RT-qPCR was performed using the same strategy as described in Fig. 1A. A, Agarose gel electrophoresis of RT-qPCR products. Total RNA of E18 rat brain was used as template in RT-qPCR with specific primers. Products were resolved on 3% agarose gel and visualized by ethidium bromide staining. Bands were at the anticipated molecular weight. Due to potential difference in PCR efficiencies, band intensity did not faithfully indicate relative amount. B, DNA sequence analysis of RT-qPCR products. RT-qPCR products were purified and subcloned into pGEM-T easy vector (Promega, Madison, WI) for sequencing with T7 primer. Shown were partial DNA sequences of respective domains.

Dual fluorescence in situ hybridization (FISH) was performed to determine which cells in the cortex express NRG1 isoforms. A, Representative images of dual FISH of VGLUT1 (red) and NRG1 isoforms (aqua). Regions in white rectangles in the upper panels were enlarged in lower panels. Arrow heads indicate cells expressing NRG1 isoforms, but not VGLUT1. Arrows indicate cells expressing both NRG1 isoforms and VGLUT1. Bar, 20 μm. B, Percentage of VGLUT1 positive cells expressing NRG1 types I (5.47 ± 1.13 %, n=18), II (31.09 ± 3.58 %, n=18), or III (51.10 ± 2.71 %, n=14). C, Representative images of dual FISH of GAD67 (red) and NRG1 isoforms (aqua). Regions in white rectangles in the upper panels were enlarged in lower panels. Arrow heads indicate cells expressing NRG1 isoforms, but not GAD67. Arrows indicate cells expressing both NRG1 isoforms and GAD67. Bar, 20 μm. D, Percentage of GAD67 positive cells expressing NRG1 types I (2.15 ± 1.29 %, n=18), II (5.62 ± 1.83 %, n=18), or III (39.56 ± 3.91 %, n=14).. E, Distribution of NRG1 isoforms in VGLUT1 (I, 46.50 ± 8.35 %, n=18; II, 82.47 ± 2.78 %, n=18; III, 58.61 ± 1.87 %, n=18) and GAD67 (I, 10.19 ± 6.20 %, n=18; II, 3.63 ± 1.23 %, n=18; III, 15.58 ± 1.64 %, n=14) positive neurons.

Total RNAs were isolated from control or KCl (50 mM, 6 hr)-treated cortical neurons (A–F, indicated ages) and astrocytes (G, H), and subjected to RT-qPCR. NRG1 types were first normalized by β-actin whose levels were revealed in same reactions and then normalized by control. **p<0.01

A, DNA sequences of IVwt-Luc and IV-SNP. The C-to-T mutation was indicated by red rectangles. The backbone of the Luc constructs were described in Fig. 7B. B, No effect of SNP8NRG243177 T allele on KCl-induced type IV NRG1 promoter activity. **p<0.01.

A, Diagram of NRG1 gene structure. Type-specific forward primers (indicated by arrows) for each type were located in unique exons whereas reverse primers were located in either Ig or EGF domain. Forward primers were also designed for EGF domain. B, Agarose gel electrophoresis of RT-qPCR products. Total RNA of human cerebral cortex was used as template in RT-qPCR with specific primers. Products were resolved on 3% agarose gel and visualized by ethidium bromide staining. Bands were at the anticipated molecular weight. Due to potential difference in PCR efficiencies, band intensity did not faithfully indicate relative amount. C, Different levels of NRG1 types in the cerebral cortex of a 13-year-old subject. D, Composition of NRG1 types in the cerebral cortex of a 13-year-old subject. E, Different levels of NRG1 types in the cerebral cortex of a 60-year-old subject. The copy numbers per micro gram total RNA of each type NRG1 were 1.67 ± 0.17 × 10⁴ (type I), 8.31 ± 1.17 × 10⁴ (type II), 2.93 ± 0.23 × 10⁵ (type III), 1.31 ± 0.06 × 10³ (type IV), 5.12 ± 0.73 × 10³ (type V), and 227 ± 226 (type VI). F, Composition of NRG1 types in the cerebral cortex of a 60-year-old subject.

