Synapse-specific regulation of AMPA receptor function by PSD-95 -- Béïque et al. 103 (51): 19535 Data Supplement - HTML Page - index.htslp -- Proceedings of the National Academy of Sciences

Béïque et al. 10.1073/pnas.0608492103.

Supporting Information

Files in this Data Supplement:

Supporting Figure 5
Supporting Methods
Supporting Figure 6
Supporting Figure 7
Supporting Figure 8
Supporting Figure 9
Supporting Figure 10
Supporting Figure 11
Supporting Figure 12

Supporting Figure 5

Fig. 5. Targeted disruption of the PSD-95 gene. (A) Schematic representation of PSD-95 cDNA, genomic structure, targeting vector, and targeted allele. Homologous recombination disrupts exons encoding PDZ 1 and 2 domains of PSD-95 gene resulting in an out-of-frame mutation in the PSD-95 gene. Probe (bar) and EcoRI sites that were used in Southern blot analysis are indicated. Black boxes, exons; Neo, neomycin-resistant gene; TK, thymidine kinase gene. (B) Southern blot analysis of EcoRI-digested genomic DNA using the probe located outside of targeting vector sequence. Note that WT bands (19.0 kb) are relatively weaker than mutant bands (3.5 kb) because of low transfer efficiency of high-molecular-size bands. (C) Immunoblot analysis using whole brain extracts from WT +/+, heterozygous +/-, and homozygous -/- mutant mice. Note that similar results were obtained with an array of five additional polyclonal antibodies raised against different epitopes of PSD-95 (provided by James Trimmer, University of California, Davis, CA; not shown).

Supporting Figure 6

Fig. 6. Current-voltage relationship of AMPAR-mediated synaptic currents is unchanged in PSD-95^-/- mice. (A) Current traces and I-V relationships for AMPAR-mediated EPSCs from a WT neuron and from a KO neuron are shown. (B) Average I-V relationships for AMPAR-mediated EPSCs in PSD-95^+/+ (n = 6) and PSD-95^-/- neurons (n = 6). The experiments were carried out in the presence of 100 mM (DL)-AP-5 and 30 mM bicuculline. Data were normalized for every cell to the amplitude of the synaptic current obtained at -60 mV.

Supporting Figure 7

Fig. 7. Glutamate release probability appears unchanged in PSD-95^-/- mice as determined by paired-pulse ratio analysis. (A) Superimposed current traces depicting AMPAR-mediated eEPSCs induced by paired pulses delivered at different interstimulus intervals obtained from a PSD-95^+/+ and PSD-95^-/- neuron are shown. (B) The average ratios (S2/S1) of AMPAR-mediated eEPSCs obtained with different interstimulus intervals in PSD-95^+/+ mice (n = 9) are not significantly different from those obtained in PSD-95^-/- mice (n = 10).

Supporting Figure 8

Fig. 8. In young hippocampus, the amplitude of 2P-EPSCs is correlated with spine volume but not spine length. (A) The amplitude of 2P-EPSCs is plotted against the volume (Left) and the length (Right) of the spine onto which MNI-glutamate was uncaged. The laser power was kept constant for any given dendritic segment from which spines were chosen for uncaging. Values obtained for a given dendritic segment are outlined by color coding. The amplitude of 2P-EPSCs was correlated (R² = 0.34) with spine volume but not with spine length (R² = 0.01). Note that calculating correlation between volume and amplitude for individual dendritic segment (i.e., assuring a more constant laser power) yielded an even greater correlation between the amplitude of 2P-EPSCs and spine volume (average R² = 0.59). (B) In our recording conditions, repetitive uncaging of MNI-glutamate (0.1 Hz) onto a single spine triggers stable responses with low variability.

