(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Combining nanobody labeling with STED microscopy reveals input-specific and layer-specific organization of neocortical synapses [1]

['Yeasmin Akter', 'Department Of Neuroscience', 'Rockefeller Neuroscience Institute', 'West Virginia University', 'Morgantown', 'West Virginia', 'United States Of America', 'Grace Jones', 'Grant J. Daskivich', 'Department Of Cell Biology']

Date: 2025-04 

The discovery of synaptic nanostructures revealed key insights into the molecular logic of synaptic function and plasticity. Yet, our understanding of how diverse synapses in the brain organize their nano-architecture remains elusive, largely due to the limitations of super-resolution imaging in complex brain tissue. Here, we characterized single-domain camelid nanobodies for the 3D quantitative multiplex imaging of synaptic nano-organization sing tau-STED nanoscopy in cryosections from the mouse primary somatosensory cortex. We focused on thalamocortical (TC) and corticocortical (CC) synapses along the apical-basal axis of layer five pyramidal neurons as models of functionally diverse glutamatergic synapses in the brain. Spines receiving TC input were larger than those receiving CC input in all layers examined. However, the nano-architecture of TC synapses varied with dendritic location. TC afferents on apical dendrites frequently contacted spines with multiple aligned PSD-95/Bassoon nanomodules of constant size. In contrast, TC spines on basal dendrites predominantly contained a single aligned nanomodule, with PSD-95 nanocluster sizes scaling proportionally with spine volume. The nano-organization of CC synapses did not change across cortical layers and resembled modular architecture defined in vitro. These findings highlight the nanoscale diversity of synaptic architecture in the brain, that is, shaped by both the source of afferent input and the subcellular localization of individual synaptic contacts.

Funding: Grants from the National Institutes of Health (P20GM109098) and Alzheimer’s Association (AARG-NTF-23-1150820) to M.H. supported this work. This material is based upon work supported by the National Science Foundation under Cooperative Agreement No. OIA-2242771 (M.H.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, NSF, or Alzheimer’s Association.

Copyright: © 2025 Akter et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Here, we characterized an approach for robust identification of synaptic nano-architecture by immunolabeling in brain cryosections. By combining nanobody labeling and 3D tau-STED imaging of L5 pyramidal neurons, we show that spines receiving TC input, but not those receiving CC input, exhibit variability in their modular nano-organization across cortical layers. These findings emphasize the importance of afferent input and subcellular localization in shaping synaptic nano-architecture in the brain.

Immunolabeling enables simultaneous imaging of multiple endogenous proteins, underscoring its potential to reveal the molecular complexity of synapses. However, labeling in PFA-fixed brain tissue can be challenging, especially for PSD molecules like PSD-95, NMDARs, and AMPARs, raising questions about its effectiveness in revealing synaptic nano-organization [ 12 , 39 – 42 ]. As affinity probes, single-domain nanobodies offer a promising solution to these barriers, enabling multiplexed, high-resolution imaging essential for detailed analysis of synaptic diversity [ 43 – 45 ].

Pyramidal neurons in the somatosensory cortex receive functionally diverse corticocortical (CC) and thalamocortical (TC) glutamatergic innervation. Although TC synapses are the minority of cortical synapses, they are highly efficient [ 26 – 31 ]. While some studies suggested that TC synapses onto spiny neurons in layer 4 (L4) are multi-quantal and several-fold stronger than unitary CC synaptic connections on the same neurons, others demonstrated that TC and CC synapses in L4 are indistinguishable electrophysiologically [ 32 – 34 ]. Instead, clustering of TC synapses in specific dendritic domains was suggested to synchronize feedforward activity [ 35 – 37 ]. More recently, work using expansion microscopy in the mouse visual cortex showed that spines of L2/3 pyramidal neurons receiving TC input are smaller and weaker than neighboring CC synapses [ 38 ]. These data suggest that mechanisms regulating TC synaptic function are complex and vary depending on the cell type and cortical location. Understanding the principles of nano-organization of TC and CC synapses across layers will shed light on processing of cortical information in the brain.

In dendritic spines, modular nano-architecture is thought to provide the flexibility needed for dynamic processes that underlie synaptic plasticity [ 20 – 23 ]. Key post-synaptic components—PSD-95, AMPARs, NMDARs precisely aligned to pre-synaptic Bassoon, Vesicular Glutamate Transporter 1 (VGluT1), Synaptotagmin-1 (SYT1)—conform to this scalability in vitro [ 24 , 25 ]. Whether these principles underlie the organization of diverse synapses in the brain will require 3D reconstruction of molecularly identified connections in their native environment.

