Identification of Parvalbumin Interneurons as Cellular Substrate of Fear Memory Persistence

Abstract

Parvalbumin-positive (PV) basket cells provide perisomatic inhibition in the cortex and hippocampus and control generation of memory-related network activity patterns, such as sharp wave ripples (SPW-R). Deterioration of this class of fast-spiking interneurons has been observed in neuropsychiatric disorders and evidence from animal models suggests their involvement in the acquisition and extinction of fear memories. Here, we used mice with neuron type-targeted expression of the presynaptic gain-of-function glycine receptor RNA variant GlyR α3L(185L)to genetically enhance the network activity of PV interneurons. These mice showed reduced extinction of contextual fear memory but normal auditory cued fear memory. They furthermore displayed increase of SPW-R activity in area CA3 and CA1 and facilitated propagation of this particular network activity pattern, as determined in ventral hippocampal slice preparations. Individual freezing levels during extinction and SPW-R propagation were correlated across genotypes. The same was true for parvalbumin immunoreactivity in the ventral hippocampus, which was generally augmented in the GlyR mutant mice and correlated with individual freezing levels. Together, these results identify PV interneurons as critical cellular substrate of fear memory persistence and associated SPW-R activity in the hippocampus. Our findings may be relevant for the identification and characterization of physiological correlates for posttraumatic stress and anxiety disorders.

Keywords: contextual fear memory; hippocampus; interneurons; memory extinction; parvalbumin; sharp wave ripples.

Figure 1.

Fear extinction is impaired in mice with enhanced functionality of PV interneurons (Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}). (A) Sketch of the fear conditioning protocol. Mice received 3 consecutive daily training and 5 consecutive daily extinction sessions. Memory performance was analyzed in 2 min bins. (B) A hot plate test confirmed an equal pain threshold in all 3 genotypes, as measured by latency to lick hind paws (black bar: Hprt^{α3L185L +/0} (control), gray bar: Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}, red bar: Hprt^{α3L185L +/0}; Camk2a^{Cre +/−}). (C) Baseline exploratory activity before first shock presentation. Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice were more active than mice in the other 2 groups (black bar: Hprt^{α3L185L +/0} (control), gray bar: Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}, red bar: Hprt^{α3L185L +/0}; Camk2a^{Cre +/−}). (D) Freezing duration in percent time of the first 2 min of each session is displayed as a measure of long-term fear memory. A mild reduction in freezing was observed in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice at training day T2, but genotypes performed equally upon further training. During extinction Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice showed consistently higher freezing levels than Hprt^{α3L185L +/0} (control) and Hprt^{α3L185L +/0}; Camk2a^{Cre +/−} mice. (E) Number of freezing bouts in the first 2 min of each session is displayed. Similar to the freezing duration, the number of freezing bouts was reduced in the Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} group at T2. Moreover, bouts were increased at R/E1 and remained high throughout extinction. Data represent mean ± SEM. T1–T3 = training days 1–3, R = retrieval, E1–5 = extinction days 1–5. Significant difference between Hprt^{α3L185L +/0} and Hprt^{α3L185L +/0}; Pvalb^Cre+/− mice is indicated with asterisks. Significant difference between Hprt^{α3L185L +/0}; Camk2a^{Cre +/−} and Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice is indicated using #, ##, or ### symbols. (*,#: P < 0.05; **,##: P < 0.01; ***, ###: P < 0.001).

Figure 2.

Altered within-session fear extinction in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. For a detailed analysis of freezing behavior within extinction sessions, each five 2-min intervals (i1–i5) were analyzed. Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice showed high initial freezing levels and significant decline during each session: E1–-Retrieval (A), E2–Extinction Day 2 (B), E3–Extinction Day 3 (C), E4–Extinction Day 4 (D), E5–Extinction Day 5 (E). Hprt^α3L185L+/0 and Hprt^{α3L185L +/0}; Camk2a^{Cre +/−} mice showed within-session extinction only during E1 and displayed constantly low freezing levels thereafter. (black label: Hprt^{α3L185L +/0} (control), gray label: Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}, red label: Hprt^{α3L185L +/0}; Camk2a^{Cre +/−}). Data represent mean ± SEM and are reported as percent time of the investigated 2-min time interval. Asterisks (*) indicate significant differences to the first interval within Hprt^{α3L185L +/0} genotype. The “&” symbol indicates significant difference to the first interval within Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} genotype, P < 0.05; &&, P < 0.01; &&&, P < 0.001. “#” indicates significant differences to the first interval within Hprt^{α3L185L +/0}; Camk2a^{Cre +/−} genotype, P < 0.05.

Figure 3.

Auditory cued fear conditioning and extinction are unaltered in mice with enhanced functionality of PV interneurons (Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}). (A) Sketch of the conditioning protocol. Mice received a total of 4 adaptation sessions to the background context. Conditioning to a tone was performed in 3 tone–shock pairings, followed by extinction of the tone fear memory in a neutral context on 5 consecutive days. Finally, renewal (context-dependency of an extinguished fear response) was tested by presenting the tone in the shock context. (B) Both genotypes displayed comparable post training freezing, auditory fear memory extinction and fear renewal in the original shock context. (C) Likewise, genotypes did not differ in pretraining shock context or pre-extinction neutral context freezing. Contextual freezing levels in (B) and (C) are reported in percent time of 2 min, cued freezing in percent time of the first 4 10-s stimulus presentations of each session (distributed over 2 min). Data are mean ± SEM. R = retrieval, E1–E5 = extinction days 1–5, RN = renewal, PT = post training, A = adaptation. &&& indicate a significant main effect for extinction sessions P < 0.001.