RT-qPCR was performed as described in Fig. 1A with the RNA templates of cerebral cortex of different age. A, Different expression levels of NRG1 types in E18 cerebral cortex. The copy numbers per μg total RNA of each type NRG1 were 6.81 ± 0.09 × 10⁵ (type I), 1.28 ± 0.06 × 10⁶ (type II), 7.81± 0.17 × 10⁵ (type III), 557 ± 127 (type IV), 3890 ± 850 (type V), and 31 ± 14 (type VI), whereas total NRG1 were 2.75 ± 0.08 × 10⁶. B, Composition of NRG1 types in E18 cerebral cortex. C, Different expression levels of NRG1 types in P5 cerebral cortex. The copy number per μg total RNA of each type NRG1 were 3.22 ± 0.10 × 10⁶ (type I), 9.16 ± 0.45 × 10⁶ (type II), 6.03 ± 0.40 × 10⁶ (type III), 1.43 ± 0.12 × 10³ (type IV), 2.79 ± 0.50 × 10⁶ (type V) and 1.06 ± 0.63 × 10³ (type VI) whereas total NRG1 were 2.12 ± 0.12 × 10⁷. D, Composition of NRG1 types in P5 cerebral cortex. E, Different expression levels of NRG1 types in P15 cerebral cortex. The copy number per μg total RNA of each type NRG1 were 4.51 ± 0.75 × 10⁴ (type I), 6.25 ± 0.33 × 10⁵ (type II), 2.35 ± 0.06 × 10⁵ (type III), 0 (undetectable, type IV), 1.02 ± 0.08 × 10⁴ (type V) and 3.26 ± 2.32 × 10² type VI whereas total NRG1 were 9.16 ± 0.27 × 10⁷. F, Composition of NRG1 types in P15 cerebral cortex. G, Different expression levels of NRG1 types in P70 cerebral cortex. The copy number per μg total RNA of each type NRG1 were 1.46 ± 0.21 × 10⁴ (type I) 1.46 ± 0.28 × 10⁵ (type II), 1.88 ± 0.25 × 10⁵ (type III), 0 (undetectable, type IV), 5.58 ±13.60 × 10² (type V) and 64 ± 13 (type IV) whereas total NRG1 were 3.54 ± 0.55 × 10⁷. H, Composition of NRG1 types in P70 cerebral cortex. I, Alteration of NRG1 types in developing cerebral cortex.

Rats were injected with 10 mg/ml KA to induce seizure and sacrificed 4 hr after injection. Total RNAs were subjected to RT-qPCR. NRG1 types were first normalized by β-actin whose levels were revealed in same reactions and then normalized by control. **p<0.01

A, Sequence comparison of type I 5′UTRs of human, rat, and mouse. Distal and proximal CRE sites were highlighted in red rectangles. * indicate conserved nucleotides. Underlined sequence included 20 bp upstream and 15 bp downstream of CRE sites were used for EMSA. B, Diagram of luciferase report constructs. C, The proximal CRE cis-element was necessary for activity-dependent increase of type IV expression. SH-SY5Y cells were co-transfected with indicated reporters and pRL-TK. 12 hr after transfection, cells were treated with vehicle or 50 mM KCl for 12 hr. Luciferase activity was measured as described in Methods. **p<0.01. D, Interaction between proximal CRE cis-element and CREB. Double stranded oligonucleotides containing the proximal CRE of human type IV promoter (shown in 6A) was 5′ end-labeled with IRDye 700 and incubated with recombinant CREB1 for 15 min at 30 °C. In some experiments, 200 X unlabeled oligonucleotides or anti-CREB antibody were included in the reaction. The reaction mixture was resolved on 4% nondenaturing polyarcrylamide gel. S: shift band; SS: super shift; NS: non-specific band.