Aβ antibodies target not only amyloid plaques but also distinct brain cells and vessels

Abstract

BACKGROUND

Amyloid beta (Aβ) antibodies are the only therapies to slow cognitive decline in Alzheimer's disease (AD). Yet the sites where antibodies engage Aβ in the brain and mechanisms that lower Aβ are not fully understood. Defining Aβ antibody localizations in the brain is essential to understand how immunotherapy is beneficial for AD.

METHODS

N‐terminal Aβ antibody 6E10 was injected via three different routes into different AD mouse models. N‐terminal Aβ antibodies were empirically shown as most effective in AD mice. Antibody localization was examined in the brain after injections. Glymphatic dynamics were also evaluated.

RESULTS

As expected, Aβ antibody 6E10 bound to plaques but remarkably also localized to vulnerable neurons, such as hippocampal CA1 pyramidal cells, as well as microglia, astrocytes, oligodendrocytes, perivascular macrophages, and blood vessels. Antibodies did not alter glymphatic function.

DISCUSSION

We provide detailed localizations of antibodies in AD mouse brains, offering insights into targets of Aβ antibody‐based immunotherapy.

Highlights

• Amyloid beta (Aβ) antibody 6E10 shows broad distribution across plaques and multiple cellular/vascular compartments in different Alzheimer's disease (AD) mouse models.

• Localization of antibody 6E10 in AD mice was detected in selective neurons, astrocytes, microglia, perivascular macrophages (PVMs), and oligodendrocytes/oligodendrocyte precursor cells, and at the external aspects of blood vessels.

• Intracellular human‐specific Aβ antibody 6E10 appears related to phagocytic activity, as PVMs and microglia in wild‐type mice also contained 6E10, or to intracellular Aβ, as 6E10 in neurons, and oligodendrocytes were only detected in AD mice.

• Aβ antibody injections by different routes failed to augment glymphatic circulation in AD mice.

1. BACKGROUND

Amyloid beta (Aβ) immunotherapy is the first therapy targeting disease mechanisms to show cognitive benefits in clinical trials for Alzheimer's disease (AD). ¹ The mechanism generally viewed as most important to these, albeit still modest, benefits is antibody‐mediated degradation of amyloid plaques by microglia. However, plaques are known not to correlate well with cognition, and increasingly, more soluble forms of Aβ “oligomers” have been viewed as the more pathogenic form(s). ² For example, an inducible AD mouse model showed that turning off the mutant amyloid precursor protein (APP) transgene rapidly improved behavior while plaques remained unchanged. ³ Further, reduced brain activity via sleep induction or unilateral whisker removal reduced plaques but increased synapse damage in AD mice. ⁴ Thus, there is a disconnect between Aβ‐induced synapse damage and plaques. Although it is still widely viewed that secreted extracellular Aβ initiates plaque formation ⁵ and then plaques cause damage to surrounding neurons from Aβ release, ⁶ a growing body of evidence, starting with our study, showed that Aβ begins to accumulate and aggregate within AD‐vulnerable neurons prior to plaques. This was initially shown in human Down syndrome (DS) ⁷ , ⁸ , ⁹ and AD brains, ⁷ , ¹⁰ , ¹¹ , ¹² but then more extensively studied in rodent models of AD. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷

Further clues that plaque removal may not be the only target of Aβ immunotherapy are studies that showed benefits of Aβ antibody F_ab fragments in AD mice ¹⁸ and that AD mice crossed with Fc𝛄 receptor knockout mice still benefited from Aβ immunotherapy. ¹⁹ Moreover, behavioral improvement was reported in AD mice with an Aβ‐targeting antibody that did not reduce plaques. ²⁰ Additionally, behavioral impairment was reported prior to amyloid plaques in 3×Tg‐AD mice, with improvement by Aβ immunotherapy upon reduction of intraneuronal Aβ. ²¹ Neuropathological evaluation in the initial, but aborted, active immunotherapy trial showed plaque removal but continued cognitive decline in patients who came to autopsy. ²² Thus, Fc receptor‐mediated microglia degradation of plaques does not appear to be the only route to improvement. Yet, because plaque removal correlates with success of therapeutic Aβ antibodies, the field currently emphasizes plaque reduction as the critical target. ²³

It is possible that soluble extracellular Aβ oligomers are sequestered by Aβ antibodies, ²⁴ although there is limited evidence for elevations of soluble extracellular Aβ in the brain. Studies that experimentally elevate extracellular Aβ (e.g., by exogenous injection of Aβ into rodent brain ventricles) are therefore inconsistent with the earliest validated biomarker of AD, which is a drop rather than an increase in extracellular Aβ42 in cerebrospinal fluid (CSF). ²⁵ , ²⁶ In contrast to the dearth of evidence for early increases in more soluble extracellular Aβ42, hundreds of articles have reported early intraneuronal accumulation of Aβ in human AD, and DS, and in rodent models of AD. Further, intraneuronal Aβ42 was shown by immuno‐electron microscopy to pathologically aggregate within dystrophic neurites. ¹⁶

In a cellular study, we had provided mechanistic insights into internalization of Aβ antibodies in cultured neurons from AD transgenic mice reducing intraneuronal Aβ and protecting against AD‐like synapse alterations. ²⁷ However, such cellular studies require in vivo support, but up to now, Aβ antibodies were only shown to target plaques.

Studies indicate that Aβ clearance is disrupted in AD, and accumulating evidence suggests that this impairment involves dysfunction of the glymphatic system. ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² The glymphatic pathway is regulated by astrocytic aquaporin 4 (AQP4) polarization around vessels ³³ , ³⁴ and perivascular macrophages (PVMs). ³⁵ , ³⁶ However, how Aβ immunotherapy influences glymphatic function remains unclear.

Aβ immunotherapy in AD mice had empirically demonstrated that antibodies directed at the N‐terminal Aβ domain were particularly effective; ³⁷ in contrast, antibodies to central or C‐terminal regions demonstrated poor clinical efficacy. ³⁸ , ³⁹ , ⁴⁰ We set out to more precisely identify the targets in the brain of N‐terminal Aβ antibody 6E10. Antibody 6E10 was recently used in an intracranial injection paradigm, revealing dynamic crosstalk between microglia and other immune cells during Aβ immunotherapy. ⁴¹ Prior studies showed that Aβ antibodies bind to plaques when injected into AD transgenic mice, ⁴² , ⁴³ but did not further investigate the localizations of antibodies. Here, we show that Aβ antibodies also selectively localize to specific cell populations and blood vessels in the brain of different AD mouse models.

RESEARCH IN CONTEXT

1. Systematic review: Amyloid beta (Aβ) immunotherapy is the first therapy that impacts the underlying Alzheimer's disease (AD) process. However, the mechanism(s) involved remain(s) unclear. We have followed this research area and published a related cellular study years ago. The literature was searched for what Aβ antibodies target in the brain, which showed that they bind plaques.

2. Interpretation: Our findings that Aβ antibodies also localize to diverse brain cells and blood vessels should be of significant interest to the field. To maximize benefits of this line of therapy a better understanding of what Aβ antibodies target in the brain is important.

3. Future directions: We are collaborating with BioArctic AB to work with their rodent version of lecanemab. We are considering also testing a pyroGlu3Aβ antibody to compare both currently clinically approved antibodies to Aβ antibody 6E10 used in the present study. Ultimately, we would like to help develop better Aβ immunotherapy for AD.

