Testosterone Depletion in Adult Male Rats Increases Mossy Fiber Transmission, LTP, and Sprouting in Area CA3 of Hippocampus
Introduction
Although the effects of estrogens on the hippocampus have been studied extensively (Spencer et al., 2008; Smith et al., 2009; Barha and Galea, 2010; Foy, 2010; Wojtowicz and Mozrzymas, 2010; Fester et al., 2011), how androgens affect the hippocampus remains poorly understood. Androgen receptors are abundant in the hippocampus (Clancy et al., 1992; Brown et al., 1995; Kerr et al., 1995; DonCarlos et al., 2003; Tabori et al., 2005; Sarkey et al., 2008), suggesting that androgens regulate hippocampal function. Moreover, there are many examples of sex differences in hippocampus (for review, see McEwen and Milner, 2007; Galea, 2008; McLaughlin et al., 2009; Mitsushima, 2011), which also suggest a role of androgens (Becker et al., 2008; Etgen and Pfaff, 2009; Mitsushima et al., 2009). Studies examining the effects of androgens in hippocampus have, however, yielded mixed results. In rodents, testosterone administration improves hippocampal-dependent behavior in some studies (Edinger and Frye, 2004; Edinger and Frye, 2007a,b; Frye et al., 2008) but not others (Naghdi and Asadollahi, 2004; Naghdi et al., 2005; Harooni et al., 2008). Testosterone application to hippocampal slices of adult male rats potentiates Schaffer collateral transmission in area CA1 (Smith et al., 2002), but testosterone depletion after gonadectomy (Gdx) can facilitate long-term potentiation (LTP) in area CA1 (Harley et al., 2000; Sakata et al., 2000). Testosterone administration to Gdx rats increases the number of spine synapses in stratum radiatum of area CA1 (Leranth et al., 2003), consistent with a role for androgen in maintaining normal cognitive function (MacLusky et al., 2006). However, in men, androgen ablation therapy for prostate cancer leads to few or no significant deficits in cognitive performance (Nelson et al., 2008; Alibhai et al., 2010a,b; Matousek and Sherwin, 2010). The reduced levels of androgens in aging men are often considered to be a contributing factor to diseases that increase with age, such as Alzheimer's disease (Fuller et al., 2007; Drummond et al., 2009; Barron and Pike, 2011; Luchetti et al., 2011), epilepsy (Herzog, 1991; Harden and MacLusky, 2004, 2005; Frye, 2006; Reddy, 2010; Stevens and Harden, 2011), and Parkinson's disease (Al Sweidi et al., 2011). However, direct evidence in support of this idea, such as prevention of these diseases by androgen therapy, is generally weak, although androgens ameliorate symptoms in some instances (Herzog, 1991; Almeida and Flicker, 2003; Cherrier et al., 2005; Harden and MacLusky, 2005; Driscoll and Resnick, 2007).
We hypothesized that the actions of testosterone on the hippocampus might involve both positive and negative effects, which could explain previous contradictory results. This idea originated in our findings that intact male rats exhibit low levels of hippocampal BDNF immunoreactivity compared with females, primarily in the mossy fiber (MF) pathway (Scharfman et al., 2003). Therefore, androgens may normally keep BDNF protein levels relatively low, which would maintain normal excitability but possibly limit plasticity (i.e., androgens would have positive and negative effects). In addition, this hypothesis is based on previous studies suggesting an important role for neurotrophins as mediators of gonadal steroid action (Toran-Allerand, 1992; Toran-Allerand et al., 1996) and specifically androgens and BDNF, in avian or neuromuscular systems (Louissaint et al., 2002; Ball et al., 2004; Ottem et al., 2007; Wissman and Brenowitz, 2009; Verhovshek et al., 2010). Therefore, we asked whether testosterone regulates BDNF expression in MFs, as well as anatomical and electrophysiological indices of MF transmission and plasticity in adult male rats. Our results show, remarkably, that Gdx rats exhibit increased BDNF protein expression in the MFs and increased MF plasticity.
Materials and Methods
Adult male rats (Charles River) were bred in-house, weaned at 21–23 d, and housed (2 or 3/cage) with food (Purina 5001, WF Fisher) and water ad libitum and a 12 h light:dark cycle. Chemicals were from Sigma-Aldrich unless stated otherwise. All procedures involving animals were approved by the animal care and use committee of the Nathan Kline Institute for Psychiatric Research.
Surgical procedures
Before Gdx, animals were anesthetized with 1 ml/kg ketamine-xylazine solution (80 mg/ml ketamine hydrochloride and 12 mg/ml xylazine hydrochloride). Bilateral Gdx was conducted as previously described (Edwards et al., 1999). In brief, a half-inch midline incision of the lower abdomen was followed by removal of the testes and ligation around the vasculature adjacent to each gonad with sterile sutures (Dermalon, Henry Schein). The incision was closed with wound clips (Stoelting), swabbed with betadine (Henry Schein), and animals were placed over a heating pad (Homeothermic Blanket, Harvard Apparatus) until they recovered from anesthesia. Sham surgery used the same procedures, but the testes were not removed from the abdominal cavity.
For comparison of Gdx and sham rats in the Results section, animals were subjected to surgery shortly after puberty (50–60 d old). Animals were used for experiments ∼2 weeks later to allow ample time for recovery.
Electrophysiology
Slice preparation
After deep anesthesia by isoflurane inhalation, the rat was decapitated, the brain was removed from the skull, and it was placed in ice-cold (4°C) oxygenated (95%O2/5%CO2) sucrose-based artificial CSF (ACSF, in mm: 252.0 sucrose, 5.0 KCl, 2.4 CaCl2, 2.0 MgSO4, 26.0 NaHCO3, 1.25 NaH2PO4, 10.0 d-glucose) for ∼2 min. One hemisphere was glued with cyanoacrylate (Krazy Glue) to the Teflon-coated stage of a vibrating tissue slicer (Vibroslice, Stoelting) and slices (400-μm-thick) were cut in the horizontal plane while immersed in 4°C ACSF. Slices were placed immediately in 100 ml of preoxygenated, room-temperature ACSF, and placed with a wide-bore Pasteur pipette on a nylon net of a recording chamber (modified from one previously sold by Fine Science Tools) (Scharfman et al., 2001) to provide improved flow of ACSF and increased aeration (http://www.healthresearch.org/technology-transfer/brain-and-tissue-slice-recording). Slices were perfused with oxygenated (95%O2/5%CO2), preheated (30°C) ACSF at ∼1 ml/min by a peristaltic pump (Minpuls2 Gilson). The compartment of the recording chamber where slices were placed was maintained at 31–32°C by a temperature controller (PTCO3, Scientific Systems Design). After 30 min, slices were perfused with NaCl-based ACSF (126.0 mm NaCl instead of sucrose). Recordings began 60 min after slices were first placed in the recording chamber and 30 min after the start of perfusion with NaCl-ACSF.
Recording
Similar recording procedures to those described previously (Scharfman et al., 2000, 2007; Skucas et al., 2011) were used. Recording electrodes were pulled (Model P87, Sutter Instruments) from borosilicate glass with a capillary in the lumen (1.0 mm outer diameter, 0.75 mm inner diameter, World Precision Instruments). Recording electrodes were filled with NaCl-ACSF and resistances were 3–6 MΩ. Recordings were amplified (Axoclamp 2B, Molecular Devices), digitized (Digidata 1440A, Molecular Devices), and acquired using pClamp (Molecular Devices). For recording sites, a depth in the slice was chosen where the maximal response was elicited, typically 50–100 μm below the surface. For laminar analysis, the exact depth was determined by a calibrated micromanipulator with precision to the nearest micron (Leica Microsystems) and was the same for all recording sites.
Stimulation
Monopolar stimulating electrodes were made from Teflon-coated stainless steel wire (diameter including Teflon, 75 μm; A-M Systems). Stimuli were controlled by a stimulus isolation unit (IsoFlex, AMPI Instruments) that generated current pulses (100 μA), which were varied in duration digitally (10–250 μs) to test different stimulus strengths. Stimuli were triggered every 30–40 s to test basal transmission.
For stimulation of MFs, recordings were in CA3b and the stimulating electrode was placed in the subgranular zone (SGZ) at the crest of the DG (Fig. 1) because this location avoided stimulation of CA3c pyramidal cell dendrites and activation of area CA3 recurrent collaterals to other CA3 pyramidal cells (Claiborne et al., 1993). The response to MF stimulation was recorded at many sites in stratum lucidum in CA3b at the onset of each experiment to determine the location within stratum lucidum where the maximal field EPSP (fEPSP) was recorded. Other methods to confirm that the SGZ stimulus activated MFs are shown in Figure 1. Nevertheless, we cannot exclude the possibility that a pathway in addition to the MF pathway was activated by SGZ stimulation (e.g., the backprojecting axons of CA3 pyramidal cells that extend deep into the hilus) (Scharfman, 2007).
LTP
LTP experiments required a stable baseline, which was defined by a 15 min period when the fEPSP slope varied within 95–105% of the mean of the first 3 evoked responses. For MF plasticity, a 25 Hz, 1 s train was triggered twice with a 10 s interval, and stimulus strength was 50% of the strength that elicited the maximum fEPSP. For θ-burst stimulation (TBS), 10 trains of 4 half-maximal stimuli at 100 Hz were triggered 200 ms apart. fEPSPs were normalized to the average of the 15 min baseline.
Data analysis
fEPSP.
Slope was defined as the slope between two points: the start of the fEPSP and the peak. When a fiber volley preceded an fEPSP, the first point of the fEPSP slope measurement was immediately after the fiber volley. This definition of slope was chosen because the fEPSPs in Gdx rats often had an irregular rising phase, making the maximal slope difficult to measure (see Fig. 9A2). fEPSP amplitude was defined as the difference between the prestimulus baseline and the peak of the fEPSP. Paired pulse facilitation (PPF) was defined as the measurement of the second response divided by the measurement of the first response. Latency was measured from the middle of the stimulus artifact, which was 10–285 μs in duration, to the onset of the initial negative deflection representing the rising phase of the fiber volley (for latency to the fiber volley) or the onset of the rising phase of the fEPSP (for latency to fEPSPs) (Skucas et al., 2011).
Population spike amplitude.
The amplitude of the population spike was defined as the difference between the prestimulus baseline and the peak of the population spike (Scharfman, 1997). Care was taken to record the population spike at a site in stratum pyramidale where it was maximal, which typically was on the border of the pyramidal cell layer and stratum oriens, analogous to area CA1 (Scharfman et al., 2003), rather than a site close to stratum lucidum, where the population spike can be recorded, but is not the site where it is maximal.
Current source density.
Current source density (CSD) analysis was conducted using field potential recordings that began at the border of the alveus with stratum oriens in area CA3b and perpendicular to the CA3 pyramidal cell layer, ending in stratum lacunosum-moleculare at a location that was adjacent to the lateral tip of the superior blade of the DG (Fig. 1). Recordings were ∼25 μm apart and the same depth in the slice, typically 50 μm. Intervals between sites were monitored using a dissecting microscope (Stemi SV6, Carl Zeiss). The one-dimensional CSD profile was calculated offline using MATLAB (Mathworks) by deriving the second spatial derivative of the local field potential recordings using a 3-point formula. The CSD represents the transmembrane sink and source currents that generate the recorded field potential profile.
Immunocytochemistry
Perfusion and processing
After deep anesthesia (urethane, 4 g/kg, intraperitoneally), animals were perfused through the aorta with 4% paraformaldehyde, pH 7.4, as previously described (Scharfman et al., 2002). Brains were left in the skull overnight at 4°C and removed the next day for dynorphin immunocytochemistry. They were postfixed in 4% paraformaldehyde at 4°C and cut on a vibratome into 50-μm-thick sections. For BDNF immunocytochemistry, the brains were removed after perfusion-fixation and placed in 2% paraformaldehyde at 4°C for 2 h. They were cryoprotected by immersion in 30% sucrose dissolved in 0.1 m phosphate buffer. Sections (40 μm thick) were cut on a cryostat (Minotome Plus, International Equipment Company) the next day. Sections were processed that day, as described in Immunocytochemistry: dynorphin.
The antibody to dynorphin B1–13 (kindly provided by Dr. Teresa Milner) has been used previously to evaluate the MF pathway (Pierce et al., 1999). The polyclonal antibody to BDNF (kindly provided by Amgen-Regeneron Partners) has also been used previously and shown robust BDNF immunoreactivity (ir) in the MF pathway (Conner et al., 1997; Yan et al., 1997; Scharfman et al., 2002, 2003). The monoclonal antibody to BDNF was used because previous studies have shown that staining with this antibody led to a qualitatively similar pattern of BDNF-ir as the polyclonal antibody (Scharfman et al., 2007). Notably, processing sections in the absence of the primary antibody failed to produce MF BDNF-ir.
Immunocytochemistry: dynorphin
Sections were processed with a polyclonal antibody to dynorphin B made in rabbit (Pierce et al., 1999), using free-floating sections that were initially washed twice (5 min each) in 0.1 m Tris buffer, pH 7.6, and treated with 1% H2O2 made in 0.1 m Tris buffer (pH 7.6; 30 min). Sections were then washed in 0.1 m Tris buffer (pH 7.6; 5 min) and treated with 0.25% Triton X-100 dissolved in 0.1 m Tris buffer (Tris A; 10 min), followed by 0.25% Triton X-100 and 0.005% BSA in 0.1 m Tris buffer (Tris B; 10 min). Sections were then treated with 10% normal goat serum diluted in Tris B for 45 min. Sections were subsequently washed in Tris A (10 min) followed by Tris B (10 min) and incubated in antisera (1:1000; diluted in Tris B) for 24 h on a rotating shaker. The next day, sections were treated with Tris A (10 min) followed by Tris B (10 min) and then incubated for 45 min with a biotinylated secondary antibody against rabbit IgG made in goat (1:400, Vector Laboratories). Sections were washed in Tris A (10 min), then 0.25% Triton X-100 and 0.005% BSA in 0.5 m Tris buffer (Tris D; pH 7.6; 10 min), and finally incubated for 1 h in avidin-biotin horseradish peroxidase complex diluted in Tris (ABC Standard kit, 1:1000; Vector Laboratories). Immunoperoxidase labeling was developed in DAB (50 mg/100 ml of 0.1 m Tris buffer) with 200 mg of d-glucose, 0.3 mg of glucose oxidase, and 40 mg of ammonium chloride, and then washed three times (5 min each) in 0.1 m Tris buffer, pH 7.6. Sections were mounted on subbed slides, allowed to dry overnight, and then dehydrated in a series of ethanol solutions (70%, 5 min; 95%, 5 min, 100%, 10 min, 100%, 10 min), cleared in xylene (10 min), and coverslipped with Permount (Fisher Scientific). The sections were viewed on a brightfield microscope (BX51, Olympus of America) and photographed using a digital camera (Retiga 2000R, Q Imaging) using ImagePro (Media Cybernetics). To allow comparisons in the figures, micrographs of sections from different experimental groups were taken using the same light and camera settings.
Immunocytochemistry: BDNF
Sections were first washed in 0.05 m potassium PBS (KPBS, pH 7.3) and then treated with 0.5% H2O2 in KPBS for 30 min. After a 5 min wash in KPBS, sections were incubated in 10% normal goat serum for the polyclonal antibody (10% horse serum for the monoclonal antibody), plus 1% BSA and 0.25% Triton X-100 in KPBS for 20 min. Sections were incubated in antisera (polyclonal: 1:30,000; Scharfman et al., 2002, 2003; monoclonal: 1:1000; Sigma-Aldrich; Scharfman et al., 2007) in 1% BSA and 0.25% Triton X-100 in KPBS for 48 h at 4°C and subsequently rinsed 10 times (10 min each) in 0.25% BSA and 0.025% Triton X-100 in 0.01 m KPBS. Sections were then incubated in a biotinylated secondary antibody against rabbit IgG made in goat (1:400; Vector Laboratories) for the polyclonal antibody (or biotinylated secondary antibody against mouse IgG made in horse (1:400; Vector Laboratories) for the monoclonal antibody) and 1% BSA and 0.025% Triton X-100 in KPBS for 1 h, followed by ABC (ABC Standard kit; 1:1000; Vector Laboratories) in 1% BSA in KPBS for 1 h. Sections were rinsed in KPBS followed by Tris buffer, and reacted with DAB using methods identical to those described above for dynorphin, except 10 mm NiCl2 was added to the DAB solution.
Data analysis
Quantification of immunocytochemical data used ImagePro (Media Cybernetics). The same light settings were used for all procedures. For measurement of BDNF in the MF plexus in stratum lucidum, a square region of interest (ROI; ∼50 μm) was placed over the terminal field at the border of CA3a with CA2, where immunoreactivity was greatest (Scharfman et al., 2003, 2007) (see Fig. 9H1). Three measurements (spanning the terminal field of the MFs; see Fig. 9H1) of the grayscale value (scale, 0–255; for the purposes of this analysis, 0 was defined as white and 255 was defined as black) were made for a given section and averaged. Three measurements were also made of the area in stratum radiatum beside the ROI, where immunoreactivity was weak. The average of the three measurements in stratum radiatum was subtracted from the average of the three measurements in stratum lucidum to give the final grayscale value for a given section. The procedure was repeated for two other sections at a similar septotemporal level. The average value from the three sections was used as the best estimate for that animal. A similar procedure was used to determine whether dynorphin-ir in stratum lucidum was increased or decreased in Gdx rats compared with sham rats.
To determine whether mossy fiber sprouting occurred, a different method was used: an ROI was drawn around the MF plexus of CA3a in stratum lucidum and stratum oriens in the section with the most widespread MF sprouting in stratum oriens (see Fig. 9G1). This ROI was then used as a template for all sections to measure the mean grayscale value within the ROI. The template was moved to stratum radiatum to measure the background (see Fig. 9G1), which was subtracted.
Pharmacology
Administration of drugs to hippocampal slices
Dihydrotestosterone (DHT; Sigma-Aldrich) and 5α-androstane-3α, 17β-diol (Adiol; Sigma-Aldrich) were dissolved in 100% ethanol as a 1 m stock solution and stored at 4°C; the stock solution was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (50 nm). K252a (LC Labs) was dissolved in DMSO (Sigma-Aldrich) as a 1 m stock solution and stored at 4°C. It was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (300 nm). TrkB-IgG (R&D Systems), control IgG (R&D Systems), and 2-(2,3-dicarboxycyclopropyl)-glycine (DCG-IV; Tocris Bioscience) were dissolved in 0.9% NaCl, stored at 4°C as a concentrated stock solution (1 mm), and diluted in NaCl-ACSF on the day of use. Testosterone and placebo time-release pellets (25 mg/pellet, 21 d release; #A-151 and #C-111, respectively; Innovative Research of America) were stored at room temperature in the dark. The testosterone dose was sufficient to replace normal serum testosterone levels (Fig. 2F3).
Hormone replacement
A 1/4-inch incision was made between the shoulder blades immediately after gonadal removal or sham treatment while the animal was still anesthetized. A 1/8-inch-diameter pellet was inserted under the skin. Sterile sutures were used to close the incision.
Testosterone ELISA was conducted to determine circulating levels of testosterone. Trunk blood was collected in 15 ml of centrifuge tubes at the time of death and placed at room temperature for 20–30 min. Serum was separated by centrifugation (3000 rpm, 15 min, 23°C) and frozen (−20°C) until use. A testosterone ELISA was performed using methods provided by the manufacturer (Mouse and Rat Testosterone Kit, Calbiochem). Details were provided previously (Chakraborty et al., 2012).
Statistics
Data are expressed as mean ± SEM. The p criterion was 0.05. No more than two slices were used for a given animal. Student's t test were two-tailed and conducted using Microsoft Excel 2007 (Microsoft). Fisher's exact test, χ test, ANOVA, repeated-measures ANOVA (RM-ANOVA), ANCOVA, and nonparametric Kruskal-Wallis ANOVA were performed using Prism (GraphPad). When reporting the results of ANOVA or RM-ANOVA, interactions between factors are reported in the Results only in those cases where they were statistically significant. For parametric ANOVA, where there was inhomogeneity of variance, which was determined by Bartlett's test (Prism), data were log-transformed before statistical analysis (Zivin and Bartko, 1976).
Surgical procedures
Before Gdx, animals were anesthetized with 1 ml/kg ketamine-xylazine solution (80 mg/ml ketamine hydrochloride and 12 mg/ml xylazine hydrochloride). Bilateral Gdx was conducted as previously described (Edwards et al., 1999). In brief, a half-inch midline incision of the lower abdomen was followed by removal of the testes and ligation around the vasculature adjacent to each gonad with sterile sutures (Dermalon, Henry Schein). The incision was closed with wound clips (Stoelting), swabbed with betadine (Henry Schein), and animals were placed over a heating pad (Homeothermic Blanket, Harvard Apparatus) until they recovered from anesthesia. Sham surgery used the same procedures, but the testes were not removed from the abdominal cavity.
For comparison of Gdx and sham rats in the Results section, animals were subjected to surgery shortly after puberty (50–60 d old). Animals were used for experiments ∼2 weeks later to allow ample time for recovery.
Electrophysiology
Slice preparation
After deep anesthesia by isoflurane inhalation, the rat was decapitated, the brain was removed from the skull, and it was placed in ice-cold (4°C) oxygenated (95%O2/5%CO2) sucrose-based artificial CSF (ACSF, in mm: 252.0 sucrose, 5.0 KCl, 2.4 CaCl2, 2.0 MgSO4, 26.0 NaHCO3, 1.25 NaH2PO4, 10.0 d-glucose) for ∼2 min. One hemisphere was glued with cyanoacrylate (Krazy Glue) to the Teflon-coated stage of a vibrating tissue slicer (Vibroslice, Stoelting) and slices (400-μm-thick) were cut in the horizontal plane while immersed in 4°C ACSF. Slices were placed immediately in 100 ml of preoxygenated, room-temperature ACSF, and placed with a wide-bore Pasteur pipette on a nylon net of a recording chamber (modified from one previously sold by Fine Science Tools) (Scharfman et al., 2001) to provide improved flow of ACSF and increased aeration (http://www.healthresearch.org/technology-transfer/brain-and-tissue-slice-recording). Slices were perfused with oxygenated (95%O2/5%CO2), preheated (30°C) ACSF at ∼1 ml/min by a peristaltic pump (Minpuls2 Gilson). The compartment of the recording chamber where slices were placed was maintained at 31–32°C by a temperature controller (PTCO3, Scientific Systems Design). After 30 min, slices were perfused with NaCl-based ACSF (126.0 mm NaCl instead of sucrose). Recordings began 60 min after slices were first placed in the recording chamber and 30 min after the start of perfusion with NaCl-ACSF.
Recording
Similar recording procedures to those described previously (Scharfman et al., 2000, 2007; Skucas et al., 2011) were used. Recording electrodes were pulled (Model P87, Sutter Instruments) from borosilicate glass with a capillary in the lumen (1.0 mm outer diameter, 0.75 mm inner diameter, World Precision Instruments). Recording electrodes were filled with NaCl-ACSF and resistances were 3–6 MΩ. Recordings were amplified (Axoclamp 2B, Molecular Devices), digitized (Digidata 1440A, Molecular Devices), and acquired using pClamp (Molecular Devices). For recording sites, a depth in the slice was chosen where the maximal response was elicited, typically 50–100 μm below the surface. For laminar analysis, the exact depth was determined by a calibrated micromanipulator with precision to the nearest micron (Leica Microsystems) and was the same for all recording sites.
Stimulation
Monopolar stimulating electrodes were made from Teflon-coated stainless steel wire (diameter including Teflon, 75 μm; A-M Systems). Stimuli were controlled by a stimulus isolation unit (IsoFlex, AMPI Instruments) that generated current pulses (100 μA), which were varied in duration digitally (10–250 μs) to test different stimulus strengths. Stimuli were triggered every 30–40 s to test basal transmission.
For stimulation of MFs, recordings were in CA3b and the stimulating electrode was placed in the subgranular zone (SGZ) at the crest of the DG (Fig. 1) because this location avoided stimulation of CA3c pyramidal cell dendrites and activation of area CA3 recurrent collaterals to other CA3 pyramidal cells (Claiborne et al., 1993). The response to MF stimulation was recorded at many sites in stratum lucidum in CA3b at the onset of each experiment to determine the location within stratum lucidum where the maximal field EPSP (fEPSP) was recorded. Other methods to confirm that the SGZ stimulus activated MFs are shown in Figure 1. Nevertheless, we cannot exclude the possibility that a pathway in addition to the MF pathway was activated by SGZ stimulation (e.g., the backprojecting axons of CA3 pyramidal cells that extend deep into the hilus) (Scharfman, 2007).
LTP
LTP experiments required a stable baseline, which was defined by a 15 min period when the fEPSP slope varied within 95–105% of the mean of the first 3 evoked responses. For MF plasticity, a 25 Hz, 1 s train was triggered twice with a 10 s interval, and stimulus strength was 50% of the strength that elicited the maximum fEPSP. For θ-burst stimulation (TBS), 10 trains of 4 half-maximal stimuli at 100 Hz were triggered 200 ms apart. fEPSPs were normalized to the average of the 15 min baseline.
Data analysis
fEPSP.
Slope was defined as the slope between two points: the start of the fEPSP and the peak. When a fiber volley preceded an fEPSP, the first point of the fEPSP slope measurement was immediately after the fiber volley. This definition of slope was chosen because the fEPSPs in Gdx rats often had an irregular rising phase, making the maximal slope difficult to measure (see Fig. 9A2). fEPSP amplitude was defined as the difference between the prestimulus baseline and the peak of the fEPSP. Paired pulse facilitation (PPF) was defined as the measurement of the second response divided by the measurement of the first response. Latency was measured from the middle of the stimulus artifact, which was 10–285 μs in duration, to the onset of the initial negative deflection representing the rising phase of the fiber volley (for latency to the fiber volley) or the onset of the rising phase of the fEPSP (for latency to fEPSPs) (Skucas et al., 2011).
Population spike amplitude.
The amplitude of the population spike was defined as the difference between the prestimulus baseline and the peak of the population spike (Scharfman, 1997). Care was taken to record the population spike at a site in stratum pyramidale where it was maximal, which typically was on the border of the pyramidal cell layer and stratum oriens, analogous to area CA1 (Scharfman et al., 2003), rather than a site close to stratum lucidum, where the population spike can be recorded, but is not the site where it is maximal.
Current source density.
Current source density (CSD) analysis was conducted using field potential recordings that began at the border of the alveus with stratum oriens in area CA3b and perpendicular to the CA3 pyramidal cell layer, ending in stratum lacunosum-moleculare at a location that was adjacent to the lateral tip of the superior blade of the DG (Fig. 1). Recordings were ∼25 μm apart and the same depth in the slice, typically 50 μm. Intervals between sites were monitored using a dissecting microscope (Stemi SV6, Carl Zeiss). The one-dimensional CSD profile was calculated offline using MATLAB (Mathworks) by deriving the second spatial derivative of the local field potential recordings using a 3-point formula. The CSD represents the transmembrane sink and source currents that generate the recorded field potential profile.
Slice preparation
After deep anesthesia by isoflurane inhalation, the rat was decapitated, the brain was removed from the skull, and it was placed in ice-cold (4°C) oxygenated (95%O2/5%CO2) sucrose-based artificial CSF (ACSF, in mm: 252.0 sucrose, 5.0 KCl, 2.4 CaCl2, 2.0 MgSO4, 26.0 NaHCO3, 1.25 NaH2PO4, 10.0 d-glucose) for ∼2 min. One hemisphere was glued with cyanoacrylate (Krazy Glue) to the Teflon-coated stage of a vibrating tissue slicer (Vibroslice, Stoelting) and slices (400-μm-thick) were cut in the horizontal plane while immersed in 4°C ACSF. Slices were placed immediately in 100 ml of preoxygenated, room-temperature ACSF, and placed with a wide-bore Pasteur pipette on a nylon net of a recording chamber (modified from one previously sold by Fine Science Tools) (Scharfman et al., 2001) to provide improved flow of ACSF and increased aeration (http://www.healthresearch.org/technology-transfer/brain-and-tissue-slice-recording). Slices were perfused with oxygenated (95%O2/5%CO2), preheated (30°C) ACSF at ∼1 ml/min by a peristaltic pump (Minpuls2 Gilson). The compartment of the recording chamber where slices were placed was maintained at 31–32°C by a temperature controller (PTCO3, Scientific Systems Design). After 30 min, slices were perfused with NaCl-based ACSF (126.0 mm NaCl instead of sucrose). Recordings began 60 min after slices were first placed in the recording chamber and 30 min after the start of perfusion with NaCl-ACSF.
Recording
Similar recording procedures to those described previously (Scharfman et al., 2000, 2007; Skucas et al., 2011) were used. Recording electrodes were pulled (Model P87, Sutter Instruments) from borosilicate glass with a capillary in the lumen (1.0 mm outer diameter, 0.75 mm inner diameter, World Precision Instruments). Recording electrodes were filled with NaCl-ACSF and resistances were 3–6 MΩ. Recordings were amplified (Axoclamp 2B, Molecular Devices), digitized (Digidata 1440A, Molecular Devices), and acquired using pClamp (Molecular Devices). For recording sites, a depth in the slice was chosen where the maximal response was elicited, typically 50–100 μm below the surface. For laminar analysis, the exact depth was determined by a calibrated micromanipulator with precision to the nearest micron (Leica Microsystems) and was the same for all recording sites.
Stimulation
Monopolar stimulating electrodes were made from Teflon-coated stainless steel wire (diameter including Teflon, 75 μm; A-M Systems). Stimuli were controlled by a stimulus isolation unit (IsoFlex, AMPI Instruments) that generated current pulses (100 μA), which were varied in duration digitally (10–250 μs) to test different stimulus strengths. Stimuli were triggered every 30–40 s to test basal transmission.
For stimulation of MFs, recordings were in CA3b and the stimulating electrode was placed in the subgranular zone (SGZ) at the crest of the DG (Fig. 1) because this location avoided stimulation of CA3c pyramidal cell dendrites and activation of area CA3 recurrent collaterals to other CA3 pyramidal cells (Claiborne et al., 1993). The response to MF stimulation was recorded at many sites in stratum lucidum in CA3b at the onset of each experiment to determine the location within stratum lucidum where the maximal field EPSP (fEPSP) was recorded. Other methods to confirm that the SGZ stimulus activated MFs are shown in Figure 1. Nevertheless, we cannot exclude the possibility that a pathway in addition to the MF pathway was activated by SGZ stimulation (e.g., the backprojecting axons of CA3 pyramidal cells that extend deep into the hilus) (Scharfman, 2007).
LTP
LTP experiments required a stable baseline, which was defined by a 15 min period when the fEPSP slope varied within 95–105% of the mean of the first 3 evoked responses. For MF plasticity, a 25 Hz, 1 s train was triggered twice with a 10 s interval, and stimulus strength was 50% of the strength that elicited the maximum fEPSP. For θ-burst stimulation (TBS), 10 trains of 4 half-maximal stimuli at 100 Hz were triggered 200 ms apart. fEPSPs were normalized to the average of the 15 min baseline.
Data analysis
fEPSP.
Slope was defined as the slope between two points: the start of the fEPSP and the peak. When a fiber volley preceded an fEPSP, the first point of the fEPSP slope measurement was immediately after the fiber volley. This definition of slope was chosen because the fEPSPs in Gdx rats often had an irregular rising phase, making the maximal slope difficult to measure (see Fig. 9A2). fEPSP amplitude was defined as the difference between the prestimulus baseline and the peak of the fEPSP. Paired pulse facilitation (PPF) was defined as the measurement of the second response divided by the measurement of the first response. Latency was measured from the middle of the stimulus artifact, which was 10–285 μs in duration, to the onset of the initial negative deflection representing the rising phase of the fiber volley (for latency to the fiber volley) or the onset of the rising phase of the fEPSP (for latency to fEPSPs) (Skucas et al., 2011).
Population spike amplitude.
The amplitude of the population spike was defined as the difference between the prestimulus baseline and the peak of the population spike (Scharfman, 1997). Care was taken to record the population spike at a site in stratum pyramidale where it was maximal, which typically was on the border of the pyramidal cell layer and stratum oriens, analogous to area CA1 (Scharfman et al., 2003), rather than a site close to stratum lucidum, where the population spike can be recorded, but is not the site where it is maximal.
Current source density.
Current source density (CSD) analysis was conducted using field potential recordings that began at the border of the alveus with stratum oriens in area CA3b and perpendicular to the CA3 pyramidal cell layer, ending in stratum lacunosum-moleculare at a location that was adjacent to the lateral tip of the superior blade of the DG (Fig. 1). Recordings were ∼25 μm apart and the same depth in the slice, typically 50 μm. Intervals between sites were monitored using a dissecting microscope (Stemi SV6, Carl Zeiss). The one-dimensional CSD profile was calculated offline using MATLAB (Mathworks) by deriving the second spatial derivative of the local field potential recordings using a 3-point formula. The CSD represents the transmembrane sink and source currents that generate the recorded field potential profile.
fEPSP.
Slope was defined as the slope between two points: the start of the fEPSP and the peak. When a fiber volley preceded an fEPSP, the first point of the fEPSP slope measurement was immediately after the fiber volley. This definition of slope was chosen because the fEPSPs in Gdx rats often had an irregular rising phase, making the maximal slope difficult to measure (see Fig. 9A2). fEPSP amplitude was defined as the difference between the prestimulus baseline and the peak of the fEPSP. Paired pulse facilitation (PPF) was defined as the measurement of the second response divided by the measurement of the first response. Latency was measured from the middle of the stimulus artifact, which was 10–285 μs in duration, to the onset of the initial negative deflection representing the rising phase of the fiber volley (for latency to the fiber volley) or the onset of the rising phase of the fEPSP (for latency to fEPSPs) (Skucas et al., 2011).
Population spike amplitude.
