Skip to main content
NCBI home page
eLife logoLink to eLife
. 2020 Dec 4;9:e56384. doi: 10.7554/eLife.56384

Ultrastructural comparison of dendritic spine morphology preserved with cryo and chemical fixation

Hiromi Tamada 1,2,3,4, Jerome Blanc 2, Natalya Korogod 2,5, Carl CH Petersen 1,✉, Graham W Knott 2,✉
Editors: Catherine Dulac6, Moritz Helmstaedter7
PMCID: PMC7748412  PMID: 33274717

Abstract

Previously, we showed that cryo fixation of adult mouse brain tissue gave a truer representation of brain ultrastructure in comparison with a standard chemical fixation method (Korogod et al., 2015). Extracellular space matched physiological measurements, there were larger numbers of docked vesicles and less glial coverage of synapses and blood capillaries. Here, using the same preservation approaches, we compared the morphology of dendritic spines. We show that the length of the spine and the volume of its head is unchanged; however, the spine neck width is thinner by more than 30% after cryo fixation. In addition, the weak correlation between spine neck width and head volume seen after chemical fixation was not present in cryo-fixed spines. Our data suggest that spine neck geometry is independent of the spine head volume, with cryo fixation showing enhanced spine head compartmentalization and a higher predicted electrical resistance between spine head and parent dendrite.

Research organism: Mouse

Introduction

In our previous paper (Korogod et al., 2015), we used a cryo fixation method to study the neuropil of the adult mouse cerebral cortex, comparing it with a standard chemical fixation approach that has been used for many years to preserve brain tissue for analysis with electron microscopy (e.g. Spacek and Harris, 1997; Knott et al., 2002). This study had been motivated by the work of Anton Van Harreveld (Van Harreveld et al., 1965; Van Harreveld and Malhotra, 1967; see Korogod et al., 2015 for detailed description) who had been the first to show how fresh, cryo frozen brain tissue showed appreciable amounts of extracellular space which was not present in tissue asphyxiated for 8 min. Korogod et al., 2015 elaborated on this result showing how cryo fixation preserved an extracellular space matching previous in vivo measurements, whereas chemically fixed tissue shows reduced extracellular space. Furthermore, synapse density and vesicle docking, as well as astrocytic processes close to synapses and around blood capillaries, were significantly different between the two fixation methods. This led us in this study to investigate the extent to which dendritic spines might be affected by chemical fixation.

The dendritic spine is a small dendritic protrusion that carries the majority of the excitatory synaptic connections in the adult brain (Colonnier, 1968) with a size and shape that is closely linked with its function (reviewed by: Holtmaat and Svoboda, 2009; Harris and Kater, 1994). The larger the spine head, the larger its synapse (Harris and Stevens, 1988; Arellano et al., 2007; Knott et al., 2006), the number of its receptors (Kharazia and Weinberg, 1999; Noguchi et al., 2005; Nusser et al., 1998), and its synaptic strength (Matsuzaki et al., 2001). However, the synapse on the spine head is separated from the parent dendrite by the spine neck. This is often thin, compartmentalizing biochemical signals and creating a potentially significant electrical resistance between synapse and dendrite. Changes in the morphology of spine head and neck have been seen after different forms of activity including LTP induction and tetanic stimulation (Lang et al., 2004; Matsuzaki et al., 2004; Harvey and Svoboda, 2007; Tønnesen et al., 2014; Fifková and Anderson, 1981). The understanding of dendritic spine morphology and function has come from many years of analysis using ultrastructural and in vivo analysis with electron and light microscopy. Up until now, however, all the ultrastructural data comes from chemically fixed samples. These show that the brightness of in vivo imaged spines closely correlates with their volume measured in serial electron microscopy images (Knott et al., 2006; Cane et al., 2014; Nägerl et al., 2007; El-Boustani et al., 2018). Although the resolution of light microscopy had limited the comparison of such features of spine length, and width from living neurons, the super resolution methods of STED and FRAP have been able to make these measurements in hippocampal spines in both slices and in vivo (Tønnesen et al., 2014; Pfeiffer et al., 2018). We investigated these parameters using serial section electron microscopy to compare the sizes of spine heads, their neck lengths and neck widths between cryo and chemically fixed tissue (Figure 1). We show that although cryo fixation reveals unchanged spine neck length and head volume, the spine neck is significantly narrower in cryo-fixed tissue compared to chemically fixed tissue, likely having an important impact upon consideration of the communication between a spine and its parent dendrite.

Figure 1. Serial section electron microscopy to compare the sizes of spine heads, their neck lengths and neck widths between cryo and chemically fixed tissue.

Dendritic spines in chemically fixed (A, B, C) and cryo-fixed (D, E, F) neuropil were reconstructed in 3D from serial electron micrographs. (G) An example of a 3D model showing cross-sections (purple) through the neck region. (H) These models were used to measure head volume, neck length, and average and minimum neck cross-sectional area. The neck cross-sectional area is an average of several neck cross-sectional areas (X1, X2,…, Xn) through the spine neck, and the neck diameter (Y1, Y2,…, Yn) was calculated from these values. (I) Box plot showing a difference in the coefficient of variation (%) of neck diameter for each spine. Plot shows the median, interquartile range with minimum and maximum values (p=0.0012, unpaired, Kolmogorov-Smirnov test).

Figure 1—source data 1. Coefficients of variation data used in Figure 1I.

Figure 1.

Figure 1—figure supplement 1. Comparison between FIBSEM and serial section TEM data showing no significant difference between the two imaging methods.

Figure 1—figure supplement 1.

(A) Spine neck diameter in cryo-fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.12 ± 0.07 µm; ssTEM, 0.13 ± 0.06 µm; p=0.06, K-S test). (B) Spine neck cross-sectional area in cryo-fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.014 ± 0.027 µm2; ssTEM, 0.016 ± 0.02 µm2; p=0.06, K-S test). (C) Spine head volume in cryo-fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.087 ± 0.06 µm3; ssTEM, 0.065 ± 0.062 µm3; p=0.07, K-S test). (D) Spine neck length in cryo-fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.88 ± 0.74 µm; ssTEM, 0.74 ± 0.48 µm; p=0.62, K-S test). (E) Spine neck diameter in chemically fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.19 ± 0.06 µm; ssTEM, 0.18 ± 0.05 µm; p=0.52, K-S test). (F) Spine neck cross-sectional area in chemically fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.03 ± 0.02 µm2; ssTEM, 0.027 ± 0.016 µm2; p=0.52, K-S test). (G) Spine head volume in chemically fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.089 ± 0.097 µm3; ssTEM, 0.072 ± 0.075 µm3; p=0.89, K-S test). (H) Spine neck length in chemically fixed samples was not significantly different between those measured in FIBSEM and ssTEM (FIBSEM, 0.75 ± 0.54 µm; ssTEM, 0.92 ± 0.59 µm; p=0.10, K-S test).
Figure 1—figure supplement 1—source data 1. Spine parameter data used in Figure 1—figure supplement 1.
Open in a new tab

Results

Spines were analyzed from 3D models constructed from serial images taken with either TEM (cryo fixation: n = 89 spines, N = 3 mice; chemical fixation: n = 75 spines, N = 1 mouse) or FIBSEM (cryo fixation: n = 27 spines, N = 2 mice; chemical fixation: n = 75 spines, N = 2 mice). No quantitative differences were found between the results obtained by TEM and FIBSEM, and the results were therefore pooled (Figure 1—figure supplement 1).

The tissue was prepared with identical protocols to those used in our previous paper (Korogod et al., 2015). As before, these revealed very little extracellular space with chemical fixation (Figure 1), while cryo fixation showed elements more dispersed with significant spaces separating them. The cryo-fixed membranes also had a distinct appearance from those exposed to chemical fixative. Cryo-fixed membranes were smoothly curving and lacked wrinkles typically seen after chemical fixation. Furthermore, the cryo-fixed tissue showed many thin processes, some of which were spine necks, which form the focus of the current study (Figure 1).

