A method for estimating relative changes in the synaptic density in Drosophila central nervous system
Synapse density is an essential indicator of development and functioning of the central nervous system. It is estimated indirectly through the accumulation of pre and postsynaptic proteins in tissue sections. 3D reconstruction of the electron microscopic images in serial sections is one of the most definitive means of estimating the formation of active synapses in the brain. It is tedious and highly skill-dependent. Confocal imaging of whole mounts or thick sections of the brain provides a natural alternative for rapid gross estimation of the synapse density in large areas. The optical resolution and other deep-tissue imaging aberrations limit the quantitative scope of this technique.
Here we demonstrate a simple sample preparation method that could enhance the clarity of the confocal images of the neuropil regions of the ventral nerve cord of Drosophila larvae, providing a clear view of synapse distributions. We estimated the gross volume occupied by the synaptic junctions using 3D object counter plug-in of Fiji/ImageJ®. It gave us a proportional estimate of the number of synaptic junctions in the neuropil region. The method is corroborated by correlated super-resolution imaging analysis and through genetic perturbation of synaptogenesis in the larval brain.
The method provides a significant improvement in the relative estimate of region-specific synapse density in the central nervous system. Also, it reduced artifacts in the super-resolution images obtained using the stimulated emission depletion microscopy technique.
KeywordsSquash preparation Ventral nerve cord Fiji® Synaptic junctions STED Drosophila
Synapses are characterized by the presence of presynaptic active zones (AZs) and postsynaptic densities (PSDs) separated by a synaptic cleft . AZs are the vesicle docking and neurotransmitter release sites associated with an electron dense cytomatrix [2, 3, 4, 5, 6, 7, 8, 9]. The PSDs on the postsynaptic membrane is constituted by neurotransmitter receptors, channels, adhesion molecules and signaling components . The bulk of synaptogenesis occurs during early development, but synapses can constantly form in the adult brain as well [10, 11, 12, 13]. Development of behavior requires formation of a large number of new synapses and modification of existing ones, resulting in a compact organization of central nervous system (CNS). At both stages, embryonic and adult, activity-dependent refinement of synaptic connections takes place [14, 15, 16, 17, 18, 19]. This dynamic process is assisted by molecular remodeling of AZs and PSDs as studied in both vertebrate and invertebrate systems [20, 21, 22, 23, 24, 25]. Further, these structural changes are functionally associated with alterations in neurotransmitter release during synaptic plasticity [23, 26, 27]. However, most of these studies were carried out either at neuromuscular junctions (NMJs) or in cultured hippocampal neurons, which cannot be generalized. Molecular mechanisms underlying central synaptogenesis and synaptic plasticity during development at a global scale is still poorly understood.
A major limitation is to estimate the total synapse number in the CNS. In general, synapse estimation involves serial sectioning of brain samples and imaging under transmission electron microscope (TEM). Subsequently, a sampled estimate is established from 3D-reconstructed data of the entire synaptic field [28, 29, 30, 31, 32]. Synapses are counted in the serial electron microscopic reconstructions using unbiased stereological methods like optical dissector method [33, 34], and size frequency method . However, sophisticated instrumentation and stringent sample preparation make it expensive and invasive. Generating and aligning these serial electron micrographs and their analysis is a complex procedure involving highly skilled labor and expensive instrumentation. Also, sample preparation is prone to generate artifacts. Even after such rigorous process, one ends up with a sample survey estimate with small sampling frequency . Further, the low throughput of this technique makes it difficult for making gross comparisons amongst large volume samples.
A relatively more straightforward method is to label these synapses using fluorescent markers and obtain their optical sections using light microscopy to reconstruct the entire imaging field. The primary challenge in using conventional light microscopy is the diffraction-limited size of the active synaptic zone and high synaptic density in the CNS [37, 38, 39, 40, 41, 42]. It does not allow resolution of individual synapses due to strong background signal from out of focus light. New approaches to quantify synapses and resolve their nanoscopic organization are adopted with the advent of super-resolution microscopy (SRM). Recently, few studies in Drosophila using the stimulated emission depletion (STED) and stochastic optical reconstruction microscopy (STORM) techniques were able to reveal the synaptic ultrastructure with relatively less tissue invasion [27, 37, 38, 39, 40, 41, 42, 43, 44]. Despite its capacity to resolve at the nanoscale, it was most effective in resolving NMJs or in selectively marked neurons. In CNS, high tissue thickness, density of fluorescent signal, and autofluorescence reduce the signal to noise ratio. High tissue scattering and depth aberrations also introduce specific imaging artifacts. Furthermore, SRM requires fluorophores with efficient binding properties [45, 46], and expensive instrumentation.
A few studies have attempted to combine the advantages of both light and electron microscopy by applying correlative light and electron microscopy (CLEM) technique to brain tissue . Moreover, automation of tissue sectioning in block-face scanning EM (BF-SEM) is a step further to create a 3D reconstruction of entire tissue to reduce time and labor requirement . The disadvantage of this technique was in image analysis of the 3D reconstructed image to identify desired structures and automate their counting . Most of the existing reports of quick estimation of gross synapse number in the CNS are descriptions of the redistribution of presynaptic proteins within neurons and their accumulation in cell bodies using relative intensity estimates [48, 49, 50, 51]. For example, Dey et al.  calculated the overall volume occupied by the presynaptic markers in the neuropil region in ventral nerve cord of Drosophila to correlate the effects of altered Rab4 transport. Although it provided an indirect estimate, assuming that each synapse occupies nearly equal volume in the neuropil, it is unclear whether that would correlate to the number of synapses.
Here, we present a simple sample preparation technique useful for quick and gross assessment of synapses density in the central nervous system of Drosophila larvae with large area sampling. The squash preparation method preserves overall morphology and structural integrity of neurons in the ventral nerve cord of Drosophila. Automated morphometric analysis of confocal images of the preparations using Fiji® provided an estimate of a total number of synaptic AZs marked by Bruchpilot immunostaining within each neuromere. Bruchpilot is a key component of AZs in Drosophila which is orthologous to human ELKS/CAST family of proteins [53, 54]. It is required for tethering vesicles and clustering of Ca2+ channels at the active zones . It helps to establish the characteristic “T-bar” structure at the active zones  and has been extensively used as a bonafide marker for synapses in the NMJs as well as in the CNS of Drosophila [27, 38, 39, 40, 42, 43, 44, 53]. Our protocol provides a better clarity of synapses for morphometric analysis. Importantly, it offers a quick survey with large sampling at a relatively less labor investment.
Drosophila melanogaster adult flies were maintained in vials and bottles containing standard cornmeal media with 3:1 ratio of female to male. These flies were transferred to fresh media vials for egg laying at 25 °C. Eggs laid for the next 1 h were collected and kept at 25 °C for aging until 78 h after egg laying (h AEL).
Dissection of the larval ventral nerve cord
Larvae aged for a designated time after egg laying (h AEL) were taken out of vials by using a wet brush. They were transferred to a petri dish with a drop of water. The cleaned up larvae were transferred with a drop of 1× Phosphate Buffered Saline (PBS, pH 7.2, 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4) onto a second petri dish containing Sylgard silicone bed (Sylgard 184 Kit from Dow-Corning Inc. USA). The central nervous system (CNS) containing optic-lobes and the VNC was dissected from the larva and transferred onto a lysine coated slide with 100 µL PEM buffer (100 mM PIPES, 2 mM EGTA, 1 mM MgSO4, pH 6.95), incubated for 2 min, and further processed as described below.
Squash preparation of larval ventral nerve cord to resolve the synaptic contacts
Antibodies used to label synapse
Mouse monoclonal to Bruchpilot
Goat Anti-Mouse IgG1 Alexa Fluor 647 conjugated
Goat Anti-Mouse IgG-Abberior® STAR 635
Image acquisition and analysis
Optical slices were acquired for each neuromere hemisegment in the A3-A6 abdominal segments of the VNC using Olympus FV1200 Laser Scanning Confocal Microscope under identical acquisition conditions using a 40 × 1.3 NA objective, 3.6× zoom and a pixel resolution of 0.17 × 0.17 µm2. The acquisition parameters viz., laser power, PMT gain, scan speed, optical zoom, offset, step size, pinhole diameter, etc., were kept constant for each experimental data set. For confocal and STED comparison, images were collected using 93× glycerol 1.3 NA objective on Leica SP8 TCS STED 3X. All images were deconvolved with Huygens Professional version 16.10 (Scientific Volume Imaging, The Netherlands, http://svi.nl). Images were processed in Fiji® and analyzed using “3D object counter plugin”  of the Fiji® software as described further.
Statistical analysis and representation of the data were carried out using OriginPro® (http://www.originlab.com). All data are presented in the box and whisker plots. Box indicates the first and third quartiles. In the boxplots, the band and small square represent the median and mean, respectively. Pairwise comparisons were made using one-way ANOVA and Bonferroni’s test to test for the significance.
Squash preparation of the ventral nerve cord did not alter the morphology of tissue and distribution of synaptic proteins
Milder fixation, longer permeabilization and longer incubation with antibody gave optimum Bruchpilot staining
Next, we ascertained whether Bruchpilot antibody would label the synaptic contacts in VNCs after the squash preparations. For optimum Bruchpilot staining, we tried multiple combinations of fixation—altering the concentration of fixative, buffers and incubation times. We expressed UAS-GFP-ChAT in the cholinergic neurons using chaGal4 to compare the VNC morphology between the whole mount preparation (Fig. 1a and c) and the squash preparation (Fig. 1b and d). Initially, the brain was dissected in 1× PEM buffer and then incubated it in 4% PFA solution made in 1x PEM buffer for 10, 20, and 40 min before squashing. It was followed by a 10 min post-fixation in 4% PFA in 1× PEM buffer and three washes with 1× PBS, 1 min each. Squash preparations were then incubated with the primary antibody diluted in 0.1% PBTX for 10 min at room temperature followed by three washes in PBTX, 1 min each. The same step was repeated for labeling with the secondary antibody, and then the tissues were mounted in a drop of Vectashield® (Vector Laboratories Inc., USA) under an 18 × 18 mm2 coverslip of thickness 0.17 mm. This procedure retained the antigens but distorted the tissue morphology (Additional file 1: Figure S1B and C) and GFP-ChAT localization in neurons. The integrity of the tissue fell apart, and small gaps were observed in the tissue which increased with the pre-squash incubation time in fixative. Also, the intensity of the Bruchpilot staining was poor (Additional file 1: Figure S1E and F). In contrast, the morphology was better preserved (Additional file 1: Figure S1A) when squash was prepared using tissues without pre-fixation, although Bruchpilot staining was still poor (Additional file 1: Figure S1D).
Since prefixation before the squash preparation resulted in poor Bruchpilot staining and caused tissue distortions, we tried squash preparations without pre-fixation followed by snap freezing and postfixation for 10 min (Additional file 2: Figure S2D), 20 min (Additional file 2: Figure S2E) and 40 min (Additional file 2: Figure S2F). This time, the morphology was retained, but Bruchpilot staining was still poor and degraded further with longer pre- or post-fixation times (Additional file 2: Figure S2B, C, E, and F). To troubleshoot the problem, we tried various permeabilization treatments—0.3% PTX (Additional file 2: Figure S2G), 0.1% PTX (Additional file 2: Figure S2H) and 0.05% PTX (Additional file 2: Figure S2I) for 5 min, among which 0.3% PTX treatment permeabilized the tissue better (Additional file 2: Figure S2G). Still, it was difficult to visualize the antigen.
Estimation of synaptic contacts from squash preparation
Synaptic contacts were estimated from the centermost optical slice of each neuromere hemisegment, in which commissures were visible, using 3D object counter plugin of Fiji. The 3D object counter plugin of Fiji applies an auto-threshold-returned total number of contiguous voxel elements in the image field. The number of 3D objects and that of the contiguous voxels contributing to the 3D objects in an optical slice can be easily estimated using this algorithm (Additional file 3: Figure S3). For our purpose, a size filter was applied manually. It corresponds to the minimum volume of a diffraction-limited image, i.e., 0.09 μm2 and a maximum cut off at 6.0 μm2. We reasoned that the synaptic boutons, as previously estimated [27, 31, 38, 39, 40, 42], would occupy an area of 0.03–6 µm2 in central neuromeric slices. The confocal microscope would not allow clear resolution of the Bruchpilot punctae within a synapse or amongst several closely spaced synapses unless they are separated by 300 nm. Therefore, often multiple synapses would appear as a single 3D object in the image. Hence, the total number of such elements would vary widely depending on the squash condition. However, the total number of voxels contributed by all the 3D objects within a neuromeric volume would be consistent. Since regular optical microscopy does not resolve the synaptic junctions, this plugin only helps to estimate the gross area occupied by the synaptic contacts within a neuromere in terms of the number of voxels. We interpreted this volume as a gross estimate proportional to the number of synaptic junctions in the region. The upper limit of the single 3D object was arbitrarily chosen after observing several images.
The conjecture was further supported by the STED imaging which provided nearly 60 nm resolution. It suggested that the lower bound cut off is consistent with a single synaptic bulb composed of several Bruchpilot punctae, and upper limit included multiple adjoining synapses. However, the STED acquisition and image analysis were tedious and highly time-consuming. Therefore, the squash preparations provided a quick and reasonably refined estimate of the volume occupied by the synaptic contacts stained by the Bruchpilot in the neuropil region.
Super-resolution microscopy revealed synaptic contacts with better clarity in squash preparations
Squash preparation segregates coalesced entities into smaller discretely observable units
Rab4 activation reduces the number of synaptic contacts in the neuropil
Expression of the small GTPase, Rab4, is upregulated in patients with mild cognitive impairment and Alzheimer’s disease in basal cholinergic forebrain and has also been proposed to play a role in axon elongation in Xenopus [56, 57]. The Rab4-associated vesicles are shown to be transported by kinesin-2 in Drosophila and mammalian cells . Overexpression of the dominant-negative form of Rab4 (S22N, DN) increased the volume of the synaptic region in the ventral nerve cord. Whereas that of the constitutively-active form of Rab4 (Q67L, CA) significantly reduced the volume . The dominant negative form of Rab4 remains in the GDP-bound inactive state and does not allow its binding to the motor protein. It leads to a significant reduction in both recycling and degradation of vesicles. Also, GDP-bound form accumulates in cell bodies present in VNC cortex and reduces the localization of Rab4 to VNC neuropil. In contrast, the constitutively active form of Rab4 would keep it in the GTP-restricted form which will sequester kinesin-2, reducing the availability of the motor for the other cargos [52, 58]. The conjecture was consistent with the observation of reduced neuropil enrichment of choline acetyltransferase (ChAT) in the Rab4CA overexpression background . Both Rab4 and ChAT bind to the C-terminal tail domain of Kinesin-2α [52, 65].
Synapse formation and maturation involve enrichment of various specialized proteins, a variety of lipids and organelles at both the sides . Axonal transport ensures the continuous replenishment of different components at synaptic sites to facilitate efficient neurotransmission [24, 60]. Any defect in axonal transport machinery leads to synapse loss and neurodegenerative disorders [61, 62, 63]. Drosophila nervous system has been immensely exploited to delineate the molecular mechanism underlying synapse assembly, maintenance and plasticity. Though most of them were studied in NMJs, visual and olfactory system; they were limited to a few types of well defined neurons.
Numerous methods till date have been employed to count synapses in CNS in Drosophila. This includes EM and SRM, but there were many pitfalls in using these techniques [36, 41, 45, 46]. Further, EM was not effective in studying molecular mechanism underlying synaptogenesis at a global scale. Moreover, most of the axonal transport deficit phenotypes appeared similar in ultrastructure, so it was difficult to compare the differences among different genetic backgrounds. SRM offered several advantages over EM, but the major drawback was sample thickness and deep tissue imaging . The relatively simple method described above could provide a significant improvements in the data quality obtained from both the Super-resolution and confocal images.
Individual synaptic boutons cannot be resolved by conventional optical microscopy in the CNS due to their compact organization [39, 40, 42]. Therefore, no good assay is established yet for quick assessment of synapse number in whole CNS. Hence, results from neuromuscular studies were extrapolated to CNS, since it was assumed that central nervous system defects would be manifested in NMJs. However, that does not hold true in the case of axonal transport deficits. For example, kinesin-2 mutations show aberrant axonal transport of its cargoes viz. Choline acetyltransferase, Acetylcholinesterase and Rab4 in Drosophila sensory neurons [51, 52, 64, 65, 66]. However, they do not directly affect synaptic bouton formation in the NMJs , compelling us to study the direct impact and correlation of the synapse homeostasis in the CNS neuropil.
The combination of total neuropil volume estimate and voxel count in squash preparation of ventral nerve cord provides a gross sample estimate of synapse density in the neuromere in the larval ventral nerve cord of Drosophila. We showed that the estimate also reflects expected changes in the synaptic density in Rab4CA overexpression background. It is possible to extend the method for gross synapse count in the adult brain as well. Also, the method can be applied in diverse genetic backgrounds and for molecular screening of synaptic content. Therefore, it has a potential for wider applications outside the Drosophila brain.
In conclusion, the squash preparation method described above is likely to serve multiple purposes of gross synapse estimation to better resolution of antigenic distribution in the synaptic region using confocal and super-resolution techniques. It is suitable for application in the Drosophila larval ganglion and potentially applicable to other brain tissues.
KR and DR planned the experiments, DR and SD performed experiments, KR and DR wrote the manuscript. All authors read and approved the final manuscript.
We sincerely acknowledge the support provided by Prof. Richa Rikhy, and the Super Resolution Imaging Facility, IISER, Pune, with the STED microscopy; the Developmental Studies Hybridoma Bank (DSHB), Iowa University, for antibodies; and Drosophila Stock Center, Bloomington, Indiana, for fly stocks.
The authors declare that they have no competing interests.
Availability of data and materials
All raw data supporting the results described above will be deposited to the BioArxViv.
Consent for publication
Ethics approval and consent to participate
Experiments were conducted on Drosophila under the guidelines of TIFR Bio-Safety Committee (IBSC).
Intramural funding by TIFR, Department of Atomic Energy, Government of India, supported this work.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 4.Südhof TC. The synaptic vesicle cycle. Annu Rev Neurosci. 2004;27:509–47. https://doi.org/10.1146/annurev.neuro.26.041002.131412.CrossRefPubMedGoogle Scholar
- 12.Waites CL, Craig AM, Garner CC. Mechanisms of vertebrate synaptogenesis. Annu Rev Neurosci. 2005;28:251–74. https://doi.org/10.1146/annurev.neuro.27.070203.144336.CrossRefPubMedGoogle Scholar
- 34.Gundersen HJG, Bagger P, Bendtsen TF, Evans SM, Korbo L, Marcussen N, et al. The new stereological tools: disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. APMIS. 1988;96:857–81. https://doi.org/10.1111/j.1699-0463.1988.tb00954.x.CrossRefPubMedGoogle Scholar
- 52.Dey S, Banker G, Ray K, Dey S, Banker G, Ray K. Anterograde transport of Rab4-associated vesicles regulates synapse organization in drosophila article anterograde transport of Rab4-associated vesicles regulates synapse organization in Drosophila. Cell Reports. 2017;18:2452–63. https://doi.org/10.1016/j.celrep.2017.02.034.CrossRefPubMedPubMedCentralGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.