Enriched expression of NF1 in inhibitory neurons in both mouse and human brain.

Neurofibromatosis type 1 (NF1) is an autosomal dominant disease caused by loss-of-function mutations in NF1 gene, which encodes a GTPase activating protein for RAS. NF1 affects multiple systems including brain and is highly associated with cognitive deficits such as learning difficulties and attention deficits. Previous studies have suggested that GABAergic inhibitory neuron is the cell type primarily responsible for the learning deficits in mouse models of NF1. However, it is not clear how NF1 mutations selectively affect inhibitory neurons in the central nervous system. In this study, we show that the expression level of Nf1 is significantly higher in inhibitory neurons than in excitatory neurons in mouse hippocampus and cortex by using in situ hybridization. Furthermore, we also found that NF1 is enriched in inhibitory neurons in the human cortex, confirming that the differential expressions of NF1 between two cell types are evolutionarily conserved. Our results suggest that the enriched expression of NF1 in inhibitory neurons may underlie inhibitory neuron-specific deficits in NF1.

Neurofibromatosis type 1 (NF1) is an autosomal dominant disorder caused by loss of function mutations in NF1 gene, which occurs in approximately 1 of 3000 births [1]. NF1 affects multiple organs, mainly skin, bone, and brain, and is diagnosed by café-au-lait spots, neurofibromas, optic glioma, Lisch nodules in iris, bone malformations [1-3]. NF1 is most abundantly expressed in the nervous system [4]. Subsequently, a wide range of cognitive deficits is associated with NF1, which include deficits in visuospatial perception, executive functioning, attention, social function and learning [5-7]. NF1 gene encodes neurofibromin (NF1) which is a GTPase-activating protein (GAP) for RAS [8-10]. Thus, loss of function mutations in NF1 gene cause increases in the activation of RAS and its downstream signaling cascades [11]. Studies using mouse models of NF1 have shown that the enhanced activation of RAS-extracellular signal-related kinase (ERK) signaling is responsible for the learning deficits in NF1 [11-14]. Nf1 heterozygous knockout mice showed deficits in spatial learning and working memory, which can be rescued by attenuating RAS activation [12, 14]. Interestingly, elegant studies by Silva and colleagues have shown that gamma-aminobutyric acidergic (GABAergic) inhibitory synaptic function is altered in both hippocampus and cortex of Nf1+/− mice [12, 13, 15]. To define the cell type responsible for the learning deficits in Nf1+/− mice, Cui and colleagues deleted Nf1 selectively in excitatory neurons, inhibitory neurons, or glia and found that deleting Nf1 only in inhibitory neurons can recapitulate behavioral and cellular phenotypes shown in Nf1+/− mice such as deficits in spatial learning and long-term synaptic plasticity [13]. Since NF1 was shown to be ubiquitously expressed in adult neurons, oligodendrocytes, and Schwann cells [4, 16], it is intriguing that deleting Nf1 selectively affect inhibitory neurons. Recently, we have shown that the genes in RAS-ERK signaling network are differentially expressed between excitatory and inhibitory neurons in mouse hippocampus by performing cell type-specific transcriptome analyses [17]. Interestingly, Nf1 expression was found to be higher in vesicular gamma-aminobutyric acid transporter (vGAT)-positive neurons than in alpha Ca2+/calmodulin-dependent kinase II (αCaMKII)-positive neurons in mouse hippocampus by using cell type-specific RNA-sequencing (RNA-seq) analysis [17], which suggest that inhibitory neuron-enriched expression of NF1 may underlie the cell type-specific pathophysiology of NF1.

To confirm the expression pattern of Nf1 in mouse brain (male C57Bl/6 J, 7–8 weeks) by using a different method, we performed fluorescent in situ hybridization. We used a gene-specific probe for mouse Nf1 together with probes for αCamkII and vGAT as markers for excitatory and inhibitory neurons, respectively. Consistent with the previous RNA-seq result [17], we found that the Nf1 expression level is significantly higher in inhibitory neurons than in excitatory neurons in the mouse hippocampus (Fig. 1a and b). The area of Nf1 mRNA particles in vGAT+ neurons were significantly larger than in αCamkII+ neurons in hippocampal CA1 region (Area of Nf1 particles: αCamkII+, 3.97 ± 0.16 μm2; vGAT+, 8.25 ± 1.24 μm2; unpaired t-test, ****p < 0.0001; Fig. 1a and b). Next, we examined the expression of Nf1 in mouse cortex (Fig. 1c and d). As in the hippocampus, total area of Nf1 mRNA particles were bigger in vGAT+ neurons than in αCamKII+ neurons in the parietal cortex (Area of Nf1 particles: αCamkII+, 3.21 ± 0.21 μm2; vGAT+, 6.1 ± 0.46 μm2; unpaired t-test, ****p < 0.0001; Fig. 1c and d). Thus, these results show that the Nf1 is enriched in vGAT+ inhibitory neurons in the mouse hippocampus and cortex, which are hubs of spatial learning and higher-level executive brain function. This inhibitory neuron-enriched expression of Nf1 might explain how inhibitory synaptic function is selectively affected in Nf1 mutant mice.

Fig. 1

In situ hybridization of Nf1 in mouse and human brain. a Representative merged image of triple fluorescent in situ hybridization probed for Nf1 (red), αCamkII (green) and vGAT (white) in hippocampal CA1 region. Higher-magnification images of the boxed area in (a) were also shown. White arrows indicate double-positive cells for Nf1 and vGAT. Sections were also stained with DAPI (blue). In situ hybridization was performed following the manufacturers’ manual (RNAscope Multiplex Fluorescent Reagent Kit, Advanced Cell Diagnostics) and the following probes (Advanced Cell Diagnostics) were used: mouse Nf1, catalog #417351; mouse Slc32a1-C2, #319191-C2; mouse CamK2-C3, #445231-C3. Images were acquired by using Axio scan Z1 (Zeiss) and analyzed by using ImageJ (NIH). Scale bar, 10 μm. b Average particle size in αCamkII+ neurons or vGAT+ neurons in CA1. Data were collected from 627 αCamkII+ cells and 76 vGAT+ cells in hippocampal CA1 area. c Representative merged image of triple fluorescent in situ hybridization probed for Nf1 (red), αCamkII (green) and vGAT (white) in the perietal cortex. Scale bar, 10 μm. d Average particle size in αCaMKII+ neurons or vGAT+ neurons in the cortex. Data were collected from 127 αCamkII+ cells and 74 vGAT+. Data is expressed as means ± SEM. Unpaired t-test, ****p < 0.0001. e, g Representative merged image of duplex chromogenic in situ hybridization probed for NF1 (blue) and αCaMKII (red) or NF1 (blue), vGAT (red) and hematoxylin for counter-staining (light-purple color) in human cortex [e, sample #20399, 3 years old female diagnosed with focal cortical dysplasia type I (temporal cortex); g sample #17490, 2 years old male diagnosed with focal cortical dysplasia type I (frontal cortex)]. Black arrows indicate co-stained cells for either NF1 and αCaMKII or NF1 and vGAT. Chromogenic detection methods according to the manufacture’s instruction (RNAscope duplex chromogenic detection Kit, Advanced Cell Diagnostics). Gene specific probes for human NF1, αCAMKII, and vGAT (human NF1, catalog # 419731; human SLC32A1-C2, #415681-C2; human CAMK2-C2, #521261-C2) were used. Images were acquired by using Aperio scan (Leica Biosystems) and analyzed by using ImageJ (NIH). Images were separated into 3 determined colors (red, green and blue) by ‘colour deconvolution’ plugin which transforms multiple-color images into separated single color channels. Scale bar, 10 μm. f and h Average NF1 particle size in αCaMKII+ neurons or vGAT+ neurons from #20399 or #17490. Data were collected from 99 αCaMKII+ cells and 92 vGAT+ cells in #20399; 142 αCaMKII+ cells and 98 vGAT+ cells in #17490. Data is expressed as means ± SEM. Unpaired t-test, *p < 0.01, ***p < 0.001

To verify that NF1 expression is also higher in inhibitory neurons than in excitatory neurons in human, we examined the NF1 mRNA expression in human cortex. Since the human tissues showed strong auto-fluorescent signals probably due to the fix condition, we used dual color chromogenic in situ hybridization system: NF1 was co-stained with either αCaMKII (also known as CaMK2A) or vGAT (also known as SL32A1). We used cortical biopsy samples from two human subjects who underwent surgery for focal cortical dysplasia type I. Normal cortical tissues around the affected area were used in this study. As previously reported [18], NF1 was detected in neurons in human brains (Fig. 1e and g). To examine whether NF1 is also enriched in inhibitory neurons in human, we analyzed the area of NF1 mRNA particle in vGAT+ or αCaMKII+ neurons. Consistent with our finding in mouse cortex, area of human NF1 particle in each cell was also significantly larger in vGAT+ neurons than in αCaMKII+ neurons in both samples (Area of NF1 particles: #20399, αCamkII+, 10.31 ± 0.4 μm2; vGAT+, 12.14 ± 0.74 μm2; unpaired t-test, *p < 0.0001; #17490, αCamkII+, 10.21 ± 0.48 μm2; vGAT+, 15.76 ± 0.98 μm2; unpaired t-test, *p < 0.0001; Fig. 1f and h). Collectively, we found that NF1 mRNA is enriched in vGAT+ neurons compared to αCaMKII+ neurons both in mouse and human brain. Importantly, hybridizations using a negative control probe targeting a bacterial gene dihydrodipicolinate reductase (DapB) did not show any nonspecific background signal in mouse and human cortex (Additional file 1: Figure S1). To further examine the specificity of the probes, we performed cross-species hybridization experiments in which we used the mouse Nf1 probe on human tissue and the human NF1 probe on mouse tissue. We detected almost no signals compared to those from species-matching conditions (Additional file 1: Figure S1).

Initially, Costa and colleagues found that monosynaptically evoked inhibitory postsynaptic potentials are significantly larger in Nf1+/− mice compared to those in wild type littermates, which was reversed by reducing Ras activity, showing that Nf1 deletion increases synaptic inhibition through Ras activation [12]. Later, Cui and colleagues showed that neurofibromin regulates ERK and synapsin phosphorylation in GABAergic neurons [13]. However, it is not clear how neurofibromin mainly regulates inhibitory synaptic functions. Recently, Omrani and colleagues showed that neurofibromin interacts with hyperpolarization-activated cyclic nucleotide-gated (HCN) channel which is enriched in parvalbumin (PV)-expressing interneurons [19]. HCN current is selectively reduced in PV-expressing interneurons, resulting in hyperexcitability in PV-expressing inhibitory neurons both in Nf1+/− and Nf1 mice in which the neuron-specific exon 9 is deleted, which may contribute to inhibitory neuron-specific phenotypes in NF1 mouse models [19].

In our previous study, we showed that major components in RAS-ERK signaling pathway including Nf1 are differentially expressed between excitatory and inhibitory neurons in mouse hippocampus, proposing that this cell type-specific distribution of signaling molecules may underlie cell type selective pathophysiology observed in NF1 and other Rasopathies such as Noonan syndrome [17]. The expression pattern of NF1 has been extensively studied both in rodents and human brains, which have revealed that NF1 is expressed in neurons, oligodendrocytes, and Schwann cells, and even astrocytes depending on the conditions and the isoforms [4, 16, 18, 20–22]. However, it was not clear in which neuronal type NF1 is enriched. Our results clearly demonstrate that NF1 is enriched in inhibitory neuron in mouse and human brain. Specific inhibitory neuronal types in which NF1 is enriched remains to be determined. Taken together, our results strongly suggest that the cell type-specific RAS-ERK signaling networks might be, at least for several molecules, evolutionarily conserved between mouse and human and therefore, the mechanisms for cognitive deficits revealed in NF1 mouse models may turn out to be also true in human NF1.

Figure S1. Representative image of in situ hybridization by using a negative control probe or cross-species probes in human and mouse cortex. (A-B) Human (A) or mouse (B) cortex samples were hybridized with a probe targeting DapB by using a fluorescent or chromogenic hybridization method, respectively. No DapB signal (blue in A, red in B) was detected. Slices were counter-stained with hematoxylin or DAPI. (C-D) Cross-species hybridization. (C) Human cortex sample was hybridized with the mouse Nf1 probe (blue). (D) Mouse cortex sample was hybridized with the human NF1 probe (blue). Slices were counter-stained with hematoxylin. Scale bars, 50 μm. (PDF 131 kb)