Quantifying spine geometry, we found that the mean spine neck cross-sectional area was significantly narrower in the cryo-fixed samples (cryo fixation, 0.0159 ± 0.0185 µm2; chemical fixation, 0.0284 ± 0.0197 µm2; unpaired Kolmogorov-Smirnov test, p<0.0001; Figure 2A). The same was seen for the minimum cross-sectional area (cryo fixation, 0.01 ± 0.016 µm2; chemical fixation, 0.017 ± 0.014 µm2; unpaired Kolmogorov-Smirnov test, p<0.0001; Figure 2B). The spine neck diameter was calculated from the measurements of spine neck cross-sectional area. The mean spine neck diameter was 128.0 ± 62.8 nm in cryo-fixed tissue, which was significantly smaller than the mean spine neck diameter of 182.4 ± 54.5 nm found for chemically fixed tissue (unpaired Kolmogorov-Smirnov test, p<0.0001; Figure 2C), and similarly for the minimum spine neck diameter (cryo fixation, 0.07 ± 0.057 µm; chemical fixation, 0.107 ± 0.037 µm; unpaired Kolmogorov-Smirnov test, p<0.0001; Figure 2D). Spine length measurements showed a large diversity in each group with no statistical difference between the two fixation conditions (cryo fixation, 0.770 ± 0.550 µm; chemical fixation 0.836 ± 0.569 µm; unpaired Kolmogorov-Smirnov test, p=0.53; Figure 2E). Measurements of the spine head volume also showed no difference between chemically fixed and cryo-fixed tissue (cryo fixation, 0.069 ± 0.062 µm3; chemical fixation, 0.081 ± 0.087 µm3; unpaired Kolmogorov-Smirnov test, p=0.65; Figure 2F).

Figure 2. Cryo-fixed dendritic spines show reduced neck widths.

(A) Distribution of neck cross-sectional area measurement in cryo-fixed (n = 116 spines, N = 5) and chemically fixed (n = 150, N = 3) cortex (p<0.0001, Kolmogorov-Smirnov). Curves show log-normal fits (R2 values of 0.92 for cryo-fixed spines; 0.94 for chemically fixed). Inset shows a box plot indicating median, interquartile range and minimum and maximum values. (B) The distribution of minimum neck cross-sectional area measurements in cryo-fixed and chemically fixed cortex (p<0.0001, Kolmogorov-Smirnov). Log-normal curves have R2 values of 0.98 for cryo-fixed spines and 0.98 for chemically fixed. (C) The distribution of neck diameter calculated from the cross-sectional area measurements in cryo-fixed and chemically fixed cortex (p<0.0001, Kolmogorov-Smirnov). Log-normal curves have R2 values of 0.95 for cryo-fixed spines and 0.96 for chemically fixed. (D) The distribution of minimum neck diameter in cryo-fixed and chemically fixed cortex (p<0.0001, Kolmogorov-Smirnov). Log-normal curves have R2 values of 0.90 for cryo-fixed spines and 0.89 for chemically fixed. (E) The distribution of neck length measurements in cryo-fixed and chemically fixed cortex (p=0.53, Kolmogorov-Smirnov). Log-normal curves have R2 values of 0.80 for cryo-fixed spines and 0.76 for chemically fixed. (F) Distribution of head volume measurement in cryo-fixed and chemically fixed cortex (p=0.65, Kolmogorov-Smirnov). Log-normal curves have R2 values of 0.89 for cryo-fixed spines and 0.86 for chemically fixed.

Figure 2—source data 1. Spine parameter data used in Figure 2.

Figure 2.

Figure 2—figure supplement 1. Dendritic spine head volume correlates with the size of the synapse in cryo-fixed neuropil.

Figure 2—figure supplement 1.

Plot of synapse area against spine head volume for 33 spines (R2 = 0.73).
Figure 2—figure supplement 1—source data 1. Spine parameter data used in Figure 2—figure supplement 1.
Open in a new tab

Previously, we found that the synapse size, or area, was not different comparing the two fixation conditions (Korogod et al., 2015). Here, we further report that synapse area correlates with the volume of the spine head in cryo-fixed tissue (R2 = 0.73, n = 33 spines) (Figure 2—figure supplement 1), similar to previous reports studying chemically fixed tissue (reviewed by Bourne and Harris, 2008).

We next examined correlations between spine head volume, spine neck diameter, and spine neck length (Figure 3). Similar to a previous study (Arellano et al., 2007), we found a weak correlation in the chemically fixed spines between head volume and neck diameter (r = 0.36; p<0.0001; non-parametric Spearman test). This was not observed, however, in the cryo-fixed spines (r = 0.14; p=0.12; non-parametric Spearman test). We also found no correlation between the head volume and neck length in either cryo-fixed or chemically fixed spines (cryo fixation, r = 0.124, p=0.18; chemical fixation, r = 0.044, p=0.59, non-parametric Spearman test) and also no correlation between the neck length and the neck diameter (cryo fixation, r = −0.020, p=0.83; chemical fixation, r = −0.155, p=0.059, non-parametric Spearman test).

Figure 3. Chemically fixed, but not cryo-fixed, spines show a correlation between neck width and head volume.

Figure 3.

Open in a new tab

(A) Correlation between neck diameter and neck length for cryo-fixed spines (Spearman r = −0.020, p=0.83, n = 116 spines, N = 5 mice) and chemically fixed spines (Spearman r = −0.155, p=0.059, n = 150, N = 3 mice). (B) Correlation between neck diameter and head volume; (cryo fixed; r = 0.145, p=0.121; chemically fixed; r = 0.358, p<0.0001, Spearman). (C) Correlation between head volume and neck length; cryo fixed; r = 0.124, p=0.184, chemically fixed; r = 0.0441, p=0.592.

Figure 3—source data 1. Spine parameter data used in Figure 3.

The correlation of spine head volume and spine neck diameter in chemically fixed tissue could be due to significant structural differences in the spine neck when comparing the larger spines with smaller spines. Previous EM analyses have shown that spines with larger heads tend to be those where membranous structures, such as endoplasmic reticulum (ER), are present (Spacek and Harris, 1997). If, by their presence, these structures enlarge the neck then this may have an influence on the correlation between head volume and neck thickness. We investigated this question by grouping dendritic spines (chemical fixation, n = 75; cryo fix, n = 98) according to whether or not membranes were present in the neck (Figure 4). We scored this as ER, and did not classify them further as cisterns or vesicles as unequivocal classification in the cryo-fixed material was not possible. The cryo preparation method, where there is less extraction of molecules results in less contrast between the membranes, cytoplasm and extracellular compartments. Also, despite the sub-nanometre resolution of TEM imaging, the section thickness of 50 nm gives an obscured view of the details within the spine necks (Figure 4A’, B’, C’). FIBSEM imaging generates images with less contrast at a lower resolution. This analysis showed that spine neck length and the spine neck diameter were not different if membranes were present or absent, neither in cryo-fixed tissue nor in chemically fixed tissue (neck length, chemical fixation p=0.95, cryo fixation p=0.99; spine neck diameter, chemical fixation p=0.08, cryo fixation p=0.81; Figure 4D,E). Therefore, the presence of intracellular compartments within the neck do not appear to determine its thickness. However, spines in which ER is present in the spine neck do have larger heads after both cryo and chemical fixation (chemical fixation p<0.0001, cryo fixation p=0.0074; Tukey’s multiple comparison test; Figure 4F), in agreement with a previous study of chemically fixed spines (Spacek and Harris, 1997).

Figure 4. Dendritic spine geometry and the presence of membranes in the spine neck.

Figure 4.

Open in a new tab

(A, B, C) show examples of cryo-fixed spines, and higher magnification views of their necks (A’, B’, C’). (D) Spine neck length and, (E) spine neck diameter were not significantly different between spines with (+ER) and those without (-ER) membranes in the neck (spine neck length: cryo-fixed p=1.0, chemical fixation p=0.95; spine diameter: cryo fixation p=0.81, chemical fixation, p=0.08; Tukey’s multiple comparison n = 98 cryo-fixed spines, n = 75 chemically fixed spines). (F) Spine head volume was larger in spines with membranes in the neck (cryo fixation p=0.007; chemical fixation p<0.0001; Tukey’s multiple comparison). Box plots showing median, interquartile range, minimum, and maximum values, and mean (cross). Each group is divided according to whether or not membranes were present in the spine neck (+ER, or -ER). Scale bar in A is 500 nm and in A’ 200 nm.

Figure 4—source data 1. Spine parameter data used in Figure 4.

Finally, we considered how the electrical resistance of the spines would change on the basis of the different morphological parameters measured (Figure 5). We used Ohm's law and the formula R = ρ × spine neck length/spine neck cross-sectional area, with a resistivity constant ρ = 109 Ω cm (Cartailler et al., 2018). The mean value of the resistance of spines in cryo-fixed cortex was 109.7 ± 113.4 MΩ (spine number n = 116, N = 5), and in chemically fixed cortex was 44.1 ± 37.8 MΩ (spine number n = 150, N = 3) showing a significant difference in their distributions (p<0.0001, unpaired, Kolmogorov-Smirnov test; Figure 5A). We found obvious correlations between resistance and spine neck cross-sectional area (cryo, r = −0.74, p<0.0001, n = 116 spines, N = 5 mice; chemical, r = −0.64, p<0.0001, n = 150 spines, N = 3 mice, correlation, non-parametric Spearman; Figure 5B).

