Transcriptomic and spatial GABAergic neuron subtypes in zona incerta mediate distinct innate behaviors

Abstract

Understanding the anatomical connection and behaviors of transcriptomic neuron subtypes is critical to delineating cell type-specific functions in the brain. Here we integrated single-nucleus transcriptomic sequencing, in vivo circuit mapping, optogenetic and chemogenetic approaches to dissect the molecular identity and function of heterogeneous GABAergic neuron populations in the zona incerta (ZI) in mice, a region involved in modulating various behaviors. By microdissecting ZI for transcriptomic and spatial gene expression analyses, our results revealed two non-overlapping Ecel1- and Pde11a-expressing GABAergic neurons with dominant expression in the rostral and medial zona incerta (ZIr^Ecel1 and ZIm^Pde11a), respectively. The GABAergic projection from ZIr^Ecel1 to periaqueductal gray mediates self-grooming, while the GABAergic projection from ZIm^Pde11a to the oral part of pontine reticular formation promotes transition from sleep to wakefulness. Together, our results revealed the molecular markers, spatial organization and specific neuronal circuits of two discrete GABAergic projection neuron populations in segregated subregions of the ZI that mediate distinct innate behaviors, advancing our understanding of the functional organization of the brain.

Fig. 1. Single-nucleus transcriptomic sequencing reveals major cell type composition in zona incerta.

a Schematic diagram showing the workflow for microdissection of zona incerta from brain slices in green fluorescent protein knock-in transporter line Gad67-GFP mice, single nucleus isolation and RNA sequencing by 10x Genomics. UMAP plot of the processed dataset containing 15,375 single nuclei collected from two independent cohorts of 10x Genomics experiments (n = 6 mice for each cohort). The cell clusters were annotated according to canonical cell type-specific marker genes. Scale bar, 1 mm. b UMAP plot showing re-clustering the excitatory and inhibitory neurons into ten glutamatergic neuron clusters (Glu_N1 to Glu_N10) and twelve GABAergic neuron clusters (GABA_N1 to GABA_N12). c Heatmap showing the differentially expressed genes (DEGs) among neuron clusters from snRNA-seq dataset. The regional identity (top) of the neuron clusters were annotated by the expression of region specific DEGs (right). TH, Thalamus; STN, Subthalamus nucleus; PVN, Paraventricular nucleus; Hypo, Hypothalamus; ZI, Zona incerta; PIL, Posterior intralaminar thalamus nucleus; TRN, Thalamic reticular nucleus; UnA, Un-Annotated region. d UMAP plot showing the GABAergic neuron clusters (GABA_N4 to GABA_N9) associated with zona incerta. e Dotplot showed the differential expressed genes (DEGs) of GABAergic neuron clusters associated with zona incerta. Source data are provided as a Source Data file.

Fig. 2. Region-specific genes expression of GABAergic neuron subtypes in subregions of zona incerta.

a Violin Plots showed the expression level of selected DEG genes of GABAergic neuron clusters in zona incerta. b Representative images (b) and statistical results (c) showing the density of Ecel1 (ZIr vs ZIm, ****P < 0.0001; ZIr vs ZIc, ****P < 0.0001; ZIm vs ZIc, *P = 0.0443), Cdh23 (ZIr vs ZIm, **P = 0.0075; ZIr vs ZIc, **P = 0.0039), Pax6 (ZIr vs ZIm, ***P = 0.0004; ZIr vs ZIc, ***P = 0.0007), Pde11a (ZIr vs ZIm, **P = 0.0038; ZIm vs ZIc, ***P = 0.0006), Ptprk (ZIr vs ZIm, **P = 0.0020; ZIr vs ZIc, *P = 0.0119), Nos1 (ZIr vs ZIm, **P = 0.0062; ZIr vs ZIc, **P = 0.0092) in the ZIr, ZIm and ZIc, n = 9 brain slices from 3 mice per group. Scale bar, 200 μm. One-way ANOVA and tukey’s multiple comparisons test. d Representative images (d) and statistical results (e) showing the overlap between Ecel1, Pde11a and Ptprk in different brain sections of ZIr, ZIm and ZIc. The cell numbers in ZIr (Ecel1⁺: 410.0 ± 9.539 cells; Ecel1⁺Pde11a⁺: 9.33 ± 1.20 cells; Ecel1⁺Ptprk⁺: 43.33 ± 5.49 cells; Pde11a⁺Ptprk⁺: 4.33 ± 0.88 cells; Ecel1⁺Pde11a⁺Ptprk⁺: 1.67 ± 0.33 cells), ZIm (Pde11a⁺: 361.0 ± 23.81 cells; Ecel1⁺Pde11a⁺: 1.67 ± 0.67 cells; Ecel1⁺Ptprk⁺: 4.67 ± 0.88 cells; Pde11a⁺Ptprk⁺: 2.67 ± 0.33 cells; Ecel1⁺Pde11a⁺Ptprk⁺: 0 ± 0 cells) and ZIc (Ptprk⁺: 132.3 ± 16.13 cells; Ecel1⁺Pde11a⁺: 3.33 ± 0.33 cells; Ecel1⁺Ptprk⁺: 7.33 ± 0.88 cells; Pde11a⁺Ptprk⁺: 2.00 ± 0.58 cells; Ecel1⁺Pde11a⁺Ptprk⁺: 2.33 ± 0.88 cells) were shown respectively. N = 9 brain slices from 3 mice per group. Scale bars, 100 μm and 20 μm (zoom-in image). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 3. Generation and validation of transgenic mice Ecel1-Cre and Pde11a-Cre.

a Representative images showing the expression of Ecel1 (a) and Pde11a (b) in Slc32a1 (Vgat) positive neurons in ZIr or ZIm. Scale bar, 50μm. c The percentage of Ecel1⁺Slc32a1⁺ cells relative to all Ecel1⁺ or Slc32a1⁺ cells in ZIr (Ecel1⁺Slc32a1⁺/Ecel1⁺, 96.39 ± 1.19%; Ecel1⁺Slc32a1⁺/Slc32a1⁺, 59.69 ± 6.28%) and the percentage of Pde11a⁺Slc32a1⁺ cells relative to all Pde11a⁺ or Slc32a1⁺ cells in ZIm (Pde11a⁺Slc32a1⁺/Pde11a⁺, 97.7 ± 2.3%; Pde11a⁺Slc32a1⁺/Slc32a1⁺, 33.92 ± 9.67%). n = 9 brain slices from 3 mice per group. d Schematic diagram showing the generation of Ecel1-Cre or Pde11a-Cre knock-in mice. e Representative images showing the Cre and Ecel1 or the Cre and Pde11a expression in ZIr, ZIm and ZIc. Scale bars, 100μm and 20μm (zoom-in image). f The percentage of Ecel1⁺Cre⁺ relative to all Cre⁺ or Ecel1⁺ neurons in ZIr (Ecel1⁺Cre⁺/Cre⁺,93.87 ± 1.45%; Ecel1⁺Cre⁺/Ecel1⁺, 96.41 ± 0.92%), ZIm (Ecel1⁺Cre⁺/Cre⁺, 91.11 ± 2.81%; Ecel1⁺Cre⁺/Ecel1⁺, 86.46 ± 4.39%) and ZIc (Ecel1⁺Cre⁺/Cre⁺, 78.41 ± 4.75%; Ecel1⁺Cre⁺/Ecel1⁺, 78.00 ± 5.95%) of Ecel1-Cre mice. n = 9 brain slices from 3 mice per group. g Statistical results showing the expression and distribution of Cre in the ZIr (71.95 ± 1.12%), ZIm (12.76 ± 0.67%) and ZIc (15.28 ± 0.49%) of Ecel1-Cre mice. ZIr vs ZIm, ****P < 0.0001; ZIr vs ZIc, ****P < 0.0001. N = 9 brain slices from 3 mice per group. h The percentage of Pde11a⁺Cre⁺ relative to all Cre+ or Pde11a⁺ neurons in ZIr (Pde11a⁺Cre⁺/Cre⁺, 97.70 ± 2.30%; Pde11a⁺Cre⁺/Pde11a⁺, 99.48 ± 0.52%), ZIm (Pde11a⁺Cre⁺/Cre⁺, 99.70 ± 0.15%; Pde11a⁺Cre⁺/Pde11a⁺, 98.91 ± 0.22%) and ZIc (Pde11a⁺Cre⁺/Cre⁺, 93.00 ± 4.72%; Pde11a⁺Cre⁺/Pde11a⁺, 89.73 ± 3.06%) of Pde11a-Cre mice. n = 9 brain slices from 3 mice per group. i Statistical results showing the expression and distribution of Cre in the ZIr (4.43 ± 2.54%), ZIm (90.33 ± 2.91%) and ZIc (5.24 ± 1.27%) of Pde11a-Cre mice. ZIr vs ZIm, ****P < 0.0001; ZIm vs ZIc, ****P < 0.0001. N = 9 brain slices from 3 mice per group. One-way ANOVA and tukey’s multiple comparisons test for (g) and (i). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 4. Distinct efferent projections of Ecel1 positive neurons in the rostral ZI and Pde11a positive neurons in the medial ZI.

a Schematic diagram and representative images showing the stereotaxic unilateral injection of Cre-dependent AAV mediated mGFP and synaptophysin-mRuby to specifically label Ecel1 positive neurons in the rostral ZI (ZIr^Ecel1) of Ecel1-Cre mice (a) or Pde11a positive neurons in the medial ZI (ZIm^Pde11a) of Pde11a-Cre mice (b). Scale bar, 250 μm. The experiment was independently repeated 3 times with similar results for each mouse strain. c Representative images showing dense distribution of mGFP and mRuby positive axons from the ZIr^Ecel1 neurons to the PAG and from ZIm^Pde11a neuron to the PRF. Scale bar, 100 μm. d Quantification of integrated fluorescence intensity of mRuby and mGFP-labeled axon terminals in PAG or PRF of ZIr^Ecel1 and ZIm^Pde11a neurons. *P = 0.015, **P = 0.0032, n = 3 mice. e Quantification of relative fluorescent density of mRuby and mGFP-labeled axon terminals of ZIr^Ecel1 and ZIm^Pde11a neurons in the target brain regions. N = 9 slices from 3 mice for each brain region. Abbreviations: LS lateral septal nucleus; POA preoptic area; PVT paraventricular thalamic nucleus; Re reuniens thalamic nucleus; LH lateral hypothalamic area; MD mediodorsal thalamic nucleus; PO posterior thalamic nucleus; Auv secondary auditory cortex; APTD anterior pretectal nucleus; Dk nucleus of Darkschewitsch; RN red nucleus; PAG periaqueductal gray; PRF pontine reticular nucleus, oral part; SC superior colliculus. Two-tailed unpaired t test for (d). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 5. Optogenetic activation of ZIr^Ecel1 neurons and the ZIr^Ecel1 to PAG pathway both induce self-grooming.

a Schematic diagram showing optogenetic stimulation of the ChR2-EYFP or EGFP expressing ZIr^Ecel1 neurons in Ecel1-Cre mice. b Schematic diagram showing the behavioral paradigm for measurement of self-grooming behavior upon optogenetic stimulation. c Statistic analysis of time spent for self-grooming (left) and grooming bouts (right) in 3 min before, during and after optogenetic stimulation of ZIr^Ecel1 neurons in ChR2-group (c) and EGFP-group (d), respectively. Data are presented as mean ± SEM. **P = 0.0014, ****P < 0.0001. Blue column indicates laser stimulation period (5 Hz, 3 min). e Schematic diagram showing the recording of inhibitory postsynaptic currents (IPSCs) from PAG neurons evoked by optogenetic stimulation of ChR2-EYFP expressing axon terminals from the ZIr^Ecel1 to the PAG in acute brain slices. f Representative traces and statistical analysis of light-evoked IPSCs recorded in PAG neuron in ACSF and following sequential application of TTX (f, 1 μM. **P = 0.0082), TTX + 4-AP (f, 100 μM. ****P < 0.0001), or bicuculline (g, 20 μM. **P = 0.0039). N = 9 neurons from 5 mice, two-tailed paired t test. Blue bar indicates optogenetic stimulation (10 ms). h Schematic diagram and representative images showing optogenetic stimulation of the ChR2-EYFP or EGFP expressing terminals from ZIr^Ecel1 to PAG pathway in Ecel1-Cre mice. Scale bar, 100 μm. i Statistic analysis showing time spent for self-grooming (left) and grooming bouts (right) in 3 min before, during or after optogenetic stimulation of ZIr^Ecel1 to PAG pathway in ChR2-group (i) and EGFP-group (j), respectively. **P = 0.0013, ****P < 0.0001. Blue column indicates laser stimulation period (20 Hz, 3 min). One-way ANOVA for (c), (d), (i) and (j). Two-tailed paired t test for (f) and (g). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 6. Calcium response patterns of Ecel1 positive neurons in the rostral ZI in self-grooming behavior.

a Schematic (left) and representative images showing GCamP6s and mCherry expression in ZIr of Ecel1-Cre mice (right). Scale bars, 500 μm and 250 μm (zoom-in image). The experiment was independently repeated 5 times with similar results. b Schematic diagram showing the behavioral paradigm for fiber photometry recording in freely moving mice. c Schematic diagram (c) and statistic analysis (d) showing water spray toward the face effectively induced self-grooming behaviors. **P = 0.0013, n = 5 mice. e Heatmap (e) and average calcium transients (f) of ZIr^Ecel1 neurons before, during and after water-spray induced self-grooming. Red trace: recording of GCaMP6s fluorescent signal; Gray trace: recording of control mCherry fluorescent signal. The dashed lines indicate the start or the end of self-grooming. Shaded areas around means indicate standard error of mean (SEM). g Statistic analysis showing area under curve (AUC) of average calcium transients of GCamP6s channel compared with mCherry control channel before grooming starts (*P = 0.0156) and after grooming starts (*P = 0.0152), before grooming ends (*P = 0.0385) and after grooming ends (*P = 0.0114) in water-spray induced self-grooming. N = 5 mice per group. Two-tailed paired t-test for (d) and (g). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 7. Ecel1 positive neurons in the rostral ZI are required for water spray induced self-grooming.

a Schematic diagram and representative images showing the stereotaxic bilateral injection of AAV-DIO-hM3Dq-mCherry or AAV-DIO-mCherry in the ZIr of Ecel1-Cre mice. Scale bar, 100 μm. The experiment was independently repeated 5 times with similar results. b–d Statistic analysis of time spent for self-grooming (b), grooming bouts (c) and grooming duration per bout (d) in 20 min following injection of CNO in hM3Dq-group mice compared with mCherry-group mice. **P = 0.0097, *P = 0.0336. e Schematic diagram and representative images showing the injection of AAV-DIO-hM4Di-mCherry to chemogenetically inhibit ZIr^Ecel1 neurons of Ecel1-Cre mice. Scale bar, 100 μm. The experiment was independently repeated 7 times with similar results. f The behavioral paradigm for measurement of water-spray induced self-grooming behavior following injection of CNO in hM4Di-group or mCherry-group mice. g–i Statistic analysis of time spent for water-spray induced self-grooming (g), grooming bouts (h) and grooming duration per bout (i) in 20 min following injection of CNO in hM4Di mice. *P = 0.0117, **P = 0.0018. j Schematic diagram showing the stereotaxic bilateral injection of AAV-DIO-DTA to ablate the ZIr^Ecel1 neurons of Ecel1-Cre mice. k–m Statistic analysis of water-spray induced self-grooming time (k), grooming bouts (l) and grooming duration per bout (m) in 20 min following water spray in DTA-group mice. ****P < 0.0001, **P = 0.0036. Two-tailed unpaired t-test for (b–d), (g–i) and (k–m). Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 8. Calcium response patterns of Pde11a positive neurons in the medial ZI during sleep-wakefulness cycle.

a Schematic and representative images (right) showing in vivo recording of calcium response pattern of GCamP6s- or mCherry-expressing Pde11a positive neurons in ZIm. Gray column indicates the position of optical fiber. Scale bars, 500 μm and 250 μm (zoom-in image). The experiment was independently repeated 5 times with similar results. b Schematic diagram showing the behavioral paradigm for fiber photometry, electroencephalogram (EEG) and electromyographic (EMG) recordings in freely moving mice. c Sample trace showing the changes of fluorescent calcium signal of ZIm^Pde11a neurons aligned to sleep- wakefulness state transitions. NREM (yellow bar), non-rapid eye movement; REM (green bar), rapid eye movement; Wake (blue bar). d Average trace and heatmap showing changes of the fluorescent calcium signal of ZIm^Pde11a neurons during sleep-wakefulness state transitions. The red trace showed average fluorescent GCaMP6s or control mCherry (gray) signals, while shaded areas indicate standard error of mean (SEM). e Statistical analysis of the area under curve (AUC) of average fluorescent calcium signal of ZIm^Pde11a neurons during sleep-wakefulness state transitions. N = 5 mice per group. *P = 0.0207 (NREM to REM), *P = 0.0332 (REM to Wake), *P = 0.0171 (Wake to NREM), two-tailed paired t-test. Data are presented as mean ± SEM. Source data are provided as a Source Data file.

Fig. 9. Optogenetic activation of ZIm^Pde11a neurons promotes wakefulness through the ZIm^Pde11a to PRF pathway.

a Optogenetic stimulation of ChR2-EYFP or EGFP expressing ZIm^Pde11a neurons and the behavioral paradigm for EEG-EMG recordings of freely moving Pde11a-Cre mice (b). c Sample EEG power spectrum, EMG traces and brain states (color coded) during optogenetic activation of ZIm^Pde11a neurons. Dark blue: wake; Yellow: NREM; Green: REM. Freq: frequency. d The percentage of wake, NREM, or REM states before, during, and after optogenetic stimulation of ZIm^Pde11a neurons in ChR2-group (left, n = 5 mice) and EGFP-group (right, n = 6 mice). Shaded areas around means indicate standard error of mean (SEM). e Probability change of each state 120 s before and during optogenetic stimulation of ZIm^Pde11a neurons in ChR2-group compared with EGFP-group. ****P < 0.0001, **P = 0.0021. f–h Schematic diagram (f) and representative traces (g and h left) showing the recording of evoked IPSCs in PRF neurons in acute brain slices. TTX, *P = 0.0146; TTX + 4-AP, *P = 0.0129 (n = 6 neurons from 4 mice); Bicuculline, *P = 0.0215 (n = 7 neurons from 4 mice). Blue bar indicates laser stimulation period (10 ms). i Optogenetic stimulation of ChR2-EYFP or EGFP expressing terminals of ZIm^Pde11a to PRF pathway together with EEG-EMG recordings in Pde11a-Cre mice. Scale bar, 100μm. The experiment was independently repeated 5 times with similar results. j Sample EEG power spectrum, EMG traces and brain states during optogenetic activation of ZIm^Pde11a to PRF pathway. k The percentage of wake, NREM, or REM states before, during, and after optogenetic stimulation of ZIm^Pde11a to PRF pathway in ChR2-group (left, n = 5 mice) and EGFP-group (right, n = 5 mice). Shaded areas around means indicate standard error of mean (SEM). l Probability change of each state 120 s before and during optogenetic stimulation of ZIm^Pde11a to PRF pathway in ChR2-group compared with EGFP-group. ****P < 0.0001. Two-tailed unpaired t test for (e) and (l). Two-tailed paired t test for (g) and (h). Data are presented as mean ± SEM. Blue bar indicates laser stimulation period (20 Hz, 120 s). Source data are provided as a Source Data file.

Fig. 10. Selective chemogenetic activation or ablation of ZIm^Pde11a neurons bidirectionally regulate sleep-wakefulness cycle.

a Schematic diagram and representative images showing the injection of AAV-DIO-hM3Dq-mCherry or AAV-DIO-mCherry into the ZIm of Pde11a-Cre mice. Scale bar, 250μm. The experiment was independently repeated 10 times with similar results. b Schematic showing sleep recording conditions with saline- or CNO-injection in hM3Dq-group mice. c–e The percentage of time spent in wake (c ZT7, ***P = 0.0008), NREM sleep (d ZT7, ***P = 0.0007) and REM sleep (e ZT7, *P = 0.0117) in 2-h bins following injection of saline or CNO in hM3Dq-group mice. Wake, **P = 0.0056 (c); NREM, **P = 0.0080 (d); REM, **P = 0.009 (e). f–h Time spent in wake (f), NREM sleep (g) and REM sleep (h) across the 24-hour sleep-wake cycle in hM3Dq-group mice. Wake, **P = 0.0083 (f); NREM, *P = 0.0180 (g); REM, *P = 0.0304 (h). i Number and duration (j) of wake, NREM sleep and REM sleep episodes during the light phase. Number: Wake, **P = 0.0027; NREM, **P = 0.0026. Duration: NREM, ****P < 0.0001. k Schematic diagram showing the injection of AAV-DIO-DTA to ablate the ZIm^Pde11a neurons. l–n The percentage of time spent in wake (l ZT21, *P = 0.0165), NREM sleep (m ZT21, *P = 0.0244) and REM sleep (n ZT21, *P = 0.0427) in 2-h bins in DTA-group mice compared with mCherry-group mice. REM, **P = 0.0040 (n). o–q Time spent in wake (o), NREM sleep (p) and REM sleep (q) across the 24-hour sleep-wake cycle in DTA-group mice compared with mCherry-group mice. Wake, **P = 0.0045 (o); NREM, **P = 0.0082 (p); REM, **P = 0.0013 (q). r Number and duration (s) of wake, NREM sleep and REM sleep episodes during the dark phase. Number: Wake, ***P = 0.0004; NREM, ***P = 0.0005; REM, **P = 0.0016. Duration: Wake, ***P = 0.0004. Two-way ANOVA and Sidak’s multiple comparisons test for (c-e) and (l-n). Two-tailed paired t test for (c–j). Two-tailed unpaired t test for (o–s). Data are presented as mean ± SEM. Gray bar indicates the dark phase (ZT12-ZT0). Source data are provided as a Source Data file.