Supporting Figure 9

Fig. 9. PSD-95 gene deletion and spine volume and length. (A) We calculated spine length for the population of spines onto which we uncaged glutamate and found that it was slightly longer in PSD-95^-/- mice than in WT mice (PSD-95^+/+, 0.98 ±0.06 mm, n = 20 spines; PSD-95^-/- 1.27 ±0.07 mm, n = 25 spines; P < 0.01). (B) To circumvent any potential selection bias, we again measured spine length from a much broader population of spines and again found that spines were longer in PSD-95^-/- mice (PSD-95+/+, 0.77 ± 0.01 mm, n = 271 spines, 7 cells, 4 mice; PSD-95^-/- 0.89 ± 0.01 mm, n = 509 spines, 12 neurons, 5 mice; P < 0.01). (C) The 2P-AMPAR/NMDAR ratio obtained from uncaging onto individual spines was plotted against the length of those spines. The AMPA/NMDA ratio was uncorrelated with spine length in both WT (R² = 0.03) and KO (R² = 0.01) mice. We also include here cumulative distribution plots for spine volume of the spines onto which we have carried out 2P-uncaging (D₁; WT, n = 20; KO, n = 25) and for a much broader population (D₂; WT, n = 271; KO, n = 509). Likewise, cumulative distribution plots are shown for spine length for the spines onto which we have carried out 2P-uncaging (E₁) and from a broader population (E₂).

Supporting Figure 10

Fig. 10. Parameters of NMDAR-mediated synaptic current decay kinetics. The decay kinetics of isolated NMDAR-mediated synaptic currents was best fitted by a double exponential with decay time constants _fast and _slow. The weighted decay time constant (_w) was calculated as outlined in the supplemental Methods section. % fast indicates the relative proportion of the fast component to the peak current amplitude. These parameters are plotted for WT and PSD-95^-/- mice. (B) The decay kinetics of the NMDAR portion of the current induced by 2P-uncaging of MNI-glutamate (as determined in Fig. 4E) was plotted against the volume of spine from which the 2P-current was elicited. No correlation was found between these two parameters, both in WT (R² = 0.05) and KO (R² = 0.01).

Supporting Figure 11

Fig. 11. The magnitude of LTP is greater in PSD-95^-/- mice. Extracellular recordings of fEPSPs in slices before and after a theta burst stimulation of Schaffer collaterals show that the magnitude of LTP obtained in KO mice (n = 9) was (P < 0.01) greater than that obtained in WT (n = 10). Sample traces represent averaged field potentials before and 30-40 min after tetanization.

Supporting Figure 12

Fig. 12. Linear correlation between the true radius (R) of a sphere and the FWHM of its fluorescence intensity profile. The FWHM value of the fluorescent profiles obtained for different R values were predicted for our imaging conditions from equation (1) of Supporting Methods. For R values between 0.3 mm and 0.6 mm, R was linearly correlated with FWHM (r² = 0.99). The resulting function was expressed by the equation shown in the Inset.