Figure 5. Computed spine neck resistances are higher in cryo-fixed cortex compared to chemically fixed cortex.

Figure 5.

Open in a new tab

(A) Dendritic spine neck resistances were calculated with resistivity constant of 109 (Ω cm) for cryo-fixed and chemically fixed cortex (cryo-fixed spines n = 116 spines, N = 5 mice; chemically fixed spines n = 150 spines, N = 3 mice). Cryo-fixed spines have higher resistance, box plots showing median, range, and distribution of resistances (p<0.0001, unpaired, Kolmogorov-Smirnov test). (B) Correlation between neck cross-sectional area and resistance (cryo fixed, r = −0.74, p<0.0001; chemical fixation, r = −0.64, p<0.0001).

Figure 5—source data 1. Spine parameter data used in Figure 5.

Discussion

The aim of this study was to gain a better understanding of the native ultrastructure of dendritic spines and to investigate if chemical fixation alters the morphology of dendritic spines, a neuronal structure that for many years has been the focus of considerable research due to its significant role in neuronal connectivity and plasticity (Harris and Kater, 1994; Holtmaat and Svoboda, 2009). We found that the volume of spine heads and the length of spine necks are unaltered by chemical fixation, and only the size of the spine neck is changed, being 42% thicker in diameter in chemically fixed tissue compared to cryo-fixed tissue. The thinner spine necks observed in cryo-fixed tissue suggest a larger electrical resistance coupling spines to parent dendrites than previously indicated from analysis of chemically fixed tissue.

Several of the parameters we assessed here appear to show that spines remain largely unchanged between chemical and cryo fixation, in agreement with previously published work. Spine head volume in cryo-fixed tissue was 0.069 µm3, which was not different to the spine volume in chemically fixed tissue of 0.081 µm3 (Figure 2). Our spine volume measurements are similar to other studies in the same region of the mouse brain (0.06 ± 0.04 µm3, layer 4 [Rodriguez-Moreno et al., 2018]; 0.075 ± 0.005 µm3 - 0.079 ± 0.006 µm3, layer 1, adult and aged mice, layer 1 [Calì et al., 2018]). Our measurement of mean spine length after cryo-fixation of 0.77 ± 0.55 µm was not different from the length of 0.84 ± 0.57 µm in chemically fixed tissue (Figure 2), and is similar to other studies, for example rat CA1 hippocampus, 0.45 ± 0.29 µm (Harris and Stevens, 1989) rat cerebellum, 0.66 ± 0.32 µm (Harris and Stevens, 1988) and mouse layer 2/3 pyramidal cells, 0.66 ± 0.37 µm (Arellano et al., 2007). Despite the fact that chemical fixation causes important changes to the extracellular space and astrocytic arrangement (Korogod et al., 2015), the overall structure of dendritic spines appears to be resilient to the alterations that the aldehydes are causing.

On the other hand, our results suggest that spine neck thickness is affected by chemical fixation, with a mean diameter of 128 nm in cryo-fixed tissue compared to 182 nm in chemically fixed tissue (Figure 2). This finding raises the question as to what could be happening during the aldehyde preservation method that could explain the widening. Spines are highly dynamic structures, with their shape changing during development and also adulthood (Nimchinsky et al., 2002), and in concert with changes in synapse size (Sala and Segal, 2014; Knott et al., 2006). To explain dynamic spine morphology, we need to consider the structural components. Actin is present throughout spines, with branched and linear actin being in different proportions in the neck and head (Matus, 2000; Korobova and Svitkina, 2010). This cytoskeleton determines the size and shape of the spine, with synaptic activation capable of altering the quantities of filamentous actin, and changing spine shape (Honkura et al., 2008). Modeling of dendritic spines using the physical properties of different structural components suggests the presence and arrangement of different proteins, such as actin in the neck, maintain its diameter (Miermans et al., 2017). This is supported by evidence of a clear periodic arrangement of actin rings in the spine neck (Bär et al., 2016) as well as various experiments showing that disruption of the actin regulatory pathways lead to changes in the spine shape (Basu and Lamprecht, 2018). However, preserving the integrity of actin is difficult, due to its sensitivity to aldehyde fixatives such as the ones used here (Boyles et al., 1985). Therefore, the shape of spines appears to be critically dependent on the arrangement of actin, which is dismantled on exposure to aldehydes, and this is likely to contribute to the differences in spine necks comparing cryo-fixed and chemically fixed tissue.

Some spine necks have also been shown to be rich in other proteins, such as the GTPases from the septin family (Tada et al., 2007), thought to be involved in the restriction of movement of certain membrane proteins out of the spine head (Ewers et al., 2014). A differential protein concentration within this narrowed part of the spine could exert on oncotic pressure pulling water into this region during chemical fixation.

We cannot exclude the possibility that chemical fixation leaves the spine neck unchanged with the differences measured occurring during the process of tissue extraction, before cryo fixation. However, we think this unlikely, as although the effects of hypoxia on dendrites have been investigated - showing changes such as swelling, the formation of filopodia, and the elongation of immature spines - these were seen after 15 min (Jourdain et al., 2002; Segura et al., 2016). Our workflow vitrifies samples within 90 s of the heart stopping and within the first 30 s the tissue had been extracted and placed on a cooled surface. This method showed that in vivo levels of the extracellular space were preserved (Korogod et al., 2015) and there was no evidence of the compact neuropil, showing little extracellular space that is a feature of asphyxiated tissue as described by Van Harreveld and Malhotra, 1967.

Our data support the observation from light microscopy showing that the morphology of the spine neck is independent of the spine head volume (Araya et al., 2014; Tønnesen et al., 2014). Previous electron microscopy analysis had suggested a correlation, although only a weak one, between these parameters (Arellano et al., 2007; Bartol et al., 2015), which we also show here with chemical fixation. However, cryo-fixed spines, presumed to be in their native state, show no correlation. Spine neck thickness, therefore, appears to be independent of the spine head volume, and therefore synapse size.

Functionally, the smaller spine widths observed in cryo-fixed tissue without change in spine neck length suggest that spine neck resistances might be higher than that implied by previous studies of chemically fixed tissue. Assuming cytosolic resistivity of 109 Ω cm (Cartailler et al., 2018), we estimate a mean spine neck resistance of 110 MΩ ranging from 2 MΩ to 595 MΩ in cryo-fixed tissue compared to 44 MΩ ranging from 1 MΩ to 206 MΩ in chemically fixed tissue. It is important to note that these are likely lower bound estimates, since the presence of intracellular organelles, and large macromolecular complexes, in the spine neck will further increase spine neck resistance. Our results are consistent with estimates from physiological analyses (e.g. Bloodgood and Sabatini, 2005; Grunditz et al., 2008) as well as morphological analyses in vivo using STED microscopy of hippocampal dendrites in CA1 showing neck widths of 147 nm (Pfeiffer et al., 2018). The same imaging approach on cultured hippocampal slices gives similar results and estimated that half of spines had a resistance of greater than 56 MΩ. Here, the median value is 80 MΩ for cryo fixed, and 34 MΩ for chemically fixed spines.

At the upper end of the range of spine neck resistance, there could be a considerable impact upon spine voltage dynamics during synaptic signaling, with a 1 GΩ spine neck resistance causing a 1 mV drop in potential from spine to parent dendrite for each 1 pA of synaptic current. Given that synaptic currents are typically considered to be in the range of tens of pA, our data are consistent with tens of mV difference between spine head and dendrite for the smallest spine neck diameters. For these spines, regulation of spine neck diameter could therefore contribute to functional synaptic plasticity as previously proposed (Bloodgood and Sabatini, 2005; Grunditz et al., 2008). Our data showing smaller spine neck width in cryo-fixed tissue therefore suggest that spine neck resistance may play a more important role in synaptic signaling compared to estimates based on chemically fixed tissue.

These results, and those from our previous paper (Korogod et al., 2015), indicate how cryo fixation gives a truer representation of neuronal structure, at least at the level of dendritic spines. In future studies, it will be of great interest to examine axonal structure comparing cryo and chemical fixation. However, achieving a reliable preservation of neuronal tissue in its native state, and in significant volumes, is not possible with freezing due to the limited thickness to which vitrification can occur in samples with such high water content. Therefore, chemical fixation is still the best method for the preservation of significant volumes of nervous tissues, and variations to this approach that address some of its drawbacks have been used, such as alterations to osmolarity to correct for extracellular space loss (Pallotto et al., 2015). Our cryo fixation also uses a low-temperature embedding method with the replacement of water with solvents containing heavy metal stains. Advances in focused ion beam microscopy at cryo temperatures have shown that these steps may be avoidable in the future, removing any possible structural changes they may cause (Schertel et al., 2013; Zachs et al., 2020).

