Reduction in parvalbumin expression not loss of the parvalbumin-expressing GABA interneuron subpopulation in genetic parvalbumin and shank mouse models of autism

Abstract

Background: A reduction of the number of parvalbumin (PV)-immunoreactive (PV(+)) GABAergic interneurons or a decrease in PV immunoreactivity was reported in several mouse models of autism spectrum disorders (ASD). This includes Shank mutant mice, with SHANK being one of the most important gene families mutated in human ASD. Similar findings were obtained in heterozygous (PV+/-) mice for the Pvalb gene, which display a robust ASD-like phenotype. Here, we addressed the question whether the observed reduction in PV immunoreactivity was the result of a decrease in PV expression levels and/or loss of the PV-expressing GABA interneuron subpopulation hereafter called "Pvalb neurons". The two alternatives have important implications as they likely result in opposing effects on the excitation/inhibition balance, with decreased PV expression resulting in enhanced inhibition, but loss of the Pvalb neuron subpopulation in reduced inhibition.

Methods: Stereology was used to determine the number of Pvalb neurons in ASD-associated brain regions including the medial prefrontal cortex, somatosensory cortex and striatum of PV-/-, PV+/-, Shank1-/- and Shank3B-/- mice. As a second marker for the identification of Pvalb neurons, we used Vicia Villosa Agglutinin (VVA), a lectin recognizing the specific extracellular matrix enwrapping Pvalb neurons. PV protein and Pvalb mRNA levels were determined quantitatively by Western blot analyses and qRT-PCR, respectively.

Results: Our analyses of total cell numbers in different brain regions indicated that the observed "reduction of PV(+) neurons" was in all cases, i.e., in PV+/-, Shank1-/- and Shank3B-/- mice, due to a reduction in Pvalb mRNA and PV protein, without any indication of neuronal cell decrease/loss of Pvalb neurons evidenced by the unaltered numbers of VVA(+) neurons.

Conclusions: Our findings suggest that the PV system might represent a convergent downstream endpoint for some forms of ASD, with the excitation/inhibition balance shifted towards enhanced inhibition due to the down-regulation of PV being a promising target for future pharmacological interventions. Testing whether approaches aimed at restoring normal PV protein expression levels and/or Pvalb neuron function might reverse ASD-relevant phenotypes in mice appears therefore warranted and may pave the way for novel therapeutic treatment strategies.

Fig. 1

Representative PV⁺ and VVA⁺ cells from PND25 mouse cortex. a, b Single channel acquisition of (a) PV⁺ (magenta) and (b) VVA⁺ (green) cells. (c, d) Merged images showing PV (magenta) and VVA (green) overlapping in the PND25 cortex of a WT mouse, in (d) additionally with DAPI (light blue) counterstaining. In PV⁺ cell bodies (magenta), where most of the cell including the nucleus evidenced by DAPI staining (blue) was within the thickness of the section, the PNN (green) surrounding the cell was clearly visible. a–d) Low magnification and e–h) High magnification images. Arrowheads indicate the PV⁺VVA⁺ double-positive neurons that are in focus. In the overlay image, the rim of the cells was lighting up in white, indicative for “co-localization” of PV and VVA. i–l) High magnification images of PV⁺ (magenta) and VVA⁺ (green) cells from PND25 PV-/- mouse cortex. The morphology of PNNs in PV-/- animals does not differ from the one of WT mice. Scale bar: 100 μm (a–d), 50 μm (e–l)

Fig. 2

PV immunofluorescence images from mPFC of a PV+/- (a), SSC of a Shank1-/- (b) and striatum of a Shank3B-/- (c) mouse in comparison to the same regions of a WT mouse. PV expression levels (≈ signal intensity) varied considerably between individual PV⁺ neurons. Note the generally weaker somatic staining in the mutant mice (right panels), also evident by the fainter staining of the PV-ir neuropil. The weaker staining results in a lower number of neurons considered as positive for PV as shown in Figs. 3, 4 & 6. Scale bars: 100 μm in a, c; 200 μm in b

Fig. 3

a Left: Stereological estimations of PV⁺ (light gray) and VVA⁺ (dark gray) cells in mPFC (upper row), SSC (middle row) and striatum (lower row) of PND25 WT, PV+/- and PV-/- male mice. Significant differences are observed in PV⁺ cells between WT, PV+/- and PV-/- animals (p-value <0.05). *Significant vs. WT mice. Asterisks represent *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, respectively. Middle: stereological estimation of double-labeled (PV⁺ VVA⁺) cells of PND25 WT and PV+/- mice (white bars); Right: percentage of PV⁺ cells surrounded by VVA (light gray) and VVA⁺ cells showing PV expression (dark gray) in WT and PV+/- mice. b Quantitative Western blot analysis of forebrain samples of P25 WT PV+/- and PV-/- mice. A representative Western blot (left) and the quantification of PV protein levels in WT and PV+/- forebrain are shown (right). No PV signal was detectable in PV-/- mice. The Ponceau Red-stained protein markers loaded on the same membrane as the brain extract samples are shown with their respective molecular mass on the left. Data are from three independent experiments and are shown as mean ± SEM. Results are expressed as a percentage of normalized PV levels measured in control (WT), defined as 100 %. GAPDH or calbindin D-28k (CB) signals served as loading controls and were used for the normalization of the PV signals. Both, CB and GAPDH expression levels were unchanged in PV+/- and Shank mutants compared to WT mice (data not shown). c qRT-PCR values from P25 PV+/- mice representing Pvalb mRNA levels were normalized to 18S mRNA levels and expressed as fold change compared to WT. Data from three independent experiments were pooled together and are shown as mean ± SD. In all graphs, asterisks indicate statistical significance vs. WT (p-value <0.05, p = 0.0003)

Fig. 4

a Left: stereological estimations of PV⁺ (light gray) and VVA⁺ (dark gray) cells in SSC of WT and Shank1-/- PND25 male mice. Middle: estimation of double-labeled cells in SSC of PND25 WT and Shank1-/- mice. Right: percentage of PV⁺ cells surrounded by VVA (light gray) and percentage of VVA⁺ cells displaying PV expression and thus identified as PV⁺ cells (dark grey) in WT and Shank1-/- mice. b Representative Western blot and quantification of PV protein levels in Shank1-/- mice. CB signals were used as loading controls and served for the normalization of the PV signals. Data are from three independent experiments and are shown as mean ± SEM. Results are expressed as percentage of normalized PV levels measured in control (WT) samples, defined as 100 %. c Quantitative RT-PCR analysis of forebrain samples of Shank1-/- mice. PV (Pvalb) mRNA levels were normalized to 18S mRNA levels and expressed as fold change. Data from three independent experiments were pooled together and are shown as mean ± SD. In all graphs, asterisks indicate statistical significance vs. WT (p-value <0.05, p = 0.0006)

Fig. 5

Representative PV immunofluorescence images from the striatum of a WT and a Shank1-/- mouse. No qualitative differences in the number and signal intensities of PV⁺ neurons were observed between genotypes. Scale bar: 200 μm

Fig. 6

a) Left: stereological estimations of PV⁺ (light gray) and VVA⁺ (dark gray) cells in the striatum of WT and Shank3B-/- PND25 male mice. Middle: estimation of double-labeled cells in SSC of PND25 WT and Shank3B-/- mice. Right: percentage of PV⁺ cells surrounded by VVA (light gray) and VVA⁺ cells with PV expression (dark grey) in WT and Shank3B-/- mice. b) Representative Western blot and quantification of PV protein levels in Shank3B-/- mice. CB was used as loading control for the normalization of the PV signal, since CB expression levels were unchanged in Shank3B-/- mice (data not shown). Data are from three independent experiments and are shown as mean ± SEM. Results are expressed as a percentage of normalized PV levels measured in control (WT), defined as 100 %. c) Quantitative RT-PCR analysis of forebrain samples of Shank3B-/- mice. PV (Pvalb) mRNA levels were normalized to 18S mRNA levels and expressed as fold change. Data from three independent experiments were pooled together and are shown as mean ± SD. In all graphs, asterisks indicate statistical significance vs. WT (p-value <0.05, p = 0.0002)

Fig. 7

Representative PV immunofluorescence images (a–d) and Western blot analyses (e) from cortex and hippocampus of WT (a, c) and Shank3B-/- (b, d) mice. No apparent differences in the number or signal intensity of PV⁺ neurons in both brain regions were evident between WT and Shank3B-/- mice. Scale bars in a and b: 100 μm; in c and d: 200 μm. e) Western blot signals from brain homogenates containing cortex and hippocampus showed relatively high signal variability independent of the genotype. For the normalization of the PV signal either the GAPDH Western blot signal (upper part) or the intensity of the Ponceau Red-stained membrane (PR) was used. Results from 3 WT and 3 Shank3B-/-mice are depicted. The positions of the molecular weight markers are indicated. Due to the large difference in signal intensities for GAPDH and PV, the membrane was cut and exposed individually