Nanoscale alterations in GABAB receptors and GIRK channel organization on the hippocampus of APP/PS1 mice

Reduction of plasma membrane GIRK2 on the hippocampus of aged APP/PS1 mice

The GIRK2 subunit is an essential component of the functional channel in the hippocampus as it determines both the channel assembly and plasma membrane targeting [14, 39]. Thus, we aimed at investigating whether Aβ pathology affects the spatial organization of the GIRK2 subunit. Hippocampal sections from 12-month-old wild type and APP/PS1 mice were processed for SDS-FRL [35] and the GIRK2 subcellular distribution in pyramidal neurons from the distal part of the stratum radiatum of the CA1 region was assessed (Fig. 1). In wild type animals, immunoparticles for GIRK2 were detected in postsynaptic elements, namely, at the extrasynaptic plasma membrane of spines in contact with axon terminals, which are likely deriving from Schaffer collaterals, and dendritic shafts of CA1 pyramidal cells (Fig. 1A–C), as recently reported by pre-embedding experiments [20]. Importantly, immunoparticles for GIRK2 were found mostly in clusters containing 3 or more particles, and in a lesser extent scattered as single gold particles outside the clusters (Fig. 1A–C). Conversely, a reduced density of GIRK2 immunoparticles and fewer channel clusters was found along the membrane surface of CA1 pyramidal cells from APP/PS1 mice (Fig. 1D–F). No labelling was observed on the E-face or on cross-fractures (Fig. 1A–F). Indeed, the quantitative analysis of the images demonstrated that GIRK2 density was significantly reduced in oblique dendrites (oDen) and spines in APP/PS1 mice (oDen = 10.02 ± 1.28 immunoparticles/μm2; spines= 30.87 ± 3.09 immunoparticles/μm2) when compared to age-match wild type mice (oDen= 41.93 ± 2.86 immunoparticles/μm2; spines= 75.35 ± 9.53 immunoparticles/μm2) (Mann–Whitney test, ****p<0.0001) (Fig. 1G).

Fig. 1figure 1

Postsynaptic reduction of GIRK2 in CA1 pyramidal cells of APP/PS1 mice. Electron micrographs showing postsynaptic immunoparticles for GIRK2 in the stratum radiatum of the hippocampal CA1 field at 12 months of age, as detected using the SDS-FRL technique in wild type and APP/PS1 mice. AC In wild type, GIRK2 immunoparticles were detected forming clusters (blue circles) or scattered (blue arrows) and associated with the P-face in dendritic spines (s) and oblique dendrites (oDen) of CA1 pyramidal cells. DF In APP/PS1, very low frequency of clusters or scattered (blue arrows) GIRK2 immunoparticles were observed in dendritic spines (s) and oblique dendrites (oDen) of CA1 pyramidal cells. The E-face is free of any immunolabelling. G Quantitative analysis showing that the density of surface GIRK2 immunoparticles was significantly reduced in the APP/PS1 mice compared to age-matched wild type controls in the two subcellular compartments analysed (Mann–Whitney test, ****p < 0.0001). Error bars indicate SEM. Scale bars: AF, 0.2 μm

In addition to somato-dendritic domains of CA1 pyramidal cells, immunoparticles for GIRK2 were also present in axon terminals (Fig. 2), as previously reported [16, 20]. In wild type mice, GIRK2 immunoparticles were mainly detected at extrasynaptic sites and the immediate perisynaptic region of the active zone and, to a lesser extent, within the active zone, as identified by the concave shape of the P-face and the accumulation of IMPs (Fig. 2A, B). GIRK2 immunoparticles were detected forming clusters or scattered at extrasynaptic sites, the active zone, and the edge of the active zone (Fig. 2A, B). In APP/PS1 mice, GIRK2 immunoparticles were found in the same subcellular compartments as in wild type, but less frequently detected along the surface (Fig. 2C). Our quantitative analysis showed that the density of GIRK2 in presynaptic nerve terminals was significantly reduced in extrasynaptic and the active zone in APP/PS1 mice (extra = 15.66 ± 3.70 immunoparticles/μm2; AZ= 61.04 ± 14.69 immunoparticles/μm2) compared to age-matched wild type mice (extra = 44.38 ± 5.80 immunoparticles/μm2; AZ= 156.82 ± 50.41 immunoparticles/μm2) (Mann–Whitney test, *p < 0.05; ***p < 0.001) (Fig. 2D).