Materials and methods

The sample preparation and imaging methods were identical to those used in our previous paper (Korogod et al., 2015) and described briefly below.

Preparation of cryo-fixed tissue

This study was performed in strict accordance with the rules issued by the Swiss Federal Veterinary Office, under authorization 1889 issued by the 'Service de la consommation et des affaires vétérinaires' of the Canton de Vaud, Switzerland. The animals were handled according to approved institutional guidelines and under the experimentation license 1889.3 (Swiss Federal Veterinary Office). Adult mice (male, C57BL/6J, 6–10 weeks old, N = 5 mice) were deeply anesthetized with isoflurane (experimentation license 1889.3; Swiss Federal Veterinary Office). After decapitation, the brain was rapidly removed from the skull, placed on the top surface of a closed, cooled glass Petri dish filled with ice. The tissue was not exposed to any solutions. Razor blades were then used to slice small (approximately 2 mm x 2 mm x 0.2 mm) pieces of cortex. This was achieved by using two razor blades, one in each hand, and drawing them against each other, in opposite directions, against the surface of the glass to slice the tissues into smaller and smaller pieces. Once a piece of a suitable size is cut, it is then placed inside the 3 mm diameter and 200 µm cavity of an aluminium sample holder followed by a small drop of 1-hexadecene to remove all the air from around the piece of tissue. This was then inserted into the high-pressure freezer (HPM 100, Leica Microsystems). This entire procedure was completed in less than 90 s from the moment of decapitation.

The frozen samples, inside the aluminium carriers, were transferred to a low-temperature embedding device (ASF2; Leica Microsystems) and at −90°C the frozen pieces of tissue were transferred into a solution of 0.1% tannic acid in acetone. They were then left for 24 hr at −90°C, followed by 7 hr in 2% osmium tetroxide in acetone at −90°C. The temperature was then raised to −20°C during 14 hr, and left at this temperature for 16 hr, and then raised to 4°C during 2.4 hr and left for a further 1 hr. Next, the liquid was replaced with pure acetone and rinsed at 4°C. Finally, the tissue samples were mixed with increasing concentrations of epoxy resin (Durcupan) and left overnight at room temperature in 90% resin before 100% resin was added the next day for 6 hr, and then cured at 60° C for 24 hr.

Preparation of chemically fixed tissue

Mice (male, C57BL/6J, 6–10 weeks old, N = 3 mice) were anesthetized with an overdose of isoflurane and transcardially perfused with 50 ml of 2.5% glutaraldehyde and 2.0% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). After 2 hr, the brain was removed and vibratome sliced at a thickness of 80 µm. Slices were then washed thoroughly with cacodylate buffer (0.1 M, pH 7.4), post-fixed for 40 min in 1.0% osmium tetroxide with 1.5% potassium ferrocyanide, and then 40 min in 1.0% osmium tetroxide alone. They were finally stained for 30 min in 1% uranyl acetate in water before being dehydrated through increasing concentrations of alcohol and then embedded in Durcupan ACM (Fluka, Switzerland) resin. The sections were then placed between glass microscope slides coated with mold releasing agent (Glorex, Switzerland) and left to harden for 24 hr in a 65°C oven.

Electron microscopy imaging

Serial, ultrathin sections were cut at 50 nm thickness, and collected on copper grids, contrasted with lead citrate and uranyl acetate, and imaged with a transmission electron microscope (TEM) operating at 80 kV (Tecnai Spirit, FEI Company with Eagle CCD camera) with a pixel size of 2.3 nm. Additionally, focused ion beam scanning electron microscopy (FIBSEM) imaging was carried out using the same imaging parameters as described previously (Korogod et al., 2015). Briefly, this used an imaging voltage of 1.5 kV, and milling voltage of 30 kV and milling current of 80 pA. The pixel size was 5 nm.

The chemically fixed spines were sampled from three mice, two imaged with FIBSEM, and one with serial section TEM. The volumes imaged were 125 µm3, 125 µm3, and 504 µm3, respectively. The cryo-fixed spines were sampled from five mice, two imaged with FIBSEM, and three with serial section TEM. The volumes imaged were 125 µm3, 1112 µm3, 504 µm3, 450 µm3, 446 µm3, and 446 µm3.

Image processing, analysis, 3D reconstruction, and morphometric measurements

Image series from both the TEM and FIBSEM were registered using the alignment functions in the TrakEM2 plugin of FIJI (Cardona et al., 2012, http://fiji.sc). Spines and parts of the dendrite were segmented with the same software using the arealist function. All spines in each of the imaged volumes were reconstructed if their entire structure was clearly visible in all the serial images. The models were then exported as. obj files into the Blender 3D modeling software (Blender; https://www.blender.org) in which the NeuroMorph tools plugin (Jorstad et al., 2015; Jorstad et al., 2018; https://neuromorph.epfl.ch) was used to measure the spine neck cross-sectional area (minimum and average), spine neck diameter (minimum and average), spine neck length, and spine head volume (Figure 1).

To generate the average and minimum cross-sectional area of the spine neck, we first generated a centerline through the axis in the spine, using the software Vascular Modeling Toolkit (VMTK.org). This centerline is then used by the ‘Centreline Processing’ tool in the NeuroMorph addons and produces a cross-section of the spine neck at regular, defined intervals of approximately 150 nm. We defined the boundary between the spine neck and its head and the parent dendrite as the region where the spine neck expands significantly. As this measure could be considered somewhat arbitrary, we also measured the minimum spine neck thickness. The coefficient of variation of the spine neck width along the spine length was not significantly different between the two groups (cryo fixation, 21 ± 14%; chemical fixation, 15 ± 7%; unpaired Kolmogorov-Smirnov test, p=0.0012; Figure 1I). Therefore, there was some variation in spine neck width along their length. For this reason, both the mean and the minimum values of cross-sectional area and diameter are presented.

Statistical analysis

All statistical analyses and plots were generated using Graph Pad Prism software. The statistical comparisons of chemically fixed vs cryo-fixed spine neck diameter, cross-section area, length and head volume were performed with unpaired, Kolmogorov-Smirnov tests. Correlation analyses between the quantified parameters were performed with non-parametric Spearman tests.

Acknowledgements

We thank Lucie Navratilova, at the Center of Electron Microscopy EPFL, for help with the FIBSEM imaging; Catherine Maclachlan, Stéphanie Clerc-Rosset and Marie Crosier at BioEM for help with sample preparation. Funding of this work was supported by Young Researchers Exchange Programme between Japan and Switzerland (Japanese-Swiss Science and Technology Programme) (HT), JSPS KAKENHI grant #JP17KK0191 (HT), and Swiss National Science Foundation grants: #31003A_182010 (CCHP), #31003A_170082 (GWK) and CRSII3_154453 (CCHP and GWK).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Carl CH Petersen, Email: carl.petersen@epfl.ch.

Graham W Knott, Email: graham.knott@epfl.ch.

Catherine Dulac, Harvard University, United States.

Moritz Helmstaedter, Max Planck Institute for Brain Research, Germany.

Funding Information

This paper was supported by the following grants:

  • Swiss National Science Foundation 31003A_182010 to Carl CH Petersen.

  • Swiss National Science Foundation 31003A_170082 to Graham William Knott.

  • Japanese Society for the Promotion of Science JP17KK0191 to Hiromi Tamada.

  • Swiss National Science Foundation CRSII3_154453 to Carl CH Petersen, Graham William Knott.

Additional information

Competing interests

No competing interests declared.

Author contributions

Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft.

Formal analysis, Investigation.

Conceptualization, Methodology, Writing - review and editing.

Conceptualization, Resources, Funding acquisition, Validation, Writing - review and editing.

Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing.

Ethics

Animal experimentation: This study was performed in strict accordance with the rules issued by the Swiss Federal Veterinary Office, under authorization 1889 issued by the 'Service de la consommation et des affaires vétérinaires' of the Canton de Vaud, Switzerland. The animals were handled according to approved institutional guidelines and under the experimentation license 1889.3 (Swiss Federal Veterinary Office).

Additional files

Transparent reporting form

Data availability

All data generated during this study are included in the manuscript and the supporting files. Source data files are provided for all results. These are: Figures 1, 2, 3, 4 and 5 and Figure supplements for Figure 1 and 2.

References

  1. Araya R, Vogels TP, Yuste R. Activity-dependent dendritic spine neck changes are correlated with synaptic strength. PNAS. 2014;111:E2895–E2904. doi: 10.1073/pnas.1321869111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Arellano JI, Benavides-Piccione R, Defelipe J, Yuste R. Ultrastructure of dendritic spines: correlation between synaptic and spine morphologies. Frontiers in Neuroscience. 2007;1:131–143. doi: 10.3389/neuro.01.1.1.010.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bär J, Kobler O, van Bommel B, Mikhaylova M. Periodic F-actin structures shape the neck of dendritic spines. Scientific Reports. 2016;6:37136. doi: 10.1038/srep37136. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bartol TM, Bromer C, Kinney J, Chirillo MA, Bourne JN, Harris KM, Sejnowski TJ. Nanoconnectomic upper bound on the variability of synaptic plasticity. eLife. 2015;4:e10778. doi: 10.7554/eLife.10778. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Basu S, Lamprecht R. The role of actin cytoskeleton in dendritic spines in the maintenance of Long-Term memory. Frontiers in Molecular Neuroscience. 2018;11:143. doi: 10.3389/fnmol.2018.00143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bloodgood BL, Sabatini BL. Neuronal activity regulates diffusion across the neck of dendritic spines. Science. 2005;310:866–869. doi: 10.1126/science.1114816. [DOI] [PubMed] [Google Scholar]
  7. Bourne JN, Harris KM. Balancing structure and function at hippocampal dendritic spines. Annual Review of Neuroscience. 2008;31:47–67. doi: 10.1146/annurev.neuro.31.060407.125646. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Boyles J, Anderson L, Hutcherson P. A new fixative for the preservation of actin filaments: fixation of pure actin filament pellets. Journal of Histochemistry & Cytochemistry. 1985;33:1116–1128. doi: 10.1177/33.11.3902963. [DOI] [PubMed] [Google Scholar]
  9. Calì C, Wawrzyniak M, Becker C, Maco B, Cantoni M, Jorstad A, Nigro B, Grillo F, De Paola V, Fua P, Knott GW. The effects of aging on neuropil structure in mouse somatosensory cortex-A 3D electron microscopy analysis of layer 1. PLOS ONE. 2018;13:e0198131. doi: 10.1371/journal.pone.0198131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cane M, Maco B, Knott G, Holtmaat A. The relationship between PSD-95 clustering and spine stability in vivo. Journal of Neuroscience. 2014;34:2075–2086. doi: 10.1523/JNEUROSCI.3353-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cardona A, Saalfeld S, Schindelin J, Arganda-Carreras I, Preibisch S, Longair M, Tomancak P, Hartenstein V, Douglas RJ. TrakEM2 software for neural circuit reconstruction. PLOS ONE. 2012;7:e38011. doi: 10.1371/journal.pone.0038011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cartailler J, Kwon T, Yuste R, Holcman D. Deconvolution of voltage sensor time series and Electro-diffusion modeling reveal the role of spine geometry in controlling synaptic strength. Neuron. 2018;97:1126–1136. doi: 10.1016/j.neuron.2018.01.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Colonnier M. Synaptic patterns on different cell types in the different laminae of the cat visual cortex. an electron microscope study. Brain Research. 1968;9:268–287. doi: 10.1016/0006-8993(68)90234-5. [DOI] [PubMed] [Google Scholar]
  14. El-Boustani S, Ip JPK, Breton-Provencher V, Knott GW, Okuno H, Bito H, Sur M. Locally coordinated synaptic plasticity of visual cortex neurons in vivo. Science. 2018;360:1349–1354. doi: 10.1126/science.aao0862. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Ewers H, Tada T, Petersen JD, Racz B, Sheng M, Choquet D. A Septin-Dependent diffusion barrier at dendritic spine necks. PLOS ONE. 2014;9:e113916. doi: 10.1371/journal.pone.0113916. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fifková E, Anderson CL. Stimulation-induced changes in dimensions of stalks of dendritic spines in the dentate molecular layer. Experimental Neurology. 1981;74:621–627. doi: 10.1016/0014-4886(81)90197-7. [DOI] [PubMed] [Google Scholar]
  17. Grunditz A, Holbro N, Tian L, Zuo Y, Oertner TG. Spine neck plasticity controls postsynaptic calcium signals through electrical compartmentalization. Journal of Neuroscience. 2008;28:13457–13466. doi: 10.1523/JNEUROSCI.2702-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Harris KM, Kater SB. Dendritic spines: cellular specializations imparting both stability and flexibility to synaptic function. Annual Review of Neuroscience. 1994;17:341–371. doi: 10.1146/annurev.ne.17.030194.002013. [DOI] [PubMed] [Google Scholar]
  19. Harris KM, Stevens JK. Dendritic spines of rat cerebellar purkinje cells: serial electron microscopy with reference to their biophysical characteristics. The Journal of Neuroscience. 1988;8:4455–4469. doi: 10.1523/JNEUROSCI.08-12-04455.1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Harris KM, Stevens JK. Dendritic spines of CA 1 pyramidal cells in the rat Hippocampus: serial electron microscopy with reference to their biophysical characteristics. The Journal of Neuroscience. 1989;9:2982–2997. doi: 10.1523/JNEUROSCI.09-08-02982.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Harvey CD, Svoboda K. Locally dynamic synaptic learning rules in pyramidal neuron dendrites. Nature. 2007;450:1195–1200. doi: 10.1038/nature06416. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Holtmaat A, Svoboda K. Experience-dependent structural synaptic plasticity in the mammalian brain. Nature Reviews Neuroscience. 2009;10:647–658. doi: 10.1038/nrn2699. [DOI] [PubMed] [Google Scholar]
  23. Honkura N, Matsuzaki M, Noguchi J, Ellis-Davies GC, Kasai H. The subspine organization of actin fibers regulates the structure and plasticity of dendritic spines. Neuron. 2008;57:719–729. doi: 10.1016/j.neuron.2008.01.013. [DOI] [PubMed] [Google Scholar]
  24. Jorstad A, Nigro B, Cali C, Wawrzyniak M, Fua P, Knott G. NeuroMorph: a toolset for the morphometric analysis and visualization of 3D models derived from electron microscopy image stacks. Neuroinformatics. 2015;13:83–92. doi: 10.1007/s12021-014-9242-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Jorstad A, Blanc J, Knott G. NeuroMorph: a software toolset for 3D analysis of neurite morphology and connectivity. Frontiers in Neuroanatomy. 2018;12:59. doi: 10.3389/fnana.2018.00059. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Jourdain P, Nikonenko I, Alberi S, Muller D. Remodeling of hippocampal synaptic networks by a brief anoxia-hypoglycemia. The Journal of Neuroscience. 2002;22:3108–3116. doi: 10.1523/JNEUROSCI.22-08-03108.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kharazia VN, Weinberg RJ. Immunogold localization of AMPA and NMDA receptors in somatic sensory cortex of albino rat. The Journal of Comparative Neurology. 1999;412:292–302. doi: 10.1002/(SICI)1096-9861(19990920)412:2&#x0003c;292::AID-CNE8&#x0003e;3.0.CO;2-G. [DOI] [PubMed] [Google Scholar]
  28. Knott GW, Quairiaux C, Genoud C, Welker E. Formation of dendritic spines with GABAergic synapses induced by whisker stimulation in adult mice. Neuron. 2002;34:265–273. doi: 10.1016/S0896-6273(02)00663-3. [DOI] [PubMed] [Google Scholar]
  29. Knott GW, Holtmaat A, Wilbrecht L, Welker E, Svoboda K. Spine growth precedes synapse formation in the adult neocortex in vivo. Nature Neuroscience. 2006;9:1117–1124. doi: 10.1038/nn1747. [DOI] [PubMed] [Google Scholar]
  30. Korobova F, Svitkina T. Molecular architecture of synaptic actin cytoskeleton in hippocampal neurons reveals a mechanism of dendritic spine morphogenesis. Molecular Biology of the Cell. 2010;21:165–176. doi: 10.1091/mbc.e09-07-0596. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Korogod N, Petersen CC, Knott GW. Ultrastructural analysis of adult mouse neocortex comparing aldehyde perfusion with cryo fixation. eLife. 2015;4:e05793. doi: 10.7554/eLife.05793. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Lang C, Barco A, Zablow L, Kandel ER, Siegelbaum SA, Zakharenko SS. Transient expansion of synaptically connected dendritic spines upon induction of hippocampal long-term potentiation. PNAS. 2004;101:16665–16670. doi: 10.1073/pnas.0407581101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Matsuzaki M, Ellis-Davies GC, Nemoto T, Miyashita Y, Iino M, Kasai H. Dendritic spine geometry is critical for AMPA receptor expression in hippocampal CA1 pyramidal neurons. Nature Neuroscience. 2001;4:1086–1092. doi: 10.1038/nn736. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Matsuzaki M, Honkura N, Ellis-Davies GC, Kasai H. Structural basis of long-term potentiation in single dendritic spines. Nature. 2004;429:761–766. doi: 10.1038/nature02617. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Matus A. Actin-based plasticity in dendritic spines. Science. 2000;290:754–758. doi: 10.1126/science.290.5492.754. [DOI] [PubMed] [Google Scholar]
  36. Miermans CA, Kusters RP, Hoogenraad CC, Storm C. Biophysical model of the role of actin remodeling on dendritic spine morphology. PLOS ONE. 2017;12:e0170113. doi: 10.1371/journal.pone.0170113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Nägerl UV, Köstinger G, Anderson JC, Martin KA, Bonhoeffer T. Protracted synaptogenesis after activity-dependent spinogenesis in hippocampal neurons. Journal of Neuroscience. 2007;27:8149–8156. doi: 10.1523/JNEUROSCI.0511-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Nimchinsky EA, Sabatini BL, Svoboda K. Structure and function of dendritic spines. Annual Review of Physiology. 2002;64:313–353. doi: 10.1146/annurev.physiol.64.081501.160008. [DOI] [PubMed] [Google Scholar]
  39. Noguchi J, Matsuzaki M, Ellis-Davies GC, Kasai H. Spine-neck geometry determines NMDA receptor-dependent Ca2+ signaling in dendrites. Neuron. 2005;46:609–622. doi: 10.1016/j.neuron.2005.03.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Nusser Z, Lujan R, Laube G, Roberts JD, Molnar E, Somogyi P. Cell type and pathway dependence of synaptic AMPA receptor number and variability in the Hippocampus. Neuron. 1998;21:545–559. doi: 10.1016/S0896-6273(00)80565-6. [DOI] [PubMed] [Google Scholar]
  41. Pallotto M, Watkins PV, Fubara B, Singer JH, Briggman KL. Extracellular space preservation aids the connectomic analysis of neural circuits. eLife. 2015;4:e08206. doi: 10.7554/eLife.08206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pfeiffer T, Poll S, Bancelin S, Angibaud J, Inavalli VK, Keppler K, Mittag M, Fuhrmann M, Nägerl UV. Chronic 2P-STED imaging reveals high turnover of dendritic spines in the Hippocampus in vivo. eLife. 2018;7:e34700. doi: 10.7554/eLife.34700. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Rodriguez-Moreno J, Rollenhagen A, Arlandis J, Santuy A, Merchan-Pérez A, DeFelipe J, Lübke JHR, Clasca F. Quantitative 3D ultrastructure of thalamocortical synapses from the "Lemniscal" Ventral Posteromedial Nucleus in Mouse Barrel Cortex. Cerebral Cortex. 2018;28:3159–3175. doi: 10.1093/cercor/bhx187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Sala C, Segal M. Dendritic spines: the locus of structural and functional plasticity. Physiological Reviews. 2014;94:141–188. doi: 10.1152/physrev.00012.2013. [DOI] [PubMed] [Google Scholar]
  45. Schertel A, Snaidero N, Han HM, Ruhwedel T, Laue M, Grabenbauer M, Möbius W. Cryo FIB-SEM: volume imaging of cellular ultrastructure in native frozen specimens. Journal of Structural Biology. 2013;184:355–360. doi: 10.1016/j.jsb.2013.09.024. [DOI] [PubMed] [Google Scholar]
  46. Segura I, Lange C, Knevels E, Moskalyuk A, Pulizzi R, Eelen G, Chaze T, Tudor C, Boulegue C, Holt M, Daelemans D, Matondo M, Ghesquière B, Giugliano M, Ruiz de Almodovar C, Dewerchin M, Carmeliet P. The oxygen sensor PHD2 controls dendritic spines and synapses via modification of filamin A. Cell Reports. 2016;14:2653–2667. doi: 10.1016/j.celrep.2016.02.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Spacek J, Harris KM. Three-dimensional organization of smooth endoplasmic reticulum in hippocampal CA1 dendrites and dendritic spines of the immature and mature rat. The Journal of Neuroscience. 1997;17:190–203. doi: 10.1523/JNEUROSCI.17-01-00190.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Tada T, Simonetta A, Batterton M, Kinoshita M, Edbauer D, Sheng M. Role of septin cytoskeleton in spine morphogenesis and dendrite development in neurons. Current Biology. 2007;17:1752–1758. doi: 10.1016/j.cub.2007.09.039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Tønnesen J, Katona G, Rózsa B, Nägerl UV. Spine neck plasticity regulates compartmentalization of synapses. Nature Neuroscience. 2014;17:678–685. doi: 10.1038/nn.3682. [DOI] [PubMed] [Google Scholar]
  50. Van Harreveld A, Crowell J, Malhotra SK. A study of extracellular space in central nervous tissue by freeze-substitution. Journal of Cell Biology. 1965;25:117–137. doi: 10.1083/jcb.25.1.117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Van Harreveld A, Malhotra SK. Extracellular space in the cerebral cortex of the mouse. Journal of Anatomy. 1967;101:197–207. [PMC free article] [PubMed] [Google Scholar]
  52. Zachs T, Schertel A, Medeiros J, Weiss GL, Hugener J, Matos J, Pilhofer M. Fully automated, sequential focused ion beam milling for cryo-electron tomography. eLife. 2020;9:e52286. doi: 10.7554/eLife.52286. [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision letter