Synapses in the central nervous system are small, highly specialized cell-cell junctions designed to rapidly transfer and process information. A characteristic feature of central synapses is their remarkable structural and functional variability [ 1 , 2 ]. Each pyramidal neuron receives thousands of synapses with unique signaling properties that form the basis for cortical computation and information storage in the brain [ 3 – 10 ]. At each synapse, diverse proteins with distinct lifetimes organize at the nanoscale in an ordered manner to build active zones and post-synaptic densities (PSDs) that endow synapses with exquisite regulation of synaptic function [ 11 – 19 ]. Despite their importance for brain function, plasticity, and aging, how functionally diverse synapses are organized at nanoscale and how the source of afferent input might shape synaptic nano-organization is not understood.

Results

PSD-95 antibody, but not nanobody, shows improved labeling in glyoxal-fixed tissue Recently, glyoxal fixation was shown to improve antibody labeling of synaptic proteins [41]. We therefore compared the efficiency of nanobodies and antibodies in glyoxal-fixed cryosections. We focused on PSD-95, VGluT1, and SYT1 for which both antibodies and nanobodies are available (S4A, S4B Fig). Consistent with published literature, glyoxal fixation significantly improved detection of PSD-95 using two different antibodies, which resulted in enhanced colocalization of PSD-95 with Bassoon clusters (S4C, S4G Fig). Notably, in both 4% PFA and glyoxal, nanobody reliably labeled PSD-95, and PSD-95 cluster densities identified by nanobody were significantly higher than for either of the two PSD-95 antibodies (Figs 1C and S4C). VGluT1 nanobody also showed consistent labeling in both fixatives (Fig 1D). In contrast to PFA, VGluT1 antibody-labeled fewer clusters than nanobody in glyoxal (S4D Fig). We obtained similar results for SYT1 (S4E Fig). Notably, in both fixatives, the colocalization of PSD-95 with Bassoon, VGluT1, and SYT1 was more robust when nanobodies were used to label these proteins (Figs 1G, 1H and S4G-S4I). In glyoxal, nanobody-labeled PSD-95 and SYT1 clusters appeared brighter than clusters labeled with antibodies (S4A, S4B Fig) [46,47]. Consistent with their sub-nanomolar affinities, nanobodies bound to expressed PSD-95 and SYT1 in HEK 293T cells over a broad range of concentrations, indicating that immunolabeling with nanobodies can be highly quantitative (S5 Fig). Altogether, our results suggest that staining of PSD-95, VGluT1, and SYT1 with nanobodies results in reliable, high-affinity labeling of putative synapses, in both PFA and glyoxal.

