Merkel cell mechanotransduction facilitates adult neurogenesis and cognition in an enriched environment

Significance

Tactile perception reshapes the brain when exploring a novel environment. Our study has demonstrated that Piezo2-mediated Merkel cell mechanotransduction facilitates environmental enrichment-induced neurogenesis and cognitive enhancement. This process is dependent on an EE-activated neuraxis, in which the dentate gyrus received tactile input through a circuit that originated from the somatosensory cortex and relayed via dopaminergic neurons of the substantia nigra pars compacta. Overall, these findings unravel the mechanism underlying regulation of brain function via intact tactile processing.

Abstract

The tactile system empowers us to act on and interact with the changes of the external environment. In rodents, tactile sensation, a fundamental sense, is largely mediated via the vibrissae and the barrel cortex. However, it remains unclear how tactile perception reshape the brain when exploring a novel environment. Here, we showed that exposure to an enriched environment (EE) failed to enhance adult neurogenesis and cognition in the mice with defective touch perception due to loss of a mechanotransduction channel Piezo2 in Merkel cells. Moreover, we found an EE-activated neuraxis, in which the dentate gyrus received tactile input through a circuit that originated from the somatosensory cortex (S1) and relayed via dopaminergic neurons of the substantia nigra pars compacta (SNc). Defective touch perception diminished the S1 to SNc afferent, thereby reducing dopamine release. Notably, stimulation of the S1 to SNc afferent restored EE-induced adult neurogenesis and cognition in the mice with defective touch perception. Therefore, our study highlighted the important role of intact tactile processing in brain function.

The tactile system is the first sense to develop prenatally, which is fundamental to our daily life (1). It empowers us to identify objects, to discriminate material properties, and to act on and interact with the changes of the external environment. In mammals, the skin, the largest organ in the body, serves as the resident site of innervation for the tactile sensors (2). Merkel disc, also known as Merkel cell–neurite complex, is the tactile end organ in the skin, which consists of Merkel cells and Aβ-afferent nerve endings (3). Recently, Merkel cells rather than Aβ-afferent nerve endings have been demonstrated to transduce touch stimuli into mechanically activated currents via Piezo2 ion channel (4–6). Through the ascending somatosensory pathway, the external haptic information is transmitted to the somatosensory cortex, where perception is generated (7). Tactile sensation, as a fundamental sense for mice, is mediated via the vibrissae and the barrel cortex (8). Notably, under pathological conditions such as peripheral neuropathy, tactile sensation can be either exaggerated to result in tactile allodynia or reduced to cause numbness (9). Furthermore, tactile processing abnormalities have long been identified in autism spectrum disorders and attention-deficit/hyperactivity disorders (10, 11). Therefore, understanding tactile processing not only provides us with fundamental knowledge about neural function but also sheds light on the etiology of neurological disorders.

The brain is a dynamic structure that constantly undergoes cellular and molecular changes in response to the environment, which ultimately modify and shape behavior. The enriched environment (EE) is a key experimental paradigm to understand how environmental complexity changes the structure and function of the brain across the lifespan of an animal (12). Exposure to EE has been known to elicit a number of neuroanatomical and behavioral changes, including dendritic and synaptic complexity and improved learning and memory (13). Although EE influences many regions of the brain, it is clear that it has a particularly profound effect on the hippocampus (12–14). The dentate gyrus (DG) of the adult hippocampus continuously generates new neurons from resident neural stem cells (NSCs) throughout life (15, 16). This process, named adult hippocampal neurogenesis, is a striking form of brain plasticity (17). Adult hippocampal neurogenesis is thought to play a fundamental role in hippocampus-dependent learning, which is highly modulated by EE (18, 19). Due to the complexity of EE, there are many components, including social, sensory, spatial, and physical stimulation that may participate in the observed effects (12–14). However, it is still unknown how tactile perception mediates EE-induced effects in the brain.

Results

Exposure to EE Does Not Enhance Cognitive Function in P2-cKO Mice.