Editor: Moritz Helmstaedter1
Reviewed by: U Valentin Nägerl2

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Thank you for submitting your article "Ultrastructural comparison of dendritic spine morphology preserved with cryo and chemical fixation" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Catherine Dulac as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: U. Valentin Nägerl (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

This manuscript analyzes the morphology of dendritic spines in cryo-fixated brain tissue using electron microscopy (EM). Cryo fixation is known to provide a geometric arrangement of neurites that more closely resembles the in-vivo condition. In particular, it is expected to correct the reduction of extracellular space volume that the conventional aldehyde-based fixation approaches yield in EM. Interestingly, while the morphology of the spine heads (the sites of most synapses in the cerebral cortex) was largely unaltered in cryo fixated tissue, the diameter of spine necks was reduced by about 30%. Since the spine neck constitutes a biophysically relevant electrical resistance for transmission of synaptic signals towards the neuron's cell body, such a reduced diameter carries significance for arguments about spine necks as a potential regulator for effective synapse strength. Together, this work provides a focused quantitative morphological description of one of the main sites of synaptic processing in mammalian cortex.

The reviewers agreed on a few requests, all can be addressed by additional analyses or text edits:

– More careful analysis of the spine neck effect: especially comparison to thin axons to understand whether the diameter reduction is a general effect for thin neuronal processes.

– More accurate methodological descriptions, see reviews below.

– Discussion whether cryo fixation may specifically alter thin processes (i.e. could it be that cry fixation provides an overall more realistic geometric configuration, but alters thin processes disproportionally?) For this, a comparison to STED data (obtained in vivo, thus considered to be as realistic) is most relevant, see reviews below.

– It should be made clear that the comparison of cry-fixated tissue with aldehyde-based fixation is not a novel approach of the authors but was carried out by van Harreveld in the 70s. This prior art must be cited.

– It would be important to control for spine density of the parent dendritic shafts (see reviewer 1 below).

Reviewer #1:

The manuscript compared 3D EM results of spine morphology between chemical and cryo fixed samples. The most striking finding was 30% reduction in the spine neck width but with unchanged spine head volume after cryo fixation. Since the size changes appeared in a non-uniform fashion, the previously reported correction between spine neck width and spine head volume was no longer seen in the cryo fixed samples. Moreover, the modeling suggested that thinner spine neck would lead to higher electrical resistance between spine head and parental dendrite. I believe these results, when justified properly, will have considerable impact on synaptic signaling.

1) Description of individual data set was missing. How large was each data set and acquired at what resolution? How many spines were analyzed in each data set, respectively?

2) Were all spines included in the analysis or randomly sampled? If not all spines, authors need to provide more calibrations (e.g. spine density on parental dendrites). Because the traceability of thin structures was found more difficult in data set without extracellular space (Pallotto et al., 2015), sampling bias might result in a larger average spine neck width in chemical fixed samples.

