Guan Hock Khoo1&Yu-Ting Lin1,2&Tsung-Chih Tsai2&Kuei-Sen Hsu1,2
Received: 1 November 2018 / Accepted: 8 February 2019 / Published online: 2 March 2019
# Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
Long-term depression (LTD) of synaptic efficacy is widely regarded as a cellular basis of learning and memory. The magnitude of hippocampal CA1 LTD induced by low-frequency stimulation (LFS) declines with age, but the mechanisms involved remain poorly understood. Perineuronal nets (PNNs) are specialized extracellular matrix structures that function in dampening synaptic plasticity during postnatal development, suggesting that PNN formation may restrict LTD induction in the adult hippocampus.
Here, we show that PNNs tightly enwrap a subpopulation of parvalbumin (PV) interneurons in the hippocampal CA1 region and enzymatic removal of PNNs with the chondroitinase ABC alters the excitatory/inhibitory synaptic balance toward more excita-tion and restores the ability of LFS to induce an N-methyl-D-aspartate receptor-dependent LTD at Schaffer collateral-CA1 synapses in slices from male adult mice. Early interference with depolarizing GABA with Na+-K+-2Cl−cotransporter inhibitor bumetanide impairs the maturation of PNNs and enhances LTD induction. These results provide novel insights into a previously unrecognized role for PNNs around PV interneurons in restricting long-term synaptic plasticity at excitatory synapses on hippocampal CA1 neurons in adulthood.
Keywords Perineuronal nets . Long-term depression . Excitatory/inhibitory synaptic balance . Parvalbumin interneurons . Hippocampus
Introduction
Long-term depression (LTD) is a persistent activity-dependent reduction in synaptic efficacy that, together with its counter-part, long-term potentiation (LTP), has long been implicated in information storage and adaptation to external stimuli [1–3]. Over the past decades, LTD received much less atten-tion than LTP among those who study synaptic plasticity.
While LTD can be induced in many brain regions, much of our current understanding of the properties and its functional relevance comes from studies in the hippocampus. In the hip-pocampus, LTD can be experimentally induced by several different types of electrical and pharmacological stimulation protocols [1, 4,5]. The most commonly used protocol for inducing LTD involves prolonged low-frequency stimulation
(LFS) at 0.5–5 Hz. In the hippocampal CA1 region, LFS-induced LTD (LFS-LTD) requires the activation of N-meth-yl-D-aspartate receptors (NMDARs), a rise in postsynaptic intracellular Ca2+, and through consequential activation of serine-threonine protein phosphatase cascades [6,7]. Despite considerable progress in understanding the cellular and mo-lecular mechanisms underlying LTD, it was intriguing to find that LTD is difficult to elicit and less robust in hippocampal slices from adult animals when compared to slices from their young counterparts [8–13]. However, the mechanisms under-lying the age-related decline in the magnitude of LTD remain elusive.
Perineuronal nets (PNNs) are lattice-like extracellular matrix structures that wrap around the soma and proximal dendrites of subpopulations of neurons throughout the brain and spinal cord [14–16]. PNNs are formed during postnatal development and completed by early adulthood [17]. The importance of PNNs is highlighted by the fact that the emergence of PNNs coincides with closure of the critical period plasticity within the visual, motor, and somatosensory systems [17–20], suggesting a possi-ble role of PNNs in restricting brain plasticity and stabilizing neural network. Parvalbumin (PV)-expressing interneurons are the predominant neuronal subtype that is enveloped by PNNs.
* Kuei-Sen Hsu
1 Department of Pharmacology, College of Medicine, National Cheng Kung University, No. 1, University Rd., Tainan City 70101, Taiwan
2 Institute of Basic Medical Sciences, College of Medicine, National Cheng Kung University, Tainan 70101, Taiwan
Molecular Neurobiology(2019) 56:6436–6450 https://doi.org/10.1007/s12035-019-1526-1
Author's personal copy
The role of PNNs in the regulation of PV interneuron function has not reached a consensus. Degradation of PNNs has been shown to increase [21,22] or decrease [23] the intrinsic excit-ability of PV interneurons. PNNs have also been shown to con-trol synaptic plasticity in the amygdala and hippocampus. In the amygdala, PNN deposition is followed by fear memory persis-tence, and conversely, disruption of PNNs by chondroitinase ABC (ChABC) renders fear memories susceptible to erasure [24]. Additionally, in the hippocampus, PNNs have been shown to modulate LTP at Schaffer collateral-CA1 synapses [25,26]
and contribute to contextual fear memory [26], as well as sup-press LTP at excitatory synapses on CA2 pyramidal neurons [27]. These findings strongly imply a potential role for PNNs in regulating LTD induction, but direct and causal evidence for this association is currently unavailable. Therefore, in the present study, we hypothesized that PNNs may act as a molecular brake on LTD induction and that the progressive developmental in-crease in PNNs may be associated with age-related dein-crease in the magnitude of LTD at Schaffer collateral-CA1 synapses.
Althoughγ-aminobutyric acid (GABA) is a major inhibitory neurotransmitter in the adult brain, GABA excites immature neu-rons due to high expression of the Na+-K+-2Cl−cotransporter (NKCC1) in the developing brain [28]. GABA-mediated depo-larization has been implicated in a series of developing processes, including neuronal migration and morphological differentiation, and the plasticity of neuronal circuits [29,30]. Previous work demonstrated that blocking early GABA-mediated depolariza-tion with selective NKCC1 antagonist bumetanide results in last-ing disruption of cortical excitatory synapse formation [31].
Interestingly, a recent study indicated that early bumetanide treat-ment impairs PNN developtreat-ment around PV interneurons and prolongs critical period plasticity in the rat visual cortex [32].
On the basis of these observations, we also hypothesized that bumetanide treatment during early postnatal development may affect the maturation of PNNs, which may then alter the critical period for LTD induction. Here, our results demonstrate, for the first time, that PNNs function to restrict CA1 LTD induction in adulthood and that PNN disruption reinstates juvenile-like states of synaptic plasticity. In addition, the effect of PNNs on LTD induction emerged from maintaining excitation/inhibition synap-tic balance in CA1 network of the hippocampus. Moreover, early-life bumetanide treatment impairs the maturation of PNNs and enhances LTD induction in adult slices.
Materials and Methods
AnimalsYoung (P14-28) and adult (P56-84) male C57BL/6 mice were used in experiments. Mice were housed in groups of four under a 12-h light/dark cycle (lights off at 7:00 P.M.) with access to food and drinking water ad libitum. All experimental
procedures were conducted in accordance with the National Institutes of Health guidelines for the care and use of labora-tory animals and were approved by the Institutional Animal Care and Use Committee of National Cheng Kung University.
In early-life bumetanide treatment experiments, male litter-mates were injected intraperitoneally twice daily, from P3 to P8, with vehicle [0.01% dimethyl sulfoxide (DMSO) in phys-iological solution] or bumetanide (0.2 mg/kg body weight;
Sigma-Aldrich, Saint Louis, MO), according to previously described treatment regimen [32].
Hippocampal Slice Preparations and Electrophysiological Recordings
Acute hippocampal slices were prepared using standard pro-cedures as described previously [33]. Mice were deeply anes-thetized with 5% isoflurane and sacrificed by decapitation.
The brain was removed and quickly placed in ice-cold oxy-genated sucrose cutting solution containing the following (in mM): 234 sucrose, 2.5 KCl, 0.5 CaCl2, 7 MgCl2, 25 NaHCO3, 1.25 NaH2PO4, and 11 glucose at pH 7.3–7.4 and equilibrated with 95% O2–5% CO2. Coronal slices containing the hippo-campus (250 or 400μm) were prepared using a vibrating microtome (VT1200S; Leica Biosystems, Wetzlar, Germany) and immediately transferred to a holding chamber of artificial cerebrospinal fluid (aCSF) containing the follow-ing (in mM): 117 NaCl, 4.7 KCl, 2.5 CaCl2, 1.2 MgCl2, 25 NaHCO3, 1.2 NaH2PO4, and 11 glucose at pH 7.3–7.4 and equilibrated with 95% O2–5% CO2and then kept at room temperature (~ 25 °C) for at least 1 h before starting record-ings. For ChABC treatment experiments, slices were incubat-ed with vehicle [0.1% bovine serum albumin (BSA)] or ChABC (0.2 U/ml in 0.1% BSA; Sigma-Aldrich catalog#
C2905) for at least 2 h at 37 °C before being transferred to the recording chamber, according to previously described pro-cedure [25].
For extracellular field potential recordings, one slice was transferred to a submersion-type recording chamber and con-tinuously perfused with oxygenated aCSF at a flow rate of 2–
3 ml/min at ~ 32 °C on a fixed stage. The extracellular field potential recordings were carried out using an Axoclamp-2B a m p l i f i e r ( M o l e c u l a r D e v i c e s , S a n J o s é , C A ) . Microelectrodes were pulled from microfiber-containing glass capillary tubings (outer diameter = 1.0 mm) on a Brown-Flaming electrode puller (Sutter Instruments, Novato, CA) and were filled with 1 M NaCl. The responses were low-pass-filtered at 2 kHz, digitally sampled at 10 kHz using a Digidata 1320A (Molecular Devices), and analyzed with pCLAMP 8.0 software (Molecular Devices). Field excitatory postsynaptic potentials (fEPSPs) were evoked by electrical stimulation to Schaffer collateral/commissural fibers in the stratum radiatum of the CA1 area with a bipolar tungsten-stimulating electrode at the baseline frequency of 0.033 Hz.
Mol Neurobiol (2019) 56:6436–6450 6437
Author's personal copy
The stimulation strength was adjusted to elicit a response hav-ing amplitude that was 30–40% of the maximum spike-free response. The slope of fEPSP was measured from approxi-mately 20–70% of the rising phase using the least squares regression. Paired-pulse facilitation (PPF) was assessed by using a succession of paired pulses separated by intervals of 20, 40, 60, 80, 100, and 200 ms. LTD was induced by appli-cation of LFS at 1 Hz for 15 min (900 pulses), paired-pulse LFS (PP-LFS) at 1 Hz for 15 min (900 paired pulses, 40 ms interpulse interval), (S)-3,5-dihydroxyphenylglycine (DHPG, 50μM; Tocris Bioscience, Bristol, UK) for 5 min, or NMDA (15μM; Tocris) for 3 min. The magnitude of LTD was calcu-lated as percentage of change of fEPSP slope 50–60 min after LTD induction compared to baseline fEPSP (10 min before LTD induction).
Whole-cell patch-clamp recordings were made from visu-alized pyramidal neurons in the CA1 region of hippocampal slices using an Axopatch 200B amplifier (Molecular Devices). Data acquisition and analysis were performed using a digitizer (Digidata 1440A) and pCLAMP 9 software (Molecular Devices). Patch pipettes (3–6 MΩ) were filled with an intracellular recording solution containing the follow-ing (in mM): 130 CsMeSO4, 8 CsCl, 1 MgCl2, 0.3 EGTA, 10 HEPES, 4 Mg-ATP, 0.3 Na-GTP, 10 Na-phosphocreatine, and 1 QX-314 (pH 7.2 adjusted with CsOH; 280–290 mOsm). To evoke perisomatic inhibition, a bipolar-stimulating electrode was placed in the middle of the stratum pyramidale (SP) of CA1 region 200μm away from the recorded cell. To evoke dendritic inhibition, the stimulating electrode was placed in the stratum radiatum (SR), 200–300 μm away from the re-corded cell. All stimulation was conducted at 0.1 Hz to avoid inducing synaptic plasticity. Excitatory postsynaptic current/
inhibitory postsynaptic current (EPSC/IPSC) ratio was calcu-lated as the peak EPSC at − 65 mV divided by the IPSC amplitude at 0 mV. A total of 20 recording events with inter-vals of 30 s at each holding potential were used for analysis.
For miniature inhibitory postsynaptic current (mIPSC) record-ings, CsMeSO4was replaced with CsCl in the intracellular recording solution and tetrodotoxin (TTX, 0.5μM; Sigma-Aldrich), 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 20μM; Tocris), and D-2-amino-5-phosphonopentanoic acid (APV, 50μM; Tocris) were added to the bath. mIPSCs were recorded from CA1 pyramidal neurons held in voltage-clamp mode at a holding potential of− 60 mV and analyzed off-line using a commercially available software (Mini Analysis 4.3;
Synaptosoft, Leonia, NJ) as previously described [34].
Bicuculline methiodide (20μM; Tocris) was routinely applied at the end of the recording to verify that the mIPSCs were exclusively mediated by GABAAreceptors. Means were cal-culated from 3-min epochs recorded. Detection threshold for analysis was set at three times the root mean square of the background noise, and each event was further confirmed by visual inspection after detection. To assess cell stability, series
and input resistances were continuously monitored throughout the experiment with a 5-mV depolarizing step given after ev-ery afferent stimulus and data were excluded from analysis if resistance changed by more than 20%.
Immunohistochemistry
Mice were deeply anesthetized with a mixture of Zoletil (50 mg/kg; Virbac, Carros, France) and Rompun (0.5 mg/kg;
Bayer, Leverkusen, Germany), and were perfused transcardially with 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS), pH 7.4. After the perfusion, brains were rapidly removed, fixed in 4% PFA for 24 h at 4 °C, and then equilibrated in 30% sucrose for 48 h at 4 °C before slicing. Coronal brain slices (20 μm) containing the hippocampus were washed with 0.4% Triton X-100-contain-ing PBS and then incubated in blockX-100-contain-ing solution containX-100-contain-ing 3% goat serum in PBS. After this, sections were incubated in the primary antibodies against NeuN (1:200; Millipore cata-log# MAB377, RRID:AB_2298772, Darmstadt, Germany), calcium/calmodulin-dependent protein kinase IIα (CaMKIIα, 1:500; Novas Biologicals catalog# NB100-81830, RRID:AB_1145020, Littleton, CO), glutamic acid de-carboxylase 67 (GAD67, 1:1000; Millipore catalog#
MAB5406, RRID:AB_2278725), parvalbumin (PV, 1:1000;
Millipore catalog# MAB1572, RRID:AB_2174013), or Wisteria floribunda agglutinin (WFA, 1:1000; Vector labora-tories catalog# B-1355, RRID:AB_2336874, Burlingame, CA). Finally, sections were washed three times with 0.4%
Triton X-100 in PBS and incubated in secondary Alexa Fluor 488 (Invitrogen Molecular Probes, catalog# A150077, Eugene, OR) or Alexa Fluor 568 antibodies (Invitrogen Molecular Probes, catalog# A10037) for 2 h at room temper-ature. After being washed with PBS, sections were mounted with ProLong Gold Antifade Reagent (Invitrogen Molecular Probes) with or without 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories). Images were acquired on an Olympus FluoView FV1000 confocal microscope (Olympus, Tokyo, Japan) with sequential acquisition setting at a resolution of 1024 × 1024 pixels, z-stack with 15–20 op-tical sections. All images were analyzed by NIH ImageJ soft-ware and all the parameters used were kept consistent during capturing. For the quantification of PNN+ and PV+ neurons, the dorsal hippocampal CA1 was analyzed at − 1.94 to − 2.18 mm from bregma as described previously [35]. Three sections per mouse were acquired and analyzed. The data per mouse was the average of the sections. All counting was performed in a blind manner.
Statistical Analysis
No statistical methods were used to predetermine sample size, but our sample sizes were based on previous work of a similar
6438 Mol Neurobiol (2019) 56:6436–6450
Author's personal copy
nature by our laboratory [36,37]. The results are presented as mean ± SEM. All statistical analyses were performed using the GraphPad Prism 6 software (RRID:SCR_000306). To compare the difference between the two population means, we first determined whether the data were normally distribut-ed using the Shapiro-Wilk test. The significance of any differ-ence between two groups was calculated using the unpaired two-tailed Student t test. One-way or two-way repeated mea-sures ANOVA tests were used for multiple groups’ compari-son and Bonferroni’s post hoc analyses were used to assess the significance between groups. Because the data of LTD mag-nitudes was not normally distributed, the Mann-Whitney U test was used to compare differences between two indepen-dent groups. N represents the number of slices or animals used. Values of p < 0.05 were considered significant.
Results
PNNs Mainly Enwrap a Subpopulation of Parvalbumin-Expressing Interneurons
We initially sought to examine the distribution of PNNs and the identity of PNN-enwrapped cells in the hippocampal CA1 region of adult mice. PNN-enwrapped cells, as detected by WFA labeling [25], were found predominantly in the stratum oriens (SO) and SP of the CA1 region. Double immunofluo-rescent staining with the neuronal marker NeuN revealed that nearly all PNN-enwrapped cells were positive for NeuN (∼
95.6%), indicating that these cells were neurons (Fig.1a, c).
To determine the identity of PNN-enwrapped neurons, double immunofluorescent staining was performed with antibody against CaMKIIα, a marker of excitatory neurons.
Immunoreactivity for CaMKIIα was not detected at all PNN-enwrapped neurons (Fig.1a, c). In contrast, we observed that the large majority (∼ 90.5%) of PNN-enwrapped neurons express GAD67, a marker of GABAergic neurons (Fig.1b, c).
Moreover, consistent with previous work [22,38], the vast majority (89.8 ± 4.8%) of PNN-enwrapped neurons were immunopositive for PV; however, only 40.3 ± 5.6% of PV-expressing (PV+) neurons were labeled with WFA (Fig.1d).
The density of WFA + PV+ neurons was highest in the SP, less in the SO and SR, but not in the stratum lacunosum-moleculare (SLM) (Fig. 1e). These results indicate that PNNs are predominantly expressed around a subpopulation of PV+ GABAergic interneurons in the hippocampal CA1 region.
An Inverse Correlation Between PNNs and the Magnitude of LTD
To explore the interaction between PNNs and LTD induction in early life, we measured the expression of PNNs and the
magnitude of LTD in the hippocampal CA1 region of mice at three different developmental stages. If an increase in PNNs is a critical contributor to dampen LTD, then the time course of these changes should directly overlap in time with changes in LTD magnitude. In agreement with previous findings [39, 40], we observed that the densities of PV+ (Fig.2a, b) and WFA+ neurons (Fig.2a, c) in the CA1 SP and SO increased gradually with age during early postnatal development.
Accordingly, there was a progressive increase in the density of WFA + PV+ neurons across postnatal age from postnatal day (P) 14 to P28 (Fig.2a, d). To assess age-related alterations in LTD induction, we chose a prolonged LFS (1 Hz for 15 min) protocol particularly efficient in inducing robust LTD at Schaffer collateral-CA1 synapses in hippocampal slices from younger animals [12,41]. As expected, a signifi-cant age-related decrease in the magnitude of CA1 LTD (50–
60 min after the end of LFS) was observed (one-way ANOVA, F(2,17)= 25.23, p < 0.0001), with 33.2 ± 3.5% de-pression in P14 pups, 23.8 ± 2.6% in P21 juveniles, and 5.8
± 1.4% in P28 young adults (Fig.2e, f). Moreover, a statisti-cally significant inverse correlation was observed between the density of WFA + PV+ neurons and the magnitude of LFS-LTD (r = 0.99, p = 0.045; Fig.2g).
PNNs Restrict LFS-Induced LTD
Having observed an inverse relationship between PNNs and the magnitude of LFS-LTD during early postnatal development, we next asked whether PNNs act to restrict LFS-LTD induction in the hippocampal CA1 region of adult mice. To test this, we degraded PNNs with the enzyme ChABC, which digested and removed glycosaminoglycan chains from the core proteins in PNNs [42], in acute hippocampal slices from adult mice (P56-P84) and attempted to induce LFS-LTD at Schaffer collateral-CA1 synapses. Enzymatically treated slices were morphological-ly and functionalmorphological-ly intact (on the basis of normal action potential firing, Fig.6a). As demonstrated previously [25], we found that 2-h treatment of slices with ChABC (0.2 U/ml in 0.1% BSA) ablated WFA-labeled PV+ neurons in the hippocampal CA1 region (Fig.3a). Treatment of slices with ChABC did not affect basal synaptic transmission because the stimulus-response curves for fEPSPs were similar between vehicle (0.1% BSA)- and ChABC-treated slices (Fig.3b). In addition, we found no signif-icant difference between vehicle- and ChABC-treated slices in PPF of fEPSPs, an indirect measure of presynaptic release prob-ability, at any of the interpulse intervals examined (Fig.3c). As reported previously for untreated slices [12,41], LFS at 1 Hz for 15 min failed to elicit a reliable LTD at Schaffer collateral-CA1 synapses in vehicle-treated slices. However, a 2-h incubation with ChABC enabled LTD induction in slices from adult mice (Fig.3d). Notably, LTD in slices treated with ChABC was sig-nificantly greater than LTD in slices treated with vehicle (p = 0.04, Mann-Whitney U test; Fig. 3g). To more thoroughly
Mol Neurobiol (2019) 56:6436–6450 6439
Author's personal copy
examine whether PNNs also regulate the induction of other forms of CA1 LTD induced by different stimulation protocols, we recorded LTD induced by group I metabotropic glutamate receptor (mGluR) agonist DHPG and PP-LFS, respectively, in slices that were treated with vehicle or ChABC. As shown in Fig.
3e, bath application of DHPG (50μM) for 5 min induced a reliable LTD of fEPSPs. We detected no significant difference in the magnitude of DHPG-induced LTD between vehicle- and ChABC-treated slices (Fig.3g). Moreover, unlike that in LFS-LTD, there was no age-related loss of LTD induced by PP-LFS (Fig.3f). There was no difference in the magnitude of PP-LFS-induced LTD between vehicle- and ChABC-treated slices (Fig.
3g). Collectively, these results indicate that PNNs specifically restrict LFS-LTD at mature Schaffer collateral to CA1 synapses.
Mechanisms Underlying the Induction of LFS-LTD in ChABC-Treated Slices
We then set out to uncover the possible mechanism underlying LFS-induced LTD in ChABC-treated slices from adult mice.
In slices from young animals, it is well known that hippocam-pal CA1 LFS-LTD is dependent on the activation of
In slices from young animals, it is well known that hippocam-pal CA1 LFS-LTD is dependent on the activation of