PSD-95 nanobody reliably identifies trans-synaptic nanodomains in situ using STED microscopy STED nanoscopy is well suited for probing nanoscale organization of synapses in brain tissue due to its near-infrared (775 nm) STED that attenuates light absorption and scattering [49]. Furthermore, tau-STED enables fast, simultaneous imaging of multiple fluorophores at reduced STED powers, minimizing bleaching without compromising resolution (Fig 2) [24]. We reasoned that combining tau-STED nanoscopy with nanobody labeling of cryosections might increase the accuracy of trans-synaptic NC (nanomodules) identification in situ. We labeled 6 µm Thy1-YFP-H brain cryosections with either PSD-95 nanobody or antibody and used Bassoon antibody to visualize active zones. We imaged YFP-labeled spine morphology using confocal mode, while PSD-95 and Bassoon were imaged using a 3D resolved two-color tau-STED method (Fig 3A). We segmented and analyzed X, Y, and Z-resolved Bassoon and PSD-95 NCs in an unbiased manner using the DiAna macro in ImageJ [24,50]. We 3D-rendered confocal and tau-STED images in Neurolucida 360 to aid in the visualization of PSD-95 and Bassoon in YFP-labeled spines. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. PSD-95 nanobody reliably identifies trans-synaptic nanodomains in situ using STED microscopy. (A) Two-channel tau-STED images and 3D reconstructions of PSD-95 (green) labeled by nanobody or antibody and Bassoon (red) in dendritic spines (YFP, dotted outline) in cryosections fixed in 4% PFA. Arrows indicate aligned PSD-95 and Bassoon. (B) Average center-to-center distances between Bassoon and the nearest PSD-95 nanoclusters (NCs) labeled with antibody (n = 14 images) or nanobody (n = 16, ***p < 0.0001, unpaired Student’s t test). (C) Cumulative distribution of nearest Bassoon – PSD-95 NC centers labeled with PSD-95 antibody (n = 68,806 clusters) or nanobody (n = 39,405 clusters, ***p < 0.0001, Kolmogorov–Smirnov test). (D) Average center-to-center distances between colocalized Bassoon and PSD-95 NCs labeled by PSD-95 antibody and nanobody (p = 0.9037, unpaired Student’s t test). (E) Cumulative frequency distributions of center-to-center distances between trans-synaptic Bassoon – PSD-95 NCs labeled by PSD-95 antibody (n = 12,992 clusters) and nanobody (n = 26,092, p = 0.4483, Kolmogorov–Smirnov test). (F) Percent colocalization between Bassoon and antibody- or nanobody-stained PSD-95 NCs (***p < 0.0001, unpaired Student’s t test). (G) Average Bassoon NC density per 25 × 25 µm image (p = 0.2518, unpaired Student’s t test). (H) Two-channel tau-STED images and 3D reconstruction of Bassoon (red) and PSD-95 (green) labeled by nanobody or two antibodies in dendritic spines (YFP, dotted outline) from cryosections fixed in glyoxal. Arrows indicate aligned PSD-95 and Bassoon. (I) Average center-to-center distances between Bassoon and the nearest PSD-95 NCs labeled with antibody (IgG1, n = 8; IgG2A, n = 6 images) or nanobody (n = 11, * p < 0.05, one-way ANOVA, Tukey’s post hoc). (J) Cumulative distribution of nearest Bassoon and PSD-95 NC centers, labeled with PSD-95 antibody (IgG1, n = 26,673; IgG2A, n = 32,533 clusters) or nanobody (n = 59,420, ***p < 0.0001, Kruskal–Wallis, Dunn’s post hoc). (K) Average trans-synaptic Bassoon and PSD-95 center-to-center distances (**p < 0.0005, one-way ANOVA, Tukey’s post hoc). (L) Cumulative frequency distributions of aligned Bassoon – PSD-95 center-to-center distances in PSD-95 antibody (IgG1, n = 20,656; IgG2A, n = 17,709 clusters) and nanobody (n = 43,476, ***p < 0.0001, Kruskal–Wallis, Dunn’s post hoc). (M) Percent colocalization between Bassoon and PSD-95 NCs (*p < 0.05, one-way ANOVA, Tukey’s post hoc). (N) Average Bassoon NC density (*p < 0.05, one-way ANOVA, Tukey’s post hoc). Bar graphs represent mean ± SEM. Data were collected from a minimum of three different neurons (dots) acquired from two biological replicates. Scale bar: 500 nm applies for A, H. The source data for panels B-G and I-N can be found in S2 Table. https://doi.org/10.1371/journal.pbio.3002649.g003 We quantified the degree of PSD-95 localization near active zones by comparing the nearest-neighbor distances between the centers of Bassoon and PSD-95 NCs to the trans-synaptic distance of colocalized PSD-95 and Bassoon (Fig 3B-3E). Bassoon’s nearest PSD-95 neighbor was, on average, within 200 nm when we used the PSD-95 nanobody. In contrast, the nearest Bassoon neighbors were significantly further (350 nm) when PSD-95 was labeled with the antibody. One explanation for this increased distance is that the nearest Bassoon/PSD-95 neighbors are located at different synapses rather than across the synaptic cleft. We next analyzed the center-to-center distances of only those PSD-95/Bassoon NCs that overlapped at least by one pixel, which we designated as trans-synaptic NCs (S8A Fig). Although the trans-synaptic distances between the centers of Bassoon and PSD-95 were ~140 nm in both conditions, they more closely approximated the nearest-neighbor distances between Bassoon and PSD-95 nanobody-labeled NCs (Fig 3D, 3E) [24]. Indeed, labeling with the PSD-95 antibody identified fewer trans-synaptic NCs compared to the nanobody (Fig 3F). Since Bassoon NC density was similar in both conditions, the large difference in nearest-neighbor distances between PSD-95 and Bassoon is likely due to the lower efficiency of PSD-95 antibody labeling in PFA-fixed brains (Fig 3G). Indeed, glyoxal fixation improved PSD-95 antibody colocalization with Bassoon, with nearest-neighbor distances more closely approximating trans-synaptic distances (Fig 3H-3L ). As a result, the proportion of trans-synaptic PSD-95/Bassoon pairs was indistinguishable from PSD-95 nanobody-labeled trans-synaptic nanodomains (Fig 3M, 3N). Notably, glyoxal fixation did not further improve PSD-95 nanobody labeling as indicated by the similar nearest-neighbor distances between PSD-95 and Bassoon NCs and the comparable percentage of trans-synaptic nanomodules in PFA-fixed brain sections. Thus, while glyoxal enhances trans-synaptic nanodomain detection for the PSD-95 antibody, the PSD-95 nanobody reliably labels PSD-95 opposite Bassoon NCs in both PFA- and glyoxal-fixed brains. Because of anisotropic X, Y, and Z resolution, nanomodule orientation may potentially affect the distances between pre- and post-synaptic proteins (S8 Fig). Therefore, we determined the distances between the centers of PSD-95 and Bassoon NCs aligned in either the XY or Z plane and found no significant differences (S8B Fig). Thus, center-to-center distances allow reliable identification of aligned NCs regardless of their orientation. Moreover, Z-projected clusters and spines were less abundant, potentially minimizing the impact of anisotropy in our data set (S8C-S8E Fig). Altogether, our tau-STED imaging of brain cryosections enables simultaneous visualization of multiple endogenous synaptic molecules to assign their nano-organization to spines of varying morphologies with high reliability.

VGluT2 + spines with multiple PSD-95/Bassoon nanomodules are abundant on apical dendrites of L5 pyramidal cells Layer 5 pyramidal cells also receive the POm TC input in their apical domains (L1-3) [31,51]. Therefore, we subjected apical dendrites of L5 pyramidal neurons to five-channel confocal and tau-STED imaging of TC and CC synapses (Fig 5A-5E and S1-S3 Videos). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. VGluT2 + spines with multiple PSD-95/Bassoon nanomodules are abundant on apical dendrites of L5 pyramidal cells. (A) Five-channel confocal/tau-STED images of YFP-labeled dendritic spines on apical dendrites of L5 pyramidal neurons in S1 of Thy-1-YFP-H mice. YFP (white), VGluT1 (red), and VGluT2 (green) were imaged in confocal mode (left), while PSD-95 (magenta) and Bassoon (cyan) were imaged in STED mode (right). (B-D) 3D reconstruction with Neurolucida 360 to classify VGluT1 + or VGluT2 + spines and localize PSD-95 and Bassoon nanoclusters (NCs). Scale bar (A-D), 5 µm. (E) Sub-stacks of individual dendritic spines receiving either VGluT1 (red) or VGluT2 (green) input (top row). Tau-STED-resolved PSD-95 (magenta) and Bassoon (cyan) NCs (arrows) in the same spines (middle row, dotted outlines). 3D-reconstructed spines with corresponding PSD-95 and Bassoon NCs (bottom row). High-contrast images are shown. Scale bars, 1 µm (tau-STED), 500 nm (3D render). (F) Quantification of the fraction of TC and CC synapses determined by VGluT1 and VGluT2 staining of 606 spines. (G) Quantification of the fraction of VGluT1 + (n = 507) and VGluT2 + (n = 99) spines with single or multiple aligned PSD-95 and Bassoon nanomodules. (H) Distributions VGluT1 + or VGluT2 + spines with the indicated numbers of PSD-95 or Bassoon NCs. (I) Linear correlation between spine size and PSD-95 NC numbers in VGluT1 + (R2 = 0.2629, slope = 3.122 ± 0.464) and VGluT2 + (R2 = 0.1776, slope = 1.833 ± 0.799) spines (**p = 0.0037, ANCOVA). (J) Comparison of spine size in VGluT1 + and VGluT2 + spines with one (VG1: n = 329 spines, VG2: n = 49 spines, p = 0.5722), two (VG1: n = 133 spines, VG2: n = 39 spines, p = 0.0931) and three PSD-95 NCs (VG1: n = 31 spines, VG2: n = 8 spines, p = 0.1368, unpaired Student’s t test). (K) Average sizes of all VGluT1 + and VGluT2 + spines regardless of the number of NCs (**p = 0.0014, unpaired Student’s t test). Bars represent the mean ± SEM acquired from spines (dots in J, K) on 37 apical dendritic segments (dots in H) of L5 neurons from two biological replicates. The source data for panels H-K can be found in S2 Table. https://doi.org/10.1371/journal.pbio.3002649.g005 Of the 606 dendritic spines we imaged on apical dendrites, only 16.3% were TC synapses, while 83.7% were classified as CC synapses (Fig 5F). The nano-organization of VGluT1 + spines in L1-3 was similar to spines in L5, with most spines displaying a single aligned PSD-95/Bassoon nanomodule. In contrast, fewer VGluT2 + spines (51%) contained a single aligned PSD-95/Bassoon, while 46% had multiple nanomodules (Fig 5G). Thus, the distribution of PSD-95 and Bassoon NCs in VGluT2 + spines shifted relative to VGluT1 + spines, resulting in nearly equal proportions of spines with one or two NCs (Fig 5H). VGluT2 + and VGluT1 + spines with three and four NCs were uncommon (<6%). It is important to note that we saw only one or two VGluT2 + spines with any given nano-organization per dendritic segment due to their low abundance in the cortex. Thus, even though proportionally there were more VGluT2 + spines with multiple aligned PSD-95/Bassoon nanomodules in L1-3, VGluT1 + spines with two or more nanomodules were still more abundant on apical dendrites of L5 cells. Given the differences in the pattern of PSD-95 and Bassoon nano-organization in VGluT1 + and VGluT2 + spines, we wondered whether the relationship between nanomodule numbers and spine size might also differ. Both VGluT1 + and VGluT2 + spine sizes showed a linear correlation with PSD-95 NC numbers (Fig 5I). However, the slope of this linear relationship was significantly lower for VGluT2 + spines compared to VGluT1 + spines (p = 0.0037, ANCOVA). We reasoned that this is due to fewer VGluT2 + spines with one PSD-95 NC and more VGluT2 + spines with two PSD-95 NCs compared to VGluT1 + synapses. To shed light on this skewed relationship, we compared sizes of VGluT1 + and VGluT2 + spines with one, two, and three PSD-95 NCs (Fig 5J). We found no significant differences in the average volumes of VGluT1 + and VGluT2 + spines within each category, indicating that apical VGluT1 + and VGluT2 + spines with equivalent numbers of PSD-95 NCs are similar with respect to their size. Yet, when comparing all spines in L1-3 regardless of their PSD-95 NC numbers, VGluT2 + spines were significantly larger on average than VGluT1 + spines (Fig 5K). Thus, CC and TC synapses on the apical dendrites of L5 neurons exhibit comparable organizational principles. However, a key distinction is the greater prevalence of VGluT2 + spines with multiple nanomodules, which are larger.

Proteome expansion confirms distinct populations of L1/3 dendritic spines with discrete VGluT1 and VGluT2 clusters We next undertook deep structured illumination microscopy (SIM) of Thy-1-YFP-H brain sections in which we expanded the proteome to verify the accurate assignment of clusters to individual YFP-labeled spines and to define their relationship to pre-synaptic input [56]. Following the expansion of brain sections, we labeled them with VGluT1 and VGluT2 antibodies along with appropriate secondary antibodies to visualize the vesicle clusters on individual spines (S9A Fig). Using 3D projections, we verified that VGluT1 and VGluT2 contacted specific YFP-labeled dendritic spines (S9B, S9C Fig, and S4 Video). Consistent with our confocal imaging, 73% of spines received VGluT1 input, while only 19% were contacted by VGluT2 (S9E Fig). Of the 364 spines, only seven were contacted by both VGluT1 and VGluT2, further supporting the accurate attribution of nanodomains to specific spines. Importantly, both VGluT1 and VGluT2 formed discrete clusters on the surface of YFP-labeled spines (S9D Fig). Similar to our STED analysis of aligned PSD-95/Bassoon nanomodules, most VGluT1 + spines (~69%) contained a single VGluT1 cluster. In contrast, nearly half of VGluT2 + spines (45%) had multiple (two and three) vesicle clusters (S9F Fig). These results complement our findings that many VGluT2 + spines on apical dendrites of L5 neurons form synapses with multiple nanodomains and suggest that these synaptic sites also contain multiple vesicular clusters.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002649

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/