3) What was the inter-section distance in Figure 1G and H? Were they equally distributed along spine neck? What was the criterion of compartmentalizing a spine? It would be also interesting to report width variation along spine neck.

4) Spine neck width was found around 150 nm using STED microscopy in acute brain slice (Tonnesen et al., 2014) and in vivo (Pfeffer et al., 2018). The authors should at least discuss this data.

5) In the subsection “Preparation of cryo-fixed tissue”: was the petri dish filled with ice or like Korogod et al., 2015, with cold ASFC? Would the pre-cooling alter the spine neck width, compared to that measured in slice and in vivo at room or physiological temperature?

6) Were the 200 μm cuts made by vibratome (as in the Korogod et al., 2015) or razor blade? Because I find 90 seconds were quite tight for a vibratome cut, considering tissue mounting, adjusting cutting windowing and knife approaching. A more detailed description of this procedure is appreciated.

Reviewer #2:

The study is an extension of the authors' 2015 eLife paper, where they showed that cryofixation does a much better job at preserving the extracellular space of brain tissue and astroglial morphology than standard chemical fixation. In the present study, they focus on dendritic spines of cortical neurons and compare their morphology in EM images obtained after cryogenic and chemical fixation.

The main result is that spine necks are thinner (by around one third) in cryofixed than in chemically fixed samples, most (but not all) other experimental parameters being the same. By contrast, no differences were seen for spine head size, which is reassuring.

Accurate and reliable information about the key microanatomical parameters and relationships of synapses is critical for deciphering the relationship between synaptic structure and function, and this study contributes to this effort in an important way, raising our awareness of the potential artifacts and pitfalls of the traditional approach to achieve ultrastructural perseveration for EM analyses. This is particularly relevant in a time of major efforts to reconstruct brain anatomy by EM connectomics, where people probably claim too glibly to be reporting the anatomical 'ground truth'.

The observation that spine head size is the same for both procedures indicates (but does not prove) that the measurements are valid, whereas the reported differences in spine neck diameter merely says there is a problem, but a priori we cannot tell, which procedure captures the reality more accurately, even if it seems likely that cryofixation is more trustworthy.

The paper is well written and the data are convincing. Nevertheless, it is at present a bit skinny and piecemeal. To make it more nourishing, we ask the authors to consider to also compare the shape or presence of ultrastructural features of cell-biological interest in their images, such as ER and mitochondria in spines, which may be differentially affected (relocate or remodel) in important but unknown ways during the fixation. Maybe this is possible in the spine head.

Also, it would also be very interesting to obtain analogous morphological information about axonal boutons and shafts, which may also respond differentially to the fixation procedures. The authors discuss the possibility that disruption of actin organization or oncotic pressure might account for the differences in spine neck diameters. If true, we can expect to see these effects also in other thin structures like axon shafts, which also feature actin rings, or spinules reportedly projecting out from a large fraction of spine heads. The authors should also keep the discussion more open about the 'culprit' and consider other factors (than aldehydes), e.g. potential differential ischemic conditions between the fixation procedures that may also explain the different outcomes.

eLife. 2020 Dec 4;9:e56384. doi: 10.7554/eLife.56384.sa2

Author response

The reviewers agreed on a few requests, all can be addressed by additional analyses or text edits:

– More careful analysis of the spine neck effect: especially comparison to thin axons to understand whether the diameter reduction is a general effect for thin neuronal processes.

In the revised manuscript, we have added further careful analyses of spine neck geometry, including: minimum spine neck width and cross-sectional area, more images of cryo-fixed spines, and further details of a comparison between FIBSEM and serial section TEM imaged samples.

We agree with the reviewers that a complete and quantitative appraisal of all the elements in brain tissue should be carried out. This study, therefore, only represents the first part of this quest. We have focused initially on dendritic spines because these structures have taken centre stage for a long time in a vast literature examining the structural plasticity of the brain. Much of this work has used live imaging methods, and some chemical fixation, but without any clear understanding as to what happens to spine shape during the preservation process using aldehydes. This was the motivation of our study. We agree with the reviewers that investigating the effect of chemical fixation on axons is also of great interest, and, indeed, we are collaborating with the group of Shigeki Watanabe at John’s Hopkins on exactly this topic. This collaborative project is looking at how different axons are altered by chemical fixation and together we hope to submit a paper on this topic once all the analyses are complete. This will be presented as a separate piece of work. In the revised manuscript, we now explicitly write: “In future studies, it will be of great interest to examine axonal structure comparing cryo and chemical fixation.”

– More accurate methodological descriptions, see reviews below.

As our manuscript was submitted as a Research Advance, following on from the Korogod et al. paper (Korogod et al., 2015), we applied the journal’s guidelines and did not include every methodological detail that was previously published. However, we have elaborated on some of these to ensure that the procedures can be easily repeated. In addition, the journal Bio-Protocols, that collaborates with eLife to provide detailed methodologies, has accepted an elaborated write-up of this workflow, and I understand this can be linked to any final publication.

– Discussion whether cryo fixation may specifically alter thin processes (i.e. could it be that cry fixation provides an overall more realistic geometric configuration, but alters thin processes disproportionally?) For this, a comparison to STED data (obtained in vivo, thus considered to be as realistic) is most relevant, see reviews below.

The reviewers raise an interesting question. The distribution of spine neck areas after chemical fixation appears to closely reflect a right-ward shift compared to the distribution of spine neck areas in cryo-fixed tissue. There is, therefore, no indication of disproportionate effects. However, we can also not exclude such a phenomenon. We agree with the reviewer that STED imaging followed by electron microscopy would be an exciting future research project. In the revised manuscript, we now write on :

“Our results are consistent with estimates from physiological analyses (e.g. Bloodgood and Sabatini, 2005; Grunditz et al., 2008) as well as morphological analyses in vivo using STED microscopy of hippocampal dendrites in CA1 showing neck widths of 147 nm (Pfeiffer et al., 2018).”

– It should be made clear that the comparison of cry-fixated tissue with aldehyde-based fixation is not a novel approach of the authors but was carried out by van Harreveld in the 70s. This prior art must be cited.

We now include some statements and references in the first paragraph of the Introduction that cite the previous work of Anton Van Harreveld). This work was also cited in our parent paper, and, as mentioned above, in this Research Advance manuscript we did not want to repeat explanations that had already been given.

– It would be important to control for spine density of the parent dendritic shafts (see reviewer 1 below).

Detailed comments are given below.

Reviewer #1:

The manuscript compared 3D EM results of spine morphology between chemical and cryo fixed samples. The most striking finding was 30% reduction in the spine neck width but with unchanged spine head volume after cryo fixation. Since the size changes appeared in a non-uniform fashion, the previously reported correction between spine neck width and spine head volume was no longer seen in the cryo fixed samples. Moreover, the modeling suggested that thinner spine neck would lead to higher electrical resistance between spine head and parental dendrite. I believe these results, when justified properly, will have considerable impact on synaptic signaling.

1) Description of individual data set was missing. How large was each data set and acquired at what resolution? How many spines were analyzed in each data set, respectively?

These details are now given as follows:

“The chemically fixed spines were sampled from three mice, two imaged with FIBSEM, and one with serial section TEM. […]. The volumes imaged were 125 µm3, 1112 µm3, 504 µm3, 450 µm3, 446 µm3, and 446 µm3.”

2) Were all spines included in the analysis or randomly sampled? If not all spines, authors need to provide more calibrations (e.g. spine density on parental dendrites). Because the traceability of thin structures was found more difficult in data set without extracellular space (Pallotto et al., 2015), sampling bias might result in a larger average spine neck width in chemical fixed samples.

We included all spines that were completely contained in the volumes sampled, and which could also be clearly visualized and completely segmented. We considered this to be an unbiased approach to assessing the difference between the two fixation approaches. We accept that we could not analyse all the spines in the volumes of cryo fixed material. Some spines, and in particular those with very thin necks, were difficult to follow and reconstruct in the serial images, but not to detect. Pallotto et al., 2015, used block face scanning electron microscopy (SBEM; pixel resolution of 9.8 nm); a lower resolution imaging approach, for capturing very large volumes, but it is not our understanding, or experience, that TEM or FIBSEM gives images in which spines could not be detected. As the membrane contrast in the cryo-fixed tissue is lower than in chemically fixed, however, due to the contrasting reaction taking place at cryo temperatures and less extraction of native molecules, if any thin spines are missed, this would be more likely to occur in the cryo-fixed material.