Fig. 2figure 2

Reduction of GIRK2 immunoparticles in presynaptic compartments of APP/PS1 mice. Electron micrographs showing presynaptic immunoparticles for GIRK2 in the stratum radiatum of the hippocampal CA1 field at 12 months of age, as detected using the SDS-FRL technique in wild type and APP/PS1 mice. A, B In wild type, GIRK2 immunoparticles were mostly found along the extrasynaptic site (blue arrows) of axon terminals (at). Although few immunoparticles were detected within the active zone (az, purple overlay), recognized by the concave shape of the P-face and the accumulation of IMPs, many were detected at the border between the active zone and extrasynaptic sites. In these compartments, GIRK2 immunoparticles were observed forming clusters (blue ellipses/circles) and scattered (blue arrows) outside the clusters. C In APP/PS1, fewer GIRK2 immunoparticles forming clusters or scattered (blue arrows), were detected at extrasynaptic sites of axon terminals (at), at the edge of the active zone (az), or within the active zone (red overlay). D Quantitative analysis showing that the density of GIRK2 immunoparticles was significantly reduced in the APP/PS1 mice compared to age-matched wild type controls in the two presynaptic compartments analysed (Mann–Whitney test, *p < 0.05; ***p<0.001). Error bars indicate SEM. Scale bars: AD, 0.2 μm

GIRK2 differentially co-clusters with other GIRK channel subunits

It is well known that GIRK2 may assemble with other GIRK subunits to form heteromeric channels [2]. Thus, we assessed GIRK2 co-clustering with other GIRK subunits by double-labelling SDS-FRL experiments. First, we provided morphological insights into the GIRK1-GIRK2 association (Fig. 3A–C). Immunoparticles for GIRK2 co-clustered with those for GIRK1 in all neuronal compartments including the extrasynaptic plasma membrane of spines, dendritic shafts, and axon terminals (Fig. 3A–C). Next, we performed double labelling to investigate the spatial relationship between GIRK2 and GIRK3 (Fig. 3D–G). Although many immunoparticles for GIRK2 were distributed close to GIRK3 immunoparticles, most of them were not co-clustering in spines, dendritic shafts, or axon terminals (Fig. 3D–G).

Fig. 3figure 3

Differential co-clustering of GIRK2 with other GIRK subunits in CA1 pyramidal cells. Electron micrographs of the stratum radiatum of the hippocampal CA1 field showing double labelling for GIRK2 (5 nm)/GIRK1 (10 nm) and GIRK2 (5 nm)/GIRK3 (10 nm), as detected using the SDS-FRL technique. AC Immunoparticles for GIRK2 co-clustered with those for GIRK1 (green ellipses/circles) in dendritic spines (s), oblique dendrites (oDen), and axon terminals (at). DG Immunoparticles for GIRK2 (blue ellipses/circles) were segregated from GIRK3 clusters (black ellipses/circles) or immunoparticles (black arrows) in dendritic spines (s), oblique dendrites (oDen), and axon terminals (at). H Quantitative analysis using the SDS-FRL technique, showing the nearest neighbour distance (NND) between immunoparticles for GIRK2 to GIRK1 and immunoparticles for GIRK2 to GIRK3 in spines and dendritic shafts, as well as presynaptically in axon terminals and their active zones. The distances between immunoparticles for GIRK2 and GIRK1 are significantly shorter in all neuronal compartments than those between GIRK2 and GIRK3 (Mann–Whitney test, ****p < 0.0001). Thus, this analysis demonstrated a spatial association between GIRK2 and GIRK1, but not between GIRK2 and GIRK3. Scale bars: AG, 0.2 μm

