Altered inhibitory function in hippocampal CA2 contributes in social memory deficits in Alzheimer's mouse model

Abstract

Parvalbumin (PV)-expressing interneurons which are often associated with the specific extracellular matrix perineuronal net (PNN) play a critical role in the alteration of brain activity and memory performance in Alzheimer's disease (AD). The integrity of these neurons is crucial for normal functioning of the hippocampal subfield CA2, and hence, social memory formation. Here, we find that social memory deficits of mouse models of AD are associated with decreased presence of PNN around PV cells and long-term synaptic plasticity in area CA2. Furthermore, single local injection of the growth factor neuregulin-1 (NRG1) is sufficient to restore both PV/PNN levels and social memory performance of these mice. Thus, the PV/PNN disruption in area CA2 could play a causal role in social memory deficits of AD mice, and activating PV cell pro-maturation pathways may be sufficient to restore social memory.

Keywords: Behavioral neuroscience; Cellular neuroscience; Molecular neuroscience.

Graphical abstract

Figure 1

Reduced number of PV+ cells associated with PNN loss in the area CA2 of Tg2576 mice (A) Stainings for PV (green) and PCP4 (blue) in area CA2 of 9-month-old NTg and Tg2576 mice (left) and stereological estimation of PV+ cells in area CA2 (right) reveal reduced numbers of PV+ cells in Tg2576 (n = 7) compared to NTg (n = 7) mice. ∗: p < 0.05, unpaired t test between genotypes. (B) Stainings for WFA (PNN, magenta) and PCP4 (blue) in area CA2 and stereological estimation of absolute number indicate a decrease of PV+/PNN+ cells in area CA2 of Tg2576 (n = 7) compared to NTg (n = 7) mice. ∗∗∗: p < 0.005, unpaired t test between genotypes. (C) For assessment of PV intensity in area CA2, PV intensity of fluorescence of PV-/PCP4+ pyramidal neuron soma (dotted lines with ¤) is used as baseline for normalization of PV intensity of PV+/PCP4- cells (with ∗). (D) Normalized PV intensity in the soma of PV+ cells from SP of area CA2 shows lower PV signal in PV+ cells of Tg2576 mice (n = 89 cells from seven mice) compared to those of NTg mice (n = 102 cells from seven mice). Bars represent the median with interquartiles and each dot represents a PV+ cell. ∗∗∗: p < 0.005 between genotype, Mann-Whitney test. (A and B) Bar graphs represent the mean and error bars the SEM Scale bars = 50 μm. See also Figure S1.

Figure 2

Inhibitory transmission and PV-dependent long-term plasticity are altered in area CA2 of Tg2576 mice (A) Whole-cell recordings of CA2 pyramidal neurons during stimulation of CA3 axons reveal an increased amplitude of the postsynaptic potential (PSP), a composite of excitatory and inhibitory potentials, in slices prepared from Tg2576 mice (red) as compared to those of NTg mice (black). Average sample traces are shown on top. (B) The paired-pulse ratio (PPR) of two inhibitory postsynaptic currents (IPSC) recorded in CA2 neurons with CA3 axonal stimulation is increased in slices of Tg2576 as compared to NTg mice. (C) Averaged population spike (PS) amplitudes measured extracellularly in the pyramidal layer of area CA2 reveal that stimulation of CA3 inputs more readily results in action potential firing in Tg2576 than in NTg mice. Average sample traces are shown on top. (D) The time course of normalized IPSC amplitudes recorded in CA2 pyramidal neurons before and after high frequency stimulations indicates that iLTD can be triggered in slices of NTg mice but only a transient depression is observed in Tg2576 mice. (E) Averaged sample traces at the time indicated by number are shown on top. (E) The PPR of two IPSCs recorded in CA2 pyramidal neurons is increased following HFS (2) in samples prepared from NTg mice (black, left) (n = 6 recordings from four mice), but remains unchanged between before (1) and after (2) HFS in Tg2576 mice (red, right). (F) Amplitude of the field excitatory PSP in area CA2 shows a smaller increase following a high-frequency stimulation in Tg2576 compared to NTg mice. Average sample traces of fEPSP at the time indicated by number are shown on the right. See also Figure S2.

Figure 3

Social recognition and social memory are impaired in Tg2576 mice (A) Sociability test in the 3-chamber paradigm indicates that mice from both genotypes spend more time with a conspecific than an empty cage. ∗∗∗: p < 0.005 repeated-measures two-way ANOVA followed by Sidak’s post hoc test between cages. (B) Social recognition test reveals that NTg mice spend more time with a new mouse than with the familiar one, whereas Tg2576 mice spend an equal amount of time with both mice. ∗∗∗: p < 0.005 repeated-measures two-way ANOVA followed by Sidak’s post hoc test between cages. (C) Social memory assessed with the 5-trial test reveals impairment in Tg2576 mice. ∗: p < 0.05, ∗∗∗: p < 0.005 between trials for NTg and Tg2576 mice, repeated-measures two-way ANOVA followed by Sidak’s post hoc test. (D) Normalized measures show that the investigation time of NTg mice decreases significantly from trial two to 4. In contrast, no significant change in investigation time is observed across trials in Tg2576 mice. Repeated-measures one-way ANOVA followed by Sidak’s post hoc test. ∗∗p < 0.01 and ∗∗∗p < 0.005 between trial n and trial 1. Bar graphs represent the mean and error bars SEM See also Figure S3.

Figure 4

PNN disruption in area CA2 is sufficient to decrease PV presence and to abolish social memory capacity (A) Staining for WFA (PNN, magenta) 7 days after local injections of PBS or ChABC solution in area CA2 shows a large decrease of PNN presence in area CA2 (∗∗∗: p < 0.005 for unpaired t test with Bonferroni’s correction for multiple comparisons between treatment), while sparing CA1 and CA3 areas. (B) Parvalbumin intensity in PV+ cells from SP of area CA2 is reduced after ChABC injection. Bars represent the median with interquartiles and each dot represents a PV+ cell. ∗∗∗: p < 0.0001, Mann-Whitney test. (C) The total duration of investigation of an unknown mouse reveals alteration of social memory formation in WT mice following ChABC injection in area CA2. ∗: p < 0.05, ∗∗∗: p < 0.005 between trials for NTg and Tg2576 mice, RM two-way ANOVA followed by Sidak’s post hoc test. (D) Normalized data show a significant decrease of the investigation time of PBS mice from trial two to 4. In contrast, no significant change in investigation time is observed in ChABC mice. ∗p < 0.05, ∗∗∗p < 0.005 between trial n and trial 1, repeated-measures one-way ANOVA followed by Sidak’s post hoc test. (A, C, and D) Bar graphs represent the mean and error bars the SEM. Scale bar = 50 μm. See also Figure S4.

Figure 5

Local NRG1 restores PV+/PNN+ cells and social memory formation in Tg2576 mice (A) Quantification of PV+ cells 5 days after NRG1 injection in area CA2 reveals a restoration of the number of detectable PV+ cells in area CA2 of Tg2576 mice injected with NRG1 compared to Tg2576-PBS mice, abolishing the difference between genotypes. ∗: p < 0.05, ∗∗∗: p < 0.005, two-way ANOVA followed by Sidak’s post hoc test. (B) The increased number of PV+ cells following NRG1 injection was accompanied by a restored presence of PNN around PV cells in Tg2576 mice compared to NTg ones. ∗∗: p < 0.05, ∗∗∗: p < 0.005, ns: p > 0.05, two-way ANOVA followed by Sidak’s post hoc test. (C) PV intensity of the soma of PV+ cells from area CA2 SP shows that PV signal is higher in PV+ cells of Tg2576-NRG1 mice compared to those of Tg2576-PBS mice. Bar represents the median and interquartiles and each dot represents one PV+ cell. ∗: p < 0.05, ∗∗∗: p < 0.005, Kruskal-Wallis followed by Dunn’s post hoc test. (D) Social memory test indicates that local injections of NRG1 in area CA2 induce a restoration of social memory formation in Tg2576 mice. ∗∗∗: p < 0.005 repeated-measures two-way ANOVA. (E) Normalized data relative to mean interaction time during trial one shows a significant decrease in investigation time in Tg2576 mice injected with NRG1 from trial two to four similar of NTg groups. In contrast, no significant change in investigation time is observed in Tg2576 mice injected with PBS. ∗∗∗p < 0.005 between trial n and trial 1, repeated-measures one-way ANOVA followed by Sidak’s post hoc test. (F) Spatial memory was assessed with an object location test. (G) Preference index of the investigation time between displaced and fixed object during retention phase indicate that spatial memory is altered in Tg2576 mice, even following NRG1 injections. ∗∗: p < 0.01, ∗∗∗: p < 0.005 for one-sample t test against chance (50%). (A, B, E, and G) Bar graphs represent the mean and error bars of the SEM See also Figure S5.