We must also point out that the depth to which vitrification occurs is limited due to the high water content of adult brain tissue. Only a few microns are well preserved before ice crystal damage is clearly evident. Therefore, we have only small volumes in which to sample spines. The idea of reconstructing pieces of dendrite and making estimates of spine density is a good one, but was not possible in volumes with such limited thickness.

We now write:

“All spines in each of the imaged volumes were reconstructed if they their entire structure was clearly visible in all of the serial images.”

3) What was the inter-section distance in Figure 1G and H? Were they equally distributed along spine neck? What was the criterion of compartmentalizing a spine? It would be also interesting to report width variation along spine neck.

We agree that including the variation of the width would provide a better understanding of the spine neck morphology in the two groups of spines. This we have added to Figure 1 with a clearer explanation of how we define the neck (subsection “Image processing, analysis, 3D reconstruction, and morphometric measurements”). The cross-sectional areas that were calculated along the spine neck were always spaced equally, and were approximately 150 nm apart. Establishing an objective, and precise, measure of where the spine neck finishes and the spine head starts, across all spines, was not possible as the heads of some spines would not be detected, i.e. those whose necks expand gradually to form the head. We have tried this for the detection of axonal boutons in axonal reconstructions (Gala et al., 2017 eLife) but this only works for clearly defined boutons, such as those of excitatory axons. For this reason, we used the average neck cross sectional area, defining the neck as the region attaching the spine head to the shaft, and finishing where the neck’s width starts to expand appreciably to form the head the dendritic shaft. We do agree that this could be considered somewhat arbitrary, which would hold for all other published spine morphology studies to date. Therefore, we have now included of minimum neck width measurement. Both graphs indicated the same result that chemical fixed spines have wider necks.

4) Spine neck width was found around 150 nm using STED microscopy in acute brain slice (Tonnesen et al., 2014) and in vivo (Pfeffer et al., 2018 ). The authors should at least discuss this data.

The STED microscopy analysis of live dendritic spines is clearly important data for the assessment of spine resistances. We now cite these papers in the Discussion:

“Our results are consistent with estimates from physiological analyses (e.g. Bloodgood and Sabatini, 2005; Grunditz et al., 2008) as well as morphological analyses in vivo using STED microscopy of hippocampal dendrites in CA1 showing neck widths of 147 nm (Pfeiffer et al., 2018).”.

5) In the subsection “Preparation of cryo-fixed tissue”: was the petri dish filled with ice or like Korogod et al., 2015, with cold ASFC? Would the pre-cooling alter the spine neck width, compared to that measured in slice and in vivo at room or physiological temperature?

The use of cold ACSF in the study by Korogod et al. was only for the part of the study that estimated how the shape of the brain changed during the chemical fixation. The ACSF was used as a buffer for cutting fresh vibratome sections. It was not used for the cryo fixation of fresh brain tissue. We realise that this may have been confusing and now write:

“After decapitation, the brain was rapidly removed from the skull, placed on the top surface of a closed, cooled glass Petri dish filled with ice. The tissue was not exposed to any solutions.”

6) Were the 200 μm cuts made by vibratome (as in Korogod et al., 2015) or razor blade? Because I find 90 seconds were quite tight for a vibratome cut, considering tissue mounting, adjusting cutting windowing and knife approaching. A more detailed description of this procedure is appreciated.

This study did not use a vibratome, and the cuts were made using razor blades. The extraction of the tissue was rapid, and the cuts made quickly to ensure that the tissue was frozen as soon as possible to avoid the effects of anoxia. More details are now given in the Materials and methods section:

“Razor blades were then used to slice small (approximately 2 mm x 2 mm x 0.2 mm) pieces of cortex. This was achieved by using two razor blades, one in each hand, and drawing them against each other, in opposite directions, against the surface of the glass to slice the tissues into smaller and smaller pieces.”

Reviewer #2:

The study is an extension of the authors' 2015 eLife paper, where they showed that cryofixation does a much better job at preserving the extracellular space of brain tissue and astroglial morphology than standard chemical fixation. In the present study, they focus on dendritic spines of cortical neurons and compare their morphology in EM images obtained after cryogenic and chemical fixation.

The main result is that spine necks are thinner (by around one third) in cryofixed than in chemically fixed samples, most (but not all) other experimental parameters being the same. By contrast, no differences were seen for spine head size, which is reassuring.

Accurate and reliable information about the key microanatomical parameters and relationships of synapses is critical for deciphering the relationship between synaptic structure and function, and this study contributes to this effort in an important way, raising our awareness of the potential artifacts and pitfalls of the traditional approach to achieve ultrastructural perseveration for EM analyses. This is particularly relevant in a time of major efforts to reconstruct brain anatomy by EM connectomics, where people probably claim too glibly to be reporting the anatomical 'ground truth'.

The observation that spine head size is the same for both procedures indicates (but does not prove) that the measurements are valid, whereas the reported differences in spine neck diameter merely says there is a problem, but a priori we cannot tell, which procedure captures the reality more accurately, even if it seems likely that cryofixation is more trustworthy.

The paper is well written and the data are convincing. Nevertheless, it is at present a bit skinny and piecemeal. To make it more nourishing, we ask the authors to consider to also compare the shape or presence of ultrastructural features of cell-biological interest in their images, such as ER and mitochondria in spines, which may be differentially affected (relocate or remodel) in important but unknown ways during the fixation. Maybe this is possible in the spine head.

Also, it would also be very interesting to obtain analogous morphological information about axonal boutons and shafts, which may also respond differentially to the fixation procedures. The authors discuss the possibility that disruption of actin organization or oncotic pressure might account for the differences in spine neck diameters. If true, we can expect to see these effects also in other thin structures like axon shafts, which also feature actin rings, or spinules reportedly projecting out from a large fraction of spine heads. The authors should also keep the discussion more open about the 'culprit' and consider other factors (than aldehydes), e.g. potential differential ischemic conditions between the fixation procedures that may also explain the different outcomes.

The reviewer raises several excellent points. As we mentioned above, this was a follow-on study (eLife Research Advance format) from the paper by Korogod et al., 2015. We are in fact engaged in a larger study with the group of Shigeki Watanabe in the analysis of the effect of chemical fixation on the morphology of axons of different types. The reviewer’s suggestion of studying intracellular organelles is of great interest. However, there were no mitochondria within any of the spines and we were not able to detect any spinules. We study effects of the presence of ER in Figure 4, finding bigger spine head volumes in the presence of ER, but no difference in spine neck length or diameter.

We also agree that it is perhaps speculative to assume that the changes are only due to the aldehyde fixation and have now added some appropriate text into the Discussion. As our previous work had shown how the extracellular compartment was the same as physiological measurements, and this had been shown, by Van Harreveld, to reduce after 8 minutes of asphyxiation we assume the morphology seen after cryo fixation was close to native, though of course we cannot be certain. We also cite papers that show changes in immature dendrites occurring after 15 minutes of hypoxia:

“We cannot exclude the possibility that chemical fixation leaves the spine neck unchanged with the differences measured occurring during the process of tissue extraction, before cryo fixation. […] This method showed that in vivo levels of the extracellular space were preserved (Korogod et al., 2015) and there was no evidence of the compact neuropil, showing little extracellular space that is a feature of asphyxiated tissue as described by Van Harreveld and Malhotra, 1967.”

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Figure 1—source data 1. Coefficients of variation data used in Figure 1I.
    elife-56384-fig1-data1.xlsx (26.7KB, xlsx)
    Figure 1—figure supplement 1—source data 1. Spine parameter data used in Figure 1—figure supplement 1.
    elife-56384-fig1-figsupp1-data1.xlsx (13.7KB, xlsx)
    Figure 2—source data 1. Spine parameter data used in Figure 2.
    elife-56384-fig2-data1.xlsx (37.1KB, xlsx)
    Figure 2—figure supplement 1—source data 1. Spine parameter data used in Figure 2—figure supplement 1.
    elife-56384-fig2-figsupp1-data1.xlsx (11KB, xlsx)
    Figure 3—source data 1. Spine parameter data used in Figure 3.
    elife-56384-fig3-data1.xlsx (33.7KB, xlsx)
    Figure 4—source data 1. Spine parameter data used in Figure 4.
    elife-56384-fig4-data1.xlsx (18.6KB, xlsx)
    Figure 5—source data 1. Spine parameter data used in Figure 5.
    elife-56384-fig5-data1.xlsx (31.7KB, xlsx)
    Transparent reporting form
    elife-56384-transrepform.pdf (320.4KB, pdf)

    Data Availability Statement

    All data generated during this study are included in the manuscript and the supporting files. Source data files are provided for all results. These are: Figures 1, 2, 3, 4 and 5 and Figure supplements for Figure 1 and 2.

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES