Transgenic 5XFAD mice obtained from Jackson Laboratories were maintained at Vanderbilt University Animal Care Unit under standard conditions, in a 12-h light/dark cycle and with free access to food and water. These animals were genotyped using tail or ear derived DNA to confirm the expression of genes, 377 bp amyloid precursor protein (APP) and 608 bp presenilin 1 (PSEN1) using polymerase chain reaction based 1% agarose gel electrophoresis. From these animals, nine Ergo-treated 5XFAD (4–5 months old, three females and six males), nine non-treated 5XFAD (4–5 months old, five females and four males), and eight non-treated C57BL/6J wildtype (WT controls) (4–5 months old, four females and four males) mice were used for the investigations. C57BL/6J is the background strain for 5XFAD transgenic mice. Therefore, in addition to Ergo-treated vs. non-treated 5XFAD, C57BL/6J wildtype was included as a control to determine the level of expression of Aβ (refer Supplementary File which describes 6E10 immunoreactivity in transgenic and WT mice) as well as other protein markers, and to explore the existing Aβ clearance mechanisms.

Commercially purchased (Cayman Chemical, Ann Arbor, MI, USA) L-Ergo, formulated in sterilized water was administrated orally without purifications steps as recommended by the manufacturer. Nine 5XFAD mice, starting at the age of 1.5 months, were subjected to oral gavage treatment with high doses of Ergo solution at a concentration of 50 mg/kg 3 times per week over the course of 8 weeks. For every gavage treatment, freshly prepared Ergo solution was used. This procedure involves passing a reusable oral gavage needle through the mouth and placing the rounded tip atop of the esophagus of an awake animal in a manner to encourage the animal to swallow the formulation voluntarily. The curvature of the needle fitting with oral cavity, along with the extra smooth round ball stainless steel tip ensured minimal discomfort to the treated animals. To ameliorate potential compounding stress-induced discrepancies, non-treated 5XFAD and WT controls similarly underwent oral gavage procedure with sterilized water. Animals were sacrificed at the age of 4–5 months for subsequent histopathological studies.

Cardiac perfusion was carried out under ice-cold conditions. A sharp incision was made into the abdomen of the anesthetized mouse, followed by a longitudinal cut with a scalpel to open the thoracic cavity, which then was stabilized with a retractor. Perfusion began by inserting a 20-gauge syringe containing ice-cold 1× phosphate-buffered saline (PBS) (30 ml, pH 7.4) in the left ventricle while the atrium was snipped off, and then followed by injection of 4% paraformaldehyde solution. Once the perfusion was completed, the animals were decapitated, eyes were harvested immediately. Whole eye globes were then fixed in 4% paraformaldehyde at 4°C for a period of 1–2 weeks. Right eye or oculus dexter (OD) was used for wholemounts, while left eye or oculus sinister (OS) was used for the cross-sectional studies.

Briefly, the cornea was removed by cutting in a circular path along with the ora serrata using small scissors while holding the eye at the limbus with forceps. Next, the lens and vitreous body were removed with blunt-edged forceps. The neuroretina was then carefully detached from the eyecup using fine-tipped sable brushes positioned between the neuroretina and the eyecup. The optic nerve was cut at the location between the neuroretina and the eyecup to ease the separation of the neuroretina from the eyecup. The neuroretina was then cut into half to make two hemi-retinal wholemounts, rinsed with PBS (pH 7.2–7.4) and the thin strands of vitreous body were removed under a dissection microscope.

For this work, nine Ergo-treated 5XFAD (4–5 months old, three females and six males), nine non-treated 5XFAD (4–5 months old, five females and four males), and eight WT controls (4–5 months old, four females and four males) were used. To determine Aβ burden, we used 6E10 primary antibody (mouse monoclonal, Cat# 803001, BioLegend) which is a marker for 1–16 amino acid residues of Aβ peptide (refer Supplementary File which describes 6E10 immunoreactivity in transgenic and WT mice). Other primary antibodies included glial fibrillary acidic protein (GFAP, rabbit poly clonal, Cat# Z0334, Dako), a type III intermediate filament protein and a marker for astrocytes and ionized calcium-binding adapter molecule 1 (IBA1, rabbit polyclonal, Cat# 019-19741, Wako), a marker for microglia/macrophage-specific calcium-binding protein. Secondary antibodies Alexa Fluor™ 546 goat anti-mouse IgG1 (Cat# 21123) and Alexa Fluor^® 488 goat anti-rabbit IgG (Cat# 11070) were used for fluorescence confocal microscopy.

In this work, one half of the neuroretina (i.e., hemi-retinal wholemount) was used for 6E10 and IBA1 double immunostaining and the other half was used for 6E10 and GFAP double immunostaining. Briefly, neuroretinas were placed carefully in 48-well plates containing PBS and rinsed 3 times, each for 5 min. Formic acid (88%) was then added to each well for antigen retrieval and incubated for 5 min at room temperature (RT). Samples were then washed in PBS. After the washing steps with PBS (3 times, each for 5 min), samples were incubated with blocking buffer (3% normal goat serum plus 0.3% Triton X-100 in PBS) for 20 min at RT. Primary antibodies 6E10 (1:100) and GFAP or IBA1 (1:100) diluted in 3% blocking buffer were then added to each well and incubated at RT for 1 h, followed by 72 h at 4°C with gentle agitation. Next, the wholemounts were washed with PBS (4 times, each for 5 min), and then secondary antibodies, Alexa Fluor™ 546 (1:500) and Alexa Fluor^® 488 (1:500) diluted in 1% blocking buffer were added to each well and incubated at RT for 45 min. Next, the wholemounts were washed with PBS (4 times, each for 15 min), and the nuclei were stained with 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI) (1:500) for 10 min, followed by washing with PBS for 4 times, each 15 min. Finally, neuroretinas were mounted onto the glass slides using mounting medium containing 80% glycerol and 20% PBS, covered with #1.5 thickness coverslips for confocal microscopy. Negative control slides were simultaneously processed, but with the omission of primary antibodies and washed separately to avoid the cross contaminations throughout the remaining steps.

For this work, six Ergo-treated 5XFAD (4–5 months old, two females and four males), six non-treated 5XFAD (4–5 months old, three females and three males), and six WT controls (4 months old, three females and three males) were used. In addition to 6E10, GFAP, and IBA1, primary antibodies OCTN1 (rabbit polyclonal, Cat# ACT-014, Alomone Labs), a marker for Ergo transporter and AQP4 (rabbit polyclonal, Cat# AQP-004, Alomone Labs), a major membrane water channel in the CNS were used in the eye cross-sections. Six micrometer thick eye cross-sections, after a series of deparaffinization and rehydration steps, underwent pre-treatment with 88% formic acid for 5 min at RT for double immunostaining with 6E10 and then rinsed in PBS. Antigen retrieval was carried out with 0.05% proteinase K in Tris-EDTA buffer (pH 8.0) for all 5 primary antibodies and incubated for 10 min at RT. Next, sections were incubated in blocking buffer (3% normal goat serum and 0.3% Triton X-100 in PBS) for 20 min at RT. Primary antibodies OCTN1 (1:200), 6E10 (1:200), GFAP (1:200), IBA1 (1:200), and AQP4 (1:100) diluted in 3% blocking buffer were then added to each section and incubated at RT for 1 h, followed by overnight at 4°C. Following washings with PBS, secondary antibodies, Alexa Fluor™ 546 (1:500) and Alexa Fluor^® 488 (1:500) diluted in PBS were added to each section and incubated at RT for 45 min. After washing steps, nuclei were stained with DAPI. Sections were finally mounted using 80% glycerol plus 20% PBS mounting medium with #1.5 thickness coverslips and stored at 4°C for confocal microscopy. Negative control slides were simultaneously processed, but with the omission of primary antibodies and washed separately to avoid the cross contaminations throughout the remaining steps.

All fluorescent images were captured using Zeiss LSM 800 confocal microscope with ZEN 2.6 (blue edition) software. Cy3 immunofluorescence from the 6E10 antibody was imaged at 543 nm, Alexa Fluor^® 488 immunofluorescence from the GFAP, IBA1, AQP4, and OCTN1 antibodies was imaged at 488 nm, and the nuclear labeling with DAPI was imaged at 405 nm for 200× (scale bar 50 μm) and 400× (scale bar 20 μm) magnifications. Confocal settings including laser, pinhole, and master gain were same across the three animal groups for each marker whereas digital gain and offset kept at default settings throughout the study. For the double labeled wholemounts (6E10 and GFAP or IBA1), a series of z-stack images (200×) were taken in an average of five non-overlapping fields focusing on the level of retinal ganglion cell layer (GCL). The GCL was identified by sequentially imaging through the thickness of the wholemount, starting from inner limiting membrane (ILM)/nerve fiber layer (NFL), then proceeding to the GCL (in which, DAPI labeled nuclei are abundant), inner plexiform layer (IPL, a cell free layer) and finally into the inner nuclear layer (INL, in which DAPI labeled nuclei are abundant). The series of z-stack images confirmed that the antibody incubations (72 h) penetrated into the thick wholemounts, as labeling was observed as deep as the INL of z-stack images. For the 6E10 and GFAP or IBA1 double labeled eye cross-sections, z-stack images (400×) were taken in a minimum of two central and two peripheral fields of the retina starting at the level of the ILM/NFL to outer nuclear layer (ONL). For the OCTN1 labeled eye cross-sections, z-stack images (400×) were taken from anterior segments of the eye such as cornea, lens and the ciliary body, and posterior segments of the eye such as central and peripheral retinas and retinal pigment epithelium (RPE)/Choroid regions. And, for the 6E10-AQP4 labeling, images were captured at 200× magnification in a minimum of four central and two peripheral retinal regions.

Due to the sex-based differences in the severity of AD pathology in 5XFAD mice, we initially compared 6E10 immunoreactivity between male and female findings within Ergo-treated and non-treated 5XFAD groups using independent samples t-test (Student’s or Welch’s corrected) for both wholemounts and cross-sections (central and peripheral retinas). However, sex-based analysis did not show any significant difference (p > 0.05) in 6E10 immunoreactivity between male and female mice within Ergo-treated and non-treated 5XFAD groups for wholemounts (p = 0.147 and p = 0.613, respectively), central retina (p = 0.566 and p = 0.49, respectively) and peripheral retina (p = 0.436 and p = 0.163, respectively). Therefore, pixel counts and semi-quantitative counts were compared across WT controls, non-treated 5XFAD, and Ergo-treated 5XFAD by combining male and female data. Pixel counts were analyzed using non-parametric one-way ANOVA test (Kruskal–Wallis H test, two-sided) followed by Dunn’s post-hoc test at a Bonferroni adjusted significance level of p < 0.05 for multiple comparisons. Semi-quantitative counts were analyzed using one-way ANOVA test followed by Games-Howell post-hoc test at a significance level of p < 0.05 for multiple comparisons. Spearman’s rank-order correlation test was conducted to see the association between 6E10 and other tested protein markers at a significance level of p < 0.05 based on the planes of section. SPSS version 25.0 (IBM Corp., Armonk, NY, USA) was used for the statistical analysis, and the GraphPad Prism 9 (GraphPad Software Inc., San Diego, CA, USA) was used to generate graphs.

Organic cation transporter novel-type 1 is responsible for Ergo uptake and accumulation in eye tissues, and therefore we initially analyzed eye cross-sections to localize OCTN1 transporters (Figure 1 and Table 1). Total OCTN1 immunoreactivity in the whole eye was significantly different across the three animal groups (p = 0.002, Kruskal–Wallis H test) (Table 1). It was significantly high in WT controls compared with non-treated or Ergo-treated 5XFAD (p = 0.002 and p = 0.002, respectively Bonferroni corrected Dunn’s post-hoc test) (Figure 1T). Between anterior [mean ± standard error of mean (SEM), 1,835.1 ± 386.0] and posterior (1,238.3 ± 265.7) segments (as indicated in Figure 1S), total OCTN1 immunoreactivity was not significantly different (p = 0.19, Student’s t-test). OCTN1 immunoreactivity in anterior segments of the eye such as cornea, lens, and ciliary body, and in posterior segments of the eye such as retina and RPE/Choroid were separately compared across the three animal groups and summarized (Table 1 and Figures 1U, V). Total OCTN1 immunoreactivity in anterior segments were significantly different (p = 0.029, Kruskal–Wallis H test) across the three animal groups. OCTN1 expressions in anterior segments of the WT controls (Figures 1A, D, G), non-treated 5XFAD (Figures 1B, E, H) and Ergo-treated 5XFAD (Figures 1C, F, I) are illustrated. Between animal groups, it was significantly high in WT controls compared with non-treated 5XFAD (p = 0.015, Bonferroni corrected Dunn’s post-hoc test) (Figure 1U). Similarly, total OCTN1 immunoreactivity in posterior segments were significantly different (p = 0.029, Kruskal–Wallis H test) across the three animal groups. OCTN1 expressions in posterior segments of the WT controls (Figures 1J, M, P), non-treated 5XFAD (Figures 1K, N, Q), and Ergo-treated 5XFAD (Figures 1L, O, R) are illustrated. Between animal groups, it was significantly high in WT controls compared with Ergo-treated 5XFAD (p = 0.016, Bonferroni corrected Dunn’s post-hoc test) (Figure 1V). However, individual region-based analysis did not show any significant difference across the three animal groups (Table 1). Intergroup comparisons showed no significant difference in the level of expression of OCTN1 transporters in both anterior and posterior segments of the eye of Ergo-treated and non-treated 5XFAD mice.

Colocalization of 6E10 and GFAP or IBA1 antibodies in the wholemounts are illustrated along with negative control (Figures 2A–G), and their immunoreactivity was compared across the three animal groups (Figures 2I–K and Table 2). Existence of neuritic type Aβ plaques surrounded by reactive astrocytes (Figure 2B, white arrowheads), large Aβ deposits (Figure 2C, yellow arrowheads), GFAP+(ve) phagocytic astrocytes/PLS (Figure 2C, red boxes), IBA1+(ve) microglia (Figures 2E–G, blue arrowheads) and phagocytic IBA1(+)ve blood-derived macrophages/PLS (Figure 2G, yellow boxes) in the wholemounts are indicated (details were given in the following paragraphs). Antibody labeling was predominantly detected on ILM/NFL and GCL, and thus orthogonal projections represented the immunoreactivity that were obtained from ILM/NFL and GCL. 6E10 immunoreactivity was significantly different across the three animal groups (p < 0.001, Kruskal–Wallis H test) (Table 2). Between groups, it was significantly high in Ergo-treated 5XFAD and WT controls (p < 0.001 and p < 0.001 respectively, Bonferroni corrected Dunn’s post-hoc test) compared with non-treated 5XFAD (Figure 2I). With respect to glial cell markers, GFAP and IBA1 immunoreactivities were significantly different across the three animal groups (p = 0.020 and p = 0.002, respectively, Kruskal–Wallis H test) (Table 2). However, between groups, GFAP and IBA1 immunoreactivities were significantly high in WT controls compared with non-treated 5XFAD (p = 0.006 and p = 0.001, respectively, Bonferroni corrected Dunn’s post-hoc test) (Figures 2J, K). Although it was statistically non-significant, GFAP immunoreactivity was reduced in the wholemounts Ergo-treated 5XFAD compared with WT controls (Figure 2J).

A significantly high level of 6E10 immunoreactivity on the surface of wholemounts of Ergo-treated 5XFAD and WT controls suggests the existence of effective Aβ clearance mechanisms. This could be possibly contributed by various systems and cells that involved in Aβ clearance such as phagocytic IBA1+(ve) resident and/or blood-derived macrophages, GFAP+(ve) phagocytic astrocytes/PLS, and glymphatic and perivascular drainages. To support the above suggestion, we provided the evidence for the existence of effective Aβ clearance mechanisms on the surface of wholemounts (Figures 3, 4). Negative control slide was prepared without primary antibodies to distinguish the debris or autofluorescence bodies (Figures 3A–D) in which DAPI uptake was poor (white circles in Figure 3D). IBA1(+)ve PLS (Figures 3E–H, yellow box and Supplementary Videos 1, 2) and GFAP(+)ve PLS (Figures 3I–L, red boxes and Supplementary Videos 3, 4) are found on top of the GCL. Based on triple positive staining, functional PLS (Figures 3E–L) were distinguished from debris or autofluorescence bodies (Figures 3A–D). However, the size of the functional PLS [GFAP+(ve) or IBA1+(ve)] was varied across the three animal groups (Supplementary Image 1).

Besides, to confirm the existence of effective Aβ clearance stimulated by Ergo treatment, a semi-quantification protocol was adapted to count 6E10+(ve) Aβ plaques or large deposits, GFAP+(ve) PLS and IBA1+(ve) PLS on the surface of wholemount neuroretinas manually (Figures 2H, 3M, N). Mean counts (mean ± SEM) were significantly low in Ergo-treated (3.74 ± 0.37) compared with non-treated (5.83 ± 0.75) 5XFAD (p = 0.036, Games-Howell post-hoc test) (Figure 2H). Whereas mean counts of 6E10+(ve) Aβ deposits in WT controls (4.17 ± 0.43) were non-significant when compared with Ergo-treated or non-treated 5XFAD. Mean counts of GFAP+(ve) PLS were 1.32 ± 0.34, 1.37 ± 0.27, and 0.76 ± 0.22 in Ergo-treated 5XFAD, non-treated 5XFAD and WT controls, respectively though it was statistically non-significant across the three animal groups (p = 0.308, one-way ANOVA test) [Figure 3M, each red box was counted as one GFAP+(ve) PLS]. Mean counts of IBA1+(ve) PLS were significantly high in Ergo-treated 5XFAD (1.29 ± 0.21) and WT controls (1.69 ± 0.31) compared with non-treated 5XFAD (0.65 ± 0.14) (p = 0.038, and p = 0.009, respectively, Games-Howell post-hoc test) [Figure 3N, each yellow box was counted as one IBA1+(ve) PLS]. GFAP+(ve) or IBA1+(ve) PLSs were identified by 6E10+(ve) Aβ deposits that were engulfed or surrounded by DAPI+(ve) clumps of nuclei.

In addition to PLS, existence of glymphatic drainage, mediated by tightly attached astrocytic end-feet (white arrowheads) toward proximal end of the optic nerve (white asterisk) was identified on the surface of wholemounts (Figures 4A–D and Supplementary Videos 5, 6). Its existence at the surface level was demonstrated via 3D projections (Figures 4E, F, white arrowheads). Similarly, terminal end of the capillary beds (yellow arrowheads), facilitating Aβ influx into perivascular drainage were identified on the surface of wholemounts as demonstrated via 3D projections (Figures 4G–K and Supplementary Video 7).Visually, Aβ clearance via glymphatic drainage mediated by astrocytes and perivascular drainage mediated by capillary beds were mostly identified in WT controls and Ergo-treated 5XFAD, respectively. However, their existence was also noticed in the remaining animal groups in varying degrees (Supplementary Image 2).

In addition to wholemounts staining, eye cross-sections were used to screen the 6E10, AQP4, GFAP, and IBA1 expression patterns across the retinal layers (Tables 3, 4). Colocalization of 6E10 and AQP4 markers in the central and far peripheral retinal cross-sections are illustrated (Figures 5A–F). 6E10 and AQP4 immunoreactivities were significantly different across the three animal groups (p < 0.001 and p < 0.001, respectively, Kruskal–Wallis H test) (Table 3). Between groups, 6E10 immunoreactivity was significantly low in Ergo-treated 5XFAD compared with WT controls or non-treated 5XFAD (p = 0.005 and p < 0.001, respectively, Bonferroni corrected Dunn’s post-hoc test) (Table 3). Similarly, between groups, AQP4 immunoreactivity was significantly low in Ergo-treated 5XFAD compared with WT controls or non-treated 5XFAD (p < 0.001 and p < 0.001, respectively, Bonferroni corrected Dunn’s post-hoc test) (Table 3). 6E10 and AQP4 expressions were further analyzed in detail for central and peripheral retinas (Table 3 and Figures 5G, H). In both central and peripheral retinas, immunoreactivities were significantly low for 6E10 (p < 0.001 and p = 0.003, respectively, Bonferroni corrected Dunn’s post-hoc test) and AQP4 (p < 0.001 and p = 0.002, respectively, Bonferroni corrected Dunn’s post-hoc test) in Ergo-treated compared with non-treated 5XFAD. Whereas 6E10 and AQP4 immunoreactivities were significantly low only in the central retina of Ergo-treated 5XFAD compared with WT controls (p = 0.004 and p < 0.001, respectively, Bonferroni corrected Dunn’s post-hoc test) (Table 3).

GFAP immunoreactivity in the retinal cross-sections were significantly different across the three animal groups (p < 0.001, Kruskal–Wallis H test) whereas IBA1 immunoreactivity was not significantly different (p = 0.873, Kruskal–Wallis H test) (Table 4). Colocalization of 6E10 and GFAP or IBA1 markers in the central and far peripheral retinal cross-sections are illustrated (Figures 6A–F, 7A–F, respectively). GFAP immunoreactivity was significantly low in Ergo-treated or non-treated 5XFAD compared with WT controls (p < 0.001 and p = 0.003, respectively, Bonferroni corrected Dunn’s post-hoc test). Moreover, between Ergo-treated and non-treated 5XFAD, GFAP immunoreactivity was significantly high in non-treated 5XFAD (p = 0.001, Bonferroni corrected Dunn’s post-hoc test). GFAP expression was further analyzed in detail for central and peripheral retinas (Table 4 and Figures 6G, H). In both central and peripheral retinas, immunoreactivities were significantly low in Ergo-treated 5XFAD compared with WT controls (p < 0.001 and p < 0.001, respectively Bonferroni corrected Dunn’s post-hoc test). Whereas, between Ergo-treated and non-treated 5XFAD, GFAP immunoreactivity was significantly low only in the peripheral retina of Ergo-treated 5XFAD (p = 0.007, Bonferroni corrected Dunn’s post-hoc test). Further, detail analysis of IBA1 immunoreactivity in central and peripheral retinas did not show any significant difference across the three animal groups (Figures 7G, H).

In addition to multiple comparisons, Spearman’s rank-order correlation test was conducted to see the strength of association between 6E10 and IBA1 or 6E10 and GFAP expressions in the hemi-retinal wholemounts (Figures 8A–F). Correlation coefficients (ρ) obtained between 6E10 and IBA1 expressions were significant and positively related in all three groups, however the associations were stronger in WT controls (p < 0.001, ρ = 0.603) followed by Ergo-treated 5XFAD (p < 0.001, ρ = 0.517) compared with non-treated 5XFAD (p = 0.018, ρ = 0.359) (Figures 8A–C). Whereas ρ values obtained between 6E10 and GFAP expressions were statistically non-significant in all three animal groups (Figures 8D–F). Spearman’s correlation test was also conducted to see the strength of association between 6E10 and IBA1 or GFAP or AQP4 expressions in the retinal cross-sections (Figures 8G–O). Between 6E10 and IBA1 or 6E10 and AQP4 expressions, ρ values were statistically non-significant in all three animal groups (Figures 8G–I, M–O). Similarly, between 6E10 and GFAP expressions, ρ values were statistically non-significant in WT controls and Ergo-treated 5XFAD (Figures 8J, L), whereas it was significant and positively correlated in non-treated 5XFAD (p = 0.001, ρ = 0.648) (Figure 8K).