2. METHODS

2.1. Animals

Male mice were allowed a minimum acclimation period of 5 days before any handling or experimental procedures. They were housed in a temperature‐ and humidity‐controlled environment with a 12 hour light/dark cycle, in groups of three to five per individually ventilated cage. All cages were enriched with toys to promote animal welfare, and animals had ad libitum access to standard chow and water. All mouse experiments were ethically approved by the Malmö/Lund Ethics Committee on Animal Testing (dnr 5.8.18‐13038/2024). For antibody injections, we used several different AD mouse models. (1) 5xFAD mice (Jackson Laboratory, B6SJL‐Tg [APPSwFILon, PSEN1*M146L*L286V]6799Vas/Mmjax), the familial AD (FAD) mutations APP K670N/M671L (based on the 770‐residue isoform), along with I716V and V717I, as well as presenilin 1 (PS1) mutations M146L and L286V, were introduced into the cDNAs of amyloid precursor protein (APP; 695) and PS1 through site‐directed mutagenesis; these modified sequences were then subcloned into exon 2 of the mouse Thy1 transgene cassette, as described previously, ¹⁴ and subsequently verified by sequencing using standard protocols. (2) APP ^NL‐G‐F mice (kindly provided by T. Saido, Riken Institute, but also available at Jackson Laboratory, APPtm3.1Tcs) harbor mutations: APP K670_M671delinsNL (Swedish), APP I716F (Iberian), and APP E693G (Arctic); these mutations promote Aβ pathology by increasing total Aβ production (Swedish mutation), increasing the Aβ₄₂/Aβ₄₀ ratio (Iberian mutation), and promoting Aβ aggregation through facilitating oligomerization and reducing proteolytic degradation (Arctic mutation). ⁴⁴ (3) APP ^NL‐F mice (kindly provided by T. Saido, Riken Institute, but also available at Jackson Laboratory, APPtm2.1Tcs); for the APP ^NL‐F knock‐in mice, the APP construct contains a humanized Aβ region along with two pathogenic mutations, the Swedish “NL” and the Iberian “F.” ⁴⁴ Due to the use of the endogenous mouse APP promoter, the constructs of the APP ^NL‐F and APP ^NL‐G‐F mice are expressed in appropriate cell types and temporal specificity.

2.2. Experimental injections

Antibody 6E10 is a mouse immunoglobulin G (IgG)1 monoclonal antibody widely used in AD research. It targets the N‐terminal domain of Aβ and recognizes Aβ peptides, Aβ‐containing full‐length APP, and β‐APP C‐terminal fragments. Experimental research has shown that Aβ N‐terminal domain antibodies reduce amyloid plaques and improve behavior in AD mice. ⁴⁵ , ⁴⁶ This part outlines the study design of all 6E10 antibody and bovine serum albumin conjugated to Alexa Fluor 647 (BSA‐647) administrations across the different experiments. For all the experiments, brains were carefully excised from the skull and placed in 4% paraformaldehyde (PFA) at 4°C overnight. After fixation in 4% PFA, the mice brains were immersed in 15% sucrose followed by 30% sucrose, each for 1 day. Coronal sections (40 µm) were cut using a microtome and stored in cryoprotectant (30% sucrose and 30% ethylene glycol in 0.1 M phosphate buffer [PB], pH 7.4) until use. PB was prepared by dissolving 14.42 g Na₂HPO₄·2H₂O and 2.62 g NaH₂PO₄·H₂O in 1 L of ddH₂O.

2.2.1. Unilateral intrahippocampal injections

Unilateral intrahippocampal injections were conducted as previously described. ⁴⁷ Mice were anesthetized with isoflurane (4% induction, 2% maintenance), and the head was shaved, sterilized with 70% ethanol, and fixed in the stereotaxic frame. After exposing the skull, a burr hole was drilled at the following coordinates relative to bregma: anterior–posterior (AP) 2.5 mm, medial–lateral (ML) 2.0 mm, dorsal–ventral (DV) 1.8 mm. A total of 5 µL 6E10 antibody (BioLegend #803003, 1 mg/mL) or Alexa Fluor 488–6E10 (BioLegend #803013, 1 mg/mL) was infused at 1 µL/minute using a Hamilton syringe (Hamilton 7634‐01). The needle remained in place for 5 minutes before withdrawal. Incisions were sutured, and animals were monitored until full recovery.

Injection conditions: (1) Alexa Fluor 488–6E10 in 5xFAD and APP ^NL‐F mice, administered 24 hours or 72 hours before sacrifice; (2) control group of secondary antibody (goat anti–guinea pig Alexa Fluor 633, 2 mg/mL, 5 µL) injection of 5xFAD; (3) unlabeled 6E10 (1 mg/mL, 5 µL) in 5xFAD mice, administered 24 hours or 72 hours before sacrifice.

2.2.2. Combined intrahippocampal antibody injection and cisterna magna BSA‐647 injection

To assess glymphatic influx after unilateral intrahippocampal antibody injection, 5xFAD mice received intrahippocampal 6E10 antibody injections (24 hours or 72 hours prior), followed by cisterna magna (CM) injection of BSA‐647 (10 mg/mL, bovine serum albumin, Alexa Fluor 647 conjugate, 66 kDa, Invitrogen, A34785) diluted in artificial CSF (aCSF). BSA‐647 was allowed to circulate for 30 minutes before sacrifice. aCSF was prepared as previously reported. ²⁹ The aCSF consisted of 124 mM NaCl, 3 mM KCl, 26 mM NaHCO₃, 1.24 mM NaH₂PO₄, 2 mM MgSO₄, 2 mM CaCl₂, and 10 mM glucose.

2.2.3. Intraperitoneal hypertonic and normal saline injections

To evaluate the effect of osmotic modulation on brain entry of BSA‐647 or 6E10 antibody, wild‐type (WT) and APP ^NL‐G‐F mice received intraperitoneal (IP) injections of either normal saline (0.154 M) or hypertonic saline (1 M), administered concurrently with a CM injection of BSA‐647 (10 mg/mL, 10 µL) or 6E10 antibody (1 mg/mL, 10 µL). BSA‐647 was allowed to circulate for 30 minutes, and the 6E10 antibody for 6 hours before sacrifice. Based on the finding that hypertonic saline enhanced 6E10 antibody accumulation in the brain, all subsequent CM 6E10 injection experiments were performed with concurrent IP hypertonic saline administration and a 6 hour circulation period. Solutions were prepared as previously reported. ⁴⁸ Normal saline consisted of 0.154 M NaCl in ddH₂O and was administered at 20 µL/g. Hypertonic saline (1 M NaCl in ddH₂O) was also administered at 20 µL/g to induce hyperosmolality.

2.2.4. Intracerebroventricular injections

The intracerebroventricular (ICV) injections protocol was as previously described. ⁴⁹ Mice were anesthetized with isoflurane as mentioned above and positioned in a stereotaxic frame. A burr hole was drilled at: AP 0.2 mm, ML 1.0 mm, DV 2.3 mm. A total of 5 µL Alexa Fluor 488–6E10 was infused at 1 µL/minute using a Hamilton syringe. The needle remained in place for 5 minutes to minimize backflow. After suturing, mice were monitored until full recovery.

2.2.5. ICV antibody and CM BSA‐647 injection

To determine whether prolonging brain‐wide exposure to the 6E10 antibody, rather than restricting delivery to the hippocampus, improves the assessment of glymphatic influx, 6E10 (1 mg/mL, 5 µL) was injected into the lateral ventricle of APP ^NL‐G‐F mice. Twenty‐four hours later, CM injection of BSA‐647 (10 mg/mL, 5 µL) was performed, and the BSA‐647 was allowed to circulate for 30 minutes before sacrifice.

