Acute slice preparation for electrophysiology increases spine numbers equivalently in the male and female juvenile hippocampus: a DiI labeling study

Abstract

Hippocampal slices are widely used for in vitro electrophysiological experiments to study underlying mechanisms for synaptic transmission and plasticity, and there is a growing appreciation for sex differences in synaptic plasticity. To date, several studies have shown that the process of making slices from male animals can induce synaptogenesis in cornu ammonis area 1 (CA1) pyramidal cells, but there is a paucity of data for females and other brain regions. In the current study we use microcrystals of the lipophilic carbocyanine dye DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) to stain individual neurons in the CA1 and dentate gyrus (DG) hippocampal subfields of postnatal day 21 male and female rats. We show that the preparation of sections for electrophysiology produces significant increases in spines in sections obtained from females, similar to that observed in males. We also show that the procedures used for in vitro electrophysiology also result in significant spine increases in the DG and CA1 subfields. These results demonstrate the utility of this refined DiI procedure for staining neuronal dendrites and spines. They also show, for the first time, that in vitro electrophysiology slice preparations enhance spine numbers on hippocampal cells equivalently in both juvenile females and males.

NEW & NOTEWORTHY This study introduces a new DiI technique that elucidates differences in spine numbers in juvenile female and male hippocampus, and shows that slice preparations for hippocampal electrophysiology in vitro may mask these differences.

INTRODUCTION

The ability of the brain to store and process information is thought to depend on the plastic nature of synapses, which are composed of a presynaptic axon terminals, dendritic postsynaptic spines, and associated glia (Araque et al. 1999; Bliss et al. 2007). Dendritic spines are small (1–2 μm) membrane protrusions that are present on most neurons in the brain and are believed to be a major component of learning and memory processes (Bailey et al. 2015). Interestingly, the morphology of dendritic spines is highly variable, and it is thought that their structure (i.e., spine head width, spine neck length and width) influences their neurophysiological properties (Rangamani et al. 2016; Rochefort and Konnerth 2012; Uteshev et al. 2000). In turn, the morphology of individual spines has been shown to change dramatically in response to both activity-dependent and -independent mechanisms (Harris and Stevens 1988, 1989; Kasai et al. 2010; Murthy et al. 2001; Yuste et al. 2000). Thus spine shape and numbers have become critical metrics for synaptic physiology related to learning and memory processes (Lisman et al. 2018).

The shape and number of synapses can also be modulated by hormones, and, in particular, dendritic spines in the brain have been shown to vary in density across the estrous cycle (Gould et al. 1990; Shors 2006; Woolley et al. 1990, 1997). Thus hormonal variation could lead to sex-specific differences in spine dynamics. This is an important consideration, because sex hormones act throughout the entire brain of males and females, and they can exert their effects through both genomic and nongenomic receptor actions (McEwen and Milner 2017). That is, ligand binding to a nuclear receptor can produce genomic effects by altering the transcription rate of target genes; however, these same receptors can also have nongenomic effects, such as propagating signal transduction through a kinase pathway or activating G proteins (Wilkenfeld et al. 2018). For instance, gonadotropin-releasing hormone (GnRH) can regulate estradiol synthesis, spine density, and the expression of synaptic proteins through both genomic and nongenomic mechanisms (Fester et al. 2012; Kotitschke et al. 2009). The mechanisms involved in spine generation are important to understand, because a number of sex-specific differences in synaptic plasticity have been reported in a variety of neuropathological processes that range from fetal alcohol spectrum disorders (FASD) to Fragile-X syndrome (Bostrom et al. 2015; Helfer et al. 2012, 2014; Yau et al. 2016).

The hippocampal slice preparation has become a mainstay for electrophysiologists because it provides a preparation wherein the cytoarchitecture and synaptic circuits of the hippocampus are largely left intact (Lo et al. 1994), making it ideal for neuropharmacology, neurophysiology, and even organotypic hippocampal slice culture experiments (Lein et al. 2011). A number of studies have noted that hippocampal slicing procedures can have a significant impact on spine numbers in slices obtained from male animals (Bourne et al. 2007; Kirov et al. 1999, 2004b); however, there remains a paucity of data on sex differences. This is an important consideration, because a sudden sex-specific change in spine numbers induced by the particular methodology followed could mask any synaptic plasticity-related morphological changes that may exist between males and females (Bourne et al. 2007). The use of cold immersion for cutting slices originally gained popularity because it provided a means to provide some degree of neuroprotection from the hostile environment that is created when sections are cut with a vibratome (Christie et al. 1996; Radek and Giardina 1992; Soltesz and Mody 1995). The first report that spine numbers in males were increased in slices prepared for slice electrophysiology began to question this procedure (Kirov et al. 1999). Subsequently, it was shown that spines disappear when slices from male animals are prepared in ice-cold artificial cerebrospinal fluid (aCSF) but that spines rapidly reappear, and even proliferate, within minutes of rewarming (Kirov et al. 2004b). Modifying the aCSF to reduce neuronal activity had little effect on the observed rewarming spine proliferation (Kirov et al. 2004b), and only cutting slices at warmer temperatures resulted in spine numbers approximating those observed in fixed tissue (Bourne et al. 2007). Because dendritic spines can rapidly become functional (De Roo et al. 2008), and this can have important functional considerations (Brown and Murphy 2008), it is important to establish if there are sex differences in how hippocampal slice preparations affect hippocampal spine numbers.

