The Psychiatric Risk Gene NT5C2 Regulates Adenosine Monophosphate-Activated Protein Kinase Signaling and Protein Translation in Human Neural Progenitor Cells

Background The 5′-nucleotidase, cytosolic II gene (NT5C2, cN-II) is associated with disorders characterized by psychiatric and psychomotor disturbances. Common psychiatric risk alleles at the NT5C2 locus reduce expression of this gene in the fetal and adult brain, but downstream biological risk mechanisms remain elusive. Methods Distribution of the NT5C2 protein in the human dorsolateral prefrontal cortex and cortical human neural progenitor cells (hNPCs) was determined using immunostaining, publicly available expression data, and reverse transcriptase quantitative polymerase chain reaction. Phosphorylation quantification of adenosine monophosphate-activated protein kinase (AMPK) alpha (Thr172) and ribosomal protein S6 (Ser235/Ser236) was performed using Western blotting to infer the degree of activation of AMPK signaling and the rate of protein translation. Knockdowns were induced in hNPCs and Drosophila melanogaster using RNA interference. Transcriptomic profiling of hNPCs was performed using microarrays, and motility behavior was assessed in flies using the climbing assay. Results Expression of NT5C2 was higher during neurodevelopment and was neuronally enriched in the adult human cortex. Knockdown in hNPCs affected AMPK signaling, a major nutrient-sensing mechanism involved in energy homeostasis, and protein translation. Transcriptional changes implicated in protein translation were observed in knockdown hNPCs, and expression changes to genes related to AMPK signaling and protein translation were confirmed using reverse transcriptase quantitative polymerase chain reaction. The knockdown in Drosophila was associated with drastic climbing impairment. Conclusions We provide an extensive neurobiological characterization of the psychiatric risk gene NT5C2, describing its previously unknown role in the regulation of AMPK signaling and protein translation in neural stem cells and its association with Drosophila melanogaster motility behavior.


Immunohistochemistry/fluorescence
To identify cell types associated with NT5C2 expression, human brain sections were immunolabelled with an antibody raised against the NT5C2 protein. Paraffinembedded sections (7 µm) were deparaffinized (2x 5 min xylene, 5 min IMS, 5 min 95% IMS, 70% IMS, water), and submitted to an antigen retrieval protocol (simmer cooking on microwave for 10 min in citrate buffer, pH 6). For immunohistochemistry using chromogenic labelling, the activity of endogenous peroxidases was blocked prior to chromogenic reaction. Sections were submitted to a perm/blocking step by incubation with normal rabbit serum 10% in phosphate buffer saline (PBS) containing 0.1% Triton-X for 2 hours. Next, sections were incubated with the primary antibody (1:200) overnight at 4ºC in normal rabbit serum 10% PBS, followed on the next day by chromogenic detection using the Vectastain ABC Horseradish Peroxidase Kit (Vector Labs, Peterborough, United Kingdom), according to the manufacturer's protocol.
For immunofluorescence staining, sections were deparaffinized, submitted to antigen retrieval protocol, autofluorescence removal protocol (40 min incubation in aqueous solution of 0.25% potassium permanganate, followed by brief incubation with 1% oxalic acid), blocked using 5% normal goat serum (NGS) PBS containing 0.1% Triton-X, and incubated overnight at 4ºC with primary antibodies in blocking solution. On the next day, 3x PBS washes (10 min) were performed, followed by incubation with secondary antibodies for 2 hours at room temperature, followed with 3x 10 min PBS washes. and GFAP was performed using a custom ImageJ macro (1). Total NT5C2 intensities (mean grey value) and NT5C2 intensities co-localized with neurons, interneurons, microglia, or astrocytes, were measured and calculated on z-projected image stacks using the immunolabelled-stained sections. Gaussian filter (0.5), background subtraction (50) and a threshold (Default) were applied to the z-projected images for each channel. The settings for NT5C2 were kept constant between the different colocalization experiments whilst the setting for MAP2, Parvalbumin, GFAP and IBA1 staining were adjusted for an optimal representation of neurons, interneurons, astrocytes and microglia, and then kept constant during image analyses. Colocalization was defined as pixel clusters in the NT5C2 channel that overlapped with pixel clusters in the MAP2, Parvalbumin, GFAP and IBA1 channels, with a set size cutoff at 0.05 μm 2 . Both total NT5C2 and co-localized NT5C2 intensity were averaged over the 20 images taken per technical replicate (20 fields of view), per patient (n = 4 unaffected controls).
HEK293T cells were routinely maintained in T75 Nunc flasks (Thermo Fisher Scientific) in Advanced DMEM media supplemented with 10% fetal bovine serum and 2 mM Lglutamine (Thermo Fisher Scientific), and passaged every 2-3 days using Accutase when confluency was achieved. All cell lines were kept at 37 o C, 5% CO2. transfected overnight with 6 µL Lipofectamine, 2.5 µg DNA in 500 µl OPTIMEM and 1.5 mL feeding media, and incubated in fresh medium for another day.

RNA and protein extraction
Total RNA and protein were extracted in parallel by detaching cells using Accutase, and splitting the obtained cell suspension into two tubes for harvesting. Tri-Reagent (Thermo Fisher Scientific) was used for total RNA extraction, according to the manufacturer's protocol. All samples showed absorbance ratios between 1. 8

Quantitative reverse transcription polymerase chain reaction (RT-qPCR)
Total RNA samples were treated using the TURBO DNA-free kit™ (Thermo Fisher Scientific), according the manufacturer's protocol, using the DNase Inactivation Reagent. This RNA did not yield a PCR product in the absence of a reverse transcription step. Reverse transcription was performed using 1.5 µg of DNA-free RNA (ribosomal protein L30), for which the most stably expressed genes were RPL13A and B2M (Supplemental Figure S8). Primers were designed using Primer3 (12)

Gene ontology analysis
Gene ontology (GO) was performed to determine the biological processes associated with the knockdown of NT5C2 in the neural stem cells. The overlapping gene set was subdivided into up-and downregulated gene sets, which were input into GeneMania 3.4.1 running on Cytoscape 3.4 (17), to compute networks of co-expression and colocalization, and to reveal enrichment of GO terms. The default background list for this analysis was used, which corresponds to all genes in the genome. Significant hits were defined as those that survived correction for multiple comparisons using the false discovery rate (FDR) at q < 0.05 (5%).

Fly stocks and CG32549 knockdown
To Carolina, United States), 10 mL of deionized water, and a dash of dry baker's yeast granules. Flies were routinely maintained at 25°C at a 12hr:12hr light-dark photoperiod.

CG32549 knockdown validation
Approximately 60 D. melanogaster fly heads were dissected per condition (ubiquitous knockdown vs. wild-type control). RNA extractions were carried out using PrepEase RNA Spin Kits by USB or Trizol Reagent, and eluted using molecular grade water.
cDNA synthesis was carried using 300 ng of RNA with Applied Biosystems High Capacity Reverse Transcription cDNA Synthesis Kit following the manufacturer's protocol. cDNA samples were then diluted 1:10 times before proceeding to quantitative RT-PCR (RT-qPCR) analysis. All gene expression studies were carried out using Bio-Rad CFX96 Real-Time System. cDNA (1 µl) was incubated with NT5C2 specific primers and Bio-Rad SYBR Green Supermix. Results were measured using a threshold of 1000 relative fluorescence units (RFUs). Ct values were analyzed using the DC t method, using RpL32 (CG7939) as reference (HK) gene. CG32549 expression was represented as a % of expression in w 1118 flies crossed with Act5c Gal4 driver, and analyzed using the unpaired t-test.

Negative geotaxis/climbing assay and survival analysis of D. melanogaster
Flies which had NT5C2 knocked down ubiquitously, in neurons, or in gut, were tested for differences in psychomotor behavior and survival. The negative geotaxis/climbing assay was based on Gargano and colleagues (20).

NT5C2 expression (mouse cortex single cell RNA-sequencing)
Supplemental Figure S1. Relative frequency of raw read counts from the single-cell RNA-sequencing data obtained from the mouse brain. This analysis suggests that NT5C2 expression occurs in a cell-type specific manner (Kruskal-Wallis test, P < 0.001). Post-hoc analysis confirmed that NT5C2 was more abundant in pyramidal neurons than in astrocytes (Dunn's corrected P < 0.001), and more abundance in interneurons than microglia (corrected P < 0.01) or astrocytes (corrected P < 0.001).
Supplemental Figure S2. NT5C2 antibody validation by overexpressing pNT5C2-myc in HEK293T cells. The overexpression was detected using NT5C2 and myc antibodies.
Supplemental Figure S3. Distribution of the NT5C2 protein in the adult brain. Connecting lines define co-expression (purple) and co-localization (red). Images were generated with GeneMania (22).  Legend: number of reference genes in the category (C), number of genes in the gene set and also in category (O).