Figure 4.

Incidence of SPW-R is increased in mice with enhanced functionality of PV interneurons (Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}). (A) Example traces of SPW-Rs recorded in pyramidal cell layer of CA3 in horizontal hippocampal slice preparations. (B) Representative traces of SPW-R (top trace), ripples (middle trace), and SPW (bottom trace) of each genotype. (C–G) Summary graphs depicting differences in SPW-R properties in the CA3 and CA1 regions of Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice compared with control animals (Hprt^{α3L185L +/0}). Quantitative data on SPW occurrence (C), area under curve of SPWs (D), number of ripples per SPW (E), amplitude of ripple oscillations (F), and ripple frequency (G) are shown. The data in the graphs are shown as box plots. The median value is the horizontal line. The boundary of the box closest to zero indicates the 25th percentile and the boundary of the box farthest from zero indicates the 75th percentile. Whiskers (error bars) above and below the box indicate the 90th and 10th percentiles. Data points (single dots) beyond the 5th and 95th percentiles are also displayed. Student's t-test or Mann–Whitney U test was used to compare genotype effect in each region. Significant differences between genotypes are indicated with asterisks (*P < 0.05; **P < 0.01; ***P < 0.001).

Figure 5.

CA3–CA1 network interaction during SPW-R activity is facilitated in mice with enhanced functionality of PV interneurons (Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}). (A) Example traces of low-pass filtered (3–45 Hz) SPW-Rs recorded simultaneously in pyramidal cell layer of areas CA3 (top) and CA1 (bottom) in horizontal hippocampal slice preparations. (B) The representative example of cross-correlation of SPWs indicates increased correlation and decreased latency in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. The dashed lines indicate peak and latency. (C) Summary graph indicating a significant increase in CA3–CA1 correlation of SPWs in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice (n = 15 slices, N = 4 mice) compared with Hprt^{α3L185L +/0} control mice (n = 19 slices, N = 5 mice). (D) Mean values of CA3–CA1 correlation of SPWs (y axis) plotted against mean values of signal-to-noise ratio (SNR, x axis) for both genotypes. Note that Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice exhibited a significant increase of SNR in CA3 as well as CA3–CA1 correlation of SPWs. (E) Individual values of CA3–CA1 correlation (y axis) and SNR (x axis) were plotted against each other. Note that a high CA3–CA1 correlation correlates with an increased signal-to-noise ratio in CA3 of the hippocampal slices. (F) CA3–CA1 amplitude correlation of SPWs were increased in Hprt^{α3L185L +/0}; Pvalb^Cre+/− mice. (G) CA3–CA1 SPW latency was decreased in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. (H) The ratio of failures of SPW propagation from CA3 to CA1 was decreased in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. (I) Mean values of CA3–CA1 correlation of SPWs (y axis) were plotted against the mean values of CA3–CA1 event failures (x axis) for each genotype. Note that slices with increased correlation tended to have decreased propagation failures. The data in the graphs (C, F, G, and H) are shown as box plots. The horizontal line in the middle is the median value. The boundary of the box closest to zero indicates the 25th percentile and the boundary of the box farthest from zero indicates the 75th percentile. Whiskers (error bars) above and below the box indicate the 90th and 10th percentiles. Data points (single dots) beyond the 5th and 95th percentiles are also displayed. Student's t-test or Mann–Whitney U test was used to compare genotype effect. Pearson Product Moment Correlation was used to assess the correlation between 2 parameters. Significant differences between Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} and Hprt^{α3L185L +/0} control mice are indicated with asterisks (*P < 0.05, **P < 0.01, ***P < 0.001).

Figure 6.

The strength of CA3–CA1 network interaction during SPW-R activity is correlated with fear behavior at the second extinction session, E2. (A) Example traces of low-pass filtered (3–45 Hz) SPW-Rs recorded simultaneously in pyramidal cell layer of areas CA3 (top) and CA1 (bottom) from control mice (Hprt^{α3L185L +/0}) and Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. Note the increased SPW propagation in mice with enhanced functionality of PV interneurons. (B) Freezing duration (percent time of 2 min) at E2 was increased in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} compared with the control mice (Hprt^{α3L185L +/0}) (C) Similar to naïve mice, CA3-CA1 propagation failures of SPWs were decreased in mice with enhanced functionality of PV interneurons (Hprt^{α3L185L +/0}; Pvalb^{Cre +/−}). (D) CA3–CA1 correlation of SPWs, and (E) signal-to-noise ratio (SNR) in area CA3 were increased in Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice. Significant differences between Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} and Hprt^{α3L185L +/0} control mice are indicated with asterisks (*P < 0.05, **P < 0.01).

Figure 7.

Correlation of parvalbumin signals in the CA3b region of the ventral hippocampus with fear behavior at the second extinction session, E2. (A) Representative image shows maximum intensity projections of z-stacks with parvalbumin signals in CA3b regions of the ventral hippocampus. Scale bars: 10 µm. Identified PV interneurons are indicated by asterisks. Numbering 1 to 4 of boxed regions and the insets right-hand at higher magnification show representative PV neurons classified as (from top to bottom) “high”, “intermediate high”, “intermediate low”, and “low”. (B) Freezing duration (percent time of 2 min) at E2 of mice processed for parvalbumin immunohistochemistry was significantly different between control (black bar) and Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice (gray bar). (C,D) Cumulative plots of the relative parvalbumin signal intensities from control animals and Hprt^{α3L185L +/0}; Pvalb^{Cre +/−} mice reveal significant (P < 0.001) differences in the ventral (C), not dorsal (D), hippocampus. Statistical analysis was performed using Kolmogorov–Smirnov test.