Impairment of synaptic plasticity by the stress mediator CRH involves selective destruction of thin dendritic spines via RhoA signaling

Animals

C57BL as well as B6.Cg-TgN transgenic mice, 3–4 months old, expressing YFP under the Thy1 promoter (Thy1-YFP; Jackson Lab, Bar Harbor, ME, USA) were used. Electrophysiological studies were performed using 6–10-week-old Sprague-Dawley rats (Harlan, Placentia, CA, USA). Only males were used to avoid cyclic fluctuations of spine densities that occur in females.²³ Animals were maintained in a quiet, uncrowded, National Institutes of Health (NIH)-approved facility on a 12-h light/dark cycle, with ad libitum access to lab chow and water. All experiments were approved by University Animal Care Committees and conformed to the NIH guidelines.

Slice preparation and infusion of CRH and ROCK blocker

Hippocampal slices were prepared as described previously.^24,25 Briefly, after decapitation, brains were removed and placed in artificial cerebral spinal fluid (room temperature) containing 126 mm NaCl, 3 mm KCl, 1.25 mm KH₂PO₄, 26 mm NaHCO₃, 1 mm MgSO₄, 2 mm CaCl₂, 10 mm glucose and 20 mm sucrose. A block of tissue including entorhinal cortex and hippocampus was dissected from adhering tissue. Transverse slices (300 μm) were cut (McIlwain tissue chopper; Mickle Lab, Gomshall, UK), and maintained in an infusion chamber (at 32 °C with 95% O₂/5% CO₂) with constant 30–40 ml min^–1 artificial cerebral spinal fluid perfusion. Synthetic rat/human CRH (100 nm; Bachem, King of Prussia, PA, USA), the ROCK (Rho-associated protein kinase) blocker H1152 (100 nm, EMD, Gibbstown, NJ, USA), CRH + H1152 or vehicle were infused for 60 min (0.5 ml min^–1). After a brief normal medium washout, the slices were quickly frozen and collected for pull-down analysis or fixed in fresh 4% paraformaldehyde (45–60 min) for immunocytochemistry. Each experiment included equal numbers of slices for each group.

Immunocytochemistry

Hippocampal slices were fixed in fresh 4% paraformaldehyde for 45 min and processed for sub-sectioning (20 μm) using a cryostat. Immunocytochemistry with or without tyramide signal amplification was performed on free-floating sections.^18,22,26 Antibodies included mouse anti-GFP (1:8000; Sigma, St Louis, MO, USA), mouse anti-PSD-95 (1:2000; Affinity BioReagents, Golden, CO, USA) and rabbit GluR1 (1:1000; PC246, Calbiochem, San Diego, CA, USA). The latter recognizes the extracellular moiety of the receptor subunit. Antibody binding was visualized with anti-mouse IgG conjugated to Alexa Fluor 488 (1:200; Molecular Probes, Eugene, OR, USA) or Cyanine 3 Tyramide (1:150; PerkinElmer, Waltham, MA, USA). Sections from all experimental groups were run concurrently in the same conditions, and analyzed without knowledge of treatment group.

Electrophysiology

Transverse hippocampal slices were cut at 350 μm using a McIIwain chopper and placed in ice-cold cutting solution containing 124 mm NaCl, 3 mm KCl, 1.25 mm KH₂PO₄, 5 mm MgSO₄, 26 mm NaHCO₃ and 10 mm dextrose. Slices were then transferred and maintained in an interface recording chamber (at 32 1C with 95% O₂/5% CO₂) with constant 60–70 ml min^–1 perfusion with artificial cerebral spinal fluid containing 124 mm NaCl, 3 mm KCl, 1.25 mm KH₂PO₄, 1.5 mm MgSO₄, 2.5 mm CaCl₂, 26 mm NaHCO₃ and 10 mm dextrose.²⁷ Field electrophysiology was performed using a single bipolar stimulating electrodes (65 μm twisted nichrome wire) placed in mid-proximal stratum radiatum of CA1a to stimulate independent populations of synapses recorded with a glass electrode (2 m NaCl) in CA1b.²⁷ Baseline responses were set at 40–50% of maximum spike-free field excitatory postsynaptic potentials (fEPSPs). LTP was induced using theta burst stimulation (TBS; 10 bursts of four pulses at 100 Hz with 200-ms interburst intervals). The fEPSP slope values in figures and text are measured as 10–90% fall and are normalized to 10 min before drug infusion or the TBS. Slices used for microscopic analysis were collected 10 min after TBS and fixed in 4% paraformaldehyde in 0.1 m of sodium phosphate buffer. The n values given in the figures represent slices per group.