To investigate whether tactile perception mediates cognitive enhancement in EE, we took advantage of K14-Cre::Piezo2^f/f conditional knockout mice (5) (P2-cKO), in which a mechanically activated cation channel, Piezo2, is ablated in all epithelial including Merkel cells. Because Piezo2 is restrictedly expressed in Merkel cells, P2-cKO mice exhibit deficits in gentle touch perception (4–6). The mice without Piezo2-flox alleles were used as wild-type controls (WT). We kept the mice either in standard conditions housing (SH) or in EE without a running wheel. After 3 wk, the mice were subjected to a battery of behavioral tests (Fig. 1A). Expectedly, exposure to EE was sufficient to promote cognitive function in WT mice, as shown by increasing the preferences for the novel objects in the novel object recognition (NOR) test (Fig. 1 B and C) and the novel location of the same object in the novel object location (NOL) test (Fig. 1 D and E), promoting the freezing behavior in the contexture fear conditioning test (Fig. 1 F–H), as well as increasing the alternations in the Y-maze test (Fig. 1 I–K). Although WT and P2-cKO mice housed in SH had comparable cognitive function, P2-cKO mice did not exhibit cognitive enhancement after EE exposure (Fig. 1 B–K). Nonetheless, WT and P2-cKO mice housed either in SH or in EE performed normally in the open-field test (SI Appendix, Fig. S1 A–E) and the elevated plus maze test (SI Appendix, Fig. S1 F and G), suggesting that tactile perception is involved in EE-induced cognitive enhancement rather than locomotor activity or anxiety-like behavior.

Fig. 1.

P2-cKO mice did not exhibit cognitive enhancement after EE exposure. (A) The experimental design for the behavioral assessment of WT and P2-cKO mice housed either in SH or in EE. (B) Schematic depiction of the novel object recognition test. (C) The percentage of discrimination ratio during the novel object recognition test (n = 11 mice). (D) Schematic depiction of the novel object location recognition test. (E) The percentage of discrimination ratio during the novel object location test (n = 11 mice). (F) Schematic depiction of the fear conditioning test. (G and H) The percentage of freezing behavior observed during the training (G) and testing phases (H) of the contextual fear conditioning test (n = 11 mice). (I) Schematic depiction of the Y-maze test. (J and K) The number of total entries (J) and spontaneous alternation (K) during the Y-maze test (n = 11 mice). Values represent mean ± SEM. Two-way ANOVA with Tukey’s post hoc test; **P < 0.01, ***P < 0.001, ns; not significant.

Exposure to EE Fails to Promote Hippocampal Neurogenesis in P2-cKO Mice.

Neurogenesis in the adult hippocampus is a process regulated by experience (20). In particular, EE has a global effect in promotion of adult neurogenesis (12, 13). To investigate whether tactile perception is involved in EE-induced hippocampal neurogenesis, we housed WT and P2-cKO mice either in SH or in EE, and performed a 2-h 5-Ethynyl-2′-deoxyuridine (EdU) pulse labeling after 3-wk exposure to SH or EE (Fig. 2A). WT and P2-cKO mice housed in SH exhibited a similar number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors (Fig. 2 B–E). But following EE exposure, the quantity of these cell types increased in WT mice but not in P2-cKO mice (Fig. 2 B–E). We then evaluated the differentiation of NSCs in the mice 7 d after EdU injection (Fig. 2F). WT and P2-cKO mice housed in SH had a comparable number of DCX⁺ and EdU⁺DCX⁺ immature neurons (Fig. 2 G–I). However, following EE exposure, the number of these neurons increased in WT mice but not in P2-cKO mice (Fig. 2 G–I).

Fig. 2.

P2-cKO mice did not display enhanced adult hippocampal neurogenesis after EE exposure. (A) The experimental design for proliferation analysis in WT and P2-cKO mice housed either in SH or in EE. (B) Confocal images of Nestin⁺Sox2⁺EdU⁺ NSCs and Nestin⁻Sox2⁺EdU⁺ neural progenitor cells in the DG of WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 20 μm.) (C) Number of EdU⁺ cells in the DG (n = 5 mice). (D) Number of Nestin⁺Sox2⁺EdU⁺ NSCs in the DG (n = 5 mice). (E) Number of Nestin⁻Sox2⁺EdU⁺ neural progenitor cells in the DG. (F) The experimental design for analysis of immature neurons in WT and P2-cKO mice housed either in SH or in EE. (G) Confocal images of EdU⁺ DCX⁺ immature neurons in the DG of WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 20 μm.) (H) Number of EdU⁺ DCX⁺ immature neurons in the DG (n = 5 mice). (I) Number of DCX⁺ immature neurons in the DG (n = 5 mice). (J) The experimental design for differentiation analysis in WT and P2-cKO mice housed either in SH or in EE. (K) Confocal images of EdU⁺NeuN⁺ adult-born neurons in the DG of WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 20 μm.) (L) Number of EdU⁺ cells in the DG (n = 5 mice). (M) Number of EdU⁺NeuN⁺ adult-born neurons in the DG (n = 5 mice). (N) The experimental design for dendritic development analysis in WT and P2-cKO mice housed either in SH or in EE. (O) Confocal images and representative traces of adult-born neurons in WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 50 μm.) (P) Total dendritic length of adult-born neurons (n = 50 neurons from 5 mice). (Q) Sholl analysis of dendritic complexity of adult-born neurons (n = 50 neurons from 5 mice). Values represent mean ± SEM. Two-way ANOVA with Tukey’s post hoc test; *P < 0.05, **P < 0.01, ***P < 0.001, ns; not significant.

To examine the production of adult-born mature neurons, we injected the mice with EdU four times at 12-h intervals, housed them either in SH or in EE for 3 wk, and collected brain tissues 4 wk post EdU administration (Fig. 2J). WT and P2-cKO mice housed in SH had a comparable number of EdU⁺ cells and EdU⁺NeuN⁺ mature neurons (Fig. 2 K–M). After EE exposure, the number of these cells increased in WT mice but not P2-cKO mice (Fig. 2 K–M). Next, we assessed the maturation of adult-born neurons. To this end, we stereotactically injected a retrovirus expressing GFP (RV-GFP) into the DG of the mice, and housed them either in SH or in EE for 3 wk (Fig. 2N). After 4 wk of retroviral injection, we found that the total dendritic length and dendritic complexity of RV-GFP-labeled adult-born neurons were comparable between WT and P2-cKO mice housed in SH (Fig. 2 O–Q). However, following EE exposure, WT mice exhibited a significant increase in the total dendritic length and dendritic complexity of RV-GFP-labeled adult-born neurons, while P2-cKO mice did not (Fig. 2 O–Q). Altogether, these data indicated that touch perception is required for EE-induced adult neurogenesis.

Voluntary Running Promotes Hippocampal Neurogenesis in P2-cKO Mice.

Physical exercise is a potent enhancer of adult hippocampal neurogenesis (21). Although the mice are expected to have higher levels of physical exercise in EE, the total locomotion of individual mice does not correlate with neurogenesis (22). To investigate whether tactile perception is involved in physical exercise-induced hippocampal neurogenesis, we subjected the mice to a running wheel for 7 d, and performed a 2-h EdU pulse labeling after running (SI Appendix, Fig. S2A). P2-cKO mice completed a similar number of running laps per day as WT mice did (SI Appendix, Fig. S2B). In comparison to the sedentary controls, both WT and P2-cKO mice displayed a significant increase in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors (SI Appendix, Fig. S2 C–F). Moreover, the number of DCX⁺ immature neurons was also significantly increased in both WT and P2-cKO mice after voluntary running (SI Appendix, Fig. S2 G and H). Therefore, these data suggested that touch perception is dispensable for exercise-induced adult hippocampal neurogenesis.