The amplitude of the population spike was defined as the difference between the prestimulus baseline and the peak of the population spike (Scharfman, 1997). Care was taken to record the population spike at a site in stratum pyramidale where it was maximal, which typically was on the border of the pyramidal cell layer and stratum oriens, analogous to area CA1 (Scharfman et al., 2003), rather than a site close to stratum lucidum, where the population spike can be recorded, but is not the site where it is maximal.
Current source density.
Current source density (CSD) analysis was conducted using field potential recordings that began at the border of the alveus with stratum oriens in area CA3b and perpendicular to the CA3 pyramidal cell layer, ending in stratum lacunosum-moleculare at a location that was adjacent to the lateral tip of the superior blade of the DG (Fig. 1). Recordings were ∼25 μm apart and the same depth in the slice, typically 50 μm. Intervals between sites were monitored using a dissecting microscope (Stemi SV6, Carl Zeiss). The one-dimensional CSD profile was calculated offline using MATLAB (Mathworks) by deriving the second spatial derivative of the local field potential recordings using a 3-point formula. The CSD represents the transmembrane sink and source currents that generate the recorded field potential profile.
Immunocytochemistry
Perfusion and processing
After deep anesthesia (urethane, 4 g/kg, intraperitoneally), animals were perfused through the aorta with 4% paraformaldehyde, pH 7.4, as previously described (Scharfman et al., 2002). Brains were left in the skull overnight at 4°C and removed the next day for dynorphin immunocytochemistry. They were postfixed in 4% paraformaldehyde at 4°C and cut on a vibratome into 50-μm-thick sections. For BDNF immunocytochemistry, the brains were removed after perfusion-fixation and placed in 2% paraformaldehyde at 4°C for 2 h. They were cryoprotected by immersion in 30% sucrose dissolved in 0.1 m phosphate buffer. Sections (40 μm thick) were cut on a cryostat (Minotome Plus, International Equipment Company) the next day. Sections were processed that day, as described in Immunocytochemistry: dynorphin.
The antibody to dynorphin B1–13 (kindly provided by Dr. Teresa Milner) has been used previously to evaluate the MF pathway (Pierce et al., 1999). The polyclonal antibody to BDNF (kindly provided by Amgen-Regeneron Partners) has also been used previously and shown robust BDNF immunoreactivity (ir) in the MF pathway (Conner et al., 1997; Yan et al., 1997; Scharfman et al., 2002, 2003). The monoclonal antibody to BDNF was used because previous studies have shown that staining with this antibody led to a qualitatively similar pattern of BDNF-ir as the polyclonal antibody (Scharfman et al., 2007). Notably, processing sections in the absence of the primary antibody failed to produce MF BDNF-ir.
Immunocytochemistry: dynorphin
Sections were processed with a polyclonal antibody to dynorphin B made in rabbit (Pierce et al., 1999), using free-floating sections that were initially washed twice (5 min each) in 0.1 m Tris buffer, pH 7.6, and treated with 1% H2O2 made in 0.1 m Tris buffer (pH 7.6; 30 min). Sections were then washed in 0.1 m Tris buffer (pH 7.6; 5 min) and treated with 0.25% Triton X-100 dissolved in 0.1 m Tris buffer (Tris A; 10 min), followed by 0.25% Triton X-100 and 0.005% BSA in 0.1 m Tris buffer (Tris B; 10 min). Sections were then treated with 10% normal goat serum diluted in Tris B for 45 min. Sections were subsequently washed in Tris A (10 min) followed by Tris B (10 min) and incubated in antisera (1:1000; diluted in Tris B) for 24 h on a rotating shaker. The next day, sections were treated with Tris A (10 min) followed by Tris B (10 min) and then incubated for 45 min with a biotinylated secondary antibody against rabbit IgG made in goat (1:400, Vector Laboratories). Sections were washed in Tris A (10 min), then 0.25% Triton X-100 and 0.005% BSA in 0.5 m Tris buffer (Tris D; pH 7.6; 10 min), and finally incubated for 1 h in avidin-biotin horseradish peroxidase complex diluted in Tris (ABC Standard kit, 1:1000; Vector Laboratories). Immunoperoxidase labeling was developed in DAB (50 mg/100 ml of 0.1 m Tris buffer) with 200 mg of d-glucose, 0.3 mg of glucose oxidase, and 40 mg of ammonium chloride, and then washed three times (5 min each) in 0.1 m Tris buffer, pH 7.6. Sections were mounted on subbed slides, allowed to dry overnight, and then dehydrated in a series of ethanol solutions (70%, 5 min; 95%, 5 min, 100%, 10 min, 100%, 10 min), cleared in xylene (10 min), and coverslipped with Permount (Fisher Scientific). The sections were viewed on a brightfield microscope (BX51, Olympus of America) and photographed using a digital camera (Retiga 2000R, Q Imaging) using ImagePro (Media Cybernetics). To allow comparisons in the figures, micrographs of sections from different experimental groups were taken using the same light and camera settings.
Immunocytochemistry: BDNF
Sections were first washed in 0.05 m potassium PBS (KPBS, pH 7.3) and then treated with 0.5% H2O2 in KPBS for 30 min. After a 5 min wash in KPBS, sections were incubated in 10% normal goat serum for the polyclonal antibody (10% horse serum for the monoclonal antibody), plus 1% BSA and 0.25% Triton X-100 in KPBS for 20 min. Sections were incubated in antisera (polyclonal: 1:30,000; Scharfman et al., 2002, 2003; monoclonal: 1:1000; Sigma-Aldrich; Scharfman et al., 2007) in 1% BSA and 0.25% Triton X-100 in KPBS for 48 h at 4°C and subsequently rinsed 10 times (10 min each) in 0.25% BSA and 0.025% Triton X-100 in 0.01 m KPBS. Sections were then incubated in a biotinylated secondary antibody against rabbit IgG made in goat (1:400; Vector Laboratories) for the polyclonal antibody (or biotinylated secondary antibody against mouse IgG made in horse (1:400; Vector Laboratories) for the monoclonal antibody) and 1% BSA and 0.025% Triton X-100 in KPBS for 1 h, followed by ABC (ABC Standard kit; 1:1000; Vector Laboratories) in 1% BSA in KPBS for 1 h. Sections were rinsed in KPBS followed by Tris buffer, and reacted with DAB using methods identical to those described above for dynorphin, except 10 mm NiCl2 was added to the DAB solution.
Data analysis
Quantification of immunocytochemical data used ImagePro (Media Cybernetics). The same light settings were used for all procedures. For measurement of BDNF in the MF plexus in stratum lucidum, a square region of interest (ROI; ∼50 μm) was placed over the terminal field at the border of CA3a with CA2, where immunoreactivity was greatest (Scharfman et al., 2003, 2007) (see Fig. 9H1). Three measurements (spanning the terminal field of the MFs; see Fig. 9H1) of the grayscale value (scale, 0–255; for the purposes of this analysis, 0 was defined as white and 255 was defined as black) were made for a given section and averaged. Three measurements were also made of the area in stratum radiatum beside the ROI, where immunoreactivity was weak. The average of the three measurements in stratum radiatum was subtracted from the average of the three measurements in stratum lucidum to give the final grayscale value for a given section. The procedure was repeated for two other sections at a similar septotemporal level. The average value from the three sections was used as the best estimate for that animal. A similar procedure was used to determine whether dynorphin-ir in stratum lucidum was increased or decreased in Gdx rats compared with sham rats.
To determine whether mossy fiber sprouting occurred, a different method was used: an ROI was drawn around the MF plexus of CA3a in stratum lucidum and stratum oriens in the section with the most widespread MF sprouting in stratum oriens (see Fig. 9G1). This ROI was then used as a template for all sections to measure the mean grayscale value within the ROI. The template was moved to stratum radiatum to measure the background (see Fig. 9G1), which was subtracted.
Perfusion and processing
After deep anesthesia (urethane, 4 g/kg, intraperitoneally), animals were perfused through the aorta with 4% paraformaldehyde, pH 7.4, as previously described (Scharfman et al., 2002). Brains were left in the skull overnight at 4°C and removed the next day for dynorphin immunocytochemistry. They were postfixed in 4% paraformaldehyde at 4°C and cut on a vibratome into 50-μm-thick sections. For BDNF immunocytochemistry, the brains were removed after perfusion-fixation and placed in 2% paraformaldehyde at 4°C for 2 h. They were cryoprotected by immersion in 30% sucrose dissolved in 0.1 m phosphate buffer. Sections (40 μm thick) were cut on a cryostat (Minotome Plus, International Equipment Company) the next day. Sections were processed that day, as described in Immunocytochemistry: dynorphin.
The antibody to dynorphin B1–13 (kindly provided by Dr. Teresa Milner) has been used previously to evaluate the MF pathway (Pierce et al., 1999). The polyclonal antibody to BDNF (kindly provided by Amgen-Regeneron Partners) has also been used previously and shown robust BDNF immunoreactivity (ir) in the MF pathway (Conner et al., 1997; Yan et al., 1997; Scharfman et al., 2002, 2003). The monoclonal antibody to BDNF was used because previous studies have shown that staining with this antibody led to a qualitatively similar pattern of BDNF-ir as the polyclonal antibody (Scharfman et al., 2007). Notably, processing sections in the absence of the primary antibody failed to produce MF BDNF-ir.
Immunocytochemistry: dynorphin
Sections were processed with a polyclonal antibody to dynorphin B made in rabbit (Pierce et al., 1999), using free-floating sections that were initially washed twice (5 min each) in 0.1 m Tris buffer, pH 7.6, and treated with 1% H2O2 made in 0.1 m Tris buffer (pH 7.6; 30 min). Sections were then washed in 0.1 m Tris buffer (pH 7.6; 5 min) and treated with 0.25% Triton X-100 dissolved in 0.1 m Tris buffer (Tris A; 10 min), followed by 0.25% Triton X-100 and 0.005% BSA in 0.1 m Tris buffer (Tris B; 10 min). Sections were then treated with 10% normal goat serum diluted in Tris B for 45 min. Sections were subsequently washed in Tris A (10 min) followed by Tris B (10 min) and incubated in antisera (1:1000; diluted in Tris B) for 24 h on a rotating shaker. The next day, sections were treated with Tris A (10 min) followed by Tris B (10 min) and then incubated for 45 min with a biotinylated secondary antibody against rabbit IgG made in goat (1:400, Vector Laboratories). Sections were washed in Tris A (10 min), then 0.25% Triton X-100 and 0.005% BSA in 0.5 m Tris buffer (Tris D; pH 7.6; 10 min), and finally incubated for 1 h in avidin-biotin horseradish peroxidase complex diluted in Tris (ABC Standard kit, 1:1000; Vector Laboratories). Immunoperoxidase labeling was developed in DAB (50 mg/100 ml of 0.1 m Tris buffer) with 200 mg of d-glucose, 0.3 mg of glucose oxidase, and 40 mg of ammonium chloride, and then washed three times (5 min each) in 0.1 m Tris buffer, pH 7.6. Sections were mounted on subbed slides, allowed to dry overnight, and then dehydrated in a series of ethanol solutions (70%, 5 min; 95%, 5 min, 100%, 10 min, 100%, 10 min), cleared in xylene (10 min), and coverslipped with Permount (Fisher Scientific). The sections were viewed on a brightfield microscope (BX51, Olympus of America) and photographed using a digital camera (Retiga 2000R, Q Imaging) using ImagePro (Media Cybernetics). To allow comparisons in the figures, micrographs of sections from different experimental groups were taken using the same light and camera settings.
Immunocytochemistry: BDNF
Sections were first washed in 0.05 m potassium PBS (KPBS, pH 7.3) and then treated with 0.5% H2O2 in KPBS for 30 min. After a 5 min wash in KPBS, sections were incubated in 10% normal goat serum for the polyclonal antibody (10% horse serum for the monoclonal antibody), plus 1% BSA and 0.25% Triton X-100 in KPBS for 20 min. Sections were incubated in antisera (polyclonal: 1:30,000; Scharfman et al., 2002, 2003; monoclonal: 1:1000; Sigma-Aldrich; Scharfman et al., 2007) in 1% BSA and 0.25% Triton X-100 in KPBS for 48 h at 4°C and subsequently rinsed 10 times (10 min each) in 0.25% BSA and 0.025% Triton X-100 in 0.01 m KPBS. Sections were then incubated in a biotinylated secondary antibody against rabbit IgG made in goat (1:400; Vector Laboratories) for the polyclonal antibody (or biotinylated secondary antibody against mouse IgG made in horse (1:400; Vector Laboratories) for the monoclonal antibody) and 1% BSA and 0.025% Triton X-100 in KPBS for 1 h, followed by ABC (ABC Standard kit; 1:1000; Vector Laboratories) in 1% BSA in KPBS for 1 h. Sections were rinsed in KPBS followed by Tris buffer, and reacted with DAB using methods identical to those described above for dynorphin, except 10 mm NiCl2 was added to the DAB solution.
Data analysis
Quantification of immunocytochemical data used ImagePro (Media Cybernetics). The same light settings were used for all procedures. For measurement of BDNF in the MF plexus in stratum lucidum, a square region of interest (ROI; ∼50 μm) was placed over the terminal field at the border of CA3a with CA2, where immunoreactivity was greatest (Scharfman et al., 2003, 2007) (see Fig. 9H1). Three measurements (spanning the terminal field of the MFs; see Fig. 9H1) of the grayscale value (scale, 0–255; for the purposes of this analysis, 0 was defined as white and 255 was defined as black) were made for a given section and averaged. Three measurements were also made of the area in stratum radiatum beside the ROI, where immunoreactivity was weak. The average of the three measurements in stratum radiatum was subtracted from the average of the three measurements in stratum lucidum to give the final grayscale value for a given section. The procedure was repeated for two other sections at a similar septotemporal level. The average value from the three sections was used as the best estimate for that animal. A similar procedure was used to determine whether dynorphin-ir in stratum lucidum was increased or decreased in Gdx rats compared with sham rats.
To determine whether mossy fiber sprouting occurred, a different method was used: an ROI was drawn around the MF plexus of CA3a in stratum lucidum and stratum oriens in the section with the most widespread MF sprouting in stratum oriens (see Fig. 9G1). This ROI was then used as a template for all sections to measure the mean grayscale value within the ROI. The template was moved to stratum radiatum to measure the background (see Fig. 9G1), which was subtracted.
Pharmacology
Administration of drugs to hippocampal slices
Dihydrotestosterone (DHT; Sigma-Aldrich) and 5α-androstane-3α, 17β-diol (Adiol; Sigma-Aldrich) were dissolved in 100% ethanol as a 1 m stock solution and stored at 4°C; the stock solution was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (50 nm). K252a (LC Labs) was dissolved in DMSO (Sigma-Aldrich) as a 1 m stock solution and stored at 4°C. It was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (300 nm). TrkB-IgG (R&D Systems), control IgG (R&D Systems), and 2-(2,3-dicarboxycyclopropyl)-glycine (DCG-IV; Tocris Bioscience) were dissolved in 0.9% NaCl, stored at 4°C as a concentrated stock solution (1 mm), and diluted in NaCl-ACSF on the day of use. Testosterone and placebo time-release pellets (25 mg/pellet, 21 d release; #A-151 and #C-111, respectively; Innovative Research of America) were stored at room temperature in the dark. The testosterone dose was sufficient to replace normal serum testosterone levels (Fig. 2F3).
Hormone replacement
A 1/4-inch incision was made between the shoulder blades immediately after gonadal removal or sham treatment while the animal was still anesthetized. A 1/8-inch-diameter pellet was inserted under the skin. Sterile sutures were used to close the incision.
Testosterone ELISA was conducted to determine circulating levels of testosterone. Trunk blood was collected in 15 ml of centrifuge tubes at the time of death and placed at room temperature for 20–30 min. Serum was separated by centrifugation (3000 rpm, 15 min, 23°C) and frozen (−20°C) until use. A testosterone ELISA was performed using methods provided by the manufacturer (Mouse and Rat Testosterone Kit, Calbiochem). Details were provided previously (Chakraborty et al., 2012).
Administration of drugs to hippocampal slices
Dihydrotestosterone (DHT; Sigma-Aldrich) and 5α-androstane-3α, 17β-diol (Adiol; Sigma-Aldrich) were dissolved in 100% ethanol as a 1 m stock solution and stored at 4°C; the stock solution was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (50 nm). K252a (LC Labs) was dissolved in DMSO (Sigma-Aldrich) as a 1 m stock solution and stored at 4°C. It was diluted in NaCl-ACSF on the day of the experiment to reach the final concentration (300 nm). TrkB-IgG (R&D Systems), control IgG (R&D Systems), and 2-(2,3-dicarboxycyclopropyl)-glycine (DCG-IV; Tocris Bioscience) were dissolved in 0.9% NaCl, stored at 4°C as a concentrated stock solution (1 mm), and diluted in NaCl-ACSF on the day of use. Testosterone and placebo time-release pellets (25 mg/pellet, 21 d release; #A-151 and #C-111, respectively; Innovative Research of America) were stored at room temperature in the dark. The testosterone dose was sufficient to replace normal serum testosterone levels (Fig. 2F3).
Hormone replacement
A 1/4-inch incision was made between the shoulder blades immediately after gonadal removal or sham treatment while the animal was still anesthetized. A 1/8-inch-diameter pellet was inserted under the skin. Sterile sutures were used to close the incision.
Testosterone ELISA was conducted to determine circulating levels of testosterone. Trunk blood was collected in 15 ml of centrifuge tubes at the time of death and placed at room temperature for 20–30 min. Serum was separated by centrifugation (3000 rpm, 15 min, 23°C) and frozen (−20°C) until use. A testosterone ELISA was performed using methods provided by the manufacturer (Mouse and Rat Testosterone Kit, Calbiochem). Details were provided previously (Chakraborty et al., 2012).
Statistics
Data are expressed as mean ± SEM. The p criterion was 0.05. No more than two slices were used for a given animal. Student's t test were two-tailed and conducted using Microsoft Excel 2007 (Microsoft). Fisher's exact test, χ test, ANOVA, repeated-measures ANOVA (RM-ANOVA), ANCOVA, and nonparametric Kruskal-Wallis ANOVA were performed using Prism (GraphPad). When reporting the results of ANOVA or RM-ANOVA, interactions between factors are reported in the Results only in those cases where they were statistically significant. For parametric ANOVA, where there was inhomogeneity of variance, which was determined by Bartlett's test (Prism), data were log-transformed before statistical analysis (Zivin and Bartko, 1976).
Results
MF transmission is increased in Gdx rats
Recording procedures to study MF transmission
We first confirmed that our stimulation and recording procedures evoked responses that were mediated by MFs in adult male rats that were the same age as those that would be used for experiments (∼2 months old). As shown in Figure 1A–C, a stimulating electrode was placed at the crest of the DG in the SGZ, with recording electrodes in stratum lucidum of area CA3b. fEPSPs had the characteristics previously reported for MF transmission: (1) short latency to onset of the fEPSP (<4 ms), consistent with monosynaptic transmission; (2) robust paired pulse facilitation (Salin et al., 1996; Kamiya et al., 2002; Scott et al., 2008; Cosgrove et al., 2009); and (3) blockade by 1 μm DCG-IV, a metabotropic glutamate receptor subtype II agonist that blocks MF transmission (Figure 1D–F) (Yoshino et al., 1996). Another characteristic, a fiber volley before the fEPSP, occurred when stimulus strength was sufficiently strong (Fig. 1E, black arrow).
Laminar analysis of the response to stimulation in CA3b showed that the fEPSP recorded in stratum lucidum was followed by a later fEPSP that was only recorded in stratum radiatum, and reversed polarity in the pyramidal cell layer (Fig. 1D1,E1). CSD analysis (Fig. 1D2,E2) showed a short-latency sink in stratum lucidum and a longer latency sink in stratum radiatum, suggesting a monosynaptic fEPSP corresponding to activation of MFs, followed by a disynaptic fEPSP reflecting activation of recurrent collaterals of pyramidal cells in stratum radiatum. The field potentials recorded in stratum lucidum and the cell layer were reduced >90% by DCG-IV (1 μm; n = 8 slices, 3 rats; Fig. 1F). Notably, 1 μm DCG-IV had no detectable effect on fEPSPs evoked by Schaffer collateral stimulation, recorded in stratum radiatum of CA1 (2 slices from 2 sham controls; fEPSP slope 30 min after DCG-IV: 99.4% of pre-DCG-IV baseline).
A small fEPSP in stratum radiatum remained after DCG-IV (Fig. 1F2), which could be due to the fact that DCG-IV did not block the response entirely. It is also possible that activation of CA3 pyramidal cell axons in the hilus (Scharfman, 2007) led to activation of recurrent collaterals to area CA3 pyramidal cell dendrites in stratum radiatum. For these reasons, the term “MF fEPSP” is used below to refer only to short-latency fEPSPs recorded in stratum lucidum. “Recurrent fEPSP” is used below to refer only to the long latency fEPSPs recorded in stratum radiatum. It should be noted that the term “MF LTP” is used with caution below because we cannot rule out a possible contribution of the recurrent collateral pathway during LTP induction.
MF fEPSPs are increased in Gdx rats
For the comparisons of MF fEPSPs between Gdx and sham rats, animals had surgery at ages that were not different (Gdx: 68.4 ± 2.4-d-old; sham: 57.2 ± 3.6-d-old; n = 10 slices from 8 rats/group; Student's t test, p = 0.973), and hippocampal slices were made at similar delays after surgery (Gdx: 14.5 ± 1.6 d; sham: 13.6 ± 1.0 d; n = 10 slices from 8 rats/group; Student's t test, p = 0.736). As shown in Figure 2A, MF fEPSPs were largest in Gdx rats (n = 10 slices, 8 rats/group). There was a significant effect of Gdx on the input-output relation for fEPSP slope (two-way RM-ANOVA; F(1,198) = 5.710; p = 0.028; Fig. 2B) and a significant interaction between treatment and stimulus strength (F(11,198) = 2.566; p = 0.005) with significantly larger fEPSP slopes in Gdx rats at all stimulus strengths except the two weakest intensities (Student's t test, p < 0.05; Fig. 2B). Results were similar for amplitude; there was a significant effect of Gdx (two-way RM-ANOVA; F(1,198) = 66.086; p < 0.0001) and an interaction between Gdx and stimulus strength (F(11,198) = 14.986; p < 0.0001), with fEPSPs from Gdx rats greater in sham rats at all stimulus strengths except the weakest stimuli (Student's t test, p < 0.05).
When fiber volleys were measured, there was no significant effect of Gdx on the input-output relation by two-way RM-ANOVA (F(1,198) = 0.722; p = 0.407; Fig. 2C). However, ANOVA-based 4-parameter linear regression analysis (Scharfman et al., 2007) indicated that fiber volley amplitudes in Gdx rats were significantly higher than those for sham-operated controls (F(1,28) = 18.92; p < 0.001).
Gdx rats exhibited a greater MF fEPSP slope for a given fiber volley amplitude (Fig. 2D). Excluding the MF fEPSPs elicited at the two weakest intensities (where fiber volleys were not present), the slope of the relation was significantly greater for Gdx rats (linear regression analysis; F(1,17) = 11.486; p = 0.004; Fig. 2D).
Although fEPSPs were larger in Gdx rats, there were no differences in latencies. First, we measured the latency to onset of the peak of the maximal fiber volley and found that there were no significant differences between experimental groups (Gdx: 1.96 ± 0.33 ms; sham: 2.34 ± 0.26 ms; Student's t test; p = 0.370; Fig. 2E1). Then we measured the latencies to onset of MF fEPSPs; both minimal and maximal fEPSPs were measured because latency varies depending on the stimulus strength (latency is longest when stimuli are weak). There were no differences between fEPSPs of Gdx and sham rats when stimuli were minimal (35 μs; Gdx: 3.84 ± 0.38 ms; sham: 3.81 ± 0.23 ms; Student's t test, p = 0.952; Fig. 2E2) and no differences in the maximal fEPSPs (Gdx: 3.38 ± 0.28 ms; sham: 3.36 ± 0.12 ms; Student's t test, p = 0.957).
To confirm that effects of Gdx were the result of a reduction in serum levels of testosterone, we determined whether testosterone replacement to Gdx rats blocked effects of Gdx on MF transmission. For this purpose, a time-release pellet of either testosterone or placebo was inserted subcutaneously at the time of surgery (Fig. 2F1). There were four experimental groups: Gdx rats treated with a testosterone pellet (10 slices, 5 rats), Gdx rats that received a placebo pellet (12 slices, 6 rats), untreated Gdx rats (27 slices, 17 rats), and untreated sham controls (16 slices, 10 rats). MF fEPSPs were tested ∼2 weeks after surgery. There were differences between the four groups (one-way ANOVA; F(3,64) = 2.668; p = 0.015), with smaller fEPSP slopes in Gdx animals that received testosterone compared with Gdx rats treated with placebo (Student's t test, p = 0.008; Fig. 2F2). MF fEPSPs of Gdx animals that received placebo were not significantly different from fEPSPs of untreated Gdx rats (Student's t test, p = 0.651), and there were no significant differences between Gdx rats that received testosterone and untreated sham controls (Student's t test, p = 0.702; Fig. 2F2).
To confirm that serum testosterone levels were restored by testosterone pellets, testosterone was measured by ELISA from trunk blood collected at the time of death. Gdx rats that were untreated or treated with placebo had very low testosterone levels; those that were below the lower limit of the assay were assigned the lower limit for their testosterone value (0.10 ng/ml; Fig. 2F3). There were significant differences between the four experimental groups (Kruskal–Wallis statistic, 32.584, p < 0.0001; Fig. 2F3). Circulating testosterone levels in Gdx rats treated with testosterone (8.08 ± 1.4 ng/ml; n = 5 rats) were significantly greater than Gdx rats treated with placebo (0.16 ± 0.04 ng/ml; n = 6; Dunn's multiple-comparisons test, p > 0.05; Fig. 2F3). However, Gdx rats treated with testosterone were not significantly different from untreated sham controls (6.52 ± 0.75 ng/ml; n = 10; p > 0.05; Fig. 2F3). The mean value for serum testosterone of Gdx rats treated with placebo was not significantly different from Gdx rats that were untreated (0.19 ± 0.09 ng/ml; n = 6; p > 0.05; Fig. 2F3). These data confirmed that Gdx reduced serum testosterone to a negligible level and testosterone treatment increased it to levels that were normal for the strain and age of rats that were used.
PPF, post-tetanic potentiation, and LTP are increased in Gdx rats
PPF of MF fEPSPs was greater in Gdx rats than sham controls (Fig. 3). PPF was tested using half-maximal stimulation and a 40 ms interstimulus interval because these are parameters where PPF is typically robust (Scharfman, 1997; Scharfman et al., 2003, 2007). PPF was significantly greater in slices from Gdx rats compared with sham controls (Gdx: 166.8 ± 3.2%; n = 15 slices, 12 rats; sham: 121.0 ± 3.2%; 14 slices, 12 rats; Student's t test, p = 0.006).
We also evaluated PPF in additional rats (Gdx: 10 slices, 7 rats; sham: n = 10 slices, 6 rats) using a range of stimulus strengths because PPF has not been evaluated in detail at the ages we used, raising the possibility that PPF in sham rats might have been underestimated using a half-maximal stimulus. There was no significant difference in PPF across stimulus strengths in sham rats (one-way ANOVA; F(11,95) = 0.096; p = 0.880). PPF in Gdx rats was significantly greater than sham rats at two stimulus strengths in the middle of the input-output relation (two-way RM-ANOVA; F(11,154) = 7.907; p = 0.014; Student's t test, p < 0.05 for 110 and 135 μs; Fig. 3A2).
Next, PPF was evaluated using a range of interstimulus intervals (10, 20, 30, 40, 60, 80, 100, 150, and 200 ms; Gdx: 10 slices, 9 rats; sham: 12 slices, 7 rats) with a half-maximal stimulus strength. As shown in Figure 3B, Gdx influenced PPF (two-way RM-ANOVA; F(1,160) = 21.560; p < 0.001) with a significant interaction between interstimulus interval and effect of Gdx (F(8,160) = 3.453; p = 0.001). PPF was greater in Gdx rats when interstimulus intervals were intermediate (30, 40, or 60 ms; Student's t test, p < 0.05; Fig. 3B1,B2).
To examine post-tetanic potentiation (PTP) and LTP, we planned to use TBS, but several slices from Gdx rats exhibited spreading depression (SD) in response to TBS (described further below). Therefore, LTP was induced using two 25 Hz trains, 1 s in duration, 10 s apart; a similar procedure (25 Hz, 1 s or 25 Hz, 5 s) has been shown to elicit MF LTP in 18- to 25-d-old Wistar rats (Kwon and Castillo, 2008), which we confirmed in 30-d-old Sprague Dawley rats (147.7 ± 6.2%; n = 3 slices from 3 rats). As shown in Figure 3C1, Gdx rats exhibited LTP but sham rats did not (presumably because of their age; Gdx: 169.1 ± 20.7%; n = 9 slices, 8 rats; sham: 88.3 ± 14.1%; n = 8 slices, 7 rats; Student's t test, p = 0.008). Both groups exhibited robust PTP, but Gdx rats had greater PTP (two-way RM-ANOVA for the first 3 min after LTP induction; F(1,140) = 9.231; p = 0.009; Student's t test, p < 0.05; Fig. 3C2).
Gdx increases MF-evoked field potentials recorded outside stratum lucidum
Responses to MF stimulation were also recorded in the pyramidal cell layer to evaluate the extent that an increase in MF transmission led to enhanced pyramidal cell output, reflected by an increase in the population spike amplitude (Fig. 4A1–A2; Gdx: 17 slices, 12 rats; sham: 12 slices, 6 rats). An input-output curve for the population spike showed that there was a significant effect of Gdx (two-way RM-ANOVA; F(1,108) = 4.737; p = 0.038; Fig. 4A2). There was an interaction between surgical status (Gdx vs sham) and stimulus strength (F(4,108) = 3.749; p = 0.007), with a significant increase in population spike amplitude in Gdx rats at the highest stimulus intensities (Student's t test, p < 0.05; Fig. 4A2).
PPF of the population spike was greater in Gdx rats (half-maximal stimulus intensity; 40 ms interstimulus interval; Gdx: 137.8 ± 2.3%; n = 18 slices, 13 rats; sham: 109.1 ± 1.9%; n = 11 slices, 7 rats; Student's t test, p = 0.040; Fig. 4B,C). When multiple stimulus strengths were compared in a subset of these slices (Gdx: 10 slices, 7 rats; sham: 7 slices, 6 rats), two-way RM-ANOVA showed that there was an interaction between surgical treatment (Gdx or sham) and stimulus strength (F(3,42) = 3.257; p = 0.031) with PPF in slices from Gdx rats greater at the lowest intensity (Student's t test, p < 0.05) but not higher intensities (Student's t test, p > 0.05; Fig. 4B2).
PPF was also evaluated using a range of interstimulus intervals and a half-maximal stimulus strength (Gdx: 12 slices, 8 rats; sham: 10 slices, 6 rats). Two-way RM-ANOVA showed that there was an interaction between surgical treatment and interstimulus interval (F(1,140) = 4.072; p = 0.033) with significantly greater PPF in Gdx rats at 40 and 60 ms interstimulus intervals (Student's t test, p = 0.016 and p = 0.028, respectively; Fig. 4C1,C2). There were no significant differences between Gdx and sham rats at the short interstimulus intervals, when paired pulse inhibition typically occurs (10 or 20 ms intervals; Student's t test, p = 0.162 and p = 0.220 respectively; Fig. 4C1,C2). Therefore, Gdx rats had greater PPF of the population spike, but there was no effect on paired pulse inhibition.
Another cohort of rats was used to determine whether the delayed fEPSP recorded in stratum radiatum (referred to as the “recurrent fEPSP” below) was affected by Gdx. As shown in Figure 4D, recurrent fEPSP slopes were increased in Gdx rats (17 slices, 14 rats) compared with sham controls (16 slices, 14 rats). The input-output relation was significantly different (two-way RM-ANOVA; F(1,341) = 8.095; p = 0.008), and there was an interaction between surgical pretreatment and stimulus strength (F(11,341) = 3.269; p < 0.001) with recurrent fEPSPs of Gdx rats greater than sham at all stimulus strengths except the two lowest intensities (Student's t test, p < 0.05; Fig. 4D2). The lack of a significant effect at low stimulus intensities is consistent with the idea that recurrent fEPSPs required strong intensities of stimulation, so that pyramidal cells would reach threshold for action potential generation and activate their recurrent collaterals.
The latencies to onset of the recurrent fEPSP were not affected by Gdx. The recurrent fEPSP that was evoked by a weak stimulus (a 35 μs stimulus) had a 5.68 ± 0.16 ms latency to onset in slices from Gdx rats, and latency was 5.69 ± 0.18 ms in slices from sham rats (Student's t test; p = 0.965). The latencies to onset of the maximal recurrent fEPSP were not different either (Gdx: 5.02 ± 0.11; sham: 5.05 ± 0.13; Student's t test, p = 0.852).
PPF of the recurrent fEPSP slope (half-maximal stimulus, 40 ms interstimulus interval) was significantly greater in Gdx rats (120.7 ± 6.7%, n = 15 slices, 11 rats; sham: 98.3 ± 7.0%; n = 10 slices, 9 rats; p = 0.034).
DHT sensitivity of MF transmission and plasticity in Gdx rats
MF fEPSPs of Gdx rats are reduced by DHT
Testosterone is metabolized in the brain to DHT, a potent agonist at the androgen receptors, which are present in MFs of the adult male rat (Tabori et al., 2005). Therefore, we determined whether a deficiency in DHT was the reason for the differences in MF transmission between Gdx and sham rats. We also determined whether one of the metabolites of DHT, Adiol, would have effects on MF fEPSPs because several studies suggest a potent effect of Adiol to enhance the actions of GABA at GABAA receptors (Frye et al., 2001; Edinger et al., 2004; Frye and Edinger, 2004; Frye et al., 2004; Reddy, 2004a,b; Edinger and Frye, 2005; Kaminski et al., 2005; Frye et al., 2010; Reddy and Jian, 2010).
MF stimulation in normal ACSF was initially tested for a 10 min baseline period to confirm that fEPSP slope was stable (Fig. 5A1). Then slices were exposed to vehicle (0.0005% ethanol in ACSF), and the responses that were recorded 30 min later were compared with the baseline to confirm that vehicle had no detectable effect. Next, slices were exposed for 30 min to Adiol at a concentration that induced near-maximal potentiation of GABAA receptors in whole-cell recordings from CA1 pyramidal neurons (50 nm) (Reddy and Jian, 2010). Afterward, ACSF containing 50 nm DHT was applied for 30 min.
As shown in Fig. 5A2, MF fEPSPs of Gdx rats declined during exposure to DHT. One-way RM-ANOVA in Gdx rats showed that there was a significant change in fEPSP slope during the experiment (F(3,15) = 21.786; p < 0.001) with fEPSP slope decreasing to 72.3 ± 3.2% of control after DHT, significantly different from responses recorded before DHT (paired t test, p = 0.017). However, there were no significant effects of vehicle or Adiol (paired t tests, p > 0.05). In sham rats, there was no effect of either DHT or Adiol (one-way RM-ANOVA; F(3,11) = 0.316; p = 0.814; Fig. 5B).
These experiments suggested that DHT decreased MF transmission but Adiol did not. However, the effect of DHT could have resulted in part from pretreatment with Adiol because Adiol can be converted back to DHT in the brain (Chetyrkin et al., 2001). It is also possible that altering GABAergic transmission by preexposure to Adiol might affect subsequent responses to DHT, even though Adiol by itself had no measurable effect. To test this possibility, slices from Gdx rats were treated with 50 nm DHT immediately after the baseline (Fig. 5B1). DHT reduced the slope of the fEPSP to 74.8 ± 5.6% of control (n = 5 slices, 5 rats; paired t test, p = 0.004; Fig. 5B2). In contrast, there were no effects of Adiol after it was added (immediately after the baseline; n = 3 slices, 3 rats; Fig. 5C1,C2), and there was no significant effect of DHT in slices from sham controls (n = 4 slices, 4 rats; paired t test, p = 0.384; data not shown).
To confirm that DHT had effects on MF-evoked population spikes, and also that DHT inhibited MF transmission at multiple stimulus strengths, recordings were made in the pyramidal cell layer of Gdx rats in response to three intensities of MF stimulation: ∼30%, 50%, and 100% of the maximum. Comparisons were made between baseline and 30 min after the onset of treatment (vehicle, n = 4 slices, 4 rats; DHT, n = 4 slices, 4 rats). There was a significant effect of DHT (RM-ANOVA; F(1,12) = 40.17; p < 0.001), with amplitudes lower after DHT exposure at all stimulus intensities (Bonferroni's tests, p < 0.05). There was a significant interaction between stimulus strength and treatment (F(2,12) = 8.72; p = 0.005).
In summary, DHT reduced fEPSPs in Gdx rats, but Adiol did not, and there were no significant effects of any of the steroids in slices from sham controls. The results suggest that an androgen receptor-sensitive component of MF transmission emerges after Gdx.
MF LTP in Gdx rats is reduced by DHT
Because a DHT-sensitive component of synaptic transmission emerged after Gdx, we tested the DHT dependence of LTP in Gdx rats. After a 15 min baseline period, slices were exposed to 50 nm DHT or vehicle (Fig. 5D). LTP was induced 15 min later, the time when DHT began to exert its effects in the experiments described above. In slices where the fEPSPs declined within 15 min, an adjustment in stimulus intensity was made so that the train used to induce LTP would be half-maximal. Exposure to DHT continued until 15 min after LTP induction, and then drug-free buffer was used for the remainder of the experiment. LTP was reduced in DHT-treated slices compared with slices from Gdx rats treated with vehicle (Gdx + DHT, 85.5 ± 10.1%; Gdx + vehicle, 135.1 ± 13.1%; Student's t test; p = 0.020; Fig. 5D).
Figure 5E compares LTP in slices from Gdx rats that were preexposed to DHT before LTP was tested, Gdx rats pretreated with vehicle, slices from Gdx rats that had no pretreatment, and slices from sham rats that were untreated. LTP in DHT-treated slices from Gdx rats was comparable with LTP in slices of untreated sham rats (Student's t test, p = 0.475); LTP in vehicle-treated slices of Gdx rats was similar to untreated slices of Gdx rats (Student's t test, p = 0.565).
PTP was also reduced by DHT (Fig. 5E). Compared with vehicle-treated slices, fEPSP slopes in the first 3 min after LTP induction were significantly reduced by DHT (two-way RM-ANOVA (F(1,45) = 5.291; p = 0.047; Fig. 5F).
In summary, a DHT-sensitive component of MF PTP and LTP was present in slices from Gdx rats. Sham rats did not exhibit LTP, so the DHT-sensitive component of PTP and LTP was only evident in Gdx rats.
Hyperexcitability in Gdx rats
While recording in the pyramidal cell layer of Gdx rats, increased excitability was detected in a subset of slices (Fig. 6). The increase in excitability was manifested in several ways. For example, a single MF stimulus could evoke >1 population spike, whereas in sham rats there was only one population spike/stimulus. Also, there were small (∼1 mV) spontaneous field potentials in slices of Gdx rats when recordings were made in the pyramidal cell layer, and more unit activity than sham rats (Fig. 6A).
Another sign of abnormal excitability in slices from Gdx rats was SD in response to repetitive MF stimulation (Fig. 6B). When SD followed TBS stimulation of the MFs, it did so rapidly (within seconds; Fig. 6B). SD episodes were stereotypical: they were composed of an initial phase where small (∼1 mV) spontaneous population spikes occurred and a second phase where there was a negative shift in the DC potential, which peaked at −14 to −16 mV within ∼15 s of the onset of spontaneous activity (Fig. 6B). Subsequently, there was a recovery of the DC potential within ∼5 min (Fig. 6B). During the recovery, no response to MF stimulation could be elicited even in response to the highest stimulus strength.
The first type of hyperexcitability, where >1 population spike per stimulus occurred, was found in 10 of 45 (22%) slices from 20 Gdx rats and 0 of 56 slices in 24 sham rats (χ test, p < 0.001; Fig. 6C). SD after TBS stimulation occurred in 8 of 12 (75%) slices from 5 Gdx rats and 0 of 11 slices in 5 sham rats (Fisher's exact test, p = 0.001; Fig. 6C). Four of the 8 slices from Gdx rats that exhibited SD also showed more than one population spike/MF stimulus.
Another indication of abnormal excitability is shown in Figure 6D1. In this case, field potentials were repetitive but at a rhythm that was slower (∼20 Hz) than the multiple population spikes in Figure 6A. CSD analysis (Fig. 6D2,D3) showed the stereotypical nature of the rhythmic field potentials across laminae of CA3 and that the earliest sink was in stratum lucidum (Fig. 6D2,D3, arrow), suggesting the MF pathway initiated activity. Five of 45 (11%) slices from Gdx rats exhibited the rhythmic field potentials shown in Figure 6C, D; these were not the same slices that showed other signs of increased excitability in Figure 6A, B. Therefore, the total number of slices exhibiting hyperexcitability of one type or another was 19 of 45, or 42%.
The abnormal excitability that was recorded in Gdx rats was similar to the effects of recombinant BDNF superfused onto slices of normal adult male rats (Scharfman, 1997) and the responses to MF stimulation in rodents with elevated BDNF protein levels (Croll et al., 1999; Scharfman et al., 2003, 2007), which were blocked by the Trk receptor antagonist K252a (Scharfman, 1997; Scharfman et al., 2003, 2007). Therefore, we determined whether K252a would reduce excitability in Gdx rats.
As shown in Figure 7A, K252a reduced the first type of abnormal excitability in Gdx rats (>1 population spike/MF stimulus; Fig. 6A). Specifically, perfusion of slices with 300 nm K252a for 15 min blocked the second population spike that was elicited by MF stimulation (n = 3 slices, 2 rats; Fig. 7A), and vehicle (0.003% DMSO) had no effect (2 slices from 2 other rats; data not shown). In slices where repetitive field potentials were elicited by a single stimulus (e.g., Fig. 6D), an input-output curve for the amplitude of the first field potential was generated before and after exposure to K252a (Fig. 7B1–B3). There was a significant effect of K252a (two-way RM-ANOVA; F(1,20) = 87.483; p < 0.001) and an interaction (F(5,20) = 15.129; p < 0.001) with significant differences at all intensities above the minimum (where evoked potentials were negligible in all conditions; Student's t test, p < 0.05; Fig. 7B1–B3). There was no effect of vehicle (pre vs post vehicle, two-way RM-ANOVA; F(5,20) = 0.188; p = 0.687; Fig. 7B3).
To evaluate the effect of K252a on SD, SD was elicited by stimulus trains (1 Hz, paired pulses for 10 s) as described previously (Scharfman, 1997; Scharfman et al., 2003, 2007), which were triggered once every 15 min (Fig. 7C,D; K252a: n = 3 slices, 3 Gdx rats; vehicle: n = 3 slices, 3 other Gdx rats). After a 30 min baseline period, K252a was added to the ACSF for 15 min, followed by drug-free buffer (Fig. 7C,D). There was a significant effect of K252a on the amplitude of SD episodes (measured from baseline to the peak of the SD episode; one-way RM-ANOVA; F(9,29) = 14.268; p < 0.001) at 15 and 30 min after K252a was added (paired t test, p = 0.033 and p = 0.031, respectively; Fig. 7D2). There were no detectable effects of vehicle (one-way RM-ANOVA; F(9,29) = 1.049; p = 0.442; Fig. 7D2).Together, the results suggested that a subset of Gdx rats exhibited hyperexcitability that was dependent on Trk receptors.
BDNF protein expression increases in MFs after Gdx
Increased excitability in Gdx rats (Figs. 6 and and7)7) was similar to rats or mice with elevation of BDNF levels in MFs (Croll et al., 1999; Scharfman et al., 2003, 2007). Therefore, we determined whether MF BDNF levels were increased in Gdx rats.
Initial experiments used a well-described polyclonal antibody to BDNF (Conner et al., 1997; Yan et al., 1997) and showed that Gdx rats (which had surgery at ∼60 d of age and were evaluated ∼2 weeks later) exhibited greater BDNF-ir in the MFs than sham rats (Fig. 8A1–A2).
Similar results were obtained using a second antibody (monoclonal; Fig. 8B–D). Using the second antibody, 5 immunocytochemical procedures were conducted. For each procedure, a Gdx and sham littermate were subjected to surgery on the same day, at ∼60 d of age (61.9 ± 2.0 d, n = 5/group) followed by perfusion-fixation on the same day after surgery (12.0 ± 0.6 d), and brains were processed together. Sections from the 10 rats were quantified and analyzed in pairs (Gdx and sham). The location where the MF terminal plexus ends (Fig. 9H1) was used for quantification because BDNF-ir was always greatest there, regardless of surgical pretreatment (Fig. 8D). The timeline of the experiments is shown in Figure 8E1, and the results shown in Figure 8E2 demonstrated that MF BDNF-ir was greater in Gdx rats (Student's t test, p = 0.001).
A second experiment was conducted to determine whether the increase in BDNF protein in MFs in Gdx rats was long-lasting (Fig. 8F). For this purpose, rats were perfusion-fixed 2 months instead of 2 weeks after surgery (Fig. 8F1). Again, surgery was conducted in pairs, so the age at surgery was the same (59.8 ± 1.0 d) and delays between surgery and perfusion were the same (61.2 ± 1.9 d, n = 5/group). As shown in Figure 8F2, Gdx rats exhibited greater MF BDNF-ir than sham controls (Student's t test, p = 0.002). Therefore, the increase in MF BDNF-ir after Gdx appeared to be robust and long-lasting.
Gdx rats exhibit MF sprouting
As recordings in stratum pyramidale and stratum oriens were conducted, it became clear that Gdx rats often exhibited fiber volleys and fEPSPs in these layers in response to MF stimulation (Fig. 9A), leading to diverse sinks in these layers using CSD analysis (Fig. 9B). These observations, and the results discussed above showing that BDNF-ir in MFs was elevated after Gdx, suggested that there was sprouting of MF axons into the pyramidal cell layer and stratum oriens of Gdx rats (diagrammed schematically in Fig. 9A). Sprouting could be a response to increased BDNF levels in MFs because BDNF application to the MFs has been suggested to induce MF axon outgrowth (Lowenstein and Arsenault, 1996; Tamura et al., 2006; Gomez-Palacio-Schjetnan and Escobar, 2008; Tamura et al., 2009), although all studies do not agree (Bender et al., 1998; Qiao et al., 2001; Shetty et al., 2003).
We determined whether MF sprouting into stratum oriens occurred using dynorphin B as a marker of the MFs (Pierce et al., 1999). Immunocytochemical comparisons were made between Gdx and sham rats (n = 7/group) that were perfused ∼2 weeks after surgery (15.5 ± 2.7 d for each group). Animals were perfused and processed in pairs, with one Gdx and one sham rat subjected to surgery on the same day, perfused the same day, cut the same day, with identical delays between surgery and perfusion, and were processed at the same time (similar to the experiments discussed above for BDNF-ir). As shown in Figure 9C–F, Gdx rats exhibited a band of dynorphin-ir in stratum oriens of CA3a/b, which was not evident in sham rats.
To quantify the differences in dynorphin-ir in Gdx rats compared with sham controls, a ROI was drawn around stratum lucidum, pyramidale and oriens to encompass the area where dynorphin-ir was most robust in Gdx rats (Fig. 9G1). Then the mean grayscale value was determined for that ROI, placing the same ROI over each section from each rat so that it covered stratum lucidum, stratum pyramidale, and stratum oriens of CA3a/b (Fig. 9G1). Next, the ROI was moved to stratum radiatum (Fig. 9G1) where background staining was estimated, and subtracted from the first measurement. The results showed that the mean grayscale value for each Gdx rat was greater than the sham rat that was processed with it (paired t test, p = 0.010; Fig. 9G2).
An additional measurement was also made within the terminal plexus of the MF pathway in stratum lucidum near CA3 (Fig. 9H1) to address the possibility that dynorphin-ir in the normal MF pathway increased after Gdx. As shown in Figure 9H2, there were no significant differences in dynorphin-ir (Student's t test, p = 0.487). Therefore, an increase in dynorphin expression in Gdx rats does not explain the increased dynorphin-ir in stratum oriens of Gdx rats in Figure 9C–G. Instead, the results suggest that new MFs developed in Gdx rats in stratum oriens.
Collectively, the data suggest that Gdx rats exhibit MF sprouting ∼2 weeks after surgery. There appeared to be a functional effect of the sprouted fibers because fEPSPs evoked by MF stimulation could be recorded throughout stratum lucidum, stratum pyramidale, and stratum oriens of Gdx rats.
A Trk-dependent component of MF transmission and LTP emerges after Gdx
MF transmission in Gdx rats is reduced by K252a
Because the hyperexcitability in slices of Gdx rats was sensitive to K252a and BDNF protein was increased in MFs, we asked whether the increase in MF transmission and synaptic plasticity of Gdx rats would be K252a-sensitive. As shown in Figure 10A1, slices from Gdx and sham rats (n = 4 slices from 4 rats/group) were recorded for a 15 min baseline period, exposed to vehicle (0.003% DMSO), and the responses that were recorded 30 min later were compared with the baseline. Next, slices were exposed to 300 nm K252a for 30 min, and subsequently slices were perfused with drug-free buffer for 30 min to determine whether effects of K252a were reversible.
Drug effects were evaluated by two-way RM-ANOVA. There was a significant effect of surgical pretreatment (Gdx or sham; F(1,18) = 28.092; p = 0.002) and a significant interaction between the surgical pretreatment and responses to drug treatments (F(3,18) = 6.995; p = 0.003; Fig. 10A2). The slope of fEPSPs in Gdx rats decreased in response to K252a (one-way RM-ANOVA; F(3,15) = 6.848; p = 0.006; Student's t test, p = 0.013) to 61.0 ± 5.8% of control (Fig. 10A2). There was no significant effect of K252a in sham controls (one-way RM-ANOVA; F(3,15) = 0.098; Student's t test, p = 0.965; Fig. 10A2).
There was variable recovery after resuming drug-free buffer (Fig. 10A2; wash, ○ with large SEM), unlike shorter periods of K252a application in previous experiments where recovery was more consistent (Fig. 7; similar SE bars during the experiments). These data suggested that prolonged exposure to dilute DMSO, first as the vehicle pretreatment and then as the K252a solvent, had adverse effects, leading to inability of some slices to recover during wash. Therefore we also evaluated K252a in slices without pretreatment with dilute DMSO (n = 8 slices, 8 Gdx rats; Fig. 10B1). As shown in Figure 10B2, K252a depressed the fEPSP slope, and there was more consistent recovery. Together, these data suggest that a Trk-sensitive component of MF transmission emerged after Gdx that was not normally present in sham rats.
MF LTP in Gdx rats is reduced by K252a
To test the effect of K252a on LTP, a 15-min-long baseline was followed by exposure to 300 nm K252a for 15 min, the latency when K252a began to exert an effect in Figure 10B. After LTP induction, K252a exposure continued for another 15 min, followed by perfusion with drug-free buffer (Fig. 10C). For comparison, slices were exposed to vehicle for 15 min before and after LTP induction (Fig. 10C).
K252a-treated slices had less LTP than vehicle-treated slices: 60 min after LTP induction, fEPSP slopes in K252a-treated slices were 62.8 ± 17.1% of control (n = 6 slices, 6 Gdx rats; Fig. 10C), whereas vehicle-treated slices exhibited 113.7 ± 12.7% potentiation (n = 7 slices, 7 Gdx rats), which were significantly different (Student's t test; p = 0.031). K252a-treated slices of Gdx rats were not significantly different from slices of sham rats that were untreated (Student's t test, p = 0.439), suggesting that K252a reversed the effect of Gdx (Fig. 10C).
K252a-treated slices from Gdx rats also showed decreased PTP relative to vehicle-treated slices; two-way RM-ANOVA showed that there was an interaction between treatment (K252a vs vehicle) and PTP (F(9,99) = 2.441; p = 0.015) with K252a-treated slices showing significantly decreased PTP immediately after LTP induction (Student's t test, p = 0.006) and at 2 min (p = 0.017), but not other times.
These data were consistent with the hypothesis that a Trk-sensitive component of synaptic plasticity emerged after Gdx. However, LTP in vehicle-treated slices was reduced compared with untreated slices of Gdx rats (p = 0.036; Fig. 10C), suggesting an adverse effect of vehicle. An effect of vehicle was also suspected based on data in Figure 10A2 where there was inconsistent recovery after prolonged application of vehicle (exposure to vehicle alone, then K252a dissolved in vehicle). Therefore, we also tested the effect of a scavenger of BDNF, TrkB-IgG, where vehicle was 0.9% NaCl. TrkB-IgG-protein is composed of the recognition site of TrkB and the FC region of human IgG1 (Binder et al., 1999). Comparison was to human IgG1. The concentration of TrkB was chosen because of its efficacy in previous studies (Kang et al., 1996; Skucas et al., 2011).
As shown in Figure 10D, TrkB-IgG suppressed LTP compared with IgG. TrkB-IgG-treated slices exhibited 112.4 ± 12.0% potentiation at 60 min after LTP induction (n = 7 slices, 6 Gdx rats), compared with 141.7 ± 6.9% potentiation in IgG-treated slices (n = 9 slices, 6 Gdx rats; Student's t test, p = 0.030). LTP in IgG-treated slices of Gdx rats was not significantly different from untreated slices of Gdx rats (Student's t test, p = 0.674). PTP was reduced by TrkB-IgG compared with IgG: 1 min after LTP induction, PTP was 191.4 ± 15.7% in TrkB-IgG treated slices and 251.7 ± 21.6% in IgG-treated slices (Student's t test, p = 0.036).
Together, these data suggest that a Trk-sensitive component of MF PTP and LTP emerged after Gdx.
MF transmission is increased in Gdx rats
Recording procedures to study MF transmission
We first confirmed that our stimulation and recording procedures evoked responses that were mediated by MFs in adult male rats that were the same age as those that would be used for experiments (∼2 months old). As shown in Figure 1A–C, a stimulating electrode was placed at the crest of the DG in the SGZ, with recording electrodes in stratum lucidum of area CA3b. fEPSPs had the characteristics previously reported for MF transmission: (1) short latency to onset of the fEPSP (<4 ms), consistent with monosynaptic transmission; (2) robust paired pulse facilitation (Salin et al., 1996; Kamiya et al., 2002; Scott et al., 2008; Cosgrove et al., 2009); and (3) blockade by 1 μm DCG-IV, a metabotropic glutamate receptor subtype II agonist that blocks MF transmission (Figure 1D–F) (Yoshino et al., 1996). Another characteristic, a fiber volley before the fEPSP, occurred when stimulus strength was sufficiently strong (Fig. 1E, black arrow).
Laminar analysis of the response to stimulation in CA3b showed that the fEPSP recorded in stratum lucidum was followed by a later fEPSP that was only recorded in stratum radiatum, and reversed polarity in the pyramidal cell layer (Fig. 1D1,E1). CSD analysis (Fig. 1D2,E2) showed a short-latency sink in stratum lucidum and a longer latency sink in stratum radiatum, suggesting a monosynaptic fEPSP corresponding to activation of MFs, followed by a disynaptic fEPSP reflecting activation of recurrent collaterals of pyramidal cells in stratum radiatum. The field potentials recorded in stratum lucidum and the cell layer were reduced >90% by DCG-IV (1 μm; n = 8 slices, 3 rats; Fig. 1F). Notably, 1 μm DCG-IV had no detectable effect on fEPSPs evoked by Schaffer collateral stimulation, recorded in stratum radiatum of CA1 (2 slices from 2 sham controls; fEPSP slope 30 min after DCG-IV: 99.4% of pre-DCG-IV baseline).
A small fEPSP in stratum radiatum remained after DCG-IV (Fig. 1F2), which could be due to the fact that DCG-IV did not block the response entirely. It is also possible that activation of CA3 pyramidal cell axons in the hilus (Scharfman, 2007) led to activation of recurrent collaterals to area CA3 pyramidal cell dendrites in stratum radiatum. For these reasons, the term “MF fEPSP” is used below to refer only to short-latency fEPSPs recorded in stratum lucidum. “Recurrent fEPSP” is used below to refer only to the long latency fEPSPs recorded in stratum radiatum. It should be noted that the term “MF LTP” is used with caution below because we cannot rule out a possible contribution of the recurrent collateral pathway during LTP induction.
MF fEPSPs are increased in Gdx rats
For the comparisons of MF fEPSPs between Gdx and sham rats, animals had surgery at ages that were not different (Gdx: 68.4 ± 2.4-d-old; sham: 57.2 ± 3.6-d-old; n = 10 slices from 8 rats/group; Student's t test, p = 0.973), and hippocampal slices were made at similar delays after surgery (Gdx: 14.5 ± 1.6 d; sham: 13.6 ± 1.0 d; n = 10 slices from 8 rats/group; Student's t test, p = 0.736). As shown in Figure 2A, MF fEPSPs were largest in Gdx rats (n = 10 slices, 8 rats/group). There was a significant effect of Gdx on the input-output relation for fEPSP slope (two-way RM-ANOVA; F(1,198) = 5.710; p = 0.028; Fig. 2B) and a significant interaction between treatment and stimulus strength (F(11,198) = 2.566; p = 0.005) with significantly larger fEPSP slopes in Gdx rats at all stimulus strengths except the two weakest intensities (Student's t test, p < 0.05; Fig. 2B). Results were similar for amplitude; there was a significant effect of Gdx (two-way RM-ANOVA; F(1,198) = 66.086; p < 0.0001) and an interaction between Gdx and stimulus strength (F(11,198) = 14.986; p < 0.0001), with fEPSPs from Gdx rats greater in sham rats at all stimulus strengths except the weakest stimuli (Student's t test, p < 0.05).
When fiber volleys were measured, there was no significant effect of Gdx on the input-output relation by two-way RM-ANOVA (F(1,198) = 0.722; p = 0.407; Fig. 2C). However, ANOVA-based 4-parameter linear regression analysis (Scharfman et al., 2007) indicated that fiber volley amplitudes in Gdx rats were significantly higher than those for sham-operated controls (F(1,28) = 18.92; p < 0.001).
Gdx rats exhibited a greater MF fEPSP slope for a given fiber volley amplitude (Fig. 2D). Excluding the MF fEPSPs elicited at the two weakest intensities (where fiber volleys were not present), the slope of the relation was significantly greater for Gdx rats (linear regression analysis; F(1,17) = 11.486; p = 0.004; Fig. 2D).
Although fEPSPs were larger in Gdx rats, there were no differences in latencies. First, we measured the latency to onset of the peak of the maximal fiber volley and found that there were no significant differences between experimental groups (Gdx: 1.96 ± 0.33 ms; sham: 2.34 ± 0.26 ms; Student's t test; p = 0.370; Fig. 2E1). Then we measured the latencies to onset of MF fEPSPs; both minimal and maximal fEPSPs were measured because latency varies depending on the stimulus strength (latency is longest when stimuli are weak). There were no differences between fEPSPs of Gdx and sham rats when stimuli were minimal (35 μs; Gdx: 3.84 ± 0.38 ms; sham: 3.81 ± 0.23 ms; Student's t test, p = 0.952; Fig. 2E2) and no differences in the maximal fEPSPs (Gdx: 3.38 ± 0.28 ms; sham: 3.36 ± 0.12 ms; Student's t test, p = 0.957).
To confirm that effects of Gdx were the result of a reduction in serum levels of testosterone, we determined whether testosterone replacement to Gdx rats blocked effects of Gdx on MF transmission. For this purpose, a time-release pellet of either testosterone or placebo was inserted subcutaneously at the time of surgery (Fig. 2F1). There were four experimental groups: Gdx rats treated with a testosterone pellet (10 slices, 5 rats), Gdx rats that received a placebo pellet (12 slices, 6 rats), untreated Gdx rats (27 slices, 17 rats), and untreated sham controls (16 slices, 10 rats). MF fEPSPs were tested ∼2 weeks after surgery. There were differences between the four groups (one-way ANOVA; F(3,64) = 2.668; p = 0.015), with smaller fEPSP slopes in Gdx animals that received testosterone compared with Gdx rats treated with placebo (Student's t test, p = 0.008; Fig. 2F2). MF fEPSPs of Gdx animals that received placebo were not significantly different from fEPSPs of untreated Gdx rats (Student's t test, p = 0.651), and there were no significant differences between Gdx rats that received testosterone and untreated sham controls (Student's t test, p = 0.702; Fig. 2F2).
To confirm that serum testosterone levels were restored by testosterone pellets, testosterone was measured by ELISA from trunk blood collected at the time of death. Gdx rats that were untreated or treated with placebo had very low testosterone levels; those that were below the lower limit of the assay were assigned the lower limit for their testosterone value (0.10 ng/ml; Fig. 2F3). There were significant differences between the four experimental groups (Kruskal–Wallis statistic, 32.584, p < 0.0001; Fig. 2F3). Circulating testosterone levels in Gdx rats treated with testosterone (8.08 ± 1.4 ng/ml; n = 5 rats) were significantly greater than Gdx rats treated with placebo (0.16 ± 0.04 ng/ml; n = 6; Dunn's multiple-comparisons test, p > 0.05; Fig. 2F3). However, Gdx rats treated with testosterone were not significantly different from untreated sham controls (6.52 ± 0.75 ng/ml; n = 10; p > 0.05; Fig. 2F3). The mean value for serum testosterone of Gdx rats treated with placebo was not significantly different from Gdx rats that were untreated (0.19 ± 0.09 ng/ml; n = 6; p > 0.05; Fig. 2F3). These data confirmed that Gdx reduced serum testosterone to a negligible level and testosterone treatment increased it to levels that were normal for the strain and age of rats that were used.
Recording procedures to study MF transmission
We first confirmed that our stimulation and recording procedures evoked responses that were mediated by MFs in adult male rats that were the same age as those that would be used for experiments (∼2 months old). As shown in Figure 1A–C, a stimulating electrode was placed at the crest of the DG in the SGZ, with recording electrodes in stratum lucidum of area CA3b. fEPSPs had the characteristics previously reported for MF transmission: (1) short latency to onset of the fEPSP (<4 ms), consistent with monosynaptic transmission; (2) robust paired pulse facilitation (Salin et al., 1996; Kamiya et al., 2002; Scott et al., 2008; Cosgrove et al., 2009); and (3) blockade by 1 μm DCG-IV, a metabotropic glutamate receptor subtype II agonist that blocks MF transmission (Figure 1D–F) (Yoshino et al., 1996). Another characteristic, a fiber volley before the fEPSP, occurred when stimulus strength was sufficiently strong (Fig. 1E, black arrow).
Laminar analysis of the response to stimulation in CA3b showed that the fEPSP recorded in stratum lucidum was followed by a later fEPSP that was only recorded in stratum radiatum, and reversed polarity in the pyramidal cell layer (Fig. 1D1,E1). CSD analysis (Fig. 1D2,E2) showed a short-latency sink in stratum lucidum and a longer latency sink in stratum radiatum, suggesting a monosynaptic fEPSP corresponding to activation of MFs, followed by a disynaptic fEPSP reflecting activation of recurrent collaterals of pyramidal cells in stratum radiatum. The field potentials recorded in stratum lucidum and the cell layer were reduced >90% by DCG-IV (1 μm; n = 8 slices, 3 rats; Fig. 1F). Notably, 1 μm DCG-IV had no detectable effect on fEPSPs evoked by Schaffer collateral stimulation, recorded in stratum radiatum of CA1 (2 slices from 2 sham controls; fEPSP slope 30 min after DCG-IV: 99.4% of pre-DCG-IV baseline).
A small fEPSP in stratum radiatum remained after DCG-IV (Fig. 1F2), which could be due to the fact that DCG-IV did not block the response entirely. It is also possible that activation of CA3 pyramidal cell axons in the hilus (Scharfman, 2007) led to activation of recurrent collaterals to area CA3 pyramidal cell dendrites in stratum radiatum. For these reasons, the term “MF fEPSP” is used below to refer only to short-latency fEPSPs recorded in stratum lucidum. “Recurrent fEPSP” is used below to refer only to the long latency fEPSPs recorded in stratum radiatum. It should be noted that the term “MF LTP” is used with caution below because we cannot rule out a possible contribution of the recurrent collateral pathway during LTP induction.
MF fEPSPs are increased in Gdx rats
For the comparisons of MF fEPSPs between Gdx and sham rats, animals had surgery at ages that were not different (Gdx: 68.4 ± 2.4-d-old; sham: 57.2 ± 3.6-d-old; n = 10 slices from 8 rats/group; Student's t test, p = 0.973), and hippocampal slices were made at similar delays after surgery (Gdx: 14.5 ± 1.6 d; sham: 13.6 ± 1.0 d; n = 10 slices from 8 rats/group; Student's t test, p = 0.736). As shown in Figure 2A, MF fEPSPs were largest in Gdx rats (n = 10 slices, 8 rats/group). There was a significant effect of Gdx on the input-output relation for fEPSP slope (two-way RM-ANOVA; F(1,198) = 5.710; p = 0.028; Fig. 2B) and a significant interaction between treatment and stimulus strength (F(11,198) = 2.566; p = 0.005) with significantly larger fEPSP slopes in Gdx rats at all stimulus strengths except the two weakest intensities (Student's t test, p < 0.05; Fig. 2B). Results were similar for amplitude; there was a significant effect of Gdx (two-way RM-ANOVA; F(1,198) = 66.086; p < 0.0001) and an interaction between Gdx and stimulus strength (F(11,198) = 14.986; p < 0.0001), with fEPSPs from Gdx rats greater in sham rats at all stimulus strengths except the weakest stimuli (Student's t test, p < 0.05).
When fiber volleys were measured, there was no significant effect of Gdx on the input-output relation by two-way RM-ANOVA (F(1,198) = 0.722; p = 0.407; Fig. 2C). However, ANOVA-based 4-parameter linear regression analysis (Scharfman et al., 2007) indicated that fiber volley amplitudes in Gdx rats were significantly higher than those for sham-operated controls (F(1,28) = 18.92; p < 0.001).
Gdx rats exhibited a greater MF fEPSP slope for a given fiber volley amplitude (Fig. 2D). Excluding the MF fEPSPs elicited at the two weakest intensities (where fiber volleys were not present), the slope of the relation was significantly greater for Gdx rats (linear regression analysis; F(1,17) = 11.486; p = 0.004; Fig. 2D).
Although fEPSPs were larger in Gdx rats, there were no differences in latencies. First, we measured the latency to onset of the peak of the maximal fiber volley and found that there were no significant differences between experimental groups (Gdx: 1.96 ± 0.33 ms; sham: 2.34 ± 0.26 ms; Student's t test; p = 0.370; Fig. 2E1). Then we measured the latencies to onset of MF fEPSPs; both minimal and maximal fEPSPs were measured because latency varies depending on the stimulus strength (latency is longest when stimuli are weak). There were no differences between fEPSPs of Gdx and sham rats when stimuli were minimal (35 μs; Gdx: 3.84 ± 0.38 ms; sham: 3.81 ± 0.23 ms; Student's t test, p = 0.952; Fig. 2E2) and no differences in the maximal fEPSPs (Gdx: 3.38 ± 0.28 ms; sham: 3.36 ± 0.12 ms; Student's t test, p = 0.957).
To confirm that effects of Gdx were the result of a reduction in serum levels of testosterone, we determined whether testosterone replacement to Gdx rats blocked effects of Gdx on MF transmission. For this purpose, a time-release pellet of either testosterone or placebo was inserted subcutaneously at the time of surgery (Fig. 2F1). There were four experimental groups: Gdx rats treated with a testosterone pellet (10 slices, 5 rats), Gdx rats that received a placebo pellet (12 slices, 6 rats), untreated Gdx rats (27 slices, 17 rats), and untreated sham controls (16 slices, 10 rats). MF fEPSPs were tested ∼2 weeks after surgery. There were differences between the four groups (one-way ANOVA; F(3,64) = 2.668; p = 0.015), with smaller fEPSP slopes in Gdx animals that received testosterone compared with Gdx rats treated with placebo (Student's t test, p = 0.008; Fig. 2F2). MF fEPSPs of Gdx animals that received placebo were not significantly different from fEPSPs of untreated Gdx rats (Student's t test, p = 0.651), and there were no significant differences between Gdx rats that received testosterone and untreated sham controls (Student's t test, p = 0.702; Fig. 2F2).
To confirm that serum testosterone levels were restored by testosterone pellets, testosterone was measured by ELISA from trunk blood collected at the time of death. Gdx rats that were untreated or treated with placebo had very low testosterone levels; those that were below the lower limit of the assay were assigned the lower limit for their testosterone value (0.10 ng/ml; Fig. 2F3). There were significant differences between the four experimental groups (Kruskal–Wallis statistic, 32.584, p < 0.0001; Fig. 2F3). Circulating testosterone levels in Gdx rats treated with testosterone (8.08 ± 1.4 ng/ml; n = 5 rats) were significantly greater than Gdx rats treated with placebo (0.16 ± 0.04 ng/ml; n = 6; Dunn's multiple-comparisons test, p > 0.05; Fig. 2F3). However, Gdx rats treated with testosterone were not significantly different from untreated sham controls (6.52 ± 0.75 ng/ml; n = 10; p > 0.05; Fig. 2F3). The mean value for serum testosterone of Gdx rats treated with placebo was not significantly different from Gdx rats that were untreated (0.19 ± 0.09 ng/ml; n = 6; p > 0.05; Fig. 2F3). These data confirmed that Gdx reduced serum testosterone to a negligible level and testosterone treatment increased it to levels that were normal for the strain and age of rats that were used.
PPF, post-tetanic potentiation, and LTP are increased in Gdx rats
PPF of MF fEPSPs was greater in Gdx rats than sham controls (Fig. 3). PPF was tested using half-maximal stimulation and a 40 ms interstimulus interval because these are parameters where PPF is typically robust (Scharfman, 1997; Scharfman et al., 2003, 2007). PPF was significantly greater in slices from Gdx rats compared with sham controls (Gdx: 166.8 ± 3.2%; n = 15 slices, 12 rats; sham: 121.0 ± 3.2%; 14 slices, 12 rats; Student's t test, p = 0.006).
We also evaluated PPF in additional rats (Gdx: 10 slices, 7 rats; sham: n = 10 slices, 6 rats) using a range of stimulus strengths because PPF has not been evaluated in detail at the ages we used, raising the possibility that PPF in sham rats might have been underestimated using a half-maximal stimulus. There was no significant difference in PPF across stimulus strengths in sham rats (one-way ANOVA; F(11,95) = 0.096; p = 0.880). PPF in Gdx rats was significantly greater than sham rats at two stimulus strengths in the middle of the input-output relation (two-way RM-ANOVA; F(11,154) = 7.907; p = 0.014; Student's t test, p < 0.05 for 110 and 135 μs; Fig. 3A2).
Next, PPF was evaluated using a range of interstimulus intervals (10, 20, 30, 40, 60, 80, 100, 150, and 200 ms; Gdx: 10 slices, 9 rats; sham: 12 slices, 7 rats) with a half-maximal stimulus strength. As shown in Figure 3B, Gdx influenced PPF (two-way RM-ANOVA; F(1,160) = 21.560; p < 0.001) with a significant interaction between interstimulus interval and effect of Gdx (F(8,160) = 3.453; p = 0.001). PPF was greater in Gdx rats when interstimulus intervals were intermediate (30, 40, or 60 ms; Student's t test, p < 0.05; Fig. 3B1,B2).
To examine post-tetanic potentiation (PTP) and LTP, we planned to use TBS, but several slices from Gdx rats exhibited spreading depression (SD) in response to TBS (described further below). Therefore, LTP was induced using two 25 Hz trains, 1 s in duration, 10 s apart; a similar procedure (25 Hz, 1 s or 25 Hz, 5 s) has been shown to elicit MF LTP in 18- to 25-d-old Wistar rats (Kwon and Castillo, 2008), which we confirmed in 30-d-old Sprague Dawley rats (147.7 ± 6.2%; n = 3 slices from 3 rats). As shown in Figure 3C1, Gdx rats exhibited LTP but sham rats did not (presumably because of their age; Gdx: 169.1 ± 20.7%; n = 9 slices, 8 rats; sham: 88.3 ± 14.1%; n = 8 slices, 7 rats; Student's t test, p = 0.008). Both groups exhibited robust PTP, but Gdx rats had greater PTP (two-way RM-ANOVA for the first 3 min after LTP induction; F(1,140) = 9.231; p = 0.009; Student's t test, p < 0.05; Fig. 3C2).
Gdx increases MF-evoked field potentials recorded outside stratum lucidum
Responses to MF stimulation were also recorded in the pyramidal cell layer to evaluate the extent that an increase in MF transmission led to enhanced pyramidal cell output, reflected by an increase in the population spike amplitude (Fig. 4A1–A2; Gdx: 17 slices, 12 rats; sham: 12 slices, 6 rats). An input-output curve for the population spike showed that there was a significant effect of Gdx (two-way RM-ANOVA; F(1,108) = 4.737; p = 0.038; Fig. 4A2). There was an interaction between surgical status (Gdx vs sham) and stimulus strength (F(4,108) = 3.749; p = 0.007), with a significant increase in population spike amplitude in Gdx rats at the highest stimulus intensities (Student's t test, p < 0.05; Fig. 4A2).
PPF of the population spike was greater in Gdx rats (half-maximal stimulus intensity; 40 ms interstimulus interval; Gdx: 137.8 ± 2.3%; n = 18 slices, 13 rats; sham: 109.1 ± 1.9%; n = 11 slices, 7 rats; Student's t test, p = 0.040; Fig. 4B,C). When multiple stimulus strengths were compared in a subset of these slices (Gdx: 10 slices, 7 rats; sham: 7 slices, 6 rats), two-way RM-ANOVA showed that there was an interaction between surgical treatment (Gdx or sham) and stimulus strength (F(3,42) = 3.257; p = 0.031) with PPF in slices from Gdx rats greater at the lowest intensity (Student's t test, p < 0.05) but not higher intensities (Student's t test, p > 0.05; Fig. 4B2).
PPF was also evaluated using a range of interstimulus intervals and a half-maximal stimulus strength (Gdx: 12 slices, 8 rats; sham: 10 slices, 6 rats). Two-way RM-ANOVA showed that there was an interaction between surgical treatment and interstimulus interval (F(1,140) = 4.072; p = 0.033) with significantly greater PPF in Gdx rats at 40 and 60 ms interstimulus intervals (Student's t test, p = 0.016 and p = 0.028, respectively; Fig. 4C1,C2). There were no significant differences between Gdx and sham rats at the short interstimulus intervals, when paired pulse inhibition typically occurs (10 or 20 ms intervals; Student's t test, p = 0.162 and p = 0.220 respectively; Fig. 4C1,C2). Therefore, Gdx rats had greater PPF of the population spike, but there was no effect on paired pulse inhibition.
Another cohort of rats was used to determine whether the delayed fEPSP recorded in stratum radiatum (referred to as the “recurrent fEPSP” below) was affected by Gdx. As shown in Figure 4D, recurrent fEPSP slopes were increased in Gdx rats (17 slices, 14 rats) compared with sham controls (16 slices, 14 rats). The input-output relation was significantly different (two-way RM-ANOVA; F(1,341) = 8.095; p = 0.008), and there was an interaction between surgical pretreatment and stimulus strength (F(11,341) = 3.269; p < 0.001) with recurrent fEPSPs of Gdx rats greater than sham at all stimulus strengths except the two lowest intensities (Student's t test, p < 0.05; Fig. 4D2). The lack of a significant effect at low stimulus intensities is consistent with the idea that recurrent fEPSPs required strong intensities of stimulation, so that pyramidal cells would reach threshold for action potential generation and activate their recurrent collaterals.
The latencies to onset of the recurrent fEPSP were not affected by Gdx. The recurrent fEPSP that was evoked by a weak stimulus (a 35 μs stimulus) had a 5.68 ± 0.16 ms latency to onset in slices from Gdx rats, and latency was 5.69 ± 0.18 ms in slices from sham rats (Student's t test; p = 0.965). The latencies to onset of the maximal recurrent fEPSP were not different either (Gdx: 5.02 ± 0.11; sham: 5.05 ± 0.13; Student's t test, p = 0.852).
PPF of the recurrent fEPSP slope (half-maximal stimulus, 40 ms interstimulus interval) was significantly greater in Gdx rats (120.7 ± 6.7%, n = 15 slices, 11 rats; sham: 98.3 ± 7.0%; n = 10 slices, 9 rats; p = 0.034).
DHT sensitivity of MF transmission and plasticity in Gdx rats
MF fEPSPs of Gdx rats are reduced by DHT
Testosterone is metabolized in the brain to DHT, a potent agonist at the androgen receptors, which are present in MFs of the adult male rat (Tabori et al., 2005). Therefore, we determined whether a deficiency in DHT was the reason for the differences in MF transmission between Gdx and sham rats. We also determined whether one of the metabolites of DHT, Adiol, would have effects on MF fEPSPs because several studies suggest a potent effect of Adiol to enhance the actions of GABA at GABAA receptors (Frye et al., 2001; Edinger et al., 2004; Frye and Edinger, 2004; Frye et al., 2004; Reddy, 2004a,b; Edinger and Frye, 2005; Kaminski et al., 2005; Frye et al., 2010; Reddy and Jian, 2010).
MF stimulation in normal ACSF was initially tested for a 10 min baseline period to confirm that fEPSP slope was stable (Fig. 5A1). Then slices were exposed to vehicle (0.0005% ethanol in ACSF), and the responses that were recorded 30 min later were compared with the baseline to confirm that vehicle had no detectable effect. Next, slices were exposed for 30 min to Adiol at a concentration that induced near-maximal potentiation of GABAA receptors in whole-cell recordings from CA1 pyramidal neurons (50 nm) (Reddy and Jian, 2010). Afterward, ACSF containing 50 nm DHT was applied for 30 min.
As shown in Fig. 5A2, MF fEPSPs of Gdx rats declined during exposure to DHT. One-way RM-ANOVA in Gdx rats showed that there was a significant change in fEPSP slope during the experiment (F(3,15) = 21.786; p < 0.001) with fEPSP slope decreasing to 72.3 ± 3.2% of control after DHT, significantly different from responses recorded before DHT (paired t test, p = 0.017). However, there were no significant effects of vehicle or Adiol (paired t tests, p > 0.05). In sham rats, there was no effect of either DHT or Adiol (one-way RM-ANOVA; F(3,11) = 0.316; p = 0.814; Fig. 5B).
These experiments suggested that DHT decreased MF transmission but Adiol did not. However, the effect of DHT could have resulted in part from pretreatment with Adiol because Adiol can be converted back to DHT in the brain (Chetyrkin et al., 2001). It is also possible that altering GABAergic transmission by preexposure to Adiol might affect subsequent responses to DHT, even though Adiol by itself had no measurable effect. To test this possibility, slices from Gdx rats were treated with 50 nm DHT immediately after the baseline (Fig. 5B1). DHT reduced the slope of the fEPSP to 74.8 ± 5.6% of control (n = 5 slices, 5 rats; paired t test, p = 0.004; Fig. 5B2). In contrast, there were no effects of Adiol after it was added (immediately after the baseline; n = 3 slices, 3 rats; Fig. 5C1,C2), and there was no significant effect of DHT in slices from sham controls (n = 4 slices, 4 rats; paired t test, p = 0.384; data not shown).
To confirm that DHT had effects on MF-evoked population spikes, and also that DHT inhibited MF transmission at multiple stimulus strengths, recordings were made in the pyramidal cell layer of Gdx rats in response to three intensities of MF stimulation: ∼30%, 50%, and 100% of the maximum. Comparisons were made between baseline and 30 min after the onset of treatment (vehicle, n = 4 slices, 4 rats; DHT, n = 4 slices, 4 rats). There was a significant effect of DHT (RM-ANOVA; F(1,12) = 40.17; p < 0.001), with amplitudes lower after DHT exposure at all stimulus intensities (Bonferroni's tests, p < 0.05). There was a significant interaction between stimulus strength and treatment (F(2,12) = 8.72; p = 0.005).
In summary, DHT reduced fEPSPs in Gdx rats, but Adiol did not, and there were no significant effects of any of the steroids in slices from sham controls. The results suggest that an androgen receptor-sensitive component of MF transmission emerges after Gdx.
MF LTP in Gdx rats is reduced by DHT
Because a DHT-sensitive component of synaptic transmission emerged after Gdx, we tested the DHT dependence of LTP in Gdx rats. After a 15 min baseline period, slices were exposed to 50 nm DHT or vehicle (Fig. 5D). LTP was induced 15 min later, the time when DHT began to exert its effects in the experiments described above. In slices where the fEPSPs declined within 15 min, an adjustment in stimulus intensity was made so that the train used to induce LTP would be half-maximal. Exposure to DHT continued until 15 min after LTP induction, and then drug-free buffer was used for the remainder of the experiment. LTP was reduced in DHT-treated slices compared with slices from Gdx rats treated with vehicle (Gdx + DHT, 85.5 ± 10.1%; Gdx + vehicle, 135.1 ± 13.1%; Student's t test; p = 0.020; Fig. 5D).
Figure 5E compares LTP in slices from Gdx rats that were preexposed to DHT before LTP was tested, Gdx rats pretreated with vehicle, slices from Gdx rats that had no pretreatment, and slices from sham rats that were untreated. LTP in DHT-treated slices from Gdx rats was comparable with LTP in slices of untreated sham rats (Student's t test, p = 0.475); LTP in vehicle-treated slices of Gdx rats was similar to untreated slices of Gdx rats (Student's t test, p = 0.565).
PTP was also reduced by DHT (Fig. 5E). Compared with vehicle-treated slices, fEPSP slopes in the first 3 min after LTP induction were significantly reduced by DHT (two-way RM-ANOVA (F(1,45) = 5.291; p = 0.047; Fig. 5F).
In summary, a DHT-sensitive component of MF PTP and LTP was present in slices from Gdx rats. Sham rats did not exhibit LTP, so the DHT-sensitive component of PTP and LTP was only evident in Gdx rats.
MF fEPSPs of Gdx rats are reduced by DHT
Testosterone is metabolized in the brain to DHT, a potent agonist at the androgen receptors, which are present in MFs of the adult male rat (Tabori et al., 2005). Therefore, we determined whether a deficiency in DHT was the reason for the differences in MF transmission between Gdx and sham rats. We also determined whether one of the metabolites of DHT, Adiol, would have effects on MF fEPSPs because several studies suggest a potent effect of Adiol to enhance the actions of GABA at GABAA receptors (Frye et al., 2001; Edinger et al., 2004; Frye and Edinger, 2004; Frye et al., 2004; Reddy, 2004a,b; Edinger and Frye, 2005; Kaminski et al., 2005; Frye et al., 2010; Reddy and Jian, 2010).
MF stimulation in normal ACSF was initially tested for a 10 min baseline period to confirm that fEPSP slope was stable (Fig. 5A1). Then slices were exposed to vehicle (0.0005% ethanol in ACSF), and the responses that were recorded 30 min later were compared with the baseline to confirm that vehicle had no detectable effect. Next, slices were exposed for 30 min to Adiol at a concentration that induced near-maximal potentiation of GABAA receptors in whole-cell recordings from CA1 pyramidal neurons (50 nm) (Reddy and Jian, 2010). Afterward, ACSF containing 50 nm DHT was applied for 30 min.
As shown in Fig. 5A2, MF fEPSPs of Gdx rats declined during exposure to DHT. One-way RM-ANOVA in Gdx rats showed that there was a significant change in fEPSP slope during the experiment (F(3,15) = 21.786; p < 0.001) with fEPSP slope decreasing to 72.3 ± 3.2% of control after DHT, significantly different from responses recorded before DHT (paired t test, p = 0.017). However, there were no significant effects of vehicle or Adiol (paired t tests, p > 0.05). In sham rats, there was no effect of either DHT or Adiol (one-way RM-ANOVA; F(3,11) = 0.316; p = 0.814; Fig. 5B).
These experiments suggested that DHT decreased MF transmission but Adiol did not. However, the effect of DHT could have resulted in part from pretreatment with Adiol because Adiol can be converted back to DHT in the brain (Chetyrkin et al., 2001). It is also possible that altering GABAergic transmission by preexposure to Adiol might affect subsequent responses to DHT, even though Adiol by itself had no measurable effect. To test this possibility, slices from Gdx rats were treated with 50 nm DHT immediately after the baseline (Fig. 5B1). DHT reduced the slope of the fEPSP to 74.8 ± 5.6% of control (n = 5 slices, 5 rats; paired t test, p = 0.004; Fig. 5B2). In contrast, there were no effects of Adiol after it was added (immediately after the baseline; n = 3 slices, 3 rats; Fig. 5C1,C2), and there was no significant effect of DHT in slices from sham controls (n = 4 slices, 4 rats; paired t test, p = 0.384; data not shown).