Live imaging studies

To study the dynamic effects of CRH on dendritic spines, YFP-expressing slices were transferred to a superfusion chamber maintained at 36 °C via a heating platform (Warner Instruments, Hamden, CT, USA) to avoid hypothermia-related spine loss; this resulted in a tissue temperature of ~34 °C. The chamber was mounted on an upright Zeiss microscope, and slices were superfused with culture medium saturated with 95% O₂ and 5% CO₂. CRH (100 nm) or vehicle was infused for up to 45 min (1 ml min^–1) followed by a normal medium washout. When used, the ROCK blocker H1152 (100 nm) was infused 5–10 min before CRH application. In addition, the potential effects of H1152 itself on spines were evaluated by infusing the compound for 30 min. To avoid phototoxicity and enable imaging deep within the tissue, dendrites of YFP-expressing CA1 pyramidal neurons were imaged using a Zeiss LSM710 Meta two-photon scanning system with excitation wavelength 920 nm (Zeiss, Oberkochen, Germany). Based on the established localization of the effects of stress and CRH,^18,25 we focused on the apical dendrites of CA1 pyramidal cells and obtained z stacks (12–20 optical slices, 2 μm per slice) of 100 μm long dendritic segments of the main and oblique branches starting around 30–100 μm within stratum radiatum.

Spine analysis

The analysis was performed using two independent methods. (1) Image analysis of YFP-labeled spines (Figure 2b).^22,25 Briefly, YFP-expressing individual CA1 pyramidal neurons were reconstructed and drawn using a Zeiss LSM Image Browser and Adobe Photoshop (V. CS5). Neurons were chosen using systematic unbiased sampling from the rostral third of the anterior/posterior extent of the hippocampus (9–15 neurons per group; 3–5 independent experiments). Dendritic protrusions including spines and filopodia were identified^28–32 and reconstructed. Spine density was quantified comparing dendritic branches of the same order. Spine density was expressed as the number of spines per 10 μm of dendrite length. No correction factors were applied to the spine counts because high-magnification neuronal reconstruction permitted all spines of a given dendritic segment to be visualized. (2) Deconvolution analysis of wide-field three-dimensional images.³³ Briefly, z-series (0.2 μm steps) images were captured from the stratum radiatum in CA1 area at ×63 (NA 1.4) using a Leica microscope (DM6000, Wetzlar, Germany) and a charge-coupled device camera (Orca ER; Hamamatsu Photonics, Shizuoka, Japan). These images were collected from six sections per group per experiment in 3–5 experiments. Iterative deconvolution (99% confidence) was performed using the Volocity 4.1 software (PerkinElmer). Labeled synaptic structures were measured and counted from a 136 × 105 3 μm (x, y, z) sampling zone. Object area ( > 0.04 and < 1.2 μm³) and eccentricity criteria were applied to eliminate artifacts.

In situ labeling of F-actin

Alexa Fluor 568–phalloidin (6 mm in pipette; Invitrogen, Grand Island, NY, USA) was applied to slices (four 3-min intervals) at the end of the experiment, as described.^27,34 Values in the figure and text are presented per 550 μm² from collapsed 10 μm z stacks.

Hippocampal neuronal cultures

Primary hippocampal slice cultures were prepared as described in detail previously.³⁵ Briefly, hippocampi from postnatal 0 (P0) Sprague-Dawley rats were quickly dissected, removed from adherent meninges and incubated for 30 min in buffered salt solution containing 10 U ml^–1 papain (Worthington, Lakewood, NJ, USA). After removal of papain, cells were mechanically triturated and plated at a density of 400–600 cells μm^–2 on 12 μm coverslips that were precoated with poly-d-lysine (Sigma). Cultures were initially maintained in Neurobasal Medium (NBM) with B-27 supplement (Gibco, Grand Island, NY, USA) at 36 °C, 5% CO₂. At 3–4 h after plating, half of the culture medium was replaced with Neurobasal/B-27-based medium that was preconditioned for 24 h by 1- to 2-week-old nonneuronal cell culture prepared from P3–P4 rat cortices. Cultures were subsequently refreshed every 3–4 days with the conditioned medium. On the third day in vitro, 1 mm Cytosine-Arabinoside (Invitrogen) was added to the culture media to inhibit glial proliferation. The effects of CRH on dendritic spine density were assessed on 17–21 days in vitro. Spine density was assessed after immunocytochemistry for PSD-95 as described above.

Pull-down assays for activated Rho GTPases

Samples (6–10 pooled slices) were homogenized in Mg²⁺ lysis buffer (Millipore) with complete protease inhibitor cocktail, normalized and then assayed using RhoA or Rac/cdc42 pull-down kits (Millipore, Billerica, MA, USA), as described.³³ Samples were incubated with assay reagents for Rac/cdc42 (PAK1 PAK-PBD agarose) or Rho (Rhotekin-RBD agarose) and rocked for 2 h. Rhotekin binds strongly to RhoA and RhoC (absent in brain), so that RhoA is the only form detected in assays employed here. Agarose beads were collected by centrifugation (for 10 s at 14 000 g at 4 °C), washed, resuspended in Laemmli buffer and boiled for 5 min. Western blot analysis (12% SDS-polyacrylamide gel electrophoresis) used mouse anti-RhoA (Santa Cruz Biotechnology, Santa Cruz, CA, USA), mouse anti-Rac (Millipore) or rabbit anti-cdc42 (Millipore). GTP and GDP loading controls were incubated with 100 mm GTP-γS or 1 mm GDP for 30 min at 30 °C.

Western blot analysis

Samples (two to three pooled slices) were homogenized in radioimmunoprecipitation assay buffer containing 10 mm Tris, pH 7.2, 158 mm NaCl, 1 mm EDTA, 0.1% SDS, 1% Na-deoxycholate, 1% Triton X-100, 1× complete protease inhibitor cocktail (Roche, Indianapolis, IN, USA) and phosphatase inhibitor cocktails 1–2 (Sigma-Aldrich, St Louis, MO, USA).³³ Samples were normalized by Bio-Rad (Hercules, CA, USA), protein assay and processed for western blot analysis (12% SDS-polyacrylamide gel electrophoresis) using rabbit antisera to RhoA, Rac 1 and cdc42 (Cell Signaling Technology, Beverly, MA, USA) and the ECL Plus detection system (GE Healthcare, Piscataway, NJ, USA). Blots were re-blotted (Millipore) and probed for β-tubulin and β-actin (Sigma-Aldrich). Immunoreactive bands were measured using ImageJ (National Institutes of Health, Bethesda, MD, USA). The n values represent the numbers of samples tested.

Statistical analysis

All values in figures, figure captions and text represent group means ± s.e.m. For quantification of immunopositive elements and western blots, statistical significance (P≤0.05) was determined by analysis of variance followed by Tukey's post hoc test or two-tailed Student's t-test using SPSS 15.0 (SPSS, Chicago, IL, USA) or PRISM 5 (La Jolla, CA, USA). Curve fitting and statistical tests were performed using Matlab 7 (Natick, MA, USA).