In the current study we first present a detailed protocol that uses the carbocyanine dye DiIC₁₈ to rapidly provide high-resolution staining of dendritic spines from postnatal day (PND) 21 male and female rats prepared using procedures standard for in vitro electrophysiology (Christie et al. 1997; Helfer et al. 2014) or immunohistochemistry (Eadie et al. 2005; Kannangara et al. 2014). We then evaluate the effects of hippocampal slice preparation for electrophysiological experiments on dendritic spine density in juvenile males and females using this refined DiI procedure. The intent of this study was to develop a simple method that combines DiI staining and confocal microscopy to allow for a rapid examination of spines from 1) brain sections used for in vitro electrophysiology or 2) sections prepared using standard immunohistochemistry procedures that should better preserve physiological spine numbers and morphology.

METHODS

Animals.

All animal procedures were in accordance with the Canadian Council for Animal Care policies and were approved by the University of Victoria Animal Care Committee. Sprague-Dawley rats were obtained from Charles River Laboratories (Senneville, QC, Canada) and bred at the University of Victoria Animal Care Unit. All animals were housed in clear standard rat polycarbonate cages with Carefresh contact bedding (Absorption, Bellingham, WA). The room was maintained on a 12:12-h light-dark cycle, and the humidity and temperature (22°C) were maintained constant. Animals were assigned randomly to either the hippocampus live slice preparation (HLSP) or the hippocampus perfusion fixed (HPF) experimental groups (n = 3 animals per group and sex). All animals were euthanized on PND 21 for histology. At this age animals have not started cycling; therefore sex hormones fluctuations should not have an impact on our results and the estrous cycle stage was not considered as a variable.

Acute hippocampal slice preparation.

Hippocampal slices were prepared as described previously (Helfer et al. 2012; White et al. 2017). Isofluorane (Sigma-Aldrich, Oakville, ON, Canada) was used to anesthetize animals for decapitation, and the brain was quickly excised while submerged in aCSF containing (in mM) 125 NaCl, 2.5 KCl, 1.25 NaHPO₄, 25 NaHCO₃, 2 CaCl₂, 1.3 MgCl₂, and 10 dextrose (pH 7.2; 300 mosM), bubbled with 95% O₂-5% CO₂. A Vibratome 1500 sectioning system (Ted Pella, Redding, CA) was used to cut coronal hippocampal slices (400 μm). A custom 12-well holding chamber was used to incubate the slices in aCSF (32 ± 1°C) before electrophysiological recordings or tissue fixation for spine analysis. To examine synaptic plasticity, individual slices were transferred to a recording chamber after a recovery (1–1.5 h) incubation in aCSF. A separate set of slices from the same animals were used for DiI labeling. In this case, slices were incubated in aCSF for 1.5 h, transferred from the holding chamber to a 12-well plate containing 1.5% paraformaldehyde (PFA; 4°C) diluted in equilibrated aCSF (to avoid sudden changes in osmotic pressure that might affect cell structure), and immersion fixed for 24 h. After the fixation step, slices were rinsed twice (5 min each) in phosphate-buffered saline (PBS, pH 7.4) each and stored at 4°C until used for DiI labeling.

Field recordings.

Slices were continuously bathed in oxygenated aCSF (30 ± 1°C) at a rate of ~2 ml/min. Motorized micromanipulators (Siskiyou Design, Grants Pass, OR) and an Olympus BX51 microscope were used to visually place electrodes in the medial molecular cell layer of the dentate gyrus (DG). Stimulating and recording electrodes were placed ~200 µm apart. The medial perforant pathway (MPP) of the DG was stimulated using a concentric bipolar stimulating electrode (FHC, Bowdoin, ME) and a digital stimulus isolation unit (Getting Instruments, San Diego, CA). Field excitatory postsynaptic potentials (fEPSPs) from the DG of hippocampal slices were recorded using glass microelectrodes (0.5–1.5 MΩ) containing normal aCSF with a MultiClamp 700B microelectrode amplifier (Molecular Devices, San Jose, CA). An input-output (I/O) curve was then constructed by increasing the stimulation magnitude until a maximal fEPSP was obtained (pulse width 30–300 μs; 15-s intervals). For synaptic plasticity experiments, the current was adjusted to produce a stimulus strength of 70% the maximum fEPSP amplitude. A paired-pulse (PP) experiment was conducted at this new stimulus strength using an interpulse interval of 50 ms (5 times; 15 s between pairings). Subsequently, single-pulse simulation (0.067 Hz) was administered for a minimum of 20 min to establish a steady baseline before long-term depression (LTD) or long-term potentiation (LTP) were induced. To induce LTD, we administered a train of low-frequency stimuli (LFS; 1 Hz; 900 pulses; Christie et al. 1996). For LTP, short trains of high-frequency stimuli (HFS; 50 pulses at 100 Hz) were delivered four times with a 30-s intertrain interval (Helfer et al. 2014). Following the application of conditioning stimuli, we resumed single-pulse stimulation (0.067 Hz) for an additional 60 min.

Perfusion-fixed slice preparation.

Animals in the HPF group were euthanized with isoflurane and then perfused through the heart with 80 mL of heparinized 0.1 M PBS followed by 100 mL of 1.5% PFA in 0.1 M PBS at room temperature. Following transcardial perfusion, the brain was removed and immersed in the same fixative at 4°C for 24 h. Fixed brains were then sliced in chilled PBS (400-μm coronal sections) using a Leica VT1000S vibratome (Leica Biosystems, Concord, ON) and stored in PBS with 0.02% sodium azide at 4°C until used for DiI staining.

Tissue staining with DiI crystals.

Microscopic (5–30 μm) sized crystals of DiIC₁₈ (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate; catalog no. 468495; Millipore Sigma, Oakville, ON, Canada) were used to fluorescently label neurons. Slices were positioned in a drop of PBS on Parafilm using a paintbrush. Excess PBS was removed with a fine-tipped Pasteur pipette so that only a thin film of PBS remained on the top of the slice. The thin film helped to prevent the slice from dehydrating, which in our experience blocks DiI diffusion, and allowed the DiI crystals to be applied directly with a thin tungsten wire (Fig. 1A, inset A′) by gently depositing them on the surface of the slice in the appropriate cell layer. The placement of the DiI crystals was performed under a SMZ-168 stereomicroscope (Motic, Richmond, BC, Canada) to ensure crystals were localized to specific regions. We empirically optimized the size of the DiI crystal and the separation between crystals (see Fig. 1C, inset C′). The use of a thin tungsten wire was preferred over other applicators, such as a glass micropipette, because it was more efficient in breaking the surface tension of the PBS with the small DiI crystal remaining adhered to the tip. We also found we did not need the prolonged drying procedure required when glass electrodes are used (Kirov et al. 2004a). Slices were then covered with a small (~2 × 2 cm) piece of Parafilm (Fig. 1B) to prevent the evaporation of PBS (peripheral edges of Parafilm were rubbed gently to provide adhesion to the underlying Parafilm) and provide gentle constant pressure to the crystal-tissue interface after DiI crystals were in their final positions. To further prevent dehydration, a paper soaked with distilled water was taped to the inner face of the squared petri dish lid to maintain a humidified environment. Samples were left for 1–2 days at room temperature in this manner.

Fig. 1.

DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) crystal placement and staining. A: tip of the tungsten wire tool used to pick up DiI crystals and place them on brain sections. White arrowhead in inset (A′) indicates a small group of DiI crystals (25–35 µm) with its characteristic magenta color adhered to the very tip of the tool. The optimal diameter of DiI particles to achieve a good cell staining was around 5–20 µm. Bigger crystals (>20 µm) usually produced clumps of stained cells that were challenging to differentiate. B: example of a brain section covered with Parafilm after placement of DiI crystals. Before DiI crystals were deposited, sections were placed in a drop of PBS on top of a Parafilm base. To facilitate an easier placement of the crystals, the excess of PBS was removed using a fine-tip plastic pipette. After DiI crystals were deposited, sections were covered with a small piece of Parafilm to prevent liquid evaporation and to apply a constant and gentle pressure that prevented crystal movement. C: magnification of area depicted by the rectangle in B. Bright-field representative image shows a brain section after deposition of DiI crystals. The placement of DiI crystals was targeted as precisely as possible to the CA1 stratum (S.) pyramidale and the dentate gyrus (DG) granule cell layer. Red and green arrows correspond to the arrows in inset (C′). C′: enlargement of area depicted by the rectangle in C, showing the placement of DiI crystals in the DG granule cell layer. Green arrows point to DiI crystals with optimal size for single-cell or discrete staining. The size of these crystals is ~20 µm. Red arrow shows an example of a crystal with suboptimal size (~30 µm) for a correct staining.

After completion of the incubation period, the Parafilm was carefully removed using fine-tipped forceps. One corner of the Parafilm was gently raised while PBS drops were gently applied to the exposed slice to help prevent sticking of the Parafilm cover. If slices remained attached to the Parafilm cover, a paintbrush was used to help release them. Slices were then further fixed in 4% PFA for 30 min, washed in PBS for 10 min, counterstained with DAPI (1 μg/mL in PBS; catalog no. D1306; ThermoFisher Scientific, Surrey, BC, Canada) for 5 min and washed again with PBS (2 × 5 min each). Slices were coverslipped using Gelvatol or Vectashield (Vector Laboratories, Burlingame, CA), and edges were sealed with clear nail polish to prevent drying of slices. The slides were then stored horizontally in light-omitting slide boxes. All images were acquired within 7 days of slides being coverslipped.

Confocal microscopy imaging.

An Olympus FluoView-1000 confocal laser microscope (Olympus, Toronto, ON) was used to obtain images of DiI-stained neurons using a 561-nm laser. DAPI was visualized separately with a 405-nm laser to identify neuronal cell bodies. A 0.75-numerical aperture (NA) UPlanSApo ×20 objective (Olympus) was used to obtain low-magnification image stacks of isolated stained neurons. For these low-magnification confocal images, a 1024 × 1024-pixel frame size and 1-μm z-step size were used. These low-magnification images were used to select dendritic segments that met the following criteria: 1) secondary to quaternary dendritic segments clearly visible at a range of 50–100 μm away from the cell body; 2) segments longer than 10 μm, excluding branching points, and not overlapping with neighboring processes, and 3) signs of dendritic degeneration (i.e., swelling and blebbing) not present in the dendritic segment analyzed. Using these criteria, two to three dendritic segments for each cell were imaged using a 1.35-NA UPlanSApo ×60 oil-immersion objective and 5× digital zoom (Olympus). For these high-magnification images, a 1024 × 1024-pixel size frame and 0.3-μm z-step size were used.

Analysis.

FIJI (ImageJ; National Institutes of Health) was used to analyze .tiff stack files of high-magnification images (Schindelin et al. 2012). For all acquisitions, the corresponding image scale was standardized and set for every image. The dendritic segments were traced using the segmented line tool, and this tracing was used to obtain the length of the segment. The tracing was added to the ROI Manager tool and saved for future reference. The maximum displayed brightness was adjusted to 40% of the shaft intensity to normalize fluorescence levels between segments. Dendritic spines that protruded from the shaft a minimum of 5 pixels were counted using the Cell Counter ImageJ plug-in. Marked spines were added to the ROI Manager tool and saved for future reference. Spine density was determined by dividing the number of spines by the length of the dendritic segment.

Statistical analysis.

All statistical analyses were performed using GraphPad Prism (GraphPad Software, San Diego, CA). A one-way ANOVA test followed by a Tukey post hoc analysis was computed to compare the means of the different experimental groups and determine statistical differences. The significance criterion was set to P < 0.05. Data are means ± SD.

RESULTS

The size of the DiI crystals is critical for an optimal neuronal staining.

The size of the DiI crystal proved to be a critical determinant for whether our protocol stained individual cells or clumps of overlapping cells (Fig. 2). We determined that the optimal diameter of the DiI crystals was 5–20 µm for targeted staining. Furthermore, we obtained optimal staining when crystals were placed in the cell body layer, rather than directly in the dendritic region to be studied. Experimentally, it was challenging to manipulate small individual crystals initially, but repetition quickly and dramatically improved the success rate. We also found that when a group of small particles adhered to the tip of the probe in a cluster, it offered the opportunity to use the probe repeatedly on the same slice. In this situation, every time the probe tip gently touched the tissue section, only few of the particles on the tip of the tungsten wire would adhere to the tissue. The process was repeated until no particles were left, allowing multiple individual neurons to be stained. In contrast, placing crystals that were bigger than 20 μm normally resulted in a clump of cells being stained, and these were usually challenging to differentiate. As illustrated in Fig. 3, dendritic segments from DiI-labeled neurons provide excellent samples for confocal microscopy and subsequent rendering for three-dimensional visualization and analyses.

Fig. 2.

Size of DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) crystals is critical to achieve optimal staining. Confocal fluorescence images show the placement and staining pattern of DiI crystals with different sizes in a male perfusion-fixed hippocampus section (DAPI, blue; DiI, yellow pseudocolored). A: maximum intensity projection image of the hippocampus acquired with a ×10 objective. Red arrows indicate clumps of overstained cells in the dentate gyrus (DG) granule cell layer. This is the result of staining with suboptimally sized DiI crystals. Green arrows show granule cells stained with DiI crystals with optimal size. S., stratum. B: enlargement of area depicted by the rectangle in A. Maximum intensity projection image acquired with a ×20 objective shows again cells stained with optimally sized crystals (green arrows). White arrowheads indicate DiI crystals. C: maximum intensity projection of area depicted by the rectangle in B acquired with a ×40 objective. Green arrows show cells with optimal staining pattern; white arrowheads indicate DiI crystals (5–10 µm).

Fig. 3.

High-resolution images of DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate)-stained dendritic spines. Three-dimensional (3-D) reconstructions of a dentate gyrus dendritic segment are from a male perfusion-fixed hippocampal section. A: confocal fluorescence microscopy image rendered as a 3-D solid surface. The segment is pseudocolored with yellow (DiI emits a red-orange fluorescence). B: magnification of the area depicted by the rectangle in A. Cell membrane staining with DiI provided detailed images of different structures of the dendritic segments (i.e., dendritic shaft and dendritic spines). These solid surface-rendered images were obtained using Bitplane’s Imaris software.

Temperature was also an important factor in reducing the spread of DiI staining to adjacent neurons, with DiI diffusion being faster at warmer temperatures. One to two days at room temperature could stain the entire dendritic arbor of a given neuron. However, leaving the brain sections with crystals for periods longer than 4 days resulted in the staining of multiple cells with overlapping dendrites. Reducing the temperature to 4°C slowed down the diffusion of DiI, and thus the staining process. By initially checking the samples daily, we were able to determine when to halt the process. To block the DiI diffusion, sections were placed in 4% PFA for 30 min at room temperature. We found that when we did not re-fix tissue samples following DiI staining, the DiI would continue to diffuse for several days, even after being coverslipped.

DiI can label neurons from both perfusion-fixed hippocampus and immersion-fixed acute hippocampal slices.

The refined DiI technique presented in this report allowed us to reliably obtain discrete and single-cell staining from different neuronal types in all our experimental groups, including DG granule cells (Fig. 4, A and B) and cornu ammonis area 1 (CA1) pyramidal neurons (Fig. 4, C and D), as well as interneurons in these two regions and pyramidal cortical neurons in different cell layers. At the same time, DiI staining enabled us to acquire high-resolution imaging of dendritic spines and axonal fibers. Figure 4 shows examples of neurons from the DG and CA1 of both perfusion-fixed and acute hippocampal slices and highlights the cells that satisfy the staining pattern in order to be included in the analysis of dendritic spines (Fig. 4). In Fig. 4, it is possible to see again the importance of an adequate DiI crystal size.

Fig. 4.

Neurons stained with DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) from perfusion-fixed and acute hippocampal slices. A and B: maximum intensity projections of representative dentate gyrus (DG) granule cells stained with DiI from perfusion-fixed (A) and acute (B) hippocampal slices. Images were acquired with a ×20 objective. C and D: maximum intensity projections of representative CA1 pyramidal neurons stained with DiI from perfusion-fixed (C) and acute (D) hippocampal slices. These images were also acquired with a ×20 objective. Green arrows indicate examples of cells that satisfy the staining pattern criteria and would be selected for the subsequent spine analysis. Blue arrowheads indicate axonal fibers stained with DiI. White arrowheads indicate the presence of DiI crystals. Red arrows show the result of using a DiI crystal with excessive size. Big crystals stain a large number of cells and make the visual separation of individual cells very challenging or impossible. This example emphasizes again the critical importance of only using very small DiI crystals to achieve an optimal staining.

The procedure for generating acute hippocampal slices, while standard, is still aggressive and damages structures, particularly those close to the surface of the tissue. A significant proportion of cells close to the surface that come in contact with the cutting blade (up to 15–30 µm below the surface) can display indications of necrosis (like those illustrated in Fig. 5). This was taken into consideration when we selected cells and dendritic segments for spine analysis so that segments used in our analysis were not biased by necrotic processes (i.e., see Fig. 5B, inset B′, and Fig. 5D, inset D′).

Fig. 5.

DiI (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate) enables identification of preserved and compromised dendritic morphology. A–D: comparison of dentate gyrus (DG) granule cells (A and B) and CA1 neurons (C and D) stained with DiI from perfusion-fixed (A and C) and acute (B and D) hippocampal slices. Note that signs of dendritic swelling are only apparent in neurons from the acute slice preparation (B and D). Spines are easily visualized on perfusion-fixed neurons in insets A′ and C′ for DG and CA1, respectively, whereas a “string and bead” pattern (white arrows) can be observed in acute hippocampal slices (insets B′ and D′). Note also that healthy (green arrows) and unhealthy (red arrows) dendritic segments can be visualized with DiI labeling in acute hippocampal slices. Degenerated dendrites were excluded from the spine analysis.

Acute hippocampal slices generated viable sections that were able to show LTP and LTD.

To ensure that the procedures used to obtain acute hippocampal slices for the DiI staining also produced viable and functional tissue samples, we performed electrophysiological recordings in some sections to determine the capacity for neuronal plasticity with our HLSP in vitro cutting procedure. As is shown in Fig. 6, we were able to induce significant LTP with an HFS protocol (50 pulses at 100 Hz; 68 ± 13.5%) in all sections we tested. In a separate set of sections, we also used LFS (900 pulses at 1 Hz) and found that we were also able to induce LTD reliably (−31 ± 5.6%). These data indicate that the procedures used in this study to generate the HLSP samples also produce sections suitable for studying bidirectional synaptic plasticity in the hippocampus.