EE Neglects to Elevate Dopamine Levels in the Brain of P2-cKO Mice.

Several growth factors involved in neurogenesis, neuronal activity, and synaptic plasticity have been known to be induced by EE, in particular nerve growth factor (NGF) (23) and brain-derived neurotrophic factor (BDNF) (24). While physical exercise elevates expression of vascular endothelial growth factor (VEGF), fibroblast growth factor 2 (FGF2) and insulin-like growth factor 1 (IGF1) (25–27). We found that WT and P2-cKO mice housed in SH had similar expression levels of Bdnf, Ngf, Vegf, Fgf2, and Igf1 mRNAs in the hippocampus (SI Appendix, Fig. S3 A and B). In contrast, the expression levels of Bdnf and Ngf mRNAs, but not Vegf, Fgf2, and Igf1 mRNAs, were significantly increased in the hippocampus of both WT and P2-cKO mice after EE exposure (SI Appendix, Fig. S3B). Thus, touch perception is unlikely involved in regulation of EE-induced growth factors, such as NGF and BDNF.

Except for growth factors, neurotransmitters, such as dopamine and acetylcholine (28, 29), have been demonstrated to influence synaptic plasticity and learning/memory after EE exposure. The levels of dopamine, acetylcholine, norepinephrine, and serotonin (5-hydroxytryptamine) in the hippocampus were comparable between WT and P2-cKO mice housed in SH (SI Appendix, Fig. S3C). While WT mice housed in EE had significant increases in the levels of dopamine and acetylcholine, but not norepinephrine and serotonin (SI Appendix, Fig. S3C). Unexpectedly, exposure to EE elevated the levels of acetylcholine, but not dopamine, in the hippocampus of P2-cKO mice (SI Appendix, Fig. S3C). Hence, these data suggested that touch perception is required for EE-induced neurotransmitter dopamine expression.

Tactile Perception Mediates EE-Induced Dopaminergic Neuron Activation in the SNc.

To monitor EE-induced dopamine levels in vivo, we injected an adeno-associated virus (AAV) carrying a human synapsin 1 gene promoter (hSyn)-driven dopamine sensor gDA3m (AAV-hSyn-gDA3m) (30) into the DG of the mice to record dopamine levels of freely moving mice housed either in SH or in EE (Fig. 3 A and B). Fiber photometry recording showed that gDA3m fluorescent signals were significantly induced in WT mice when switching them from SH to EE (Fig. 3 C–F). However, the effect of EE on gDA3m fluorescent expression could not be induced in P2-cKO mice (Fig. 3 C–F), suggesting that touch perception is essential for EE-induced dopamine release in the brain.

Fig. 3.

Exposure to EE did not induce the activation of the SNc-derived dopaminergic neurons in P2-cKO mice. (A) The experimental design for viral grafting and fiber photometry recording analysis in WT and P2-cKO mice housed either in SH or in EE. (B) Schematic illustration of the stereotactical injection of virus and fiber photometry recording of dopamine levels in the DG. (Scale bar, 100 μm.) (C) Representative traces of dopamine levels in the DG of WT and P2-cKO mice housed either in SH or in EE. (D–F) Quantification of events (D), average peak ΔF/F (E), and area under the curve (AUC) (F) for dopamine levels (n = 8 mice). (G) The experimental design for c-Fos analysis in WT and P2-cKO mice housed either in SH or in EE. (H) Schematic depiction of the SNc.(I) Confocal images of c-Fos ⁺TH⁺ cells in the SNc of WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 20 μm.) (J) Percentage of c-Fos⁺TH⁺ cells among TH⁺ cells in the SNc (n = 5 mice).(K) Schematic depiction of the VTA. (L) Confocal images of c-Fos⁺TH⁺ cells in the VTA of WT and P2-cKO mice housed either in SH or in EE. (Scale bar, 20 μm.) (M) Percentage of c-Fos⁺TH⁺ cells among TH⁺ cells in the VTA (n = 5 mice). (N) The experimental design for viral grafting and fiber photometry recording analysis in WT and P2-cKO mice housed either in SH or in EE. (O and T) Schematic illustration of the stereotactical injection of viruses and fiber photometry recording of dopaminergic neuron activity in the SNc (O) and the VTA (T). Confocal images depict the fiber recording sites. (Scale bar, 100 μm.) (P and U) Representative traces of population activity of dopaminergic neurons in the SNc (P) and the VTA (U) of WT and P2-cKO mice housed either in SH or in EE. (Q–S) Quantification of events (Q), average peak ΔF/F (R), and percentage of above threshold activity (S) in dopaminergic neurons of the SNc (n = 8 mice). (V–X) Quantification of events (V), average peak ΔF/F (W), and percentage of above threshold activity (X) in dopaminergic neurons of the VTA (n = 7 mice). Values represent mean ± SEM. Two-sided paired t test or two-way ANOVA with Tukey’s post hoc test; *P < 0.05, ***P < 0.001, ns; not significant.

The ventral tegmental area (VTA) and substantia nigra pars compacta (SNc) are the nuclei associated with dopamine release (31). To assess whether touch perception mediates EE-induced dopaminergic neuron activation in the VTA and the SNc, we analyzed the expression of activity-dependent immediate early gene (c-Fos) in the mice housed either in SH or in EE (Fig. 3G). WT and P2-cKO mice housed in SH had a comparable number of c-Fos⁺TH⁺ cells in the SNc and the VTA (Fig. 3 H–M). However, the number of c-Fos⁺TH⁺ cells in the SNc, but not in the VTA, was significantly increased in WT mice after EE exposure (Fig. 3 H–M). Notably, the effect of EE on the induction of c-Fos⁺TH⁺ cells was not appeared in the SNc of P2-cKO mice (Fig. 3 H–J), suggesting that EE-induced dopaminergic neuron activation in the SNc is impeded in the mice with defective touch perception. To consolidate these findings, we injected an AAV carrying a Cre-dependent EF1α-driven calcium sensor GCaMP6f (AAV-EF1α-DIO-GCaMP6f) into the SNc or the VTA of DAT-Cre mice (Fig. 3N). Due to Cre recombinase is driven by the transcriptional control of the dopamine transporter (DAT) promoter, GCaMP6f was specific expressed in the dopaminergic neurons. We validated that our injection site was restricted to the SNc (Fig. 3O) or the VTA (Fig. 3T). By performing fiber photometry recording of the calcium signals in freely moving mice housed either in SH or in EE, we found that GCaMP6f fluorescent signals in the SNc (Fig. 3 P–S), but not in the VTA (Fig. 3 U–X), were significantly induced in WT::DAT-Cre mice when switching them from SH to EE. However, the effect of EE-induced GCaMP6f fluorescent signals were not appeared in the SNc of P2-cKO::DAT-Cre mice (Fig. 3 P–S), indicating that touch perception is required for EE-induced dopaminergic neuron activation in the SNc.

Inhibition of Dopaminergic Neurons in the SNc Prevents EE-Induced Adult Neurogenesis and Cognition Enhancement.

To examine whether dopaminergic innervation from the SNc projects to the DG, we injected an AAV carrying a Cre-dependent hSyn-driven mCherry (AAV-hSyn-DIO-mCherry) into the SNc of DAT-Cre::Nestin-GFP mice (SI Appendix, Fig. S4A). We validated that our injection site was restricted to the SNc (SI Appendix, Fig. S4B). Meanwhile, a dense of mCherry⁺ neuronal projections was observed around the DG, where the distal thin processes of Nestin-GFP⁺ NSCs were closely associated with the mCherry⁺ fibers (SI Appendix, Fig. S4B). To validate the above findings, we subsequently employed the retrograde tracing method by injecting a retroAAV2-hSyn-driven-Cre (retroAAV2-hSyn-Cre) into the DG of Ai14 reporter mice (SI Appendix, Fig. S4C). A significant amount of SNc projections to the DG was observed (SI Appendix, Fig. S4D). These data suggested that the SNc-derived dopaminergic projections are involved in controlling of dopamine release and regulating neurogenesis in the DG.

To investigate the requirement of the SNc-derived dopaminergic afferent for EE-induced dopamine release into the DG, we injected a retroAAV2 carrying a Cre-dependent EF1α-driven flippase (retroAAV2-DIO-Flp) into DG of DAT-Cre, and simultaneously injected an AAV carrying a Flp-dependent hSyn-driven inhibitory DREADDs (AAV-hSyn-fDIO-hM4Di-mcherry) into the SNc of the same mice (SI Appendix, Fig. S5 A and B), which allowed us to specifically inhibit the activity of the SNc-derived dopaminergic neurons upon application of clozapine N-oxide (CNO). In order to record the dopamine levels in freely moving mice, we then injected an AAV-hSyn-gDA3m into the DG of above mice (SI Appendix, Fig. S5 A and B). Fiber photometry recording showed that gDA3m fluorescence signals were significantly increased in vehicle-treated DAT-Cre mice when switching them from SH to EE (SI Appendix, Fig. S5 C–F). However, the effect of EE on gDA3m fluorescent expression was not induced in CNO-treated DAT-Cre mice (SI Appendix, Fig. S5 C–F).

To assess the role of the SNc-derived dopaminergic neurons in EE-induced adult neurogenesis and cognition enhancement, we used above strategy to inhibit the activity of the SNc-derived dopaminergic neurons and then subjected the mice to EE. By performing 2-h EdU pulse labeling after EE exposure (SI Appendix, Fig. S5G), we found that CNO-treated DAT-Cre mice exhibited a decrease in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors compared to vehicle-treated DAT-Cre mice (SI Appendix, Fig. S5 H–J). Furthermore, CNO-treated DAT-Cre mice generated fewer DCX⁺ immature neurons than vehicle-treated DAT-Cre mice (SI Appendix, Fig. S5K). With the aid of RV-GFP to label adult-born neurons in the DG, we found that CNO-treated DAT-Cre mice displayed a decrease in the total dendritic length and dendritic complexity of RV-GFP-labeled adult-born neurons compared to vehicle-treated DAT-Cre mice (SI Appendix, Fig. S5 L–N). By subjecting the mice into cognition-related behavioral tests after EE exposure (SI Appendix, Fig. S5O), we found that CNO-treated DAT-Cre mice exhibited lower performances than vehicle-treated DAT-Cre mice during the NOR test (SI Appendix, Fig. S5P), the NOL test (SI Appendix, Fig. S5Q), the contexture fear conditioning test (SI Appendix, Fig. S5R), as well as the Y-maze test (SI Appendix, Fig. S5S). Therefore, the activity of the SNc-derived dopaminergic neurons is involved in EE-induced adult neurogenesis and cognitive enhancement

Stimulation of Dopaminergic Neurons in the SNc Enhances Adult Neurogenesis and Cognition in P2-cKO Mice Housed in EE.

To investigate whether stimulation of SNc-derived dopaminergic neurons is sufficient to promote dopamine release in the mice with defective tactile perception when exposed to EE, we injected an AAV carrying a Flp-dependent hSyn-driven excitatory DREADDs (AAV-hSyn-DIO-hM3Dq-mcherry) into the SNc of P2-cKO::DAT-Cre mice and simultaneously injected a retroAAV2-DIO-Flp to the DG of the same mice, which allowed us to stimulate the activity of the SNc-derived dopaminergic neurons upon CNO administration (Fig. 4 A and B). In order to record the dopamine levels in freely moving mice, we injected an AAV-hSyn-gDA3m into the DG of above mice and then subjected them to EE (Fig. 4 A and B). Fiber photometry recording showed that CNO-treated P2-cKO::DAT-Cre mice increased gDA3m fluorescence signals compared to vehicle-treated P2-cKO::DAT-Cre mice (Fig. 4 C–F). Thus, stimulating the SNc-derived dopaminergic neurons is sufficient to elevate dopamine levels in the brains of mice with impaired touch perception when they are exposed to EE.

Fig. 4.

Stimulation of the SNc-derived dopaminergic neurons promoted EE-induced adult neurogenesis and cognition enhancement in P2-cKO mice. (A) The experimental design for viral grafting, CNO administration, and fiber photometry recording analysis in P2-cKO mice during EE exposure. (B) Schematic illustration of the stereotactical injection of viruses and fiber photometry recording of dopamine levels. (C) Representative traces of dopamine levels in P2-cKO mice during EE exposure after vehicle or CNO treatment. (D–F) Quantification of events (D), average peak ΔF/F (E), and AUC (F) for dopamine levels in the DG (n = 8 mice). (G) The experimental design for viral grafting, CNO administration, and proliferation analysis in P2-cKO mice after EE exposure. (H) Number of EdU⁺ cells in the DG (n = 5 mice). (I) Number of Nestin⁺Sox2⁺EdU⁺ NSCs in the DG (n = 5 mice). (J) Number of Nestin⁻Sox2⁺EdU⁺ neural progenitor cells in the DG (n = 5 mice). (K) Number of DCX⁺ cells in the DG (n = 5 mice). (L) The experimental design for viral grafting, CNO administration, and dendritic development analysis in P2-cKO mice after EE exposure. (M) Total dendritic length of adult-born neurons (n = 50 neurons from 5 mice). (N) Sholl analysis of dendritic complexity of adult-born neurons (n = 50 neurons from 5 mice). (O) The experimental design for viral grafting, CNO administration, and the behavioral assessment in WT and P2-cKO mice after EE exposure. (P) The percentage of discrimination ratio during the novel object recognition test (n = 12 mice). (Q) The percentage of discrimination ratio during the novel object location test (n = 12 mice). (R) The percentage of freezing behavior observed during contextual testing phases of the contextual fear conditioning test (n = 12 mice). (S) Number of spontaneous alternations during the Y-maze test (n = 12 mice). Values represent mean ± SEM. Student’s t test or two-way ANOVA with Tukey’s post hoc test; *P < 0.05, **P < 0.01, ***P < 0.001.

To assess whether stimulation of the SNc-derived dopaminergic neurons is sufficient to promote adult neurogenesis and cognition in the mice with defective touch perception after EE exposure, we used above strategy to stimulate the activity of the SNc-derived dopaminergic neurons and then subjected the mice to EE. By performing 2-h EdU pulse labeling after EE exposure (Fig. 4G), we found that CNO-treated P2-cKO::DAT-Cre mice exhibited an increase in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors compared to vehicle-treated P2-cKO::DAT-Cre mice (Fig. 4 H–J). Furthermore, CNO-treated P2-cKO::DAT-Cre mice generated more DCX⁺ immature neurons than vehicle-treated P2-cKO::DAT-Cre mice (Fig. 4K). With the aid of RV-GFP to label adult-born neurons in the DG, we found that CNO-treated P2-cKO::DAT-Cre mice displayed an increase in the total dendritic length and dendritic complexity of RV-GFP-labeled adult-born neurons compared to vehicle-treated P2-cKO::DAT-Cre mice (Fig. 4 L–N). By subjecting the mice into cognition-related behavioral tests after EE exposure (Fig. 4O), we found that CNO-treated P2-cKO::DAT-Cre mice performed better than vehicle-treated P2-cKO::DAT-Cre mice during the NOR test (Fig. 4P), the NOL test (Fig. 4Q), the contexture fear conditioning test (Fig. 4R), as well as the Y-maze test (Fig. 4S). