Altered Relationship Between Parvalbumin and Perineuronal Nets in an Autism Model

Abstract

Altered function or presence of inhibitory neurons is documented in autism spectrum disorders (ASD), but the mechanism underlying this alternation is poorly understood. One major subtype of inhibitory neurons altered is the parvalbumin (PV)-containing neurons with reduced density and intensity in ASD patients and model mice. A subpopulation of PV⁺ neurons expresses perineuronal nets (PNN). To better understand whether the relationship between PV and PNN is altered in ASD, we measured quantitatively the intensities of PV and PNN in single PV⁺ neurons in the prelimbic prefrontal cortex (PrL-PFC) of a valproic acid (VPA) model of ASD at different ages. We found a decreased PV intensity but increased PNN intensity in VPA mice. The relationship between PV and PNN intensities is altered in VPA mice, likely due to an "abnormal" subpopulation of neurons with an altered PV-PNN relationship. Furthermore, reducing PNN level using in vivo injection of chondroitinase ABC corrects the PV expression in adult VPA mice. We suggest that the interaction between PV and PNN is disrupted in PV⁺ neurons in VPA mice which may contribute to the pathology in ASD.

Keywords: autism spectrum disorder; chondroitinase ABC; parvalbumin; perineuronal net; valproic acid.

Figure 1

Density of parvalbumin (PV)⁺ neurons and Wisteria floribunda agglutinin (WFA)⁽⁺⁾ perineuronal nets (PNN) in prelimbic prefrontal cortex (PrL-PFC) across development in saline and valproic acid (VPA) mice. (A–F) Representative images of PV (green) and PNN (red) staining in the PrL-PFC of saline and VPA mice. Scale bar, 100 μm. (G) The density of PV⁺ neurons (left) and PNN (right) in PrL-PFC. (H) Percentage of PV⁺ neurons ensheathed by PNN (left) and PNN wrapped PV⁺ neurons (right). Blue, green, black, and red points in panel (G) represent the density of PV⁺ neurons in the saline group and VPA group, the density of PNN in the saline group and VPA group, respectively. Blue, green, black, and red circles in panel (H) represent the percentage of PV⁺ neurons ensheathed by PNN in the saline group and VPA group, PNN wrapped PV⁺ neurons in the saline group and VPA group, respectively. Black lines indicate Mean ± SEM. n = 5 mice for each group. Linear regression models were applied to compared various outcomes between groups using mice as the statistical units. *P < 0.05, **P < 0.01.

Figure 2

Comparisons of PV and PNN intensities in the PrL-PFC between saline and VPA mice. (A) Kernel density estimation of PV intensity in saline and VPA mice for different ages. Insert: Beeswarm plots of PV intensity. (B) Kernel density estimation of PNN intensity in saline and VPA mice for different ages. Insert: Beeswarm plots of PNN intensity. Yellow lines in inserts indicate geometric means of PV and PNN intensities. Kolmogorov–Smirnov tests were employed to assess disparities in the distribution of PV and PNN intensities between saline and VPA groups using cells as the statistical units. Linear mixed models were applied to compare the means of logarithm transformations of PV intensity [log(PV)] and PNN intensity [log(PNN)] between groups using cells as the statistical units. N = 287, 366, and 351 cells in P22, P35, and adult saline groups, respectively. N = 242, 305, and 348 cells in P22, P35, and adult VPA groups, respectively. *P < 0.05, **P < 0.01.

Figure 3

Relationship between PV and PNN intensities for a given neuron at different ages in saline and VPA mice. Logarithmic scales were used to present the values of PV and PNN intensities. Lines represent the relationship between the logarithm transformations of PV intensity [log(PV)] and PNN intensity [log(PNN)]: log(PV) = a×log(PNN) + b. (A) At P22, a = 0.37, b = 2.92 in saline group; a = 0.29, b = 3.02 in VPA group. (B) At P35, a = 0.22, b = 3.36 in saline group; a = 0.13, b = 3.45 in VPA group. (C) In adult, a = 0.13, b = 3.78 in saline group; a = 0.05, b = 3.88 in VPA group. Linear mixed models were applied to explore the correlation between log(PV) and log(PNN) using cells as the statistical units. N = 287, 366, and 351 cells in P22, P35, and adult saline groups, respectively. N = 242, 305, and 348 cells in P22, P35, and adult VPA groups, respectively.

Figure 4

Different assumptions of a distinct subpopulation in VPA adults. (A1,B1,C1,D1) Relationships between PV and PNN intensities for two subpopulations in adult VPA mice. Dark green and orange dots are “abnormal” and “normal” subpopulations. Dark green and orange lines represent the relationship between the logarithm transformations of PV intensity [log(PV)] and PNN intensity [log(PNN)]: log(PV) = a×log(PNN) + b. For (A1), “normal” subpopulation: a = 0.25, b = 0.04; “abnormal” subpopulation: a = −0.27; b = 0.21. For (B1), “normal” subpopulation: a = 0.24, b = 0.04; “abnormal” subpopulation: a = −0.19; b = 0.06. For (C1), “normal” subpopulation: a = 0.18, b = 0.04; “abnormal” subpopulation: a = −0.17; b = 0.10. For (D1), “normal” subpopulation: a = 0.15, b = 0.04; “abnormal” subpopulation: a = −0.20; b = 0.07. (A2,B2,C2,D2) Boxplots of fractions of “abnormal” subpopulation in adult mice for saline and VPA groups. Thick lines in boxes indicate medians of fractions. The lower and upper bounds of boxes represent the 1st (Q1) and 3rd quartiles (Q3) of fractions. IQR = Q3 − Q1. Thin lines located outside boxes are the minimum fractions and the smaller of the maximum fractions and Q3+1.5 × IQR. Circles were used to indicate values outside the ranges between Q1–1.5 × IQR and Q3 + 1.5 × IQR. Fractions of “abnormal” subpopulation in adult mice of VPA and saline groups were compared using the Mann–Whitney U test. *P < 0.05, ***P < 0.001.

Figure 5

Restored PV expression in ChABC-treated VPA adult mice. (A) Beeswarm plots of PV intensity from all neurons analyzed in saline and VPA mice treated by Veh and ChABC, respectively. (B) Kernel density estimation of PV intensity for all neurons analyzed in saline and VPA mice treated by Veh and ChABC, respectively. Yellow lines in panel (A) indicate geometric means of PV intensity. Linear mixed models were applied to compare the means of logarithm transformations of PV intensity [log(PV)] between groups using cells as the statistical units. Kolmogorov–Smirnov tests were employed to assess disparities in the distribution of PV intensity between groups using cells as the statistical units. N = 512, 448, 381, and 411 cells in Saline-vehicle, Saline-chABC, VPA-vehicle, and VPA-chABC groups, respectively. *P < 0.05.

Figure 6

Altered intensities of PV and PNN in VPA mice during postnatal development. (A) PV and PNN intensities in PrL-PFC of P22, P35, and adult saline and VPA mice. (B) Kernel density estimation of the PV and PNN intensities in PrL-PFC of P22, P35, and adult mice. Points in panel (A) represent observed PV and PNN intensities, while rectangles indicate geometric means. Linear mixed models were applied to compare the means of logarithm transformations of PV intensity [log(PV)] and PNN intensity [log(PNN)] between groups using cells as the statistical units. Kolmogorov-Smirnov tests were employed to assess disparities in the distribution of PV intensity between groups using cells as the statistical units. N = 287, 366, and 351 cells in P22, P35, and adult saline groups, respectively. N = 242, 305, and 348 cells in P22, P35, and adult VPA groups, respectively. **P < 0.01, ***P < 0.001. ^$statistical significance between P22 and P25; ^#significance between P35 and adult. P-values in Supplementary Table 3.