Prolonged treatment with ‘stress levels’ of CRH impairs synaptic function

Stress lasting longer than seconds or a few minutes generally leads to memory problems that involve the hippocampus^{1–3,6,12,14,22} and to related abnormal synaptic plasticity.^1,14,22 Therefore, we tested if treatment with CRH for similar durations produces comparable effects, cognizant of the fact that in vivo, generation of high concentrations of CRH at dendritic spines likely requires sustained stress-induced release for some time, unlike the in vitro condition. Infusion of the hormone onto adult hippocampal slices had pronounced effects on synaptic physiology. Field EPSPs began to decline ~75 min into the infusion period by 21% (to 79% of preinfusion baseline values), whereas responses in slices from the same animals treated with a control peptide (adrenocorticotropic hormone (ACTH) 100 nm) and in non-treated slices were stable across this period (Figure 1a). Notably, the waveforms of the fEPSPs in the CRH-treated slices were not noticeably different from those recorded before hormone infusion. Short-term potentiation induced by one train of naturalistic TBS³⁶ was greatly reduced in the CRH-treated slices (Figure 1a). The initial enhancement of fEPSPs at 2 min after TBS, normalized to the fEPSP values collected for 10 min before delivery of the stimulation bursts, was 133 ± 5% (mean ± s.e.m.) in control slices and 95 ± 5% in the CRH-treated cases (P = 0.0015, two-tailed t-test). A comparable difference was found at 10 min after TBS: potentiation was 90 ± 6% above baseline for controls and 56 ± 5% for the CRH slices (P = 0.006). The control peptide, ACTH, did not influence potentiation. Thus, the STP process leading to persistent potentiation was severely impaired in the CRH-treated group.

Polymerization of actin within dendritic spines carrying excitatory synapses occurs within 2 min of the patterned stimulation that promotes synaptic potentiation, and contributes critically to maintaining the potentiated state.^24,37,38 Therefore, we examined actin polymerization following TBS and found that it was greatly reduced in CRH-treated slices relative to controls (Figure 1b). This suggested that the observed actions of CRH on synaptic responses and potentiation were associated with disruption of the actin cytoskeleton.

Selective reduction of thin dendritic spine density by CRH

To investigate the basis of the depression of fEPSPs in CRH-treated slices, we evaluated the actions of the peptide on dendritic spines that carry the postsynaptic elements of excitatory synapses in hippocampal CA1 (Figures 1c–e). We initially carried out a dose-finding study of CRH, aiming to approximate levels considered by others to be released during stress.^39,40 Exposure to 30 nm CRH had an insignificant effect (not shown), whereas 60 and 100 nm CRH reduced spine density of hippocampal neurons in a dose-dependent manner (Figure 2a). Therefore, we employed 100 nm CRH for the subsequent studies. A 1-h infusion of 100 nm CRH onto hippocampal slices prepared from adult YFP-expressing mice caused a significant decrease in the overall density of spines on pyramidal cell dendrites in CA1 stratum radiatum (Figures 1c–e), consistent with reports for field CA3.²² Closer analysis showed that this effect was pronounced for the large apical branches (Figure 1c), and also involved the oblique branches (Figure 1d). The net effect across the entire sampling field was ~22% (P < 0.01; Figure 1e).

It is reported that morphological and functional changes that accompany memory processes take place in specific types of dendritic spines, and especially in thin spines.^30–32,41 Therefore, we asked if the CRH-provoked losses were restricted to particular subtypes of spines. For consistency across measurements, we focused the analysis on apical oblique branches of comparable size. CRH caused a strongly selective loss of thin spines in YFP-expressing adult mice (Figures 2b–d). The selective reduction of thin spines was reflected in a change of the percent contributions of thin and mushroom-type spines to the overall spine population (Figure 2e). Whereas mushroom-type spines were minimally affected by CRH (Figure 2d), there was a significant increase in the relative contribution of these spines, which are not considered to undergo further changes upon LTP induction, to the overall spine population. Commensurately, there was a major reduction in the availability of thin spines. The surprisingly discrete effect on thin dendritic spines also confirmed that the application of CRH does not produce global injury in adult pyramidal neurons.

CRH directly affects discrete sub-populations of dendritic spines

Spine shape and size are dynamic.^30–32 Thus, the above results, showing reduced numbers of thin spines after exposure to the stress-hormone CRH, could arise from either a CRH-induced transformation of thin spines into mushroom-type spines^30–32 or from a direct destruction of thin spines. To distinguish between these possibilities, we conducted experiments using live, multiphoton microscopy of hippocampal slices from YFP-expressing adult mice, spanning the duration in which CRH exerts its physiological effects. These experiments demonstrated the basic stability of both thin (arrows) and mushroom-type (arrowheads) spines under the conditions of the experiment (~75 min; Figures 3a and b). CRH (100 nm) led to selective and progressive disappearance of thin spines over the same time course (Figure 3c). These data indicate that rather than potentially shifting the size/shape profile of populations of dendritic spines, CRH directly affected a distinct sub-population.

Loss and size shift of synapses accompany disappearance of thin dendritic spines

If the loss of thin spines, found using direct visualization of YFP-expressing adult mice, accounts for the observed depression of synaptic responses, then it should be accompanied by a measurable loss of synapses. We tested this prediction using an independent method, wide-field three-dimensional deconvolution microscopy followed by immunostaining for PSD-95, an integral protein of glutamatergic synapses.^22,24,30,41 Previously published work has demonstrated the identity of PSD-95-labeled puncta of precisely defined size and shape as synapses in hippocampal stratum radiatum.^24,33,42 CRH caused an ~28% reduction in excitatory synapses (Figure 4b), a value similar to that for the net loss of spines in the same dendritic subfield (Figure 2d). Plotting the size distribution of the PSD-95-expressing puncta demonstrated a prominent reduction of the small (0.1–0.2 μm) puncta in CRH-treated slices (Figure 4c), and a distinct preservation of the large ones (Figure 4c, inset). This suggests that the population remaining after CRH application was enriched with spines with large heads and thus large synapses.^{24,30,32,41,43} Together, the results from two very different measurements support the conclusion that prolonged exposure to nanomolar concentrations of the hippocampal stress hormone CRH reduces the number of thin spines carrying small synapses.

It is generally considered that thin spines, and small synapses, are relatively poor in the amino-3-hydroxy-5-methyl-4-isoxazole propionic (AMPA) receptor subunit GluR1.^44–47 When these synapses and spines receive patterned afferent stimulation, they may undergo structural and functional plasticity that contributes to the process of memory;^48,49 that is, synapses expand^{24,30,31,41,43} and GluR1 receptors are added,^30,44,46,47 concomitant with growth and transformation of thin spines into larger, mushroom-type spines.^31,32,41 These facts predict that selective removal of thin spines, as described here, should have relatively little effect on GluR1-immunoreactive clusters in the size range of synapses. We tested this point using antisera selective to surface GluR1 followed by quantitative immunofluorescence microscopy. We found that the ratio of GluR1-labeled/PSD-95-labeled puncta was ~0.5 (Figures 4b and d), consistent with the percentage of mushroom-type spines (Figure 2e). In accord with the apparent selective effects of CRH on thin spines found above, a 1-h infusion of the peptide onto adult hippocampal slices did not reduce the number of visible GluR1-labeled puncta in the size range of synapses (Figure 4d), nor did it alter the size profile of these puncta (Figure 4e). These results provide additional evidence that CRH has little influence on the spines and synapses that remain after elimination of thin, GluR1-poor entities.

Loss of thin spines induced by CRH requires RhoA signaling

What type of mechanism underlies the selective spine losses caused by CRH? One possibility is suggested by the observation that the peptide blocks theta burst-driven actin polymerization. This polymerization was previously shown to depend on stimulation of an actin signaling pathway involving the small GTPase RhoA and its downstream effector cofilin, a protein involved in actin filament assembly.^24,25,33,50 Testing if CRH affects the level of RhoA activation, we discovered that a 60-min infusion led to a marked (threefold) increase in the concentration of active, GTP-bound RhoA as assessed in immunoblots (Figure 5a). These effects were relatively selective in that CRH led to only a modest reduction (~25%) in active Rac1 and little change in the activation of cdc42 (Figures 5b and c), two other members of the Rho family of small GTPases. Because the Rho GTPases may compete for the same pool of GTP, major increases in the activation of one member are not uncommonly associated with modestly reduced activation of a second member.³³ If abnormal RhoA activation contributes importantly to the effects described here, then inhibitors of RhoA kinase (ROCK), the principle effector of this GTPase, should protect thin spines from the deleterious effects of CRH. We therefore infused a highly selective inhibitor of the kinase (H1152) along with CRH onto adult hippocampal slices from YFP-expressing mice, and carried out the analyses of spine numbers and types. The results were dramatic; in the presence of H1152, CRH had no effect on the number or size distribution of spines (Figures 5d and e).