To quantitatively assess the extent of the spatial relationship between GIRK channel subunits, we measured the nearest neighbour distances (NNDs) between the 5-nm gold particles (GIRK2) and the 10-nm immunoparticles (GIRK1 or GIRK3). The medians of the NNDs between GIRK1 and GIRK2 immunoparticles were 60.6 nm (interquartile range, 25.8–92.2 nm) in spines, 41.5 nm (interquartile range, 24.19–60.62 nm) in dendritic shafts, 45.7 nm (interquartile range, 27.2–76.7 nm) in extrasynaptic axon terminals, and 43.1 nm (interquartile range, 27.5–75.1 nm) in the active zone of axon terminals (Fig. 3H). The medians of the NNDs between GIRK3 and GIRK2 immunoparticles were 161.6 nm (interquartile range, 115.7–244.3 nm) in spines, 156.5 nm (interquartile range, 108.5–237.8 nm) in dendritic shafts, 162.2 nm (interquartile range, 123.3–221.9 nm) in extrasynaptic axon terminals, and 170.4 nm (interquartile range, 128.8–242.6 nm) in the active zone of axon terminals (Fig. 3I). These distances were significantly shorter in all neuronal compartments for GIRK2 and GIRK1 than those for GIRK2 and GIRK3 (Mann–Whitney test, ****p < 0.0001). Overall, these results suggest that GIRK1 but not GIRK3 subunits are specifically targeted to have a close association with GIRK2 channels both at postsynaptic and presynaptic compartments.

Reduction of GIRK1 density, but not GIRK3, in aged APP/PS1 mice

Since GIRK1 and GIRK3 are differentially associated to GIRK2, we next interrogated whether Aβ pathology also affects differently to the spatial organization of these subunits. Thus, the nanoscale organization of GIRK1 and GIRK3 in different compartments of CA1 pyramidal cells was assessed as described above. In wild type mice, immunoparticles for GIRK1 were mostly distributed at the extrasynaptic plasma membrane of spines and dendritic shafts of CA1 pyramidal cells, either forming clusters or scattered as single gold particles outside the clusters (Fig. 4A, B). GIRK1 immunoparticles were also detected along the extrasynaptic site of axon terminals or around the edge of the active zone of axon terminals, and less frequently within the active zone (Fig. 4C). Conversely, in APP/PS1 mice, few immunoparticles for GIRK1 were detected in clusters or scattered along the membrane surface of spines and dendritic shafts of CA1 pyramidal cells or at presynaptic sites (Fig. 4D–F). No labelling was observed on the E-face or on cross-fractures (Fig. 4A–F). Importantly, the quantitative analysis of the images revealed that GIRK1 density was significantly reduced in oblique dendrites (oDen) and spines in APP/PS1 mice (oDen = 9.94 ± 0.98 immunoparticles/μm2, n=35 dendrites; spines= 25.72 ± 1.82 immunoparticles/μm2, n=37 spines) compared to age-match wild type mice (oDen = 28.13 ± 2.79 immunoparticles/μm2, n=36 dendrites; spines= 52.74 ± 5.05 immunoparticles/μm2, n=24 spines) (Mann–Whitney test, ****p<0.0001) (Fig. 4G–F). At presynaptic sites, the density of GIRK1 was also significantly reduced at extrasynaptic sites and in the active zone in APP/PS1 mice (extra= 14.26 ± 3.16 immunoparticles/μm2, n = 24 extrasynaptic sites; AZ = 95.12 ± 13.41 immunoparticles/μm2, n= 22 active zones) compared to age-match wild type mice (extra= 36.03 ± 4.36 immunoparticles/μm2, n = 29 extrasynaptic sites; AZ = 261.30 ± 47.74 immunoparticles/μm2, n= 19 active zones) (Mann–Whitney test, ***p<0.001; ****p< 0.0001) (Fig. 4). The densities of immunoparticles for GIRK1 on the three compartments were significantly different (Mann–Whitney test, p < 0.001) from the background labelling determined on the surrounding E-face plasma membranes (0.91 ± 0.08 immunoparticles/μm2).

Fig. 4figure 4

Reduction of GIRK1 immunoparticles in post- and presynaptic compartments of APP/PS1 mice. Electron micrographs showing immunoparticles for GIRK1 in the stratum radiatum of the hippocampal CA1 field at 12 months of age, as detected using the SDS-FRL technique in wild type and APP/PS1 mice. AC In wild type, immunoparticles for GIRK1 were detected forming clusters (pink circles) or scattered (pink arrows) in dendritic spines (s) and oblique dendrites (oDen) of CA1 pyramidal cells, as well as presynaptically along the extrasynaptic site and in the active zone (az, pink overlay) of axon terminals (at). DF In APP/PS1, very low frequency of clusters or scattered (purple arrows) GIRK1 immunoparticles were observed in dendritic spines (s), oblique dendrites (oDen), or axon terminals (at). The E-face is free of any immunolabelling. G Quantitative analysis showing that the density of GIRK1 immunoparticles was significantly reduced in the APP/PS1 mice compared to age-matched wild type controls in the post- and presynaptic subcellular compartments analysed (Mann–Whitney test, ***p < 0.001; ****p<0.0001). Error bars indicate SEM. Scale bars: AF, 0.2 μm

In contrast to GIRK2 and GIRK1, fewer immunoparticles for GIRK3 were detected in postsynaptic and presynaptic compartments, thus showing mostly a scattered distribution as opposed to forming clusters, both in wild type and APP/PS1 mice (Fig. 5A–E). In addition, the subcellular distribution pattern and GIRK3 density observed in wild type (Fig. 5A–C) were similar to those found in APP/PS1 mice (Fig. 5D, E). Quantitative comparison of the GIRK3 densities along the neuronal surface of CA1 pyramidal cells in the stratum radiatum revealed two main findings: (1) a uniform density of GIRK3 immunoparticles in spines, dendritic shafts, and axon terminals and (2) similar low GIRK3 densities along in wild type and APP/PS1 mice (Fig. 5F). Indeed, our analysis demonstrated that GIRK3 density did not change in APP/PS1 mice (oDen = 5.11 ± 0.38 immunoparticles/μm2, n=36 dendrites; spines= 5.86 ± 0.41 immunoparticles/μm2, n=36 spines; extra = 5.18 ± 1.19 immunoparticles/μm2, n = 27 terminals; AZ= 4.29 ± 0.59 immunoparticles/μm2, n= 24 terminals) compared to age-match wild type mice (oDen = 6.12 ± 0.55 immunoparticles/μm2, n=36 dendrites; spines= 6.29 ± 0.24 immunoparticles/μm2, n=36 spines; extra= 6.24 ± 0.36 immunoparticles/μm2, n = 27 terminals; AZ = 4.63 ± 0.62 immunoparticles/μm2, n= 27 terminals) (Mann–Whitney test, p = 0.14 for spines, p = 0.35 for dendritic spines, p = 0.064 for extrasynaptic sites, p = 0.369 for active zone sites) (Fig. 5). These density values in all examined compartments were above the non-specific labelling determined on the surrounding E-face plasma membranes (background: 0.98 ± 0.09 immunoparticles/μm2; Mann–Whitney test, p < 0.01).

Fig. 5figure 5

Unaltered distribution of GIRK3 in CA1 pyramidal cells of APP/PS1 mice. Electron micrographs showing immunoparticles for GIRK3 in the stratum radiatum of the hippocampal CA1 field at 12 months of age, as detected using the SDS-FRL technique in wild type and APP/PS1 mice. AE Both in wild type and APP/PS1, immunoparticles for GIRK3 were detected mostly scattered (black arrows), or less frequently forming clusters (black circles), in dendritic spines (s) and oblique dendrites (oDen) of CA1 pyramidal cells, as well as presynaptically along the extrasynaptic site and active zone (az, pink overlay) of axon terminals (at). F Quantitative analysis showing that the density of GIRK3 immunoparticles was unaltered in wild type and APP/PS1 mice in the post- and presynaptic subcellular compartments analysed (Mann–Whitney test, p = 0.14 for spines, p = 0.35 for dendritic spines, p = 0.064 for extrasynaptic sites, p = 0.369 for active zone sites). Error bars indicate SEM. Scale bars: AF, 0.2 μm

In summary, SDS-FRL labelling for GIRK1 and GIRK3 subunits showed clear subunit- and compartment-specific differences. While GIRK1 displayed higher densities with differences along the dendritic axis of CA1 pyramidal cells, mirroring GIRK2 distribution, GIRK3 had lower density of immunolabelling and was uniform along the axo-dendritic axis.

Reduced interaction of GABAB and GIRK2 on the hippocampus of aged APP/PS1 mice

The functional and molecular coupling of GABAB receptors and GIRK channels in the hippocampus has been previously reported [16, 40]. Since a dysfunction on GABAB-GIRK signalling in AD has been postulated [9,10,11], we interrogated whether the molecular interaction would also be altered. To this end, we first demonstrated that the density of GABAB receptors was significantly reduced in spines and oblique dendrites (oDen) in APP/PS1 mice (s = 43.87 ± 7.83 immunoparticles/μm2, n=16 spines; oDen= 32.12 ± 2.32 immunoparticles/μm2, n=61 dendrites) compared to age-matched wild type mice (s = 94.39 ± 8.155 immunoparticles/μm2, n=32 spines; oDen = 86.25 ± 2.88 immunoparticles/μm2, n=22 dendrites) (Mann–Whitney test, p<0.001). The data confirmed that the membrane localization of GABAB receptors was altered in APP/PS1 mice, similarly to GIRK1 and GIRK2. Next, we assessed the GABAB-GIRK association on the hippocampus of aged APP/PS1 mice through double-labelling SDS-FRL and P-LISA. Due to the fact that our anti-GIRK1 and anti-GIRK3 antibodies were raised in the same species as our anti-GABAB1 antibody, these experiments were conducted only with the anti-GIRK2 antibody. Firstly, double-labelling SDS-FRL experiments in wild type animals revealed that GABAB1 co-clustered with GIRK2 along the extrasynaptic plasma membrane of dendritic spines (Fig. 6A, B). In dendritic shafts, clusters of GABAB1 immunoparticles appeared to be mostly segregated from those of GIRK2 (Fig. 6A). However, a low degree of co-clustering of the immunoparticles for the two proteins was also observed (Fig. 6A, C). At presynaptic sites, the channels and receptors were mainly co-clustering, both at extrasynaptic sites, around the edge of the active zone of axon terminals and within the active zone (Fig. 6D). In APP/PS1 mice, immunoparticles for GABAB1 and GIRK2 were mostly detected scattered along the membrane surface of spines and dendritic shafts of CA1 pyramidal cells and axon terminals (Fig. 6E–G). In addition, immunoparticles for GABAB1 were always segregated from those for GIRK2 along the extrasynaptic plasma membrane of spines, dendritic shafts, or axon terminals (Fig. 6E–G).

Fig. 6figure 6

Co-clustering of GABAB receptors and GIRK2 in the hippocampus of wild type and APP/PS1. Electron micrographs of the stratum radiatum of the hippocampal CA1 field at 12 months of age showing double immunogold labelling for GABAB1 (10 nm) and GIRK2 (5 nm) in pyramidal cells, as detected using the SDS-FRL technique in wild type and APP/PS1 mice. AD In wild type, immunoparticles for GABAB1 (10 nm) co-clustered with those for GIRK2 (5 nm) (green ellipses/circles) in dendritic spines (s). In oblique dendrites (oDen), double labelling revealed that many clusters (red ellipses/circles) and immunoparticles (red arrows) for GABAB1 (10 nm) were segregated, and also clusters (blue ellipses/circles) and immunoparticles (blue arrows) for GIRK2 (5 nm) could be found, although in some cases clusters of the two proteins (green ellipses/circles) were also detected. Presynaptically, immunoparticles for GABAB1 (10 nm) co-clustered (green ellipses/circles) with those for GIRK2 (5 nm) in axon terminals (at) and edge of active zone (az, pink overlay). EG In APP/PS1, immunoparticles for GABAB1 (10 nm, red arrows) were segregated from immunoparticles for GIRK2 (5 nm, blue arrows) in spines (s), oblique dendrites (oDen), or axon terminals (at). Scale bars: AG, 0.2 μm

To examine the extent of the spatial coupling between GABAB1 and GIRK2, the NNDs between immunogold particles for GIRK2 (5 nm) with immunogold particles for GABAB1 (10 nm) were measured in our double-labelled replicas in wild type and APP/PS1 mice. In dendritic spines, the medians of the NNDs between GIRK2 and GABAB1 particles were 51.5 nm (interquartile range, 14.3–196.3 nm) in wild type and 161.5 nm (interquartile range, 21.2–429.2 nm) in APP/PS1 mice (Fig. 7A). In dendritic shafts, the medians of the NNDs between GIRK2 and GABAB1 particles were 113.5 nm (interquartile range, 26.9–273.4 nm) in wild type and 261.9 nm (interquartile range, 101.7–518.6 nm) in APP/PS1 mice. These median values were significantly different (****p < 0.0001, Mann–Whitney U test) (Fig. 7A). We then conducted fitted simulations of GIRK2 immunoparticles and compared NNDs from real and simulated GIRK2 particles to real GABAB1 immunoparticles in dendritic shafts and spines (Fig. 7B). To quantify their extent of spatial relation, the NNDs between immunoparticles for GABAB1 and GIRK2 were compared with those between real GABAB1 and simulated GIRK2 particles in spines and dendritic shafts (Fig. 7B). We found significantly larger NNDs for the randomly distributed GIRK2 immunoparticles (spines, median, 118.7nm; oDen, median: 159.1 nm) compared to the real distributions (**p < 0.01, **** p < 0.0001, Mann–Whitney U test, Fig. 7B). Therefore, we found a significant association of GABAB1 with GIRK2 in spines and to a lesser extent in the oblique dendrites of the wild type mice while this association was not present in the APP/PS1 mice.

Fig. 7figure 7

Nanoscale organization of GABAB receptors and GIRK2 channels in the hippocampus of wild type and APP/PS1. A Nearest neighbour distances (NNDs) between the 10-nm gold particles (GABAB1) and the 5-nm immunoparticles (GIRK2) were measured in spines and oblique dendrites (oDen) using the GPDQ software. The distances (median values) between immunoparticles for GIRK2 and GABAB1 are shorter in the wild type than in APP/PS1 mice, both in spines and oblique dendrites (oDen) of CA1 pyramidal cells (****p < 0.0001, Mann–Whitney U test). B We compared NNDs from real and simulated GIRK2 particles to real GABAB1 immunoparticles in dendritic shafts and spines, showing significantly larger NNDs for the randomly distributed GIRK2 immunoparticles compared to the real distributions (**p < 0.01, **** p < 0.0001, Mann–Whitney U test)

Subsequently, to sustain the results obtained by quantitative SDS-FRL we performed P-LISA, a method able to identify protein-protein interactions in situ (at distances < 40 nm) [41]. Thus, GABAB1 and GIRK2 complexes were detected by P-LISA in the stratum radiatum on the hippocampus of wild type and APP/PS1 at 12 months as previously described [37]. P-LISA signal was observed as fluorescence dots representing interactions between GABAB1 and GIRK2 (Fig. 8A, B). In wild type, we found that the neuropile of the stratum radiatum was decorated with the P-LISA dots (Fig. 8A), whereas a consistent reduction of dots in APP/PS1 hippocampus was observed (Fig. 8B). When evaluating the P-LISA signal using quantitative approaches, the density of fluorescent P-LISA dots was significantly reduced in the hippocampus of APP/PS1 mice when compared to age-matched wild type mice (Fig. 8C) (p < 0.01). Overall, our data clearly support the idea that a GABAB1 and GIRK2 interaction occurs in the hippocampus and that this interaction is significantly downregulated in an animal model of AD.

Fig. 8figure 8

Detection of GABAB1-GIRK2 heterodimers in the hippocampus of wild type and APP/PS1. A, B Photomicrographs of nuclei staining (DAPI) and dual recognition in green of GABAB1 and GIRK2 with the proximity ligation in situ assay (P-LISA), in the stratum radiatum of CA1 area of WT and APP/PS1 mice. The interaction between GABAB1 and GIRK2 is observed by the green dots. C Quantification of P-LISA signals for GABAB1 and GIRK2 proximity confirmed the significant reduction in the density of dots/μm2 in APP/PS1 mice compared to age-match wild type (**p<0.01, t-test). Values correspond to the mean ± SEM. Scale bars: 20 μm

留言 (0)

沒有登入
gif