Fig. 6.

Acute hippocampal slices were viable, displayed normal electrophysiological properties, and had the capacity to exhibit long-term potentiation (LTP) and long-term depression (LTD). A: low-frequency stimulation (LFS; 900 pulses at 1 Hz) and high-frequency stimulation (HFS; 50 pulses at 100 Hz) elicited significant LTD (−31 ± 5.6%, n = 5 slices) and LTP (68 ± 13.5%, n = 5 slices). B and C: average field excitatory postsynaptic potentials (fEPSP) traces before and after HFS (B) and LFS (C). V, voltage. D: a stable baseline was required before the conditioning stimulus was applied at time 0, and the exhibited forms of plasticity were present for at least 55–60 min postconditioning.

The methodology used to generate acute hippocampal slices for in vitro electrophysiology produces a sex-independent artificial increase in dendritic spine density in both CA1 and DG.

Previous studies have shown that the procedures used for making in vitro slice preparations from male animals can induce an artificial increase in dendritic spine density in the CA1 subfield (Bourne et al. 2007; Kirov et al. 1999, 2004a). To expand our understanding of this phenomenon and to validate the use of DiI for spine analysis, we investigated whether the similar in vitro slice procedures used by our laboratory produced similar increases in spine density in the CA1 and DG subfields. As is shown in Fig. 7, dendrites selected for analysis were required to be well stained, nonoverlapping, and 50–100 μm distant from the cell body.

Fig. 7.

Dendritic segment selection criteria. A: visual example of the dendritic segment selection process of a stained dentate gyrus (DG) granule cell (this image was acquired using a ×60 oil-immersion objective). Linear, nonoverlapping, 10-µm-long dendritic segments, 50–100 µm from the cell body (secondary to quaternary branches), were selected for spine density analysis. The light semicircular shading indicates the approximate distance for dendritic segment selection. Dendritic branch order is displayed numerically (1°, 2°, 3°). B and C: high-magnification images of representative DG dendritic segments depicted by corresponding rectangles in A that were used for spine density analysis.

The HLSP produced hippocampal slices that, while suitable for electrophysiology, also exhibited a significant change in dendritic spines in both DG [F_(3,84) = 43.77, P < 0.0001] and CA1 [F_(3,104) = 59.03, P < 0.0001] compared with HPF slices. In the DG, we found that the HLSP produced an average spine density of 21.46 ± 0.86 spines/10 μm (n = 15 slices; Fig. 8A) in males. This was significantly greater than the average spine density of 15.63 ± 0.45 spines/10 μm (n = 23 slices) observed in the male HPF group [Tukey’s post hoc: P < 0.001, 95% confidence interval (CI) (8.34, 3.30)]. Similarly, females exhibited an average spine density of 22.65 ± 0.83 spines/10 μm (n = 20 slices) in the HLSP group compared with an average spine density of 14.55 ± 0.47 spines/10 μm (n = 30 slices) in the female HPF group [Tukey’s post hoc: P < 0.001, 95% CI (10.28, 5.90)]. Overall, the percent increase in DG cell spine numbers was 27.16% for males and 35.76% for females. There was no significant difference between the DG HPF males and DG HPF females [P > 0.05, 95% CI (−1.03, 3.17)] or between the DG HLSP males and the DG HLSP females [P > 0.05, 95% CI (−3.79, 1.39)].

Fig. 8.

Acute slice preparation causes an increase in dendritic spine density in both juvenile males and females. A and B: comparison of dendritic segment spine densities in perfusion-fixed and acute hippocampal slices in the dentate gyrus (DG; A) and CA1 (B). A, top: thresholded maximum intensity projections of representative dendritic segments from the DG of both sexes and both experimental conditions. A, bottom: bar graph comparing quantitative results of spine density (number of spines per 10-µm dendritic segment) for males and females in perfusion-fixed and acute slices. The analysis reveled a sex-independent 27–35% increase in spine density in the DG in acute hippocampal slices. B: a similar dendritic spine density increase (36–40%) in acute hippocampal slices was observed in the CA1. *P < 0.05.

The male CA1 HLSP group had a higher average spine density of 21.5 ± 0.80 spines/10 μm (n = 24 animals; Fig. 8B) compared with the male HPF group, which exhibited an average spine density of 13.68 ± 0.39 [n = 25 slices; Tukey’s post hoc: P < 0.001, 95% CI (10.27, 5.37)]. These numbers are similar to those reported previously by Harris and colleagues (Bourne et al. 2007; Kirov et al. 1999, 2004a) for the CA1 subfield. In this study we also show for the first time that these same procedures produce a significant increase in dendritic spines in sections obtained from age-matched female subjects. We found an average spine density of 21.77 + 0.84 spines/10 μm (n = 25 slices) in the female HLSP neurons, which was significantly greater than that of 12.92 ± 0.45 spines/10 μm (n = 34 slices) observed in female HFP sections [Tukey’s post hoc: P < 0.001, 95% CI (11.10, 6.59)]. Overall, the percent increase in females was 40.65%, compared with 36.3% in males. There was no significant difference between the CA1 HPF male vs. HPF female groups [P > 0.05, 95% CI (−1.48, 3.02)] and the CA1 HLSP male vs. CA1 HLSP female groups [P > 0.05, 95% CI (−2.70, 2.18)].

DISCUSSION

The juvenile rodent hippocampal slice preparation is one of the most broadly used and powerful experimental preparations for synaptic plasticity studies. Although there are slight variations in the synaptic inputs optimized with different cutting orientations (i.e., coronal, transverse, sagittal), hippocampal slices largely retain the cytoarchitecture and synaptic circuits of the intact hippocampus (Lein et al. 2011; Lo et al. 1994; Xiong et al. 2017). This makes it an ideal structure to examine bidirectional synaptic plasticity (Patten et al. 2015; Pinar et al. 2017) and to study pharmacological interactions (Lein et al. 2011; Oliver et al. 1977). It can also be used to perform studies that combine optical imaging and electrophysiology (Christie et al. 1995, 1996, 1997), as well as to create slice cultures that allow long-term recordings from hippocampal circuits (Lein et al. 2011). Moreover, with the introduction of fluorescent proteins into specific cell populations, the hippocampal slice has also allowed researchers to explore the development and physiological properties of new neurons in the adult brain (Overstreet et al. 2004; van Praag et al. 2002). Thus the hippocampal slice preparation has become a staple for neuroscience research. However, the work of Harris and colleagues (Bourne et al. 2007; Kirov et al. 1999, 2004b, 2004a) has highlighted the fact that the process of making acute hippocampal slices can change the ultrastructure of neurons at the level of the synapse. In the current work we extend these findings to show that preparations commonly used for making acute hippocampal slices for electrophysiological research produce equivalent effects in slices obtained from juvenile male and female animals, in both the CA1 and DG. This raises two interesting considerations. First, acute hippocampal slices may not be the optimal preparation for studying sex differences in synaptic plasticity, because the process of creating slices may mask any differences that might exist between the sexes, such as physiological variations in dendritic spines density (Gould et al. 1990; Woolley et al. 1990). However, considerations in the composition and temperature of the aCSF might help control for this phenomenon (Kirov et al. 2004b). In this study, we selected the specific age of PND 21 to be able to compare our results with previous reports and to avoid the possible confounding effect of the estrous cycle in spine numbers (animals at this age have not started cycling yet). However, a future study investigating if this methodologically induced increase of dendritic spines could mask the reported changes in spine density during the estrous cycle would be of great interest. Nevertheless, the observations in this study and in previous reports suggest that in vivo electrophysiology may be a more appropriate means to study sex differences in synaptic plasticity (Titterness and Christie 2008, 2012). A second consideration is that because the process of creating acute slices normally enhances spine proliferation, it may also bias hippocampal slices for exhibiting long-term depression of synaptic efficacy. There is some evidence for this, as the induction of homosynaptic long-term depression is much more robust in vitro than in vivo (Christie and Abraham 1992a; Pinar et al. 2017), whereas the induction of heterosynaptic depression is much more robust in vivo than in vitro (Christie and Abraham 1992b, 1994; Wickens and Abraham 1991). There remains a paucity of data for sex differences in bidirectional plasticity per se, and in particular, studies directly comparing the induction of synaptic plasticity in vivo and in vitro are lacking.

The use of DiI for studying neuronal morphology has grown significantly since the first reports of its use for studying axonal growth in the 1980s (Godement et al. 1987; Harris et al. 1987; Thanos and Bonhoeffer 1987). In the current work, we describe a fast and robust procedure to use DiI to stain individual cells. The procedure can be used in lightly aldehyde-fixed tissue and in acute hippocampal slices used for electrophysiology and allows for the resolution of dendritic spines using confocal microscopy. Long-chain dialkyl carbocyanines have become a useful tool for labeling the plasma membrane of cells because they can be constructed to contain both 1) a charged amphiphilic fluorophore (i.e., indocarbocyanine) to localize the probe to the plasma membrane surface and 2) a lipophilic aliphatic “tail” (i.e., dioctadecyl) that helps insert it into the membrane (Wiederschain 2011). Carbocyanine dyes were initially used by cell biologists to study the structure and dynamics of cell membranes and artificial lipid bilayers (Honig and Hume 1989; Jacobson et al. 1981; Schlessinger et al. 1977; Vaz et al. 1984). Later, these dyes helped to visualize the morphology of neurons in live cell culture, providing insight into their structure, their origin, and how they interacted with different cell types (Honig and Hume 1986, 1989; Schwartz and Agranoff 1981). Indeed, the ease of use and flexibility of these dyes soon saw them being used both in vivo (Harris et al. 1987; O’Leary and Terashima 1988; Stuermer 1988; Thanos and Bonhoeffer 1987) and in fixed tissue (Burkhalter and Bernardo 1989; Godement et al. 1987), where they can serve as both anterograde and retrograde tracers of axonal fibers. More recently, carbocyanine dyes have been used to visualize and analyze dendritic spines with manual staining (Atkin et al. 2015; Cheng et al. 2014; Kim et al. 2006, 2007, 2008; Mahmmoud et al. 2015; Skrzypiec et al. 2013) and with particle-mediated ballistic delivery (Batista et al. 2016; Binley et al. 2016; Chapleau et al. 2012; Cheng et al. 2014; Fortin et al. 2014; Gupta et al. 2015; Hong and Mah 2015; Staffend and Meisel 2011; Waselus et al. 2013; Xu et al. 2015). The current work demonstrates that precise manual placement of small DiI crystals can provide high-quality staining of individual neurons that allows for an excellent dendritic spine resolution.

In summary, the current work extends previous findings (Bourne et al. 2007; Kirov et al. 1999, 2004a, 2004b) by showing that standard acute hippocampal slice preparation methods for in vitro electrophysiology can dramatically alter dendritic spines in both sexes in juvenile (PND21) rats not only in the CA1 but also in the DG. Our results and previous findings suggest that this artificial increase in dendritic spine numbers caused by methodology is a generalized phenomenon affecting several regions in the hippocampus (Kirov et al. 1999, 2004a, 2004b; Popov et al. 1992; Popov and Bocharova 1992; Wenzel et al. 1994). Observations from previous studies suggest that the increase in dendritic spines is a homeostatic regulatory response to the transient absence of synaptic transmission experienced when acute slices are excised from the brain. In response to this transient period of reduced synaptic transmission, neurons in the acute slices generate new dendritic spines (Kirov et al. 1999, 2004a; Kirov and Harris 1999). Other studies also indicate that the inhibition of the Na⁺-K+-ATPase in response to reduced ATP, glycogen, and temperature during brain slicing may also contribute to this phenomenon (Bourne et al. 2007). Synaptic scaling (i.e., structural and functional synaptic adaptation to changes in neuronal activity) is a homeostatic process that has been described in multiple regions in addition to the hippocampus (Turrigiano 2008). This would suggest that the increases in dendritic spines reported in this study would be a common process for neurons throughout the brain. Future studies will need to determine if dendritic spine density is also altered in extrahippocampal areas. With the development of this technique, future studies can more easily examine if altering the composition of aCSF and/or dissection temperature will influence spine numbers equally in slices obtained from males and females and in different brain regions. The rise in popularity of carbocyanine dyes for the visualization of neuronal morphology comes in part because they have several advantages compared with other traditional techniques. Compared with Golgi staining (Gibb and Kolb 1998), carbocyanine staining (e.g., DiI) provides a faster and more targeted staining of tissue sections and does not require the handling of toxic compounds. Moreover, when combined with confocal microscopy, the fluorescent properties of DiI allow the generation of confocal image stacks that permit three-dimensional high-resolution visualization of neuronal structures. In addition, because there are several variants of the dye (i.e., DiI, DiO, DiD), individual cells can be labeled with different colors, or, unlike Golgi stains, carbocyanine labeling can be combined with immunofluorescence to co-label other structures (Rasia-Filho 2010). Similarly, compared with retroviral labeling of neurons (Vivar et al. 2016), the DiI staining technique presented in this study is less technically challenging and less invasive, and provides a cheaper alternative. Finally, although transgenic animals provide an excellent model to study dendritic spine changes, the number of available species is currently limited (e.g., animal model of neurons expressing yellow or green fluorescent protein under the Thy1 gene promoter are limited), and usually the expression of the fluorescent protein is restricted to a small subset of cells (Enikolopov et al. 2015). Thus DiI staining has the advantage of being both easier to perform and able to provide a significant sample size rapidly.

GRANTS

This work was supported by Natural Sciences and Engineering Research Council of Canada Grant RPIN249853-2013 (to B. R. Christie) and Canadian Institutes for Health Research Grant FRN125888 (to B. R. Christie).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.S.T.-P., P.C.N., and B.R.C. conceived and designed research; J.S.T.-P., P.C.N., and C.P. performed experiments; J.S.T.-P., P.C.N., and C.P. analyzed data; J.S.T.-P., P.C.N., and B.R.C. interpreted results of experiments; J.S.T.-P., P.C.N., and B.R.C. prepared figures; J.S.T.-P. drafted manuscript; J.S.T.-P., P.C.N., P.G., and B.R.C. edited and revised manuscript; J.S.T.-P., P.C.N., C.P., P.G., and B.R.C. approved final version of manuscript.