, the Universidad Miguel Hernández (IN.JSV.04.20 & PRL.DBB.JSV.231113) and the Departamento de Salud de Alicante – Hospital General Universitario Dr. Balmis de Alicante, Spain (PI2025-173), and it was carried out in accordance with the 1964 Declaration of Helsinki and its later amendments.

Brain samples (frontal cortex, Brodmann area 8) and data from patients included in this study were provided by the Biobank HUB-ICO-IDIBELL (PT20/00171), integrated in the ISC-III Biobanks and Biomodels Platform (Spain), and they were processed following standard operating procedures with the appropriate approval of the Ethics and Scientific Committees. Cases with AD-related pathology were considered those showing neurofibrillary tangles and/or senile plaques, with the distribution established by Braak and Braak at the post-mortem neuropathological examination [5]. These were categorized as Braak stages I–II [n = 13; 5 females/8 males; 68 ± 7 years (mean ± standard error of the mean)], Braak stages III–IV (n = 12; 4 females/8 males; 74 ± 5 years), and Braak stages V-VI (n = 12; 6 females/6 males; 75 ± 9 years). Special care was taken not to include cases with combined pathologies to avoid bias in the pathological series. Non-demented, control subjects consisted of 14 cases (5 females/9 males; 55 ± 11 years). The mean post-mortem interval of the tissue was 5 h in all cases, with no significant difference between the subgroups. See Supplemental Table 1 for summarized details.

The CSF collection was obtained from the Clinical Neurochemistry Laboratory (Mölndal, Sweden). The CSF samples used were deidentified leftover aliquots from clinical routine analyses, following procedures approved by the Ethics Committee at the University of Gothenburg. Patients who were designated as AD or non-AD controls had typical core CSF biomarker levels using cutoffs that are > 90% specific for AD [18]. The collection was composed of 15 non-AD control patients (9 males/6 females, 69 ± 7 years) and 15 AD patients (4 males/11 females, 72 ± 8 years). AD patients have a CSF biomarker profile indicative of AD, including increased total tau (T-tau: 750 ± 212 pg/mL) and phosphorylated tau (P-tau181: 114 ± 31 pg/ mL) together with low Aβ1-42 (Aβ42: 509 ± 93 pg/mL) concentrations in CSF. Non-AD controls have normal levels of all three CSF biomarkers [T-tau: 250 ± 67 pg/mL, P-tau181: 32 ± 8 pg/mL, (Aβ1-42) Aβ42: 964 ± 332 pg/mL)]. See Supplemental Table 2 for summarized details.

Knock-out mice, TgF344-AD rats and CSF extraction

Brain cortices and CSF samples were also collected from four (2 male/2 female) 3-month-old Mep1b knockout (KO) and four (2 male/2 female) littermates in a C57BL/6 J background [34]. The meprin-β KO animals that were born did not exhibit overt phenotypic abnormalities other that smaller size than the average wild-type litter, with no anatomical and histological differences. Microarray and western blot analyses confirmed that meprin-β mRNA and protein was absent in the KO animals. CSF (~ 4 µL) was obtained in isoflurane anesthetized animals by cisternal puncture with a needle inserted in the suboccipital region through the atlanto-occipital membrane, with a single incision of the subarachnoid space as previously described [11, 13]. CSF samples were centrifuged at 1000 × g for 10 min at 4 °C and stored at -80 °C until further analysis. For western blot analysis, we pooled and loaded the CSF from the 4 Mep1b-KO and 4 wild-type (~ 12 μL, a minimal volume demonstrated in preliminary experiments to be sufficient for the determination of meprin-β subunits in CSF).

Transgenic TgF344-AD rats (n = 8, 4 female/4 male) expressing mutant human APP (APPsw) and presenilin-1 (PS1ΔE9) genes [10] and wild-type Fischer rats (n = 8). Rats were bred in the animal research facilities at the University of Barcelona. Animals were provided with food and water ad libitum and maintained in a temperature-controlled environment in a 12/12 h light–dark cycle. CSF samples (50–100 μL) were collected from ketamine/xylazine anesthetized animals by cisternal puncture with a single incision into the subarachnoid space [25, 40]. CSF was obtained from 19–19.5 month-old animals. This study was part of a large project that included brain analysis at different stages. Animal work was performed in accordance with the local legislation, with the approval of the Experimental Animal Ethical Committee of the University of Barcelona (Spain), and in compliance with European legislation.

Neuronal differentiation from human induced pluripotent stem cells (iPSCs)

The iPSC line used in this research was the N1-001iC2 hPSCreg name UGOTSAi002-B. The donor was a male aged 75–79 years old at collection with sporadic AD. The iPSC line was maintained in Matrigel (Corning #356,234)-coated 6-well plates and fed with mTeSR™1 (StemCell Technologies) at 37 °C, 5% CO₂. Passages were done every 4–5 days (80% confluence) using ReLeSR™ solution (StemCell Technologies). Lines were routinely screened for the absence of Mycoplasma.

iPSCs were detached using Gentle Cell Dissociation Reagent (StemCell Technologies #100–0485) at 80% confluence and differentiated into neural progenitor cells (NPCs) using the STEMdiff™ SMADi Neural Induction Kit (StemCell Technologies). Then, NPCs were expanded and matured using STEMdiff™ Neural Progenitor Medium (StemCell Technologies). NPCs were passaged using accutase (Millipore-Sigma #A6964). From NPCs, cells were differentiated into neurons using BrainPhys medium (StemCell Technologies #05790). The following set of supplements were added to the BrainPhys medium to obtain the neurodifferentiation medium: NeuroCult SM1 Neuronal Supplement 2% (StemCell Technologies #05711), N2 Supplement-A 1% (STEMCELL Technologies #07152), Dibutyryl cAMP sodium salt 1 mM (Millipore-Sigma D0627), L-Ascorbic acid 200 nM (Millipore-Sigma #A0278), BDNF 20 ng/mL (Peprotech #450–02), GDNF 20 ng/mL (Peprotech #450–10), and Penicillin/Streptomicin 0.25 × (Gibco #15,140,122). The coating where NPCs were plated for differentiation consisted of Poly-L-ornithine solution diluted 1:1 in phosphate-buffered saline (PBS) overnight at 4ºC (Millipore-Sigma #P4957), followed by 4 h at 37ºC of Laminin 2020 (Millipore-Sigma #L2020). During the neural induction period, the expression of Nestin and PAX6 was checked as markers of NPCs, and the expression of MAP2 and NEUN as a marker of mature neurons. Cells were cultured for 60 days in BrainPhys medium as described [12].

For treatment with Aβ peptides, Aβ42 and the scrambled control peptide (DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA) (AnasPec, Fremont, catalog number AS-20276, CA, USA) were dissolved in sterilized distilled water at a concentration of 1 mg/mL, aliquoted, and stored at -80 °C. Suspensions of Aβ42 or the scrambled peptide corresponding to a final concentration of 2 μM were added to each culture. After 24 h of peptide treatment, the conditioned medium from each plate was removed and the cells were washed twice with PBS, harvested, and suspended in 100 μL of ice-cold buffer extraction and processed as described below.

Protein extraction from brain tissue and iPSCs

Tissues stored at − 80 °C were thawed gradually at 4 °C and pieces of human prefrontal cortex (100 mg) were homogenized (10% w/v) in ice-cold 50 mM Tris–HCl, pH 7.4/ 150 mM NaCl/0.5% Triton X-100/0.5% Nonidet P-40 supplemented with a cocktail of proteinase inhibitors and processed as previously described [43]. In brief, the homogenates were sonicated and centrifuged at 70,000 × g at 4 °C for 1 h; the supernatant was collected, aliquoted and frozen at –80 °C until use.

For iPSCs, cells were harvested in ice-cold cellular lysis buffer, 50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% (w/v) Nonidet P-40, 0.5% (w/v) Triton X-100, supplemented with a cocktail of protease inhibitors. The cell lysates were sonicated with Bioruptor® Pico sonication device from Diagenode (2 cycles of 10 s of sonication followed by 30 s of resting in a 4 °C water bath) and centrifuged at 70,000 × g at 4 °C for 45 min; the supernatant was collected, aliquoted and frozen at –80 °C until use. Total protein concentration of the supernatants was determined using the BCA Protein Assay Kit.

Immunoprecipitation assays

Brain extracts or CSF (400 μL) were incubated on a roller overnight at 4ºC with Protein A Sepharose CL-4B (100 μL, Cytiva 17,078,001) coupled to antibodies against meprin-β N-terminal (rabbit, 10 μL, Abcam ab42743). The same amount of beads without coupled antibody was used as negative control. The input, bound and unbound fractions were analysed by western blotting using the N-terminal AF2895 or NBP2 C-terminal antibodies.

Western blot

Brain and CSF samples were run on Sodium Dodecyl Sulfate–Polyacrylamide Gel electrophoresis (SDS-PAGE; 7,5% Tris–glycine) after boiling at 98ºC for 5 min in 6 × Laemmli sample buffer. Electrotransference was performed into nitrocellulose membranes for 2 h at 300 mA at 4ºC. Primary antibodies were used against meprin-β C-terminal from Novus (rabbit, 1:000, NBP2-99,024) and N-terminal from Abcam (rabbit, 1:1000, ab42743) or R&D (goat, 1:1000, AF2895), and against α-tubulin (1:4000, Sigma-Aldrich T5168) or vinculin (Sigma-Aldrich; mouse monoclonal; 1: 1000), as loading controls. Protein concentration in brain samples was measured and 30 μg were loaded per lane. For human CSF samples 30 μL were loaded per lane and total protein staining by No-Stain Protein Labeling Reagent (ThermoFischer Scientific) served to monitor potential loading inaccuracies and normalize levels. For rat CSF samples, due to volume limitation, only 12 µL were loaded per lane; all samples showed a uniform background across the blot and preserving the validity of relative comparisons. For brain as well as for CSF western blots, a lane was reserved for a sample that was present in every immunoblot to serve as an internal control and normalization between individual blots. For iPSC-derived neurons, 40 μg were loaded per lane. Primary antibody binding was visualized with fluorescent secondary antibodies (IRDye, 1: 10,000, Bonsay), and images were acquired using an Odyssey CLx Infrared Imaging system (LI-COR Biosciences GmbH). Immunoreactive band intensities were quantified using LI-COR software (Image Studio Ver. 5.2).

Quantitative polymerase chain reaction (qRT‑PCR) analysis

RNA was extracted from human brains using the TRIzol® Reagent in the PureLink™ Micro-to-Midi Total RNA Purifcation System (Life Technologies, Carlsbad, CA, USA) following the manufacturer’s instructions. SuperScript™ III Reverse Transcriptase (Life Technologies, Carlsbad, CA, USA) was used to synthesize cDNAs from this total RNA (2 μg) using random primers according to the manufacturer’s instructions. Quantitative PCR amplification was performed on a QuantStudio™ 3 Real-Time PCR System (Applied Biosystems, Thermofisher Scientific, A28567) with TaqMan probes specific for human MEP1B (assay ID: Hs00983260_m1) (Applied Biosystems, Thermofisher Scientific, Rockford, USA), and human glyceraldehyde 3-phosphate dehydrogenase (GAPDH, forward: AGAAGGTGGTGA AGCAGGCAT; reverse: AGGTCCACCACTCTGTTG CTGT) as a housekeeping gene to normalize the expression levels of the target gene by the ΔΔCt method.

Immunocytofluorescence

Cells were fixed in 4% paraformaldehyde for 15 min at room temperature (RT) and stored in PBS at 4 °C until immunostaining was performed. Cells were washed in PBS once and permeabilized using 0.1% Triton X-100 in PBS for 10 min and blocked in 1% bovine serum albumin in PBS for 1 h at RT. Then, cells were incubated with primary antibodies in blocking solution overnight at 4 °C. Primary antibodies used were MAP2 (chicken, 1:500, Thermo Fischer PA1-10,005) and NEUN (rabbit, 1:100, Thermo Fischer PA5-78,499). After three washes with PBS, secondary antibodies (Alexa FluorTM 647 goat anti-rabbit IgG (H + L), Alexa FluorTM 488 goat anti-chicken IgY (H + L), Thermo Fischer) were added in blocking solution for 1 h RT in the dark. 4',6- diamidino-2-phenylindole (DAPI, 1 μM) was added as a nuclear counterstain before the last wash with PBS. Cells were mounted on slides using ProLongTM Diamond Antifade Mountant (Invitrogen, P36961). Images were taken by a Leica SPEII confocal microscope and visualized using Imaris.

Statistical analysis

The distribution of data was tested for normality using a D’Agostino-Pearson test. We employed the ROUT method. For comparison of two groups following parametric distributions, we used Student’s t-test, and for nonparametric distributions, we used the Mann–Whitney test to determine p-values. Correlations were performed using two-tailed Pearson correlation coefficients. The results are presented as the means ± standard error of the mean (SEM), and all the analyses were performed using GraphPad Prism (version 7; GraphPad Software, Inc). p-value < 0.05 was considered significant.

Results

Characterization of meprin-β in the brain

While the predicted molecular mass of meprin-β is 84 kDa, its actual mass is expected to be at least 20–30 kDa higher due to glycosylation [7, 26]. The protein contains several potential N-glycosylation (Asn218, Asn254, Asn370, Asn421, Asn436, Asn445, Asn547, Asn592) and O-glycosylation sites (Ser593, Thr594, Thr599, and Ser603); and the O-glycosylation is thought to protect the C-terminal region from proteolytic cleavage [26]. Accordingly, the immunoreactive band corresponding to the mature species is usually observed around 95–130 kDa for both human [32, 38] and mice [34, 47], and the differences in size are attributable to differential glycosylation [38].

To firstly address the specificity of the immunoreactive band for meprin-β, we homogenized brain tissue from control and knockout (KO) mice for Mep1b, the gene encoding meprin-β, and resolved the western blot with an antibody raised against the N-terminal ectodomain (antibody AF2895; Fig. 1A), showing clear immunoreactive bands of ~ 100 and 110 kDa in the wild-type, which were absent in the Mep1b-KO mice. Diffuse additional bands around these main immunoreactive bands may indicate intermediate or alternative glycoforms. Smaller bands (~ 70–80 kDa) were also identified (Fig. 1A). Surprisingly, a very similar banding pattern was observed with a C-terminal antibody NBP2-99,024 (Fig. 1A), also for the smaller ~ 70–80 kDa bands. The immunoreactivity of the 70–80 kDa bands to antibodies against the N-terminal ectodomain and the C-terminal intracellular domain discards the possibility of being shed fragments.

Fig. 1.

Identification of meprin-β in the brain. A Western blot of brain cortex homogenates from control wild-type (Ctrl) and KO for Mep1b (Mep1b − / −) mouse resolved with meprin-β N-terminal AF2895 and C-terminal NBP2 antibodies. B Representative western blot of human frontal cortex homogenates revealed with the N-terminal and C-terminal antibodies previously described. C Immunoprecipitation (IP) of human frontal cortex homogenate against meprin-β using the alternative N-terminal antibody ab42743, and resolved with N-terminal AF2895. Input: brain homogenate; B: bound fraction from the IP; Bc: unspecific bound from control IP. Arrowheads indicate the bands attributed to the zymogen (black) and the mature species generated after cleavage of the pro-domain; (*) unspecific bands. The uncropped blots of the analyzed samples are available as Supplemental Fig. 1

Then, we analyzed human brain extracts by western blotting using the N-terminal AF2895 antibody and the intracellular C-terminal domain antibody (NBP2-99,024) (Fig. 1B). Again, several bands ranging from 70 to 120 kDa were identified, with the most prominent bands being about 100 and 110 kDa. We also combined the antibodies in immunoprecipitation assays, using an alternative ectodomain antibody, ab108252, followed by western blotting using the AF2895 antibody (Fig. 1C).

The results confirmed the existence of diverse species of full-length meprin-β in human brain. For further analysis, we focused on the ~ 110 kDa band, which we attributed to the zymogen species, and the 100 kDa band, which we attributed to the mature species generated at the plasma membrane after cleavage of the pro-domain. The smaller immunoreactive bands of uncertain identity displayed a strong, unspecific binding to the matrix used as a control in the immunoprecipitation, and were therefore not further considered.

Characterization of meprin-β in the CSF

Despite the fact that meprin-β can be shed from the plasma membrane by ADAM10, and that secreted meprin-β has been identified in the media of cultured cells [26, 38], to our knowledge, the existence of meprin-β in the CSF has not been addressed to date.

Once more, we employed samples from Mep1b-KO mice to validate the specificity of the immunoreactive bands, resolving with N-terminal AF2895 and C-terminal NBP2-99,024 antibodies (Fig. 2A). In CSF samples from wild-type mice, immunoblotting with both antibodies revealed a ~ 100 kDa band. The shorter ~ 70–80 kDa bands were also present in CSF from Mep1b-KO mice and were, therefore, considered as unspecific (Fig. 2A). In all human CSF samples analysed, the ectodomain N-terminal (AF2895) as well as the C-terminal (NBP2-99,024) antibodies detected ~ 100 kDa and ~ 110 kDa bands, consistent with those detected in brain extracts (Fig. 2B). Meprin-β species display small differences in molecular masses displayed by electrophoretic mobility in CSF compared with those from detergent-solubilized brain or cell extracts, as previously reported for other membrane proteins, likely reflecting different degrees of interaction with sodium dodecyl sulfate (SDS) [39]. Similarly, to what we did with brain samples, we demonstrated the identity of the 100 kDa and 110 kDa bands by immunoprecipitation and a combination of alternative antibodies (Fig. 2C).

Fig. 2.

Identification of meprin-β in the CSF. A Western blot of CSF obtained from the cisterna magna of control wild-type (Ctrl) and KO for Mep1b (Mep1b − / −) mouse, and resolved with meprin-β N-terminal AF2895 and C-terminal NBP2 antibodies. B Representative western blot of human lumbar CSF (control cases) revealed with the N-terminal and C-terminal antibodies previously described. C Immunoprecipitation (IP) of human frontal cortex homogenate against meprin-β using the alternative N-terminal antibody ab42743, and resolved with N-terminal AF2895. Input: brain homogenate; B: bound fraction from the IP; Bc: unspecific bound from control IP. Arrowheads indicate the bands attributed to the zymogen (black) and the mature species generated after cleavage of the pro-domain; (*) unspecific bands. The uncropped blots of the analyzed samples are available as Supplemental Fig. 2

Overall, we detected immunoreactivity for meprin-β in human frontal cortex extracts and in CSF samples compatible with the zymogen and the mature species retaining the intracellular domain, but no shedded fragments.

Meprin-β is increased in the brain of AD patients and in cellular extracts of iPSC-derived neurons treated with Aβ42

We compared meprin-β levels between control and AD cases, subgrouped by different Braak stages. We resolved the blot with the AF2895 antibody targeting the ectodomain and employed quantitative fluorescent western blotting. A representative western blot comparing meprin-β banding pattern in control and AD samples is shown in Fig. 3A. The levels of the meprin-β attributed to the mature species did not differ from those in controls in Braak I-II (78 ± 42%; p = 0.089) nor in Braak III-IV (90 ± 37%; p = 0.358), but reached a significant increase in Braak V-VI (140 ± 30%; p < 0.001) (Fig. 3B). The levels of the meprin-β attributed to the zymogen species did not differ from those in controls in Braak I-II (101 ± 43%; p = 0.925) nor in Braak V-VI (121 ± 45%; p = 0.211), but reached a significant decrease in Braak III-IV (69 ± 28%; p < 0.05). Linear regression analysis did not detect any correlation between age and meprin-β levels in the control group, nor in the Braak I-II, Braak III-IV, or Braak V-VI groups. The levels of meprin-β did not correlate with post-mortem interval or gender, either.

Fig. 3.

Meprin-β levels in the AD brain. A Representative Western blot of meprin-β in human brain cortex (30 μg loaded) resolved with meprin-β N-terminal AF2895. An internal sample was used to normalize among different membranes. B Quantification of the levels of meprin-β immunoreactive bands attributed to mature and zymogen levels was normalized to loading control α-tubulin and expressed as percentages with respect to controls. Cases control (Ctrl) n = 14, Braak I-II n = 13, Braak III-IV n = 12, and Braak V-VI n = 12. C) Quantification of meprin-β mRNA levels was normalized to GAPDH transcript levels and expressed as percentages with respect to controls. *p < 0.05, **p < 0.01, ***p < 0.001, t-test, Mann–Whitney. The uncropped blots of the analyzed samples are available as Supplemental Fig. 3

We performed RT-qPCR studies to assess possible differences in the transcript levels between control and AD individuals. RNA was extracted from brain frontal cortex tissue adjacent to that used for western blotting analysis. We found that meprin-β transcript levels were elevated in Braak III-IV (201 ± 154%; p = 0.032) and in Braak V-VI (181 ± 106%; p = 0.023) with respect to control subjects (Fig. 3C). No significant correlation was found between meprin-β mRNA levels and age when subgroups were considered separately or AD grouped together, and no correlations were found between meprin-β mRNA levels and post-mortem delay or sex.

An altered β-secretase activity can be a key molecular contributor at the crossroads of a vicious cycle contributing to Aβ generation [9, 36]. Thus, to determine the role of Aβ in the regulation of meprin-β, we generated iPSC-derived neuronal cultures and treated them with Aβ42 to model a β-amyloid stress condition. First, iPSCs were differentiated into neural progenitor cells (NPCs), to then be differentiated into mature neurons (as illustrated in Fig. 4A) using the BrainPhys medium and a customized set of supplements (see Methods). After 60 days (D60) of neuronal maturation, these neurons show complex neurites and immunostaining experiments indicate that they express markers of mature neurons, including MAP2 and NEUN (Fig. 4B). At this maturation stage, our iPSC-neurons show a highly mature phenotype displaying electric currents and calcium signal [12]. At D60, iPSC-derived neurons were treated with 2 μM Aβ42 or same concentration of a scrambled peptide (control) for 24 h; next, cells were harvested and prepared for western blot analysis. For these experiments, we selected doses of Aβ42 that do not recapitulate physiological concentrations, but rather can induce a cellular response. Therefore, this concentration does not cause neurotoxicity, but mimics pathological conditions; similar doses of Aβ42 have been used by us [14, 31] and others [45]. We demonstrated that Aβ42 can trigger an increase in the levels of the meprin-β attributed to the mature species (69 ± 15%; p = 0.001), as well for the zymogen species (50 ± 20%; p = 0.032), compared with controls (Fig. 4C).

Fig. 4.

Meprin-β in iPSC-derived neurons treated with Aβ42. A Schematic illustration of the neurodifferentiation process from iPSCs into mature neurons. B Immunocytofluorescence images showing MAP2 + (green) and NeuN + (red), as well as DAPI (blue) in mature neurons after 60 days in culture (D60). C Representative western blot of meprin-β in extracts from D60 iPSC-derived neurons treated with Aβ42 2 μM for 24 h (Aβ), or with same concentration of a scrambled peptide (Scr), resolved with meprin-β N-terminal AF2895. D Quantification of the levels of meprin-β immunoreactive bands attributed to mature and zymogen levels was normalized to loading control α-tubulin and expressed as percentages with respect to controls. Each dot in the plot corresponds to a harvested pool of 3 wells of iPSC-derived neurons from 12-well plate. *p < 0.05, **p < 0.01, t-test. The uncropped blots of the analyzed samples are available as Supplemental Fig. 4

Meprin-β is increased in the CSF of AD patients and in the CSF of an AD-Tg rat model

To assess whether similar changes occur in AD CSF, we analyzed meprin-β species levels in lumbar CSF from AD and non-AD control subjects (Supplemental Table 2), using western blotting with the ectodomain AF2895 antibody. As described above, in all CSF samples, immunoblotting revealed meprin-β immunoreactive bands attributed to the full-length zymogen and mature species (Fig. 5A). When we compared the levels of meprin-β in control and AD cases, we obtained elevated levels of the mature meprin-β in the CSF of patients with AD (171 ± 108%; p = 0.041), while the level of the zymogen remained unaltered (Fig. 5B). Linear regression analysis showed no correlation between CSF meprin-β with any of the core biomarker. No significant correlation was found between the levels of CSF meprin-β species and age or sex.

Fig. 5.

Meprin-β levels in the AD CSF. A Representative western blot of meprin-β in lumbar CSF from non-AD controls (Ctrl, n = 15) and AD cases (n = 15) resolved with N-terminal AF2895 antibody. An internal sample was used to normalize among different membranes. 1 control and 2 AD samples were detected as outliers and excluded from the analysis of the zymogen band (not represented). B Quantification of the levels of meprin-β immunoreactive bands attributed to mature and zymogen levels is expressed as percentages with respect to controls. Quantification of the levels of meprin-β immunoreactive bands attributed to mature and zymogen levels was normalized to total protein staining levels and expressed as percentages with respect to controls *p < 0.05, t-test. The uncropped blot of the analyzed samples is available as Supplemental Fig. 5

To determine whether β-amyloid pathology could account for the increased meprin-β levels observed in the CSF of patients with AD, we used a transgenic rat model of AD. The TgF344-AD rat expresses human APP with the Swedish mutation and human PSEN1 with the Δ exon 9 mutation. We examined meprin-β levels in the CSF of 19-month-old transgenic rats and wild-type littermates by western blotting using the AF2895 antibody targeting the ectodomain. In all the rats, and similarly to what was observed in mice, meprin-β appeared as a single ~ 100 kDa immunoreactive band that we attributed to the mature species (Fig. 6A). CSF meprin-β levels were higher in TgF344-AD animals than in wild-types (63 ± 25%; p = 0.026, Fig. 6B). No differences were associated with sex.

Fig. 6.

Meprin-β levels in transgenic TgF344-AD rat CSF. A Representative western blot of CSF meprin-β in TgF344-AD (Tg; n = 8) and controls (Ctrl; n = 8) obtained from the cisterna magna. Blots were resolved with meprin-β N-terminal AF2895 antibody. An internal sample was used to normalize among different membranes. B Data are expressed as percentages with respect to wild-type control rats. *p < 0.05, t-test. The uncropped blot of the analyzed samples is available as Supplemental Fig. 6

Discussion

An imbalance between α-secretases and β-secretases, competing for APP processing, can lead to dysregulated Aβ generation. While BACE1 is the primary enzyme responsible for β-secretase activity (reviewed in [23], meprin-β has emerged as an alternative β-secretase of APP [3]. However, characterizing the alterations of meprin-β in AD remains a major knowledge gap. Addressing the potential role of meprin-β in AD pathology could contribute to our understanding of APP processing regulation and influence treatment paradigms involving β-secretase inhibitors.

To assess the potential alteration of meprin-β levels, it is crucial to understand that meprin-β is synthesized as a zymogen, and therefore, the post-translational processing at the N-terminal domain is determinant. Due to the lack of antibodies against the meprin-β pro-domain region, in this study we discriminated between protein species based on small differences in molecular mass determined by electrophoresis. In previous reports, the plasma membrane-resident meprin-β, likely comprising the pro-meprin-β and the mature species, has been described to differ by 8–10 kDa in apparent molecular mass [38], consistent with the immunoreactive banding pattern obtained in our iPSC-derived neurons. Accordingly, in our analysis, the ~ 110 kDa band is attributed to the pro-meprin-β zymogen, and the ~ 100 kDa band is attributed to the mature species, the latter presumed to be the active form.

While meprin-β mRNA levels increased in AD, the most remarkable observation was the increase in the mature protein species of meprin-β in the AD brain. Treatment of human iPSC-derived neuronal cultures with Aβ42 increased meprin-β levels, suggesting that amyloid pathology may be responsible for this alteration. In agreement, meprin-β was also increased in the CSF of the animal model of amyloid pathology, the TgF344-AD rat.

The possible existence of meprin-β ectodomain fragments has been suggested [38]. In addition to ADAM10/17, the membrane-type-I matrix metalloproteinase (MT1-MMP) is another membrane-bound secretase that has demonstrated the ability to shed meprin-β from the plasma membrane in cellular models [46]. In these cellular models, MT1-MMP and ADAM10/17 exhibit a remarkable overlap in shedding properties, and mass spectrometry-based cleavage site analysis revealed that meprin-β cleavage occurs between Pro602 and Ser603 [46]. Thus, after cleavage, the resulting membrane-tethered fragment will comprise ~ 100 amino acid residues, not compromising potential N-glycosylation sites, only the O-glycosylation of the C-terminal. In fact, the resulting ectodomain fragment obtained in cellular models exhibited a ~ 15–20 kDa reduction in molecular mass compared with meprin-β membrane-bound species. Accordingly, when membrane-bound meprin-β is solubilized with papain, the resulting ectodomain fragment exhibited ~ 20 kDa reduced molecular mass [30]. As advanced, interestingly, the experiments demonstrating that meprin-β can be shed by ADAM10/17 have been performed in over-expressing cellular models, and most of them showed for the soluble species a similar molecular mass that the one exhibited in cellular extracts [20–22, 38]. More intriguingly, the ADAM10/17-mediated shedding is restricted to the inactive pro-form of meprin-β, being meprin-β activation and shedding mutually exclusive events [47]. The tissues in which this modulatory cross-talk between meprin-β and ADAM10/17 takes place and the physiological mechanism determining meprin-β shedding are currently unknown. In fact, it has been suggested that meprin-β is exclusively a membrane-bound species. In the context of AD, ADAM10 levels are compromised [44].

In this complex scenario, to our knowledge, the existence of soluble meprin-β in vivo, particularly in CSF, has not yet been demonstrated. In this study, we investigated the presence of meprin-β in CSF and whether levels are altered in AD cases. We have found in human CSF only full-length species of meprin-β, the zymogen and the mature forms, but not truncated fragments. We report that the immunoreactive band attributed to the mature species of meprin-β increases in AD CSF compared to non-AD samples, but not the pro-meprin-β zymogen. An increase in meprin-β was also described in CSF samples from TgF344-AD, a rat transgenic AD model. CSF samples from wild-type and Mep1b-KO mice served to validate the existence of meprin-β in this fluid, while the meprin-β zymogen appears not to be present in CSF from rodents.

Interestingly, BACE1 and ADAM10, type I transmembrane proteins such as meprin-β, were also detected in human CSF as full-length forms retaining the intracellular C-terminal domain, including zymogen species [29, 44]. The mechanisms by which these membrane-bound secretases reach the CSF are unknown, but passive release from brain cells or neuronal cell death may be possible, as observed for BACE1 [2]. The possible association of meprin-β with brain-derived extracellular vesicles also appears plausible. Indeed, BACE1 [28], ADAM10 and ADAM17 have also been shown to be secreted outside cells in exosomes [48]. In this context, it has been proposed that meprin-β is also released onto EVs when ADAM10/17 is inhibited or knocked out [47].

There are limitations to the current study, particularly related to a necessary and undisputable interpretation of the different meprin-β species. We verified the specificity of meprin-β immunoreactivity by using the brain extract and CSF from Mep1b-null and wild‐type mice and combining antibodies. In brain extracts, we confirmed a complete loss of meprin-β immunoreactivity in the knockout, but in CSF, we identified some bands as unspecific. CSF represents a distinct biological matrix with a markedly different protein composition, dominated by highly abundant proteins such as albumin and immunoglobulins. We prevent the misinterpretation of signals and bands that potentially co-migrate with putative fragments, and these bands were not included in the quantification. New antibodies against the meprin-β pro-domain, or a pan-specific antibody against the predicted N-terminus of the processed mature species, could be valuable for suitable discrimination. Moreover, the CSF and brain samples are not from the same subjects, limiting the interpretation of changes in CSF meprin-β as a read-out of brain changes. Lastly, our approach is based on the determination of protein levels of the mature species of meprin-β form, presumably the active species, but not on the determination of enzymatic activity. In a previous study, we determined β-secretase activity in the CSF using a peptide substrate homologous to the BACE1 cleavage site and a BACE1 inhibitor, defining enzymatic activity specifically for BACE1 as the activity that was blocked by the specific inhibitors; this study indicated that the majority of the β-secretase activity was not inhibited by specific BACE1 inhibitors [29]. Anyhow, the reliability of β-secretase enzymatic activity determination in CSF is confounded by many factors, including the differential effect of the membrane lipid microenvironment, the possible existence of heteromeric complexes, and the distorted conformational state of the membrane-resident enzyme when it is present in the CSF. Many other factors can affect the enzymatic activity of a protein when it is either in the brain or in the CSF. For example, meprin-β protein level is elevated in AD CSF, but meprin-β is a zinc-dependent metalloprotease and zinc levels appeared reduced in AD CSF [37]. The characterization of the catalytic activities of the specific β-secretase enzymes remains as a great challenge. Due to all these caveats, the present finding should be considered conservatively; independent replications in a larger cohort of cases will be of value.

In summary, the present study combines Mep1b knockout mice, immunoprecipitation assays, and alternative N-terminal and C-terminal antibodies to characterize meprin-β in the human brain and CSF. We demonstrate that meprin-β protein levels are increased in the brain and the CSF of AD patients, and also in iPSC-derived neuronal cultures after Aβ42 treatment, and in the CSF of the TgF344-AD rat. We propose that meprin-β upregulation represents a late-stage, context-dependent response to pathological Aβ unbalance acting as a part of a vicious circle of AD pathogenesis. Although our present findings provide sufficient evidence to support a role of meprin-β in AD pathology, additional studies are necessary to fully elucidate its role in AD and to develop analytical approaches.

Meover, meprin-β can be a relevant protein in AD, but it is also present in different tissues, processing different substrates and is dysregulated orin diseases associated with inflammation, neurodegeneration, cancer and fibrosis [6], and the determination of meprin-β in biological fluids could serve to monitor the disease.

Authors’ contributions

SE conducted most of the experiments and edited the manuscript. CAG conducted the experiments on the transgenic rat, contributed to iPSC-derived neuron cultures and experiments and edited the manuscript. CMM contributed to optimizing immunoprecipitation experiments. MK and CUP provided the Meprin beta-null mice. ISD and FA provided the transgenic rat. HZ provided the CSF collection. ICI and the other authors reviewed the manuscript. JSV designed the experiments, supervised the process and wrote and edited the manuscript.

Funding

This work was supported by grants from the Fondo de Investigaciones Sanitarias (PI22/01329, co-funded by the Fondo Europeo de Desarrollo Regional, FEDER “Investing in your future”), CIBERNED (Instituto de Salud Carlos III, Spain), from the Direcció General de Ciència i Investigació, Generalitat Valenciana (CIAICO/2024/313), and “Severo Ochoa” Program for Centers of Excellence in R&D (CEX2021-001165-S). SE is funded by the Instituto de Salud Carlos III (PFIS fellowship). This study was also supported by the by the Spanish Ministry of Science, Innovation and Universities MICINN/FEDER (PID2022-142187OB-I00), to FA. CAG is supported by a predoctoral contract (PRE2022-104182) funded by the Agencia Estatal de Investigación (AEI) and the Ministerio de Ciencia e Innovación (MCIN) under the Plan Estatal de I + D + I 2021–2023, co-funded by the European Social Fund (FSE).

HZ is a Wallenberg Scholar and a Distinguished Professor at the Swedish Research Council supported by grants from the Swedish Research Council (#2023–00356, #2022–01018 and #2019–02397), the European Union’s Horizon Europe research and innovation programme under grant agreement No 101053962, Swedish State Support for Clinical Research (#ALFGBG-71320), the Alzheimer Drug Discovery Foundation (ADDF), USA (#201,809–2016862), the AD Strategic Fund and the Alzheimer's Association (#ADSF-21–831,376-C, #ADSF-21–831,381-C, #ADSF-21–831,377-C, and #ADSF-24–1,284,328-C), the European Partnership on Metrology, co-financed from the European Union’s Horizon Europe Research and Innovation Programme and by the Participating States (NEuroBioStand, #22HLT07), the Bluefield Project, Cure Alzheimer’s Fund, the Olav Thon Foundation, the Erling-Persson Family Foundation, Familjen Rönströms Stiftelse, Stiftelsen för Gamla Tjänarinnor, Hjärnfonden, Sweden (#FO2022-0270), the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 860197 (MIRIADE), the European Union Joint Programme – Neurodegenerative Disease Research (JPND2021-00694), the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre, the UK Dementia Research Institute at UCL (UKDRI-1003).

Data availability

All clinical and analyses data are available from authors upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the ethics committee of both, the Universidad Miguel Hernández (IN.JSV.04.20 & PRL.DBB.JSV.231113) and the Departamento de Salud de Alicante – Hospital General Universitario Dr. Balmis de Alicante, Spain (PI2025-173), and it was carried out in accordance with the Declaration of Helsinki and its later amendments.

Competing interests

The authors declare no competing interests.