2.2.6. CM injections

CM injections were performed as previously reported. ⁴⁸ Mice were anesthetized with an IP ketamine–domitor mixture (100 mg/mL ketamine + 1 mg/mL domitor; 10 µL/g body weight). After securing the head in a stereotaxic frame, the CM was exposed. A 30‐G needle connected to a 25‐µL Hamilton syringe (Hamilton 80223) via polyethylene tubing was inserted into the CM and secured with tissue adhesive. 6E10 antibody or BSA‐647 was infused at 1 µL/minute using a microinjection pump (Science Products model nano jet, serial #47927). To minimize backflow, the needle was left in place after infusion. Mice were allowed to recover in a temperature‐controlled incubator. To avoid potential structural damage associated with intraventricular injections of 6E10, which could influence glymphatic transport measurements, a mixture of 6E10 antibody and BSA‐647 was injected into the CM of APP ^NL‐G‐F mice and allowed to circulate for 6 hours.

2.2.7. IP 6E10 antibody injections

6E10 conjugated to Alexa Fluor 647 (0.5 mg/mL) or phosphate‐buffered saline (PBS; control) was administered IP in APP ^NL‐G‐F mice at a dose of 2 mg/kg body weight. Mice were sacrificed, and brains were collected 3 days post‐injection.

All immunolabeling was performed using a free‐floating protocol. Sections were washed 3 × 5 minutes in 0.1 M PBS. For Aβ immunolabeling, sections were incubated in 88% formic acid for 8 minutes and rinsed 3 × 5 minutes in PBST (0.3% Triton X‐100). Sections were then blocked for 1.5 hours at room temperature (RT) in 3% normal donkey serum (NDS) or 3% normal goat serum (NGS) in PBST, followed by incubation with primary antibodies diluted in the same blocking solution overnight at 4°C on an agitator. After 3 × 10 minute washes in PBS at RT, sections were incubated with appropriate secondary antibodies for 2 hours at RT and washed 3 × 10 minutes. Alexa Fluor 633 Hydrazide (A30634, Thermo Fisher Scientific) was used to label elastin fibers at a dilution of 1:1000. Tissue sections were incubated with the dye for 5 minutes at room temperature, followed by washing in PBS before imaging

Primary antibodies include the following:

Alexa Fluor 488 anti‐β‐amyloid (6E10), BioLegend 803013, RRID: AB_2564765

Alexa Fluor 647 anti‐β‐amyloid (6E10), BioLegend 803021, RRID: AB_2783374

Anti‐β‐amyloid (6E10), BioLegend 803014, RRID: AB_2728527

Mouse anti‐β‐amyloid (MOAB‐2)–Alexa Fluor 647 (1:1000; Novus NBP2‐13075AF647, RRID:AB_3260696)

Rabbit anti‐amyloid fibrils (OC; 1:1000; MilliporeSigma, AB2286)

Rabbit anti‐glial fibrillary acidic protein (GFAP; 1:1000; Dako, Z0334).

Chicken anti‐MAP2 (1:1000; Abcam ab92434)

Rabbit anti‐ionized calcium‐binding adapter molecule 1 (Iba1; 1:1000; Wako 019‐19741)

Rabbit anti‐human Aβ1‐40 (1:1000; Immuno‐Biological Laboratories 18580)

Rabbit anti‐human Aβ1‐42 (1:1000; Immuno‐Biological Laboratories 18582)

Rabbit anti‐aquaporin 4 (1:1000; Merck AB3594)

Rabbit anti‐CD206 (1:1000; Cell Signaling Technology, mAb #24595)

Goat anti‐Olig2 (1:500; Biotechne, AF2418)

Rabbit anti‐α‐smooth muscle actin (1:500; Abcam ab5694)

Goat anti‐CD31 (1:800, Bio‐Techne, AF3628‐SP)

Secondary antibodies

Alexa Fluor 488 goat anti‐mouse (1:500; Invitrogen A‐11001, RRID:AB_2534069)

Alexa Fluor 568 goat anti‐rabbit (1:500; Thermo Fisher A‐11011, RRID:AB_143157)

Alexa Fluor 568 donkey anti‐rabbit (1:500; Thermo Fisher A‐10042, RRID:AB_2534017)

Alexa Fluor 488 goat anti‐rabbit (1:500; Thermo Fisher A‐11008, RRID:AB_143165)

Alexa Fluor 568 goat anti‐chicken (1:500; Thermo Fisher A‐11041, RRID:AB_2534098)

Alexa Fluor 647 donkey anti‐goat (1:500; Thermo Fisher A‐21447, RRID:AB_2535864)

Secondary antibody Alexa Fluor™ 633‐goat anti‐guinea pig IgG (Thermo Fisher, A‐21105, RRID: AB_2535757) was used for 5xFAD mouse intrahippocampal injections.

2.4. Immunohistochemistry, imaging, and analysis

Images were acquired using a Zeiss AXIO imager M2 microscope (Carl Zeiss) and processed with ZEN software (ZEN 2.6, Carl Zeiss). A 10× objective was used for overview images. Images of the hippocampus, penetrating vessels, astrocytes, and microglia were obtained using 20×, 40×, and 63× objectives. Confocal microscopy was performed using a Leica TCS SP8 laser scanning confocal microscope with LAS X software (Leica Microsystems). A 63× oil‐immersion objective (HC PL CS2, NA 1.40) was used for high‐resolution imaging. Imaris software (Imaris SingleFull with Clearview, version 10.2, Oxford Instruments) was used for 3D reconstruction.

2.5. Image analysis

The area (%) covered by MOAB‐2, IBA‐1, GFAP, and BSA‐647 was quantified within a specified region of interest (ROI: hippocampus, cortex, or whole brain) using an automated local threshold in the FIJI software that was maintained for all images. The total area covered was determined by taking the average of at least three sections per mouse. For the intrahippocampal injections, the injected hippocampus was compared to the uninjected, contralateral side. When analyzing CM, intraventricular, and IP injections, the control was a PBS‐injected mouse. To quantify intraneuronal Aβ, the neuronal MAP2 marker was used as a mask, and combined with a manually drawn ROI for the CA1 part of the hippocampus. The selected area was then thresholded, and MOAB‐2⁺ area inside of this ROI was determined.

2.6. Statistics

All data analyses were performed using GraphPad Prism 10.1.2. Normality of all datasets was assessed using the Shapiro–Wilk test, and all data were confirmed to follow a normal distribution. When these assumptions were met, group comparisons were conducted using one‐way analysis of variance (ANOVA). In cases in which the ANOVA showed significant overall group differences, post hoc multiple‐comparison testing was performed with appropriate correction for multiple testing (Tukey method). Adjusted P values from these post hoc analyses were used to determine pairwise significance. All statistical tests applied, and their P values, are reported in the figure legends. All values are displayed as mean ± standard deviation (SD), and N represents the number of mice. We set the statistical significance level at P < 0.05.

3. RESULTS

3.1. Unilateral intrahippocampal injection of Aβ antibody reduces plaques and reveals antibodies within neurons, astrocytes, and microglia in 5xFAD mice

Prior studies had shown that N‐terminal Aβ antibodies (e.g., antibody 3D6) were particularly effective in reducing amyloid plaques in AD mouse brains. ³⁷ , ⁵⁰ We first unilaterally injected either unconjugated (Figure S1A,B in supporting information) or Alexa Fluor 488‐conjugated N‐terminal Aβ antibody 6E10 into the hippocampus of 4‐month‐old 5xFAD mice. Twenty‐four or 72 hours after injection (see experimental schema, Figure 1A), we analyzed the brains for antibody distribution. The unconjugated and fluorescently conjugated antibodies showed a similar pattern of labeling. However, because the signal of the conjugated antibody was considerably stronger, we proceeded with using this antibody for further experiments. Conjugated Aβ antibody distribution after intrahippocampal injection showed prominent labeling in the injected hippocampus after 24 hours, while after 72 hours, the labeling was comparatively less pronounced (Figure 1B–D). Remarkably, after 24 hours, we detected a clear signal of antibody 6E10 inside selective neurons, particularly in hippocampal field CA1 and the subiculum (Figure 1E–G). MOAB‐2 antibody was used to detect endogenous intracellular Aβ as well as plaques. ⁵¹ Labeling the same sections with anti‐Aβ antibody MOAB‐2 revealed a substantial degree of overlap with the intraneuronal labeling of the 6E10 antibody, providing evidence that the intracellular target of the antibody is indeed Aβ (Figure 1H–J). We also detected some antibodies in astrocytes and more markedly in microglia (Figure S1C,D).

FIGURE 1.

Unilateral intrahippocampal injection of Aβ antibody 6E10 is detected in neurons, reduces hippocampal plaques, and activates astrocytes and microglia in 5xFAD mice. A, A total of 5 µL of Alexa Fluor–conjugated 6E10 antibody was injected into one side of the hippocampus of 4‐month‐old 5xFAD mice. B, C, Representative images of Aβ antibody 6E10 distribution at 24 and 72 hours after intrahippocampal injection; scale bar = 1 mm. D, At 24 hours post‐injection, Aβ antibody 6E10 distributions in the injected hippocampus. Three regions in CA1 and subiculum were used for analysis; scale bar = 200 µm. E–G, The three selected areas in (D) showed intracellular distributions of the Aβ antibody in neurons, scale bar = 20 µm. H, Antibody MOAB‐2 labeling in the CA1 for intracellular Aβ indicates that the intracellular Aβ antibody (I) colocalizes with intracellular Aβ (J), scale bar = 10 µm. The MOAB‐2 levels in the hippocampus at 24 hours (K) and 72 hours (L) after unilateral intrahippocampal injection; scale bar = 1 mm. M, Quantification of MOAB‐2 labeling in the hippocampus, one‐way ANOVA: F(3,12) = 8.775, P = 0.0024; Tukey post hoc test: uninjected‐24 hours versus injected‐24 hours, P = 0.9762, uninjected‐72 hours versus injected‐72 hours, P = 0.0356. GFAP levels in the hippocampus at 24 hours (N) and 72 hours (O) after Aβ antibody injection; scale bar = 1 mm. IBA‐1 levels in the hippocampus at 24 (P) and 72 hours (Q) after Aβ antibody injection, scale bar = 1 mm. R, Quantification of GFAP labeling in the hippocampus, one‐way ANOVA: F(3,12) = 35.43, P < 0.0001; Tukey post hoc test: uninjected‐24 hours versus injected‐24 hours, P = 0.0014, uninjected‐72 hours versus injected‐72 hours, P < 0.0001. S, IBA‐1 labeling in the hippocampus, one‐way ANOVA: F(3,12) = 129.5, P < 0.0001, Tukey post hoc test: uninjected‐24 hours versus injected‐24 hours, P < 0.0001, uninjected‐72 hours versus injected‐72 hours, P = 0.0002. N = 4 per group. Data are represented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Aβ, amyloid beta; ANOVA, analysis of variance; GFAP, glial fibrillary acidic protein; IBA‐1, ionized calcium‐binding adapter molecule 1; MOAB‐2, mouse anti‐β‐amyloid

In line with previous studies using Aβ antibodies, ⁴² the 6E10 antibody was also detected in amyloid plaques. Although no reduction (Tukey post hoc test: P = 0.9762) in Aβ plaques from the Aβ antibody was observed in the hippocampus at 24 hours post‐injection, a significant reduction (Tukey post hoc test: P = 0.0356) was detected at 72 hours on the injected compared to the uninjected side (Figure 1K–M). Similarly, at the 72 hour but not 24 hour time point, intracellular Aβ in CA1 neurons on the injected side was significantly reduced (Tukey post hoc test: uninjected‐72 hours vs. injected‐72 hours, P = 0.0012) compared to the uninjected side (Figure S1E–K). Next to the characteristic anatomy of the CA1 pyramidal cell layer, known to be mainly populated by neurons, we verified that the 6E10 antibody accumulates within neurons rather than other cells, such as microglia, by MAP2 and IBA‐1 labeling. We observed substantial intracellular antibody accumulation in MAP2⁺ CA1 neurons on the injected side of the hippocampus (Figure S2A in supporting information), but not in microglia (Figure S2B). To more clearly visualize intraneuronal 6E10 antibody accumulation, we show the MAP2 channel in Figure S2. Neurons in CA1 showed markedly higher levels of antibody 6E10 compared to almost no detectable labeling of neurons in the dentate gyrus (DG; Figure S2C,D).

To determine whether Aβ antibodies led to an activation of glial cells in the injected side, glial cell activation was assessed by labeling for GFAP, a marker of astrocytes, and microglia by labeling for Iba‐1, a microglia‐specific marker. We observed increased activation of astrocytes (Tukey post hoc test: uninjected‐24 hours vs. injected‐24 hours, P = 0.0014; uninjected‐72 hours vs. injected‐72 hours, P < 0.0001) and microglia (Tukey post hoc test: uninjected‐24 hours vs. injected‐24 hours, P < 0.0001; uninjected‐72 hours vs. injected‐72 hours, P = 0.002) on the injected side (Figure 1N–S). However, the glial activation may be unrelated to the Aβ epitope of the 6E10 antibody but just a result of injecting antibodies and/or a consequence of the tissue damage from the injection. To better ascertain this, we performed unilateral intrahippocampal injections of a fluorescent secondary antibody (Alexa Fluor 633‐goat anti‐guinea pig IgG secondary antibody), which showed a significantly different pattern compared to that of the injected Aβ antibody 6E10 (Figure S3A in supporting information). The secondary antibody was found predominantly at the surface of small blood vessels and did not visibly bind to plaques or label hippocampal neurons. Still, unilateral intrahippocampal injection of this secondary antibody induced widespread activation of astrocytes and microglia in the hippocampus to a similar extent as that observed after the Aβ antibody injection (data not shown).

3.2. Unilateral intrahippocampal injection reveals Aβ antibodies within neurons and microglia in APP ^NL‐F mice

We next injected Aβ antibody 6E10 into the hippocampus of pre‐plaque, 5‐month‐old APP ^NL‐F mice and examined the anatomical and cellular localization of Aβ antibodies before plaque formation (see experimental schema, Figure 2A). After unilateral intrahippocampal injection, antibody 6E10 was detected in the injected side at both 24 hours and 72 hours post‐injection (Figure 2B,C). Compared to our results with the 5xFAD mice, above, which had abundant plaques, we found less antibody 6E10 in the injected hippocampus of pre‐plaque APP ^NL‐F mice (Figure 2D). Remarkably, at 24 hours after Aβ antibody injection, 6E10 labeling revealed intracellular signals in CA1 neurons (Figure 2E), which corresponded to intracellular Aβ detected by MOAB‐2 (Figure 2F) and showed colocalization (Figure 2G). Similar colocalizations were observed in other CA1 neurons (Figure 2H–J) and in neurons of the deep CA1/subiculum transitional region (Figure 2K–M). We detected activated astrocytes (Tukey post hoc test: uninjected‐24 hours vs. injected‐24 hours, P = 0.0001, uninjected‐72 hours vs. injected‐72 hours, P < 0.0001) and microglia (Tukey post hoc test: uninjected‐24 hours vs. injected‐24 hours, P < 0.0001, uninjected‐72 hours vs. injected‐72 hours, P = 0.0472; Figure 2N–S) in these injected pre‐plaque mice. The marked effect on astrocytosis with the very low P values likely, as noted for the 5xFAD mouse above, is the fact that the needle injection itself induces the astrocytosis. Of note, we observed that microglia in the hippocampus contained 6E10 antibodies (Figure S3B,C), whereas astrocytes lacked intracellular antibody signals (Figure S3D–F). However, due to the needle track–induced brain damage, we cannot conclude that the hippocampal glial activation is due to the antibody. To avoid damage to brain parenchyma, we next directly delivered Aβ antibody into the CSF via the CM. In contrast to CM injection, the brain delivery efficiency of Aβ antibodies administered by IP injection is relatively low. Previous studies have reported that only ≈ 0.1% to 0.2% of circulating antibodies can cross the blood–brain barrier (BBB) and enter the brain. ⁵² Thus, to further investigate the effects of Aβ antibodies and to mitigate the potential effects of mechanical damage from the injection itself, we turned to antibody injection into the CM. Although CM injection of Aβ antibodies is a relatively invasive method, it allows for high concentrations of Aβ antibodies in the CSF, thereby increasing brain parenchymal delivery and overall bioavailability.

