Lecanemab Alters Basement Membrane Collagen IV in Viable Microvessels Isolated from Brains with High Alzheimer’s Disease Neuropathology

Abstract

Amyloid beta (Aβ) can deposit in or near the microvascular basement membrane (BM) in Alzheimer’s disease (AD). We examined the effect of the Aβ binding antibody, lecanemab, on BM collagen IV (Col-IV) using viable brain microvessels (MV) isolated from human postmortem brain tissue with high AD neuropathologic change (ADNC=3, 16F (mean 86yrs), 11M (mean 81yrs)). MVs were exposed to lecanemab or isotype for 4d and examined for Col-IV related outcomes: western blotting, capillary electrophoresis, RT-PCR, and degraded Col-IV. We find a subset of MV from some donors demonstrate Col-IV changes that could cause microvascular injury when exposed to lecanemab.

Introduction:

Brain microvascular changes occur early in Alzheimer’s disease (AD), often predating significant amyloid beta (Aβ) deposition¹. The basement membrane (BM) of the microvasculature is altered in AD, potentially reflecting deposition of Aβ in or near the vessel wall. Although there are at least 28 collagens, only collagen IV (Col-IV) is prevalent and the dominant component of brain microvascular BM^{2, 3}. Col-IV and Aβ are in close spatial proximity in a subset of microvasculature in AD brains with cerebral amyloid angiopathy (CAA)⁴. Aβ binding antibodies predispose the microvasculature to damage^5–8, a process that could reflect changes to Col-IV. We determined the effect of lecanemab (Aβ binding monoclonal antibody) on Col-IV in isolated viable human MVs from brains with high AD neuropathic change (ADNC=3) using western blotting, capillary electrophoresis, RT-PCR for Col-IV transcripts, and immunofluorescent staining for degraded collagen.

Methods:

Immunofluorescence and Immunohistochemistry

Paraffin-embedded sections of cortex from 5 donors with moderate-high ADNC and CAA were examined to evaluate spatial relationships between Col-IV and Aβ. Immunofluorescence (IF) was performed using primary antibodies to affinity purified polyclonal against Col-IV (ab6586-ABCAM) and Aβ (6E10-Biolegend) followed by secondaries (AlexaFluor®-488 Anti-Rabbit:111–545-003 and AlexaFluor®-594 Anti-Mouse: 115–585-003, Jackson, respectively)⁹. Immpress Duet double staining polymer kit (MP-7714-Vector) was used for dual-label immunohistochemistry (IHC) with Aβ and Col-IV.

MV isolation

Viable human MVs were isolated from the superior parietal lobule cortex obtained from rapid autopsies and cryostored as described¹⁰. We utilized cryovials of MV from 27 unique donors (#1–27): 16 females (age 72–99y, mean 86y) and 11 males (age 73–96y, mean 81y) (Supplement Table 1) subsequently signed out as high ADNC in all, and moderate/high CAA in 24/27¹¹.

Exposure to Lecanemab

MV were thawed in human endothelial cell serum free media with B27 (Gibco), treated once with 10μg/ml of lecanemab (Lec) (Proteogenix, MedChemExpress) or equimolar isotype human IgG (Iso) (R&D), and gently rotated in a tissue culture incubator for 4d.

Proteomic Analysis

MV were thawed and tryptic peptides were obtained using the SPEED method¹². Briefly, the MV samples were placed in 15ul of neat trifluoroacetic acid to solubilize the pellet, heated for 3min at 60°C, and then neutralized with 150ul of 2MTris (pH8.5). Disulfide bonds were reduced and the cysteines alkylated prior to in-solution tryptic digestion for 3h at 37°C. Mass spectrometric data acquisition and data analysis were described previously¹³.

Western Blotting

MV were collected, rinsed with PBS, and lysed in M-PER buffer with protease inhibitors. Equal amounts of protein were loaded on 4–20% gradient gels, transferred, and probed with Col-IV antibody (ab6586-ABCAM@1ug/ml).

Jess Capillary western blotting with automated protein separation and immunodetection

MV protein was extracted and prepared for the Jess WB system (ProteinSimple Jess, Bio-Techne) per manufacturer’s protocol with anti-Col-IV@100ug/ml and a NIR Anti-Rabbit Secondary. Peaks (64–300kda) were quantified through Compass SW software, correcting for total protein to calculate percent differences in Lec vs Iso treated samples.