To confirm that DHT had effects on MF-evoked population spikes, and also that DHT inhibited MF transmission at multiple stimulus strengths, recordings were made in the pyramidal cell layer of Gdx rats in response to three intensities of MF stimulation: ∼30%, 50%, and 100% of the maximum. Comparisons were made between baseline and 30 min after the onset of treatment (vehicle, n = 4 slices, 4 rats; DHT, n = 4 slices, 4 rats). There was a significant effect of DHT (RM-ANOVA; F(1,12) = 40.17; p < 0.001), with amplitudes lower after DHT exposure at all stimulus intensities (Bonferroni's tests, p < 0.05). There was a significant interaction between stimulus strength and treatment (F(2,12) = 8.72; p = 0.005).
In summary, DHT reduced fEPSPs in Gdx rats, but Adiol did not, and there were no significant effects of any of the steroids in slices from sham controls. The results suggest that an androgen receptor-sensitive component of MF transmission emerges after Gdx.
MF LTP in Gdx rats is reduced by DHT
Because a DHT-sensitive component of synaptic transmission emerged after Gdx, we tested the DHT dependence of LTP in Gdx rats. After a 15 min baseline period, slices were exposed to 50 nm DHT or vehicle (Fig. 5D). LTP was induced 15 min later, the time when DHT began to exert its effects in the experiments described above. In slices where the fEPSPs declined within 15 min, an adjustment in stimulus intensity was made so that the train used to induce LTP would be half-maximal. Exposure to DHT continued until 15 min after LTP induction, and then drug-free buffer was used for the remainder of the experiment. LTP was reduced in DHT-treated slices compared with slices from Gdx rats treated with vehicle (Gdx + DHT, 85.5 ± 10.1%; Gdx + vehicle, 135.1 ± 13.1%; Student's t test; p = 0.020; Fig. 5D).
Figure 5E compares LTP in slices from Gdx rats that were preexposed to DHT before LTP was tested, Gdx rats pretreated with vehicle, slices from Gdx rats that had no pretreatment, and slices from sham rats that were untreated. LTP in DHT-treated slices from Gdx rats was comparable with LTP in slices of untreated sham rats (Student's t test, p = 0.475); LTP in vehicle-treated slices of Gdx rats was similar to untreated slices of Gdx rats (Student's t test, p = 0.565).
PTP was also reduced by DHT (Fig. 5E). Compared with vehicle-treated slices, fEPSP slopes in the first 3 min after LTP induction were significantly reduced by DHT (two-way RM-ANOVA (F(1,45) = 5.291; p = 0.047; Fig. 5F).
In summary, a DHT-sensitive component of MF PTP and LTP was present in slices from Gdx rats. Sham rats did not exhibit LTP, so the DHT-sensitive component of PTP and LTP was only evident in Gdx rats.
Hyperexcitability in Gdx rats
While recording in the pyramidal cell layer of Gdx rats, increased excitability was detected in a subset of slices (Fig. 6). The increase in excitability was manifested in several ways. For example, a single MF stimulus could evoke >1 population spike, whereas in sham rats there was only one population spike/stimulus. Also, there were small (∼1 mV) spontaneous field potentials in slices of Gdx rats when recordings were made in the pyramidal cell layer, and more unit activity than sham rats (Fig. 6A).
Another sign of abnormal excitability in slices from Gdx rats was SD in response to repetitive MF stimulation (Fig. 6B). When SD followed TBS stimulation of the MFs, it did so rapidly (within seconds; Fig. 6B). SD episodes were stereotypical: they were composed of an initial phase where small (∼1 mV) spontaneous population spikes occurred and a second phase where there was a negative shift in the DC potential, which peaked at −14 to −16 mV within ∼15 s of the onset of spontaneous activity (Fig. 6B). Subsequently, there was a recovery of the DC potential within ∼5 min (Fig. 6B). During the recovery, no response to MF stimulation could be elicited even in response to the highest stimulus strength.
The first type of hyperexcitability, where >1 population spike per stimulus occurred, was found in 10 of 45 (22%) slices from 20 Gdx rats and 0 of 56 slices in 24 sham rats (χ test, p < 0.001; Fig. 6C). SD after TBS stimulation occurred in 8 of 12 (75%) slices from 5 Gdx rats and 0 of 11 slices in 5 sham rats (Fisher's exact test, p = 0.001; Fig. 6C). Four of the 8 slices from Gdx rats that exhibited SD also showed more than one population spike/MF stimulus.
Another indication of abnormal excitability is shown in Figure 6D1. In this case, field potentials were repetitive but at a rhythm that was slower (∼20 Hz) than the multiple population spikes in Figure 6A. CSD analysis (Fig. 6D2,D3) showed the stereotypical nature of the rhythmic field potentials across laminae of CA3 and that the earliest sink was in stratum lucidum (Fig. 6D2,D3, arrow), suggesting the MF pathway initiated activity. Five of 45 (11%) slices from Gdx rats exhibited the rhythmic field potentials shown in Figure 6C, D; these were not the same slices that showed other signs of increased excitability in Figure 6A, B. Therefore, the total number of slices exhibiting hyperexcitability of one type or another was 19 of 45, or 42%.
The abnormal excitability that was recorded in Gdx rats was similar to the effects of recombinant BDNF superfused onto slices of normal adult male rats (Scharfman, 1997) and the responses to MF stimulation in rodents with elevated BDNF protein levels (Croll et al., 1999; Scharfman et al., 2003, 2007), which were blocked by the Trk receptor antagonist K252a (Scharfman, 1997; Scharfman et al., 2003, 2007). Therefore, we determined whether K252a would reduce excitability in Gdx rats.
As shown in Figure 7A, K252a reduced the first type of abnormal excitability in Gdx rats (>1 population spike/MF stimulus; Fig. 6A). Specifically, perfusion of slices with 300 nm K252a for 15 min blocked the second population spike that was elicited by MF stimulation (n = 3 slices, 2 rats; Fig. 7A), and vehicle (0.003% DMSO) had no effect (2 slices from 2 other rats; data not shown). In slices where repetitive field potentials were elicited by a single stimulus (e.g., Fig. 6D), an input-output curve for the amplitude of the first field potential was generated before and after exposure to K252a (Fig. 7B1–B3). There was a significant effect of K252a (two-way RM-ANOVA; F(1,20) = 87.483; p < 0.001) and an interaction (F(5,20) = 15.129; p < 0.001) with significant differences at all intensities above the minimum (where evoked potentials were negligible in all conditions; Student's t test, p < 0.05; Fig. 7B1–B3). There was no effect of vehicle (pre vs post vehicle, two-way RM-ANOVA; F(5,20) = 0.188; p = 0.687; Fig. 7B3).
To evaluate the effect of K252a on SD, SD was elicited by stimulus trains (1 Hz, paired pulses for 10 s) as described previously (Scharfman, 1997; Scharfman et al., 2003, 2007), which were triggered once every 15 min (Fig. 7C,D; K252a: n = 3 slices, 3 Gdx rats; vehicle: n = 3 slices, 3 other Gdx rats). After a 30 min baseline period, K252a was added to the ACSF for 15 min, followed by drug-free buffer (Fig. 7C,D). There was a significant effect of K252a on the amplitude of SD episodes (measured from baseline to the peak of the SD episode; one-way RM-ANOVA; F(9,29) = 14.268; p < 0.001) at 15 and 30 min after K252a was added (paired t test, p = 0.033 and p = 0.031, respectively; Fig. 7D2). There were no detectable effects of vehicle (one-way RM-ANOVA; F(9,29) = 1.049; p = 0.442; Fig. 7D2).Together, the results suggested that a subset of Gdx rats exhibited hyperexcitability that was dependent on Trk receptors.
BDNF protein expression increases in MFs after Gdx
Increased excitability in Gdx rats (Figs. 6 and and7)7) was similar to rats or mice with elevation of BDNF levels in MFs (Croll et al., 1999; Scharfman et al., 2003, 2007). Therefore, we determined whether MF BDNF levels were increased in Gdx rats.
Initial experiments used a well-described polyclonal antibody to BDNF (Conner et al., 1997; Yan et al., 1997) and showed that Gdx rats (which had surgery at ∼60 d of age and were evaluated ∼2 weeks later) exhibited greater BDNF-ir in the MFs than sham rats (Fig. 8A1–A2).
Similar results were obtained using a second antibody (monoclonal; Fig. 8B–D). Using the second antibody, 5 immunocytochemical procedures were conducted. For each procedure, a Gdx and sham littermate were subjected to surgery on the same day, at ∼60 d of age (61.9 ± 2.0 d, n = 5/group) followed by perfusion-fixation on the same day after surgery (12.0 ± 0.6 d), and brains were processed together. Sections from the 10 rats were quantified and analyzed in pairs (Gdx and sham). The location where the MF terminal plexus ends (Fig. 9H1) was used for quantification because BDNF-ir was always greatest there, regardless of surgical pretreatment (Fig. 8D). The timeline of the experiments is shown in Figure 8E1, and the results shown in Figure 8E2 demonstrated that MF BDNF-ir was greater in Gdx rats (Student's t test, p = 0.001).
A second experiment was conducted to determine whether the increase in BDNF protein in MFs in Gdx rats was long-lasting (Fig. 8F). For this purpose, rats were perfusion-fixed 2 months instead of 2 weeks after surgery (Fig. 8F1). Again, surgery was conducted in pairs, so the age at surgery was the same (59.8 ± 1.0 d) and delays between surgery and perfusion were the same (61.2 ± 1.9 d, n = 5/group). As shown in Figure 8F2, Gdx rats exhibited greater MF BDNF-ir than sham controls (Student's t test, p = 0.002). Therefore, the increase in MF BDNF-ir after Gdx appeared to be robust and long-lasting.
Gdx rats exhibit MF sprouting
As recordings in stratum pyramidale and stratum oriens were conducted, it became clear that Gdx rats often exhibited fiber volleys and fEPSPs in these layers in response to MF stimulation (Fig. 9A), leading to diverse sinks in these layers using CSD analysis (Fig. 9B). These observations, and the results discussed above showing that BDNF-ir in MFs was elevated after Gdx, suggested that there was sprouting of MF axons into the pyramidal cell layer and stratum oriens of Gdx rats (diagrammed schematically in Fig. 9A). Sprouting could be a response to increased BDNF levels in MFs because BDNF application to the MFs has been suggested to induce MF axon outgrowth (Lowenstein and Arsenault, 1996; Tamura et al., 2006; Gomez-Palacio-Schjetnan and Escobar, 2008; Tamura et al., 2009), although all studies do not agree (Bender et al., 1998; Qiao et al., 2001; Shetty et al., 2003).
We determined whether MF sprouting into stratum oriens occurred using dynorphin B as a marker of the MFs (Pierce et al., 1999). Immunocytochemical comparisons were made between Gdx and sham rats (n = 7/group) that were perfused ∼2 weeks after surgery (15.5 ± 2.7 d for each group). Animals were perfused and processed in pairs, with one Gdx and one sham rat subjected to surgery on the same day, perfused the same day, cut the same day, with identical delays between surgery and perfusion, and were processed at the same time (similar to the experiments discussed above for BDNF-ir). As shown in Figure 9C–F, Gdx rats exhibited a band of dynorphin-ir in stratum oriens of CA3a/b, which was not evident in sham rats.
To quantify the differences in dynorphin-ir in Gdx rats compared with sham controls, a ROI was drawn around stratum lucidum, pyramidale and oriens to encompass the area where dynorphin-ir was most robust in Gdx rats (Fig. 9G1). Then the mean grayscale value was determined for that ROI, placing the same ROI over each section from each rat so that it covered stratum lucidum, stratum pyramidale, and stratum oriens of CA3a/b (Fig. 9G1). Next, the ROI was moved to stratum radiatum (Fig. 9G1) where background staining was estimated, and subtracted from the first measurement. The results showed that the mean grayscale value for each Gdx rat was greater than the sham rat that was processed with it (paired t test, p = 0.010; Fig. 9G2).
An additional measurement was also made within the terminal plexus of the MF pathway in stratum lucidum near CA3 (Fig. 9H1) to address the possibility that dynorphin-ir in the normal MF pathway increased after Gdx. As shown in Figure 9H2, there were no significant differences in dynorphin-ir (Student's t test, p = 0.487). Therefore, an increase in dynorphin expression in Gdx rats does not explain the increased dynorphin-ir in stratum oriens of Gdx rats in Figure 9C–G. Instead, the results suggest that new MFs developed in Gdx rats in stratum oriens.
Collectively, the data suggest that Gdx rats exhibit MF sprouting ∼2 weeks after surgery. There appeared to be a functional effect of the sprouted fibers because fEPSPs evoked by MF stimulation could be recorded throughout stratum lucidum, stratum pyramidale, and stratum oriens of Gdx rats.
A Trk-dependent component of MF transmission and LTP emerges after Gdx
MF transmission in Gdx rats is reduced by K252a
Because the hyperexcitability in slices of Gdx rats was sensitive to K252a and BDNF protein was increased in MFs, we asked whether the increase in MF transmission and synaptic plasticity of Gdx rats would be K252a-sensitive. As shown in Figure 10A1, slices from Gdx and sham rats (n = 4 slices from 4 rats/group) were recorded for a 15 min baseline period, exposed to vehicle (0.003% DMSO), and the responses that were recorded 30 min later were compared with the baseline. Next, slices were exposed to 300 nm K252a for 30 min, and subsequently slices were perfused with drug-free buffer for 30 min to determine whether effects of K252a were reversible.
Drug effects were evaluated by two-way RM-ANOVA. There was a significant effect of surgical pretreatment (Gdx or sham; F(1,18) = 28.092; p = 0.002) and a significant interaction between the surgical pretreatment and responses to drug treatments (F(3,18) = 6.995; p = 0.003; Fig. 10A2). The slope of fEPSPs in Gdx rats decreased in response to K252a (one-way RM-ANOVA; F(3,15) = 6.848; p = 0.006; Student's t test, p = 0.013) to 61.0 ± 5.8% of control (Fig. 10A2). There was no significant effect of K252a in sham controls (one-way RM-ANOVA; F(3,15) = 0.098; Student's t test, p = 0.965; Fig. 10A2).
There was variable recovery after resuming drug-free buffer (Fig. 10A2; wash, ○ with large SEM), unlike shorter periods of K252a application in previous experiments where recovery was more consistent (Fig. 7; similar SE bars during the experiments). These data suggested that prolonged exposure to dilute DMSO, first as the vehicle pretreatment and then as the K252a solvent, had adverse effects, leading to inability of some slices to recover during wash. Therefore we also evaluated K252a in slices without pretreatment with dilute DMSO (n = 8 slices, 8 Gdx rats; Fig. 10B1). As shown in Figure 10B2, K252a depressed the fEPSP slope, and there was more consistent recovery. Together, these data suggest that a Trk-sensitive component of MF transmission emerged after Gdx that was not normally present in sham rats.
MF LTP in Gdx rats is reduced by K252a
To test the effect of K252a on LTP, a 15-min-long baseline was followed by exposure to 300 nm K252a for 15 min, the latency when K252a began to exert an effect in Figure 10B. After LTP induction, K252a exposure continued for another 15 min, followed by perfusion with drug-free buffer (Fig. 10C). For comparison, slices were exposed to vehicle for 15 min before and after LTP induction (Fig. 10C).
K252a-treated slices had less LTP than vehicle-treated slices: 60 min after LTP induction, fEPSP slopes in K252a-treated slices were 62.8 ± 17.1% of control (n = 6 slices, 6 Gdx rats; Fig. 10C), whereas vehicle-treated slices exhibited 113.7 ± 12.7% potentiation (n = 7 slices, 7 Gdx rats), which were significantly different (Student's t test; p = 0.031). K252a-treated slices of Gdx rats were not significantly different from slices of sham rats that were untreated (Student's t test, p = 0.439), suggesting that K252a reversed the effect of Gdx (Fig. 10C).
K252a-treated slices from Gdx rats also showed decreased PTP relative to vehicle-treated slices; two-way RM-ANOVA showed that there was an interaction between treatment (K252a vs vehicle) and PTP (F(9,99) = 2.441; p = 0.015) with K252a-treated slices showing significantly decreased PTP immediately after LTP induction (Student's t test, p = 0.006) and at 2 min (p = 0.017), but not other times.
These data were consistent with the hypothesis that a Trk-sensitive component of synaptic plasticity emerged after Gdx. However, LTP in vehicle-treated slices was reduced compared with untreated slices of Gdx rats (p = 0.036; Fig. 10C), suggesting an adverse effect of vehicle. An effect of vehicle was also suspected based on data in Figure 10A2 where there was inconsistent recovery after prolonged application of vehicle (exposure to vehicle alone, then K252a dissolved in vehicle). Therefore, we also tested the effect of a scavenger of BDNF, TrkB-IgG, where vehicle was 0.9% NaCl. TrkB-IgG-protein is composed of the recognition site of TrkB and the FC region of human IgG1 (Binder et al., 1999). Comparison was to human IgG1. The concentration of TrkB was chosen because of its efficacy in previous studies (Kang et al., 1996; Skucas et al., 2011).
As shown in