FIGURE 2.

Unilateral intrahippocampal injection of Aβ antibody 6E10 in APP ^NL‐F mice results in intraneuronal antibody in neurons and activation of astrocytes and microglia before the appearance of plaques. A, A total of 5 µL of Aβ antibody 6E10 was injected into the hippocampus of 5‐month‐old APP ^NL‐F mice. B, C, Representative images of Aβ antibody 6E10 distribution at 24 and 72 hours; scale bar = 1 mm. D, Three areas were selected in CA1 and in a subset of neurons located in the deep layer of the CA1/subiculum transitional region for analysis; scale bar = 200 µm. E, Intracellular localization of Aβ antibody 6E10 in neurons, MOAB‐2 labeling for intracellular Aβ of the same subregion (F), which colocalized with intracellular 6E10 antibody (G). Similar colocalizations were observed in other CA1 neurons (H, I, J) and in neurons of the deep CA1/subiculum transitional region (K, L, M); scale bar = 10 µm. GFAP levels in the hippocampus at 24 (N) and 72 hours (O) after antibody injection; scale bar = 1 mm. IBA‐1 levels in the hippocampus at 24 (P) and 72 hours (Q) after antibody injection; scale bar = 1 mm. Quantification of the hippocampus covered by GFAP (R), one‐way ANOVA: F(3,8) = 119.3, P < 0.0001, post hoc Tukey test, uninjected‐24 hours versus injected‐24 hours, P = 0.0001, uninjected‐72 hours versus injected‐72 hours, P < 0.0001, and IBA‐1(S) labeling, one‐way ANOVA: F(3,8) = 74.16, P < 0.0001, post hoc Tukey test, uninjected‐24 hours versus injected‐24 hours, P < 0.0001, uninjected‐72 hours versus injected‐72 hours, P = 0.0472; N = 3 per group. Data are represented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Aβ, amyloid beta; ANOVA, analysis of variance; GFAP, glial fibrillary acidic protein; IBA‐1, ionized calcium‐binding adapter molecule 1; MOAB‐2, mouse anti‐β‐amyloid

3.3. CM injection reveals Aβ antibodies in different cell types of APP ^NL‐G‐F and APP ^NL‐F mice

Unlike intrahippocampal injection, CM injection should avoid the potential for mechanical damage to the brain parenchyma, and Aβ antibodies injected in CM will be able to distribute more broadly across the brain, though at lower local concentrations compared to direct intrahippocampal injection. Compounding this, a substantial portion of the CM‐injected antibody will flow out of the brain as it travels with the circulating CSF. To enhance the translocation of Aβ antibody into the parenchyma, we tested an approach used in a previous study, wherein IP injection of hypertonic saline (HS) was shown to enhance the entry of CM‐injected antibodies into the brain parenchyma. ⁴⁸ To better evaluate the effect of IP injection of HS on CSF influx, we performed CM injection of BSA‐647 in 4‐month‐old WT mice and of Aβ antibody 6E10 in 4‐month‐old APP ^NL‐G‐F mice. Concurrently, IP injections of either isotonic saline (normal saline [NS], 0.154 M, [IP‐NS]) or HS (1 M, IP‐HS) were performed at the same time as the CM injection of the BSA‐647 or Aβ antibody (Figure S4A,B in supporting information). We found that in WT mice, the distribution in the brain parenchyma of BSA‐647 in the IP‐HS group was ≈ 2.323 times greater than in the IP‐NS group (P < 0.0001). In APP ^NL‐G‐F mice, the distribution in the brain parenchyma of the 6E10 antibody in the IP‐HS group was 2.485 times that of the IP‐NS group (P = 0.0003; Figure S4C–F). 3D reconstruction revealed that IP‐HS significantly enhanced antibody 6E10 localization around blood vessels and cortical plaques (Figure S4G,H), suggesting improved antibody translocation with IP‐HS. Henceforth, we adopted this combined injection approach (IP‐HS together with CM antibody injection) in all subsequent CM antibody experiments.

We then tested antibody 6E10 in APP ^NL‐G‐F mice 6 hours after injection into CM. Mice were injected in CM with either 6E10 or PBS in addition to IP‐HS injection (see experimental schema, Figure 3A). We found that 6E10 distributed widely throughout the brain (Figure 3B), including in the hippocampus (Figure 3C). Notably, as with intrahippocampal antibody injection, CM injection led to 6E10 antibody accumulation within neurons of CA1 (Figure 3D), where MOAB‐2 labeling revealed intracellular Aβ (Figure 3E) that localized with the injected 6E10 antibody (Figure 3F). A more medial aspect of CA1 also showed neuronal antibody 6E10 (Figure 3G) colocalizing with intracellular Aβ detected by MOAB‐2 (Figure 3H). CM injection of 6E10 also significantly reduced cortical plaques (Tukey post hoc test: cortex: P = 0.0418, hippocampus: P = 0.7799; Figure S5A–C in supporting information), and increased activated microglia (Tukey post hoc test: cortex: P = 0.0168, hippocampus: P = 0.0017; Figure S5D,E) and astrocytes (Tukey post hoc test: cortex: P = 0.0005, hippocampus: P = 0.9808; Figure S5F,G) when combined with IP‐HS. However, unlike at 72 hours post‐intrahippocampal injection in 5xFAD mice, in APP ^NL‐G‐F mice, there was no significant reduction in intracellular Aβ in CA1 neurons of the hippocampus at this shorter 6 hours post‐injection time with 6E10 compared to PBS (P = 0.856; Figure S6A–E in supporting information).

FIGURE 3.

Aβ antibody 6E10 is detected in different brain cells after CM injection in 4‐month‐old APP ^NL‐G‐F mice. A, A total of 10 µL of Aβ antibody 6E10 or PBS was injected into the CM combined with an IP injection of hypertonic saline. B, The distributions of Aβ antibody 6E10 in the brain; scale bar = 1 mm. C, The distributions of Aβ antibody 6E10 in the hippocampus; scale bar = 200 µm. D, Aβ antibody 6E10 is detected in CA1 neurons; scale bar = 10 µm. E, F, Antibody MOAB‐2 labeling in CA1 shows that intraneuronal Aβ colocalizes with the injected Aβ antibody 6E10; scale bar = 10 µm. G, The Aβ antibody 6E10 in other CA1 neurons; scale bar = 10 µm. H, Antibody MOAB‐2 labeling shows that intraneuronal Aβ colocalizes with Aβ antibody 6E10. I, Aβ antibody 6E10 in penetrating vessels and plaques. Aβ antibody 6E10 in astrocytes around plaques (J), and in penetrating vessels (K) in the cortex; scale bar = 20 µm. L, AQP4 labeling reveals Aβ antibody 6E10 in astrocytic endfeet around blood vessels in the cortex; scale bar = 20 µm. Aβ antibody 6E10 in microglia around plaques (M) and in areas without plaques (N) in the cortex; scale bar = 20 µm. Aβ antibody 6E10 accumulates in the microglia around blood vessels (O), PVMs (P) of penetrating vessels, and MM (Q), and around the nuclei of oligodendrocyte‐specific antibody Olig2 (R); scale bar = 20 µm. N = 4 per group. Aβ, amyloid beta; AQP4, aquaporin‐4; CM, cisterna magna; GFAP, glial fibrillary acidic protein; IBA‐1, ionized calcium‐binding adapter molecule 1; IP, intraperitoneal; MM, meningeal macrophage; MOAB‐2, mouse anti‐β‐amyloid; PBS, phosphate‐buffered saline; PVM, perivascular macrophage

To further investigate where else antibody CM‐injected 6E10 localized, we examined glial cells near penetrating cortical vessels and large arteries. We detected a substantial amount of antibodies surrounding blood vessels and associated with plaques (Figure 3I). Astrocytes adjacent to plaques exhibited an activated morphology (astrocytes undergo morphological hypertrophy, characterized by enlarged cell bodies and thickened processes compared to the control group) and showed colocalization with the antibody (Figure 3J). We show the independent channels of 6E10, GFAP, and MOAB‐2 in Figure S7A–C in supporting information; 3D reconstruction revealed that a subset of astrocytes associated with plaques exhibited colocalization with antibody 6E10 (Figure S7D). Orthogonal XZ and YZ views reveal that a subset of GFAP+ astrocytes show colocalization with antibody 6E10 (Figure S7E). Furthermore, we noticed that in cortical penetrating vessels, astrocytes closely associated with the vasculature also contained antibody 6E10 (Figure 3K). To further examine this, we used an antibody against AQP4, a marker of astrocytic endfeet, and observed colocalization between AQP4 and the injected Aβ antibodies around blood vessels (Figure 3L). Next, we analyzed the colocalization of antibody 6E10 with microglia in different cortical regions, including plaque‐burdened (Figure 3M) and plaque‐free cortical areas (Figure 3N). We observed a substantial amount of 6E10 inside microglia. Surprisingly, this appeared to be independent of the presence of plaques in these AD mice.

Aβ is known to accumulate in blood vessels and cause cerebral amyloid angiopathy (CAA) in AD ⁵³ and a main side effect of Aβ immunotherapies for AD is microhemorrhage. ⁵⁴ Because of this, we further investigated Aβ antibody localization around blood vessels. Interestingly, perivascular microglia (Figure 3O) and PVMs showed accumulation of Aβ antibody 6E10 (Figure 3P, S8A–C in supporting information). Additionally, we detected 6E10 signal in meningeal macrophages (Figure 3Q) and around the nuclei of some oligodendrocytes/oligodendrocyte precursor cells (OPCs) (Figure 3R); in the latter we detected the injected Aβ antibody within the cytoplasm surrounding nuclei labeled with oligodendrocyte‐specific antibody Olig2. Olig2+ oligodendrocytes and/or OPCs were also found to contain Aβ42, which localized in the cells with the intracellular Aβ antibody 6E10 (Figure S8D–G). Together, our results show a remarkably widespread distribution of injected antibody 6E10 across various cell types within the brain.

We next examined the distribution of CM‐injected Aβ antibody 6E10 in the mouse brain before the appearance of plaques (Figure S9A in supporting information). Pre‐plaque 5‐month‐old APP ^NL‐F mice showed that 6E10 antibody was widely distributed throughout the brain (Figure S9B), including the hippocampus (Figure S9C). Some 6E10 antibody was detected in neurons of CA1 (Figure S9D), where MOAB‐2 labeling revealed intraneuronal Aβ (Figure S9E) that localized with 6E10 antibody (Figure S9F). A similar pattern was observed in other CA1 neurons, with 6E10 antibody accumulation (Figure S9G) localizing with MOAB‐2 labeled intraneuronal Aβ (Figure S9H,I). Furthermore, we observed injected 6E10 in PVMs (Figure S9J), around the oligodendrocyte/OPC‐specific antibody Olig2 (Figure S9K), and in microglia (Figure S9L).

To further determine how important the actual epitope of the Aβ antibody is for our findings compared to a more general effect of injecting any antibody, we took advantage of the human specificity of antibody 6E10, which does not bind to the mouse Aβ sequence in WT mice. The distribution of CM‐injected 6E10 in 5‐month‐old WT mice differed from that in AD mouse models (Figure S9M,N). We noted only a small fraction of 6E10 was associated with blood vessels, and there was no detectable injected Aβ antibody in neurons of the hippocampus of WT mice. However, we did observe antibody 6E10 in PVMs (Figure S9O) and microglia (Figure S9P) of WT mice, which we speculate may be related to the phagocytic function of these two cell types.

3.4. Aβ antibody 6E10 distribution in blood vessels after CM injection

As indicated above, Aβ antibody 6E10 injected in CM accumulated heavily around blood vessels in the APP ^NL‐G‐F mouse brain. To determine more precisely where in the vessels the injected antibody 6E10 was localized, we labeled with markers against different parts of the vessel structure, namely CD31, elastin, and α‐smooth muscle actin (α‐SMA). Firstly, labeling with CD31, which is primarily located on the surface of vascular endothelial cells, showed no colocalization with Aβ antibody 6E10 (Figure 4A). To further investigate antibody 6E10 localization within blood vessels, we used a stain for elastin, a marker for the vascular basement membrane. The endothelial cell CD31 antibody localized inside of elastin and 6E10 antibody outside of elastin (Figure 4B). Antibody 6E10 is also outside of the vascular smooth muscle layer labeled using an antibody to α‐SMA (Figure 4C). With the α‐SMA labeling, we were able to detect the presence of 6E10 also in large vessels of the circle of Willis (Figure 4D). There, α‐SMA–positive smooth muscle cells were observed surrounding the CD31‐positive endothelial cells at a distance, with 6E10 antibody located outside the SMA layer. Last, we labeled with a C terminus–specific antibody to Aβ40, which is associated with blood vessels in CAA in AD. Aβ40 mainly localized outside the vascular endothelium of vessels of the circle of Willis, and, compared to Figure 4D, Aβ40 appeared to be more enriched in the SMA‐positive layer, although most of the injected 6E10 signal did not colocalize with the Aβ40 labeling and appears external to the SMA layer (Figure 4E). However, aggregated forms of Aβ would not be effectively detected by such a C terminus–specific Aβ40 antibody. ⁵⁵

FIGURE 4.

The distributions of Aβ antibody 6E10 in penetrating vessels and large arteries of 4‐month‐old APP ^NL‐G‐F mice after CM antibody injection. A, Labeling with the endothelial cell marker CD31 shows that the Aβ antibody 6E10 is primarily located in the outer layer of the vascular endothelium; scale bar = 10 µm. B, Aβ antibody 6E10 is distributed outside of the elastin layer; scale bar = 50 µm. C, Aβ antibody 6E10 is distributed on the outer side of the vascular smooth muscle layer; scale bar = 50 µm. D, In large arteries of the circle of Willis, vascular smooth muscle cells are located on the outer layer of the endothelium but at a greater distance, while Aβ antibody 6E10 is distributed in the outermost layer of the blood vessels; scale bar = 50 µm. E, In large vessels of the circle of Willis, the localization of Aβ40 is different from that of injected Aβ antibody 6E10 within blood vessels; scale bar = 50 µm. N = 4 per group. α‐SMA, α‐smooth muscle actin; Aβ, amyloid beta; CM, cisterna magna

3.5. IP injection reveals Aβ antibodies in different brain cells and in blood vessels of APP ^NL‐G‐F mice

In clinical settings, Aβ antibodies such as lecanemab or donanemab are administered intravenously to AD patients, whereas in preclinical AD mouse models, they are typically delivered via IP injection. ¹ We used IP injection of 6E10 (2 mg/kg) to investigate its distribution in different brain cell populations and blood vessels. A single IP injection of 647–6E10 was administered, and tissue samples were harvested 3 days post‐injection (see experimental schema, Figure 5A). We found that antibodies administered via IP injection were widely distributed in the brain (Figure 5B). Remarkably and unlike CM injection, after IP injection, antibodies were not only detected in penetrating vessels but also in capillaries of the brain parenchyma. High‐magnification views of the hippocampus revealed clear Aβ antibody binding to amyloid plaques as well as distribution within capillaries in the hippocampus (Figure 5C). Moreover, we observed intracellular accumulation of Aβ antibodies particularly in neurons located in the CA1 region. IP injection led to 6E10 antibody accumulation within neurons (Figure 5D), where MOAB‐2 labeling revealed intracellular Aβ (Figure 5E) that colocalized with the injected antibody (Figure 5F). Similar colocalization was observed in other CA1 neurons (Figure 5G–I). Some IP‐injected Aβ antibody was also detected in astrocytes surrounding plaques (Figure 5J). However, unlike with CM injection, in perivascular astrocytes, the IP‐injected Aβ antibody was predominantly confined within the blood vessels and did not co‐localize with astrocyte endfeet (Figure 5K). Aβ antibodies were localized in microglia both near and distant from plaques (Figure 5L) and surrounding the nuclei (olig2) of oligodendrocytes (Figure 5M). We next examined cortical penetrating vessels (Figure 5N) and small hippocampal vessels (Figure 5O) to investigate Aβ antibody distribution in PVMs, and observed a prominent accumulation of the Aβ antibody within these cells. To more precisely determine the vascular localization of the IP‐injected 6E10 antibody, AQP4 and CD31 labeling was performed to examine its distribution in cortical penetrating vessels (Figure 5P) and small hippocampal vessels (Figure 5Q), while α‐SMA labeling was used to assess its presence in cortical penetrating vessels (Figure 5R) and arteries of the circle of Willis (Figure 5S). The Aβ antibody localized to the abluminal perivascular compartment, lying outside the endothelial (CD31) and vascular smooth muscle (α‐SMA) layers but inside the boundary formed by astrocytic endfeet (AQP4). Although Aβ antibodies administered via IP injection cross the BBB to reach the brain parenchyma, Aβ antibodies were not detected in the vascular endothelium or vascular smooth muscle layer.

FIGURE 5.

Aβ antibody 6E10 is detected in different brain cells after IP injection in 4‐month‐old APP ^NL‐G‐F mice. A, Mice received a single IP injection of Aβ antibody 6E10 at a dose of 2 mg/kg body weight, and tissues were collected 3 days post‐injection. B, The distributions of Aβ antibody 6E10 in the brain; scale bar = 1 mm. C, The distributions of Aβ antibody 6E10 in the hippocampus, two areas were selected in CA1 for analysis; scale bar = 200 µm. D, Aβ antibody 6E10 is detected in CA1 neurons. E, Aβ42 labeling for intracellular Aβ. F, Colocalization of antibody 6E10 and intracellular Aβ; scale bar = 20 µm. G–I, A similar pattern is seen in other regions of CA1; scale bar = 20 µm. J, Aβ antibody 6E10 in astrocytes around plaques, but not localized to the perivascular astrocytes (K). L, Aβ antibody 6E10 in microglia near and distant from plaques in the cortex. M, Aβ antibody 6E10 surrounds the nuclei of oligodendrocytes. Aβ antibody 6E10 in PVMs in a penetrating vessel (N) in the cortex and small vessel (O) in the hippocampus; scale bar = 10 µm. The Aβ antibody 6E10 is distributed outside of the endothelial cell layer (CD31) but not colocalized with CD31, inside of astrocytic endfeet (AQP4) in the penetrating vessel (P) and small vessel in the hippocampus (Q); scale bar = 20 µm. The Aβ antibody 6E10 is distributed outside CD31 and smooth muscle cell layer (α‐SMA) in the penetrating vessel (R) and vessels originating from the circle of Willis (S); scale bar = 20 µm. N = 2 per group. α‐SMA, α‐smooth muscle actin; Aβ, amyloid beta; AQP4, aquaporin‐4; GFAP, glial fibrillary acidic protein; IBA‐1, ionized calcium‐binding adapter molecule 1; IP, intraperitoneal; PVM, perivascular macrophage

3.6. Effect of Aβ antibody 6E10 on the glymphatic system

Finally, we examined whether injection of 6E10 would affect glymphatic circulation. To do this, we performed two subsequent injections in 5xFAD mouse brain: a unilateral intrahippocampal injection of Aβ antibody 6E10 and a CM injection with BSA‐647, but without IP‐HS, either 24 hours or 72 hours post‐antibody injection (see experimental schema, Figure 6A). We did not use IP‐HS for these experiments, because we wanted to evaluate potential effects of Aβ antibody on BSA influx, which would be strongly affected by IP‐HS. No significant changes in BSA‐647 influx were seen within the antibody‐injected hippocampus compared to the uninjected side at both the 24 hour and 72 hour timepoints (Tukey post hoc test: uninjected‐24 hours vs. injected‐24 hours, P = 0.2135, uninjected‐72 hours vs. injected‐72 hours, P = 0.8832; Figure 6B,C; S10A for analysis). We show the overlap images of 6E10 antibody and BSA‐647 (Figure S10B in supporting information). Seventy‐two hours after unilateral intrahippocampal injection of 6E10 there was somewhat more BSA‐647 adjacent but not directly within the injected hippocampus. Rather than inside the hippocampus, the slight elevation of BSA‐647 fluorescence appears toward the choroid plexus, which is in closer contact with the CSF. To extend the circulation time of the Aβ antibody in the brain, we next performed unilateral ICV injection of 6E10 in APP ^NL‐G‐F mice and allowed it to circulate for 24 hours before CM injection of BSA‐647 without IP‐HS (see experimental schema, Figure 6D). ICV Aβ antibody injection in APP ^NL‐G‐F mouse brains also did not significantly alter glymphatic system circulation compared to the ICV PBS group (P = 0.2; Figure 6E,F; S10C for analysis).

FIGURE 6.

Effects of Aβ antibody 6E10 on glymphatic function. A, Unilateral intrahippocampal injection of 5 µL Aβ antibody 6E10 in 4‐month‐old 5xFAD mice, followed by CM injection of 5 µL of BSA‐647. B, C, Representative images of BSA‐647 distribution after unilateral intrahippocampal injection of Aβ antibody 6E10 at 24 and 72 hours; scale bar = 1 mm. D, A total of 5 µL of Aβ antibody 6E10 or PBS was injected into the lateral ventricle of 4‐month‐old APP ^NL‐G‐F mice, followed by injection of 5 µL of BSA‐647 in CM. E, F, Representative images of BSA‐647 distributions after lateral ventricle injection of Aβ antibody 6E10 or PBS at 24 hours; scale bar = 1 mm. G, Co‐injection of 5 µL BSA‐647 and 5 µL Aβ antibody 6E10 or PBS in the CM, and the BSA‐647 distribution in the brain was examined after 6 hours. H, I, Representative images of BSA‐647 distributions after co‐injection of Aβ antibody 6E10 or PBS with BSA‐647; scale bar = 1 mm. N = 3 per group. Aβ, amyloid beta; BSA‐647, bovine serum albumin conjugated to Alexa Fluor 647; CM, cisterna magna; PBS, phosphate‐buffered saline.

To control for potential mechanical damage caused by intrahippocampal or ICV injections, we next injected both 6E10 and BSA‐647 exclusively into CM in another group of APP ^NL‐G‐F mice. Typically, glymphatic system function is assessed by allowing BSA‐647 to circulate for 30 minutes, ⁴⁸ but to maintain consistency with our previous Aβ antibody circulation experiments, we allowed the BSA‐647 and 6E10 mixture to circulate in the brain for 6 hours before examining BSA‐647 distribution (see experimental schema, Figure 6G). We found that with co‐injection into CM, Aβ antibody 6E10 did not affect glymphatic circulation in 4‐month‐old APP ^NL‐G‐F mice, although there was a trend toward an increase in the accumulation of vessel‐associated BSA‐647 in brain sections (P = 0.1753; Figures 6H,I; S10D for analysis). For these co‐injection experiments of Aβ antibody 6E10 and BSA‐647, we separately show the distribution of BSA‐647 (Figure S10E), Aβ antibody 6E10 (Figure S10F), and PVMs (Figure S10G) in the brain. To analyze where in the brain this BSA‐647 was, we used CD206 antibody for PVMs and detected that the accumulated BSA‐647 colocalized with CD206⁺ PVMs (Figure S10H). CD206 antibody labeling also revealed macrophages harboring BSA‐647 in cortical penetrating vessels and small blood vessels of the hippocampus (Figure S10I,J).

4. DISCUSSION

In this study, we aimed to investigate the more precise distribution of the N‐terminal Aβ domain antibody 6E10 in the brain, after its injection via different routes. We additionally examined potential antibody effects on the glymphatic system. We show that this widely used Aβ antibody in the AD research field localizes not only to plaques and blood vessels, but also to different cell types in the brain, including neurons, astrocytes, microglia, and PVMs of three different AD mice. Particularly in hippocampal field CA1 and the subiculum, including in their border region, Aβ antibody 6E10 localized with intracellular Aβ in these AD‐vulnerable neurons known to be prone to early death in AD and to contain more intraneuronal Aβ. ⁵⁶ Neurons without clear intraneuronal Aβ did not show 6E10 antibody (Figure S2C,D). Future work is needed to determine the selectivity of Aβ antibody uptake in different neuron subtypes and anatomical locations within the brain, and how injected antibody 6E10 is internalized and whether it targets the endogenous intraneuronal Aβ domain in neurons. Antibody 6E10 also localized in a cytoplasmic pattern to select oligodendrocytes and/or OPCs as identified by Olig2 immunolabeling. As in neurons, there was overlap between the injected Aβ antibody and the intracellular Aβ present in these cells. Further work is needed to also delineate the subtypes and anatomy of oligodendrocytes and/or OPCs that internalize Aβ antibodies.

Antibody 6E10 also localized to microglia both near and away from plaques. However, and in contrast to neurons, microglia also took up secondary control antibody in 5xFAD mice and antibody 6E10 in WT mice, indicating a general ability of microglia to take up antibodies independent of epitope. CM‐injected antibody 6E10 also accumulated at astrocytic endfeet at blood vessels in AD mice, although this was also apparent with 6E10 in WT mice. Like microglia, PVMs took up not only antibody 6E10 in AD mice but also secondary antibody and antibody 6E10 in WT mice. In blood vessels, Aβ antibody localized mainly in the external layer rather than in the vascular smooth muscle, elastin, or endothelial layers. However, we noted a difference in the localization of injected antibody 6E10 in vasculature depending on IP versus CM injection. The IP‐injected antibody labeled more of the small capillaries while the CM‐injected antibody was more apparent in larger penetrating vessels in the cortex. It would be interesting to know whether the route of Aβ antibody injection could impact the chances for amyloid‐related imaging abnormalities (ARIAs).

Elevated levels of intracellular Aβ in neurons are increasingly viewed as an important contributor to cognitive impairment and synaptic damage in AD mouse models. ¹¹ , ⁵⁷ , ⁵⁸ Active immunization with the Aβ derivative K6Aβ1–30‐NH₂ in a non‐human primate resulted in a marked reduction of intraneuronal Aβ accumulation in the brains of mouse lemurs. ⁵⁹ Our prior mechanistic cellular work had shown that Aβ antibodies could be internalized into neurons in culture, but not that this can also occur in vivo. ²⁷ In ongoing studies, we are now further examining the mechanism(s) of antibody internalization and cellular Aβ reduction, and how approved Aβ therapeutic antibodies compare to the results obtained with the N‐terminal Aβ antibody used in this study.

A recent study reported that the deletion of beta‐site APP cleaving enzyme 1 (BACE1) in Olig2+ oligodendrocytes reduces Aβ plaque burden in an AD mouse model; ⁶⁰ such studies have suggested that oligodendrocytes might also be active contributors to Aβ pathology in AD. ⁶⁰ , ⁶¹ Our results show that oligodendrocytes and/or OPCs can also contain injected Aβ antibodies that localize with intracellular Aβ.

Previous studies have reported that Aβ antibodies presented to the brain can reduce plaques, activate astrocytes and microglia, and enhance the phagocytic activity of microglia. ⁶² , ⁶³ , ⁶⁴ We noted a rapid micro‐ and astroglial response already 6 hours after administering Aβ antibodies. In view of the importance of neuroinflammation in AD, further studies are warranted to explore the effects of the Aβ antibody on glia and thereby also the potential secondary influence on neurons. Due to their potent phagocytic capacity, PVMs are closely associated with the clearance of systemically administered Aβ antibodies around blood vessels. Further investigation is needed to explore potential effects on PVMs after uptake of Aβ antibody.

We did not find that the Aβ antibody significantly altered glymphatic function in AD mice. A slight increase was noted in fluorescent BSA‐647 near the choroid plexus of the hippocampus after unilateral intrahippocampal antibody injection, although no statistical changes were detected within the hippocampus, and also no changes were seen in BSA‐647 distribution with intraventricular or CM injections of antibody. Our results, however, may be affected by structural disruption of the brain caused by the injection methods used. ARIA is the most important side effect of both currently US Food and Drug Administration–approved amyloid immunotherapies lecanemab and donanemab, but the molecular mechanism(s) and which cell types are most involved remain to be fully understood. We show that Aβ antibodies localize to the outermost layer of blood vessels as well as within PVMs, astrocytic endfeet, and microglia. We also detected that IP injection of antibodies led to more antibody labeling of capillaries compared to CM injection. Further studies are needed to explore the effects of Aβ antibodies on these cell types in relation to vascular amyloid.

Antibodies are increasingly also used to target intracellular proteins, including, for example, the microtubule‐associated protein tau for AD therapy. ⁶⁵ Our study underscores that a widely used Aβ antibody exhibits a broad distribution profile in various cell types within the brains of different AD mouse models. Beyond its localization to amyloid plaques, antibodies were detected in neurons, microglia, astrocytic endfeet, PVMs, and oligodendrocytes/OPCs. These findings indicate that, in the context of immunotherapy, antibody distribution is not limited to amyloid plaques and may influence various cell populations, which could represent novel therapeutic targets in future immunotherapeutic strategies for AD.

CONFLICTS OF INTEREST STATEMENT

The authors declare no competing interests. Author disclosures are available in the supporting information.

CONSENT FOR PUBLICATION

All authors have consented for the publication of this manuscript.

ETHICS APPROVAL

All mouse experiments were ethically approved by the Malmö/Lund Ethics Committee on Animal Testing (dnr 5.8.18‐13038/2024).

CONSENT STATEMENT

No human subjects were included in the study. Consent was not necessary.

Supporting information

Supporting Information

iSupporting Information

Wen G, Lindblom N, Zhan X, et al. Aβ antibodies target not only amyloid plaques but also distinct brain cells and vessels. Alzheimer's Dement. 2026;22:e71121. 10.1002/alz.71121

DATA AVAILABILITY STATEMENT

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