RNA Isolation and RT-PCR

Power SYBR® Green Cells-to-Ct™ Kit (A35379-ThermoFisher) was used to produce DNase1 digested cell lysates from MV and cDNA synthesized per manufacturer’s instructions. RT-PCR was performed using Viia7 Real-Time PCR system with SYBR Green (1725121-Bio-Rad) for transcripts corresponding to human Col-IV alpha 1 and 2 (COLIVA1, COLIVA2), and two housekeeping genes (RPL13 and β-Actin) for 40 cycles. Each sample was run in triplicate, normalized to actin, and relative quantification performed using the comparative threshold cycle method.

Collagen Hybridizing Peptide (CHP) Immunofluorescence

MV were fixed, blocked, and incubated overnight at 4°C with F-CHP (5-FAMConjugate-3DHelix@12.5uM), to detect degraded collagen, and anti-human Col-IV (AF647Conjugate-Invitrogen@10uM) to detect total Col-IV followed by DAPI (1.25uM) for 10min. 5ul of sample were placed onto slides and mounted. Imaging used a MICA Licor Microscope with consistent light intensities, exposures, and thresholds. FIJI (ImageJ) was used to quantify CHP (numerator) with total Col-IV (denominator). Each sample had 4–8 MV images per treatment and Lec/Iso fold change was calculated.

Results:

Histologic sections from n=5 donor brains with documented moderate-high ADNC and CAA showed that all microvasculature had IF staining for Col-IV. A subset of the microvasculature from all donors showed Aβ near (Fig 1A, 1D), in (Fig 1B, 1E), or co-localizing with Col-IV in the microvascular BM (Fig 1C).

Figure 1.

Col-IV and Aβ in Human Brain Microvasculature in ADNC and CAA. Shown are representative images of brain sections from donors with moderate to high ADNC or moderate to high CAA stained with antibodies to Col-IV (green) and Aβ (red) and merge (yellow) (panels A-C) to show Aβ near (A) or in (B) or co-localized (C) with Col-IV in microvascular BM. Panels D-E show ImmPRESS Duet of Aβ (pink) near (D) or in (E) MV BM with Col-IV (brown). Size bar=20μm for all images.

MVs retained their morphology in tissue culture media with no significant morphologic changes over the 4d and were confirmed to be viable by RealTime-Glo™ MT Cell Viability Assay (Promega, data not shown). After 4–5 days in culture, MV with moderate to high ADNC retained their Col-IV content, levels of PDGFRβ (a marker of pericytes), and responsiveness to exposure to PDGF-BB with respect to Col-IV and PDGFRβ levels, relative to MV from brains with no/low ADNC (data not shown).

We tested if Lec would alter basement membrane Col-IV in viable human brain MV. Treatment with Lec correlated with the presence of Aβ as determined by proteomic analysis of human MV (n=12), with tight correlation between Lec and peptides of Aβ in Lec-treated samples, but not Iso-treated samples (Fig 2A, B). We then determined if Lec would degrade MV Col-IV as detected by lower molecular weight (MW) bands of Col-IV by western blotting (WB). We treated MVs from 16 unique donors with Lec or Iso control. Representative WB (Fig 2C) from 3 donors (D-6,D-16,D-18) showed lecanemab treatment (10μg/ml) resulted in additional COLIV lower MW bands at approximately 80 kDa in 2 samples (D-6,D-18) that were not present in carrier control (C) or Iso lanes and 1 sample (D-16) that had lower molecular weight band(s) with Iso and Lec. Of 16 unique samples, 4 showed evidence of Col-IV degradation products that constituted approximately 10–15% of the total Col-IV with Lec treatment only, and 6 showed evidence of Col-IV degradation products with Iso and Lec. Exposure of MV to a broad inhibitor of matrix metalloproteinase (MMP)-2/9 activity with Lec did not alter patterns of Col-IV in 3 samples that demonstrated Col-IV degradation (D-20,D-22,D-6), as shown in a representative western blot of (D-6) (Fig 2D). As expected, MMP9 and total MMP activity were present in MV, but neither MMP9 nor total MMP activity increased with Lec or correlated with the presence of lower MW Col-IV fragments (data not shown). We also analyzed changes in total Col-IV content by automated quantitative capillary electrophoresis (Jess) in 10 individual samples, with 2 that also had traditional western blotting. We detected a 40% or greater change in total Col-IV, a change that has been deemed to be biologically significant in the brain^{14, 15}, in 4 samples with 2 increasing total Col-IV and 2 decreasing in total Col-IV (Fig 2E). Notably, the 2 samples that increased in total Col-IV (D-20,D-22) also showed degradation on traditional western blotting.

Figure 2.

MV Treated with Lec Demonstrate Changes in Col-IV. MVs were treated with either lecanemab (Lec) or an isotype (Iso) control, shown are summed chromatographic peak areas of tryptic peptides from Lec (y-axes) plotted against the summed peak areas of the two tryptic peptides (HDSGYEVHHQK and LVFFAEDVGSNK) contained within the A-beta segment of the APP protein (x-axes). Note the tight correlation with Lec treatment and Aβ peptides (A), but not Iso treatment and Aβ peptides (B). Panel C is a western blot of viable MV lysates from samples that do (D-6,D-18) or do not (D-16) demonstrate the presence of additional lower MW bands (arrow) of Col-IV that are present with lecanemab (Lec) treatment (10μg/ml), but not in isotype (Iso) controls. Lecanemab and isotype alone variably generated faint IgG cross-reactivity with the Col-IV antibody at 50kDa (arrowheads). Panel D is a representative western blot that shows no effect of a broad MMP2/9 inhibitor on Col-IV degradation in a different cryovial of MV from D-6. Panel E shows data for changes in total Col-IV by Jess with 4 of 10 samples (as indicated by dots with donor number) demonstrating a 40% or greater change when treated with Lec versus Iso (Lec/Iso), with 2 increasing total Col-IV and 2 showing a decrease in total Col-IV. Panel F shows a scatter plot of RTPCR performed to detect transcripts for the alpha 1 and alpha 2 chains (COLIVA1 and COLIVA2, respectively) that regulate synthesis of Col-IV protein. Dots with donor number highlight 6 out of 10 samples with concordant COLIVA1 and COLIVA2 with at least one of the transcripts showing a 2-fold change.

RT-PCR was performed on 12 unique samples treated with Lec or Iso to detect transcripts for COLIVA1 and COLIVA2, which regulate synthesis of Col-IV protein. In 2 of 12 samples neither Col-IV transcripts were detected. Notably, MV from the remaining 10 samples showed at least a 2-fold change in at least one transcript for Col-IV as shown in Fig 2F, which depicts log fold change Lec/Iso.

We then performed IF on MV from 10 samples for CHP, which binds only to degraded collagen. Lec, but not Iso, treatment resulted in 5 of 10 samples showing a 40% or greater increase in CHP relative to total Col-IV (Fig 3A-E).

Figure 3.

MV Treated with Lec Show Increased CHP, a Marker of Degraded Collagen. Panels A, B shows fluorescent stain for CHP (A) relative to immunofluorescence for total Col-IV in MV treated with Iso (B). Panels C, D show fluorescent stain for CHP (C) in MV relative to IF for total Col-IV in MV treated with Lec (D). Panel E is a scatter plot of the levels of CHP relative to total Col-IV with Lec versus Iso treatment for each donor. Dots with donor numbers show which meet the 40 percent or greater change of Relative Degradation (CHP/Total Col-IV) when treated with Lec versus Iso (Lec/Iso). Size bar=50μm for all images.

Although we had 27 unique samples, each has a limited number of cryovials available. Of the 11 samples in more than one outcome measure, 9 samples demonstrated changes in Col-IV that were biologically significant and 6 were concordant in 2 or more outcomes. In summary, we find that a subset of MV from samples with high ADNC showed changes in Col-IV when treated with Lec: degradation on western blotting in 10 out of 16 samples, changes in total levels by Jess in 4 out of 10 samples, differences in transcripts by RT-PCR in 10 out of 12 samples, and significant increase in Col-IV degradation by CHP in 5 out of 10 samples.

Discussion:

The microvasculature undergoes numerous alterations during AD progression, which could reflect Aβ deposition in or near the BM¹⁶. BM Col-IV is prone to changes with age and disease that confer vulnerability to degradation^17–19. Spatial relationships between Aβ and Col-IV can range from near to overlapping, with variation among donors and within microvasculature from the same donor. This results in a patchy response that occurs in the minority of the microvasculature and is difficult to predict or measure by region or in the brain as a whole^{4, 20, 21}. Moreover, it has long been known that Aβ has generalized interactions with multiple proteins of the ECM^22–26. Consequently, we posit that perturbations to Aβ can destabilize or even disrupt Col-IV and result in myriad consequences including frank brain hemorrhages²⁷. We determined if Col-IV is altered when brain MV are exposed to the Aβ binding antibody, lecanemab. The precise regions of degradation can vary^28–30, but the increasing use of these antibodies underscores the importance of studying lecanemab interactions with the BM^{8, 31, 32}.

Lec and Aβ, but not Iso and Aβ, were highly associated in the MV. The Lec treatment resulted in Col-IV degradation in a subset of MV. Lower MW bands (60kDa or 80kDa) in 6 samples with iso treatment reflect baseline levels of Col-IV degradation underscoring the vulnerability of MV in the presence of high ADNC. When measuring total Col-IV by JESS, we found that 4 of 10 samples had a 40% change, with a split between increases and decreases, a common response of Col-IV/ECM to perturbations³³. This response in a subset of samples held true for additional outcome measures such as analysis for COLIVA1 and COLIVA2 transcripts. These findings were further supported by staining for CHP, which showed a 40% or greater increase in a subset of samples exposed to Lec, but not Iso.

The variable effect of lecanemab on Col-IV was expected. Studies show that the effects of Aβ binding with lecanemab and similar antibodies differs among individuals across all outcome measures, with clinically significant detrimental effects in a minority of patients^{5, 6, 8, 32}. Heterogeneity in the MV response likely reflects Aβ amount, location, and spatial relationships with Col-IV. Clinical studies indicate a greater risk of ARIA (amyloid-related-imaging-abnormalities) in those with the APOEε4 allele, with homozygous carriers at greatest risk⁶. We detected no APOEε4 or CAA effect on Col IV outcomes, likely reflecting the high prevalence of both risk factors in our high ADNC samples. Our studies also did not determine whether Col-IV in MV with Aβ is more impacted by Aβ removal from within or outside the MV wall, nor the spatial proximity and quantity of Aβ that must be bound/removed to elicit potential changes.

Notably, we found changes to Col-IV in a subset of MV with only 4d of exposure to Lec. Changes that occur with short exposures are more likely to reflect rapid structural alterations rather than more complex aspects of ECM remodeling. Enzyme activity, especially the MMPs, is known to be induced by the presence of ADNC^{34, 35} and could act, in concert with structural changes, to promote Col-IV and BM breakdown. We did not find any differences in Col-IV degradation when a broad inhibitor of MMP2/9 was added with Lec, and neither MMP9 or total MMP activity correlated with the presence of lower MW Col-IV fragments, but recognize that there are other components that regulate degradative processes in the brain³⁶. The viable MV in this system do not retain glial cells, except for the variable presence of astrocyte foot processes, and co-culture studies^{37, 38} are needed to better mimic the brain parenchyma. Nonetheless, data presented here underscore the utility of viable brain MV to study the effects of lecanemab on the MV BM.

In summary, we propose that lecanemab binding destabilizes Col-IV in a subset of MV. Western blotting, capillary electrophoresis, transcripts for Col-IV and staining for CHP demonstrated responses in Col-IV that could result in BM alterations. These data support the hypothesis that lecanemab can disrupt Col-IV in a manner that contributes to blood vessel injury. Ongoing studies will focus on features of microvasculature that are at greatest risk and to identify enzymatic versus structural mechanisms.

Supplementary Material

SuppDemographicTable

S1 Demographic Table. Characteristics of donors from whom MV were isolated in the Col-IV outcome experiments.

SuppFullFigWesternBlots

S2 For Reviewer’s Only: Whole/uncropped Western Blot of Figure 2.

DATA AVAILABILITY STATEMENT:

Datasets generated during and/or analyzed during the current study are available from the corresponding author on request.