Supporting Methods

Generation of PSD-95-Deficient Mouse.
PSD-95 deficient mice were generated as previously described(1). Briefly, genomic DNA was isolated from 129/Svj mouse genomic DNA BAC library (Invitrogen Corporation, Carlsbad, CA). PSD-95 containing genomic DNA regions were subcloned into pBluescript (Stratagene) and targeting vector was constructed based on information from the genomic DNA database. Exons encoding PSD-95 PDZ 1 and 2 domains were replaced by a Neo (neomycin resistance gene) cassette. Homologous genomic DNA with the Neo cassette was joined with a thymidine kinase cassette for negative selection to construct targeting vector. Linearized targeting vectors were electroporated in R1 ES cells (from A. Nagy, Mount Sinai Hospital, Toronto, ON, Canada) and homologous recombinants were isolated by positive and negative selections with G418 (Invitrogen) and Gancyclovir (Roche Products, Hertfordshire, U.K.), respectively. ES clones were screened by PCR and homologous recombination was confirmed by Southern blot analysis. Correctly targeted ES clones were injected into C57 BL/6 blastocyst followed by chimera mice production at the Transgenic Facility of Johns Hopkins University School of Medicine. After germ line transmission, heterozygote mice were intercrossed to generate homozygote mice. Mice genotypes were confirmed by PCR and Southern blot analysis. For all experiments, mice with 129 and C57BL/6 hybrid genetic background were used. All experiments were done in accordance with the policies of the Animal Care and Use Committee of the Johns Hopkins University School of Medicine Western Blot Analysis.
Brain homogenates were prepared from freshly dissected mouse whole forebrain in 10-vol of RIPA buffer (50 mM Tris·Cl, pH 6.8/100 mM NaCl/2 mM EDTA/1% SDS) containing protease inhibitors. Homogenates (2.5 ml) were electrophoresed and transferred to PDVF membranes. Western blotting was performed with 5% Skim Milk in TBS containing 0.05% Tween 20. Mouse anti-PSD95 monoclonal antibody (IgG2a; Upstate), as well as three other anti-PSD-95 polyclonal antibody kindly provided by Dr James Trimmer (University of California, Davis, CA) were used in the present study. Signal was visualized by using Chemiluminescence Reagent Plus (PerkinElmer Life Science, Boston, MA). Slice Preparation.
Wild-type or knockout littermate mice aged 10-25 days (P10-25) or rats P7-8 were anaesthetized with halothane (by inhalation) and killed by decapitation. The brain was quickly removed and cooled in ice-cold Ringer solution of the following composition (mM): 119 choline Cl, 2.5 KCl, 4.3 MgSO₄, 1 CaCl₂, 1 NaH₂PO₄, 26.2 NaHCO₃ and 22 glucose, bubbled to saturation with 95% O₂/5% CO₂. Coronal slices (300 mm thick) were cut by using a vibratome (Leica VT100S) and transferred to a holding chamber where they were allowed to recover for 1 h at 33-35°C in normal Ringer's solution containing (in mM) 119 NaCl, 2.5 KCl, 1.3 MgSO₄, 2.5 CaCl₂, 1 NaH₂PO₄, 26.2 NaHCO₃ and 11 glucose. After 1 h, the holding chamber was placed at room temperature until recordings. Field Recordings.
Field synaptic responses were evoked by stimulating Schaffer collaterals with 0.2-ms pulses delivered by using a concentric bipolar stimulating electrode and recorded extracellularly in CA1 stratum radiatum. To induce LTP, four episodes of TBS were delivered at 0.1 Hz, by using the same stimulation intensity as for baseline. TBS consists of 10 stimulus trains delivered at 5 Hz with each train consisting of four pulses at 100 Hz. Whole-Cell Recordings.
Hippocampal CA1 pyramidal neurons were targeted for whole-cell patch-clamp recording using differential interference contrast imaging on a fixed-stage upright microscope (Axioscope; Carl Zeiss, Oberkochen, Germany). Electrical signals were recorded by using an Axopatch 200, an Axopatch 200B or a Multiclamp 700B amplifier (Axon Instruments, Foster City, CA). All recordings were carried out at room temperature except for mEPSCs recordings that were acquired at 33°C. Recordings were filtered at 2 KHz, digitized at 10 KHz and acquired with Clampex (Axon Instruments). The recording pipettes were pulled from borosilicate glass (1.2 mm outer diameter) by using a Flaming-Brown horizontal puller (Model P97, Sutter Instruments). Liquid junction potentials and voltages were left uncompensated. Two intracellular solutions were used in the present study (in mM): 77 Cesium gluconate, 10 tetracesium BAPTA, 5 TEA-Cl, 3 CaCl₂, 20 Hepes, 4 MgATP, 0.5 GTP, 5 QX-314 and 10 Na₂phoshocreatine or 115 Cesium methanesulfonate, 0.4 EGTA, 5 TEA-Cl, 2.8 NaCl, 20 Hepes, 4 Mg ATP, 0.5 GTP, 5 QX-314 and 10 Na₂phoshocreatine. All experiments involving evoking glutamate current at +40 mV were carried out by using the first solution. The initial characterization of 2P-uncaging were carried out with the second solution. The pH was adjusted to 7.3 and osmolality to 280-290 mOsm. In all experiments, Ringer's solution was supplemented with 30 mM (-)bicuculline. Access resistance was continuously monitored during the experiments by delivering a 3-5 mV hyperpolarizing voltage step 180 msec before the delivery of the stimulus pulse. Experiment were discarded if the access resistance changed by more than ≈25%.

Evoked excitatory postsynaptic currents (eEPSCs) were obtained by placing a bipolar stainless steel stimulating electrode in the stratum radiatum. Two methods were used to discriminate AMPAR from NMDAR-mediated synaptic responses at + 40 mV to determining AMPAR/NMDAR ratios of eEPSCs. First, bath administration of the NMDAR antagonist (DL)-AP-5 (100 mM) was used to isolate the AMPAR-mediated eEPSC, and the NMDAR-mediated component was derived by subtracting the AMPAR-mediated component from the compound eEPSC. AMPAR/NMDAR ratios were then computed from the respective peak current. The second method implied directly estimating the amplitude of AMPAR and NMDAR-mediated currents from the trace obtained at + 40 mV by virtue of their different kinetics. The time window to measure the AMPAR current at + 40 mV was determined from the eESPC trace obtained at -70 mV. The time window to measure the NMDAR current at + 40 mV was set at 3 X decay time constant of AMPAR current at -70 mV (_AMPA in ms, usually ≈10-12 msec) following the peak of the AMPAR response at -70 mV (see Fig. 1A₂ Inset).

mEPSCs were acquired at -70 mV in control Ringer's solution supplemented with 1 mM TTX and 30 mM (-)bicuculline. Analysis was carried out by using Mini Analysis Program (version 5.2; Synaptosoft, Leonia, NJ). The average amplitude and frequency of mEPSCs was first determined for each recording and then averaged together according to the experimental group. Likewise, for the ensemble amplitude cumulative distribution plots, distribution plots were first determined for each recording and then averaged together according to the experimental group. Only events greater than 7.5 pA were included in the analysis, which corresponds to the event detection limit used for the noisiest recording analyzed.

NMDAR currents were fitted in one of two ways. When mixed AMPA/NMDA responses were elicited (either synaptically or by 2P-uncaging, see below), the NMDAR-dependent portion of the traces at + 40 mV (i.e., the portion starting at 3 x _AMPA msec after peak of AMPA response determined at -70 mV onward to baseline) was fitted to a monoexponential decay by using Clampfit 8.2 (Axon Instruments). When isolated NMDAR synaptic currents were studied (in 10 mM glycine, 20 mM NBQX, 0.1 mM Mg²⁺, at -60 mV), their decay was best fitted by a double exponential equation:

I
(t) = I_f ´ exp(

_f) + I_s ´ exp(

_s),

where I_f and I_s are the amplitudes of the fast and slow decay components, and _f and _s are their respective decay time constants. The weighted decay time constant (_w) reported was calculated by:

_w = [I_f/(I_f + I_s)] ´ t_f + [I_s/(I_f + I_s)] ´ t_s.

Statistical significance was assessed by using the Student's unpaired t test. For all experimental series reported herein, ≈75% or more of the experiments and data analysis were carried out by an experimenter blind to the genotype. Moreover, great care was taken throughout to assure closely aged-matched groups (less than 1.5 days difference for all groups).

Two-Photon Uncaging of MNI-Glutamate
. Slices were placed on the stage of a Carl Zeiss LSM 510 / NLO system (Carl Zeiss MicroImaging Group, Thornwood, NY). Whole-cell recordings were obtained following standard procedures except that the intracellular recording solution was supplemented with 40 mM Alexa Fluor 594 (Molecular Probes, Eugene, OR) to outline neuronal morphology. Cells were allowed to fill at least 20 min following whole-cell access in normal Ringer. Slices were then switched to a Ringer supplemented with MNI-Glutamate (5 mM; Tocris, Ellisville, MO). Neurons were imaged by confocal imaging and/or by 2-photon imaging at 543 nm (HeNe1 laser) and 830 nm (Chameleon mode-locked Ti-Saphire laser; Coherent, Santa Clara, CA), respectively. Preliminary experiments showed that neuronal morphology was better represented at the 543-nm line by using Alexa Fluor 594 rather than Alexa Fluor 555. Clearly visible isolated spines located on secondary or tertiary apical dendrites less than ca. 75 mm from the cell body and extending in parallel to the plane of imaging were chosen for uncaging. The uncaging spot was determined from a confocal image and the spot was illuminated at 720 nm for 0.8 msec while monitoring the underlying current (2P-EPSC) in voltage-clamp mode. On the day of experiments, 543 nm and 720 nm laser lines were aligned by using a pollen grain fluorescent sample to assure complete superimposition of the uncaging spot with the confocal image. For determining AMPA/NMDA ratio of 2P-EPSC, laser pulses at 720 nm were delivered at 0.1 Hz. The AMPAR component of the 2P-EPSC was determined at -70 mV whereas the NMDAR component was determined at +40 mV. Because responses at +40 mV are typically more noisy than at -70 mV, ≈15-25 traces at +40mV were averaged for ≈5-10 traces at -70 mV. For this same reason, it was somewhat difficult to ascertain with great precision during the experiments that the NMDAR responses elicited were actually of 5-15 pA (as outlined in Results). Therefore, three different laser powers were typically delivered per spine and the appropriate laser power was determined post hoc during analysis. We could reliably obtain stable 2P-EPSCs from a single spine for >5-6 min (e.g., Fig. 8B), which typically was the time needed to carry out the experiment on a single spine. The uncaging spot was sometimes repositioned during the course of an experiment to compensate for drifting. The laser power used in the present study amounted to 10-25 mW measured at the back aperture of the objective. Image Acquisition and Spine Morphology Analysis.
After photolysis experiments, the Ti-Saphire laser was tuned to 830 nm and a z-stack images encompassing the segment of dendrite from which we uncaged was acquired. A number of 2-photon images (i.e., acquired at 830 nm) were also acquired from neurons/dentrites/spines that were not subjected to photolysis experiments or acquired before the actual photolysis experiments. Results from images obtained in these three conditions were pooled for spine analysis. Spine Head Volume Quantification.
Spine head volume estimates were derived from fluorescence images in a manner analogous to that previously described (2). All calculations were derived from 32-bit summed z-stacks images. Our measurements of spine volume, outlined below, relied on three assumptions: (i) mature spine heads are spherical, (ii) dye evenly fills spines, and (iii) the total fluorescence from a given structure is proportional to its volume. To minimize the impact of any factor potentially influencing the fluorescence intensity of our sample (e.g., unequal dye loading between experiments, unequal light scattering by tissue, etc) we estimated spine volume in the following manner:

