n was turned off for ~5 min during the application. Repetitive collagenase application was performed until the neuronal cell bodies were loosened from the surrounding connective tissue and cells. Next, we performed in vivo whole-cell patch clamp recording of randomly selected DRG neurons. The patch pipette was filled with an internal solution containing 135 K-gluconate, 0.5 CaCl2, 2 MgCl2, 5 KCl, 5 EGTA, 5 HEPES and 5 D-glucose. Somatic stimuli were applied to the plantar skin of the hindpaw. A heat stimulus was applied via a thermal stimulator at an inter-stimulus interval of 90 s. Cold stimulus was applied with ice water. An innocuous brush on the skin was performed using a cotton swab. Innocuous pressure was induced by picking up a fold of skin with a flattened alligator clip to produce a consistent strength. Noxious mild pinching at the hindpaw was performed using calibrated serrated forceps. Mechanical stimuli were applied in ascending order with an inter-stimulus interval of 30 s. Noxious stimulation was applied sparingly to avoid neuronal sensitization. Whole-cell recordings were performed using an Axopatch 200B amplifier and were digitized with a Digidata 1440A interface and controlled by pCLAMP10.1 software. The whole-cell configuration was obtained in voltage-clamp mode. In brief, we filled a glass electrode with intracellular solution and advanced it quickly to the target depth under moderate positive pressure. When the tip of the electrode was close to a neuron, the tip resistance was increased. We removed the positive pressure and applied negative pressure to aid in seal formation. We achieved the whole-cell configuration using brief bursts of negative pressure to rupture the cell membrane. For detection of responses induced by stimuli, the DRG neuron was subjected to whole-cell recording in voltage-clamp mode at -60 mV. The peripheral stimulus caused a large inward Na+ current and a small outward K+ current, and generated an action potential in responsive neurons. After recording, the internal solution was collected in a PCR tube 
containing 2 l polymerase buffer, 0.5 l oligo dT primer, 0.5 l Quantiscript RT enzyme mix and incubated for 1 h at 42 C. The cDNA product was treated with 4 l ligase buffer and 1 l ligase mix for 30 min at 24 C, and then amplified with 14.5 l REPLI-g sc reaction buffer and 0.5 l REPLI-g SensiPhi DNA polymerase for 2 h at 30 C according to the PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19822627 manufacturer’s instructions. After amplification, the reaction solution was diluted to 300 l. Then, we performed MedChemExpress 518303-20-3 quantitative real-time PCR to evaluate the gene expression level of markers, including Gal, Zcchc12, Cldn9, Nppb, Sst, Nts, Th, Mrgpra3, Mrgprb4, Mrgprd, Nxph1, Baiap2l1, Pvalb and Nefh. Recorded neurons were sorted into corresponding clusters when the expression level of the marker genes for one cluster was at least fivefold higher than that of other clusters. Recorded neurons were sorted into C6 when several marker genes were detected simultaneously. Acknowledgments We thank Drs Li Yang and Luo-Nan Chen for comments on data analysis and manuscript. This work was supported by the National Natural Science Foundation of China and the Strategic Priority Research Program of Chinese Academy of Sciences. Childhood obesity is an important public health problem with severe consequences, including an increased risk of premature death. Body mass index has a strong genetic component with some reported heritability estimates being over 80%. Large genome-wide