Presence of extraparenchymal IgM+ granular structures

A first screening of coronal sections obtained from ICR-CD1 and SAMP8 mice aged 3, 6 and 12 months immunostained with IgM permitted to observe the presence of some extraparenchymal (EP) IgM+ granular structures, designated as EP granules, that are different to the previously described PAS granules in terms of localization and clustering. These granules were located in the specific brain regions that are detailed in Fig. 1A. As shown in the figure, granules were placed along the longitudinal fissure (between the two brain hemispheres), the quadrigeminal cistern, and a region comprised between the telencephalon and the diencephalon, that will be designated henceforth as the fissura magna as it is indeed a fissure and the largest one in the brain. Note that for bregma lower than -3 mm, this fissure continues between the telencephalon and the mesencephalon.

Fig. 1

Presence of EP granules in mouse brain. A Nissl stained coronal section of a mouse brain at bregma -2,45 mm showing EP granules localization. Granules (red dots) are indicated only in the right hemisphere whereas the left hemisphere is used to indicate the name of the different structures. As can be observed, granules were placed along the longitudinal fissure (green line), the quadrigeminal cistern (QCi), and the fissura magna (red line) and concentrated in specific regions numbered from zone 1 to zone 8. B-I Representative images of EP granules in each described region in a representative 3-month-old SAMP8 mouse. (ACi) ambient cistern, (IPCi) interpeduncular cistern. Figure 1A is adapted from [26]. Scale bars: 50 µm

Although EP granules were placed all along these fissures, they usually concentrated in specific regions numbered from 1 to 8 in the Fig. 1A. The region 1 is located in the longitudinal fissure. The region 2 is located in the upper part of the quadrigeminal cistern. The region 3 is located in the contact between the quadrigeminal cistern and the fissura magna. The region 4 is placed where the hippocampal fissure arises from the fissura magna. The region 5 is located in the lateral region of contact between the fissura magna and the dentate gyrus. In the region 6, the fissura magna begins to contact with the fimbria. In the region 7, the fissura magna is adjacent to the root of the choroid plexus of the lower horn of the lateral ventricle. Finally, the region 8 is located at the inferior part of the fissura magna, near the interpeduncular cistern (Fig. 1 B-I).

Characterization of EP granules

In contrast with PAS granules, that are typically observed forming clusters in the hippocampus each associated with an astrocyte (Fig. 2E), the EP granules reported here exhibited a dispersed distribution and lack association with astrocytes (Fig. 2A).

Fig. 2

Characterization of EP granular structures in mouse brain. When sections were stained with IgM to label the EP granules and with GFAP to stain astrocytes, we observed that EP granules from the fissures showed no relation with astrocytes (A), while the clusters of hippocampal PAS granules were found in areas occupied by astrocytes (E). When staining the sections with PAS, the EP granules were not stained (B), unlike PAS granules, which, as expected, were stained with the PAS stain (F). EP granules were not stained with p62 or GS (C and D, respectively), while PAS granules of the hippocampus were positive to these stainings (G and H, respectively). Sections A, C, D, E, G and H were also stained with Hoechst (cell nuclei, blue). Scale bars: 50 µm

We conducted a PAS stain on brain slides as this method is commonly employed to identify hippocampal PAS granules. Our findings revealed that the EP granules did not exhibit these positive staining (Fig. 2B) while PAS granules in the hippocampus did (Fig. 2F). In fact, this staining defines the named PAS granules. Given the established association of the p62 protein with hippocampal PAS granules [20], we specifically investigated the presence of p62 within the EP granules found in the longitudinal fissure, the quadrigeminal cistern and the fissura magna. Immunofluorescence analysis revealed the absence of p62 within these granules (Fig. 2C), another distinction from hippocampal PAS granules, which are associated to p62 labelling (Fig. 2G). Finally, staining to detect GS also failed to reveal the presence of this protein in the EP granules (Fig. 2D) while, as described in the literature [20], it could be seen in hippocampal PAS granules (Fig. 2H).

On the other hand, considering that these granules in the fissures are recognized by IgMs, which can bind to neoepitopes, and acknowledging the glucidic nature of the neo-epitopes within PAS granules, we conducted experiments to explore whether these IgM antibodies recognize carbohydrate epitopes within EP granules (Fig. 3A-D). Preadsorption of IgM antibodies was carried out with various carbohydrates at increasing concentrations from 0 to 0.4 M. Figure 3 shows that IgM staining reveals EP granules in the fissures when the primary antibody was not preadsorbed with sugars (Fig. 3A) and the staining disappeared when sugars were added at a concentration of 0.4 M (Fig. 3B). PAS granules exhibited comparable behavior, with positive staining observed when the primary IgM antibody was not preadsorbed with sugars (Fig. 3C) and a loss of staining when sugars were added (Fig. 3D). Supplementary Fig. 1 (A-F) illustrates that the staining of these structures gradually diminishes as the sugar concentration increases.

Fig. 3

Carbohydrate characteristics of EP granules vs PAS granules. A-D Effect of the preadsorption of IgM antibodies with a 0.4 M sugar mixture in the IgM staining (red) of EP granules from the fissures (zone 2, A and B) and in PAS granules from the hippocampus (C and D). E–H Effect of the pre-treatment with 200 U/mL of γ-amylase on the staining with IgM in EP granules (zone 2, E and F) and in PAS granules from the hippocampus (G and H). I-L Effect of a 40 min boiling pre-treatment on the staining with IgM in EP granules (zone 2, I and J) and in PAS granules from the hippocampus (K and L). Scale bars: 100 µm

To characterize the glucidic composition of the EP granules, brain sections were incubated with γ-amylase. EP granules in the fissures seemed resistant to this digestion, and several of them could still be observed after the highest γ-amylase concentration (Fig. 3E and F). However, incubation with γ-amylase resulted in the elimination of IgM staining in the PAS granules of the hippocampus (Fig. 3G and H). Indeed, amylase digestions at lower concentrations had almost no impact on IgM staining of EP granules (Suppl. Figure 1 G-I), while hippocampal PAS granules were affected by all concentrations of γ-amylase (Suppl. Figure 1 J-L).

Furthermore, considering the common practice of boiling for antigen retrieval, we immunostained brain sections with IgM after boiling them at 100ºC for 40 min. Remarkably, after the 40-min boiling in citrate, the IgM staining of the EP granules in the fissures disappeared (Fig. 3I and J), as also happened with PAS granules of the hippocampus (Fig. 3K and L).

Cell populations associated to EP granules

Given the high presence of cell nuclei in the areas of the longitudinal fissure, the quadrigeminal cistern and the fissura magna where EP granules were detected, we conducted a series of immunohistochemical staining to elucidate the cell populations associated with these granules.

Firstly, brain sections were triple-immunostained for IgM, GFAP, and vimentin (Fig. 4A). IgM permitted to observe the EP granules, GFAP was used to stain the astrocytes and the glia limitans (which are GFAP positive and vimentin positive), and vimentin to stain fibroblasts (which are vimentin positive and GFAP negative) and also astrocytes. The vast majority of cells harbouring EP granules exhibited vimentin positivity but lacked GFAP expression, indicating that these cells are fibroblast and not astrocytes. Moreover, given that the astrocytes in the glia limitans separate the brain parenchyma from the pia mater and the subarachnoidal space, and because of dura and arachnoid maters do not penetrate to the fissures, these fibroblasts must be sited in the pia mater. Consistently, a layer of pial fibroblasts separates the pia and the glia limitans [27]. Therefore, we can deduce that these granules reside outside the brain parenchyma and are in contact with cells that have a fibroblastic rather than an astrocytic identity. Moreover, a few EP granules were also found in the root of the choroid plexus. To further explore these few granules, we stained additional brain sections with IgM, anti-cytokeratin (to stain choroid plexus epithelial cells) and with anti-vimentin (for fibroblasts). As can be observed in Fig. 4B, most EP granules were associated with vimentin staining, suggesting also a selective association of these granules with the fibroblasts adjacent to the choroid plexus. Therefore, all these EP granules were found outside the brain parenchyma, in areas such as the pia mater, associated with fibroblasts.

Fig. 4

Cell populations associated to EP granular structures. A Representative sections of zone 2 from a 3-month-old SAMP8 mouse immunostained with anti-GFAP (green), anti-vimentin (red), and IgM (white) antibodies, showing merged and separated channels; B Representative sections of zone 2 from a 3-month-old SAMP8 mouse immunostained with anti-cytokeratin (green), anti-vimentin (red) and IgM (white) antibodies, showing merged and separated channels (B). All sections were stained with Hoechst (cell nuclei, blue). White arrows indicate EP granules in the pia mater; orange arrows indicate EP granules in the root of the choroid plexus. Scale bars: 100 µm

All the results regarding the characterization of these EP granules and the comparison with hippocampal PAS granules are summarized in Table 1.

Table 1 Comparison of the characteristics of EP granules and PAS granulesPresence of IgM+ astrocytes

In 12-month-old aged mice from both SAMP8 and ICR-CD1 strains, we noted the presence of astrocyte-like cells exhibiting IgM staining located near the longitudinal fissure, the quadrigeminal cistern and the fissura magna. These astrocyte-like cells IgM+ were scarcely seen in 3- or 6-month-old animals.

Colocalization of IgM staining with GFAP corroborated the astrocytic identity of these cells, proving the presence of neo-epitopes in astrocytes (Fig. 5A). Notably, only the astrocytes closest to the fissures were IgM+ (Fig. 5A1 and 5A2). Although these astrocytes IgM+ were located in regions near the longitudinal fissure, the quadrigeminal cistern and the fissura magna, they were never found associated to the areas where hippocampal PAS granules were found.

Fig. 5

Presence of IgM+ astrocytes. A Representative section of zone 2 from a 12-month-old SAMP8 mouse immunostained with anti-GFAP (green), and IgM (red) antibodies (A); note that astrocytes not close to the fissures were IgM− (green-stained, empty arrowheads, A.1) while only the astrocytes closest to the fissures were IgM+ (yellow (green + red)-stained, full arrowheads, A.2). B Effect of the preadsorption of IgM antibodies with a 0.4 M sugar mixture in IgM+ astrocytes in a representative section of a 12-month-old SAMP8 mouse. C Effect of the pre-treatment with 200 U/mL of ƴ-amylase on the staining with IgM in IgM+ astrocytes in a representative section of a 12-month-old SAMP8 mouse. D Effect of a 40 min boiling pre-treatment on the staining with IgM in IgM+ astrocytes in a representative section of a 12-month-old SAMP8 mouse. Scale bars: 100 µm

As happened with the granules found in the fissures, the preadsorption of the IgM antibodies with sugars caused the disappearance of IgM staining on these astrocytes (Fig. 5B), thus indicating that these astrocytes near the fissure exhibited glucidic structures specifically recognized by IgMs. On the other hand, treatment of brain slices with γ-amylase did not affect the IgM positivity of these astrocytes for IgM (Fig. 5C). Finally, exposure of brain samples to boiling for antigen retrieval, following a similar protocol than before, diminished IgM positivity in astrocytes (Fig. 5D).

Body weight analysis

All animals were weighed just before their sacrifice and obtained values are shown in Fig. 6A. The statistical analysis performed by ANOVA, defining both the strain and the age at sacrifice as independent variables and including in the model the interaction between them, indicated that both variables and their interaction have a significant effect on the weight of the animals (p < 0.01 for strain, p < 0.05 for age and p < 0.05 for the interaction). This interaction indicates that age impacts the two strains differently. As illustrated in Fig. 6A and consistent with the characteristics of each strain [28, 29], post hoc comparisons indicate that the mean body weight of ICR-CD1 mice was higher than that of the SAMP8 animals at each time point. Moreover, the body weight of ICR-CD1 mice progressively increased from 3 to 12 months, with 12-month-old animals weighing more than their younger littermates. In contrast, SAMP8 mice exhibited an increase in weight from 3 to 6 months, followed by a decrease from 6 to 12 months. Although these differences in SAMP8 do not reach statistical significance (p = 0,08 for the factor weight in SAMP8), the decrease in their body weight at 12 months old is likely due to their accelerated senescence, as loss of body weight is associated with the aging process itself [30], resulting in worsening and weakening.

Fig. 6

A Body weight of ICR-CD1 and SAMP8 mice at the age of sacrifice. Data are shown as mean ± s.e.m. including individual values. B-D Scores obtained in ICR-CD1 and SAMP8 mice throughout the study regarding the amount of the hippocampal PAS granules (B), the amount of IgM+ astrocytes in the vicinity of the fissures (C) and the amount of EP granules in the fissures (D). Bars show means ± s.e.m. of the scores from each experimental group, and the specific scores for each animal are also shown. *p < 0.05; ** p < 0.01. a.u.: arbitrary units

Evolution of hippocampal PAS granules

Ageing in mice leads to the appearance and accumulation in the hippocampus of specific age-related polyglucosan aggregates, known as PAS granules, with an earlier increase observed in senescence-accelerated SAMP8 animals than in control ICR-CD1 mice, with non-accelerated senescence. In order to verify these events in the animals used in the present work, we quantified the number of hippocampal PAS granules for each animal (Fig. 6B). The statistical analysis performed by ANOVA, defining the variables strain and age at sacrifice as independent variables, indicated that both variables and their interaction have a significant effect on the number of PAS granules (p < 0.01 in all cases). Post-hoc comparisons indicate that SAMP8 animals had an increased amount of PAS granules respect to ICR-CD1 animals and that the amount of PAS granules increased with the ageing of the animals. The significant interaction indicates that, as expected according to the literature [13], ageing influences the two strains differently and markedly in SAMP8 animals. Specifically, 12-month-old SAMP8 mice showed more hippocampal granules than younger SAMP8 mice and, at this older stage, SAMP8 mice exhibited more hippocampal PAS granules than age-matched ICR-CD1 mice. As commented, SAMP8 animals begin to worsen and weaken before 12 months of age, which is reflected in both the decrease of their body weight and the marked increase of PAS granules in their hippocampus.

Evolution of the amount of IgM+ astrocytes

We quantified the IgM+ astrocytes associated with the fissures. Figure 6C shows the amount of IgM+ astrocytes across strains and ages. The statistical analysis performed by ANOVA, defining the variables strain and age at sacrifice as independent variables, indicated that the variable age has a significant effect (p < 0,01) on the presence of IgM+ astrocytes as well as the interaction between both variables (p < 0,05), which indicates that the number of IgM+ astrocytes increased with age differently in ICR-CD1 animals than in SAMP8 animals. Notably, 12-month-old SAMP8 mice displayed a higher number of astrocytes reactive to IgM compared to younger SAMP8 mice. Furthermore, at this advanced age, SAMP8 mice presented more IgM+ astrocytes than age-matched ICR-CD1 mice. These results were highly congruent with both the body weight evolution and the evolution of the amount of PAS granules, all reflecting the accelerated senescence in 12-month-old SAMP8 mice.

Evolution of the amount of EP granules

From each animal, we quantified the amount of EP granules in each region of interest of one coronal brain section, and we obtained the mean of the EP granules contained in the different regions, hence defining the score of EP granules for every animal. Then, using the score of EP granules from each animal, we studied the effects of age and strain in the amount of EP granules. Figure 6D shows the EP score per animal and the mean for each experimental group. The statistical analysis performed by ANOVA, defining the variables strain and age as independent variables, showed that age significantly affected EP granule score (p < 0.01) and that the interaction was also significant (p < 0.05), indicating that the effect of age on ICR-CD1 and SAMP8 strains differs. Post hoc comparisons indicate that 12-month-old SAMP8 mice show a significant decrease in these granules compared to 3- and 6-month-old animals, reflecting again the accelerated senescence in 12-month-old SAMP8 mice.

Integrative view or the results

Having observed how age and strain variables influenced the amount of PAS granules in the hippocampus, along with IgM+ astrocytes and EP granules, we analysed the correlations among these variables. While there was a negative correlation between EP granules and both PAS granules and IgM+ astrocytes (ρ = -0.533, p < 0.01, and ρ = -0.605, p < 0.01 respectively), PAS granules and IgM+ astrocytes showed a positive correlation between them (ρ = 0.787, p < 0.01). Taken all together, this indicates that the age-related increase in PAS granules in the hippocampus, which is more pronounced in SAMP8 than in ICR-CD1 animals, is accompanied by an increase in IgM+ astrocytes and a decline in EP granules in the fissures.