First, for our imaging conditions, we predicted the fluorescent profile, on the x axis, of a sphere of true radius R, by using the following equation (solved by using Mathematica 5.1 software; Wolfram Research, Inc., Champaign, IL USA):

(1)

Values for s x and s z were determined by imaging subresolution fluorescent beads (0.1 mm; Molecular Probes) and found to be 0.21 and 0.99 mm, respectively.

We then determined the FWHM value of the fluorescent profiles F(r,R), obtained for different R values, predicted from equation (1). For R values between 0.3 mm and 0.6 mm, R was linearly correlated with FWHM (r² = 0.99: Fig12), and this function was expressed by the equation:

(2) R = A ´ [1/2] FWHM + b.

In principle, we could thus predict the true radius (R) of a sphere (for 0.3 < R <0.6 mM) from the FWHM value of its intensity profile. We have verified this contention by imaging fluorescent beads of radius 0.5 mm (Molecular Probes) and determined the FWHM value of their intensity profile. From equation (2), we calculated their R value to be 0.49 ± 0.01 mm (n = 7). As such, this suggest that for a sphere (or a spine), of 0.3 < R <0.6 mm, we can adequately predict its R by measuring FWHM of its intensity profile. We can then determine the volume of spines by using the equation:

(3) V = (4/3) p R³.

Because not all spine volumes are within the linear range of equation (2), we further relied on the assumption that the volume of a given structure is proportional to its total fluorescence, such that:

(4) V = K_f ´ F_tot

where K_f is a conversion factor. Because this conversion factor K_f is likely to vary between experiments (e.g., because of unequal dye filling, different depth in tissue, etc), we determined the K_f for every field of view analyzed. To this end, we measured the FWHM value of at least five mature spines per field that were chosen based on their large size such that their R values fell within the linear range of equation (2; Fig. 12). From equation (2), the volume of these spines was determined. By further measuring their F_tot, we could deduce a K_f for each of those five spines by:

(5) K_f = V/F_tot

The K_f for that particular field of view was determined by averaging the K_f obtained for those individual five spines. Once K_f was determined for a particular field of view, we measured the F_tot of all clearly visible spine structure and determined their volume by using equation (4).

Spine Length Quantification.
Spine length was also derived from the 32-bit image summed from a z-stack. A line from the base of the dendrite to the tip of spine head was drawn and the length of the line was measured.

1. Kim JH, Lee HK, Takamiya K, Huganir RL (2003) J Neurosci 23:1119-1124.

2. Matsuzaki M, Honkura N, Ellis-Davies GC, Kasai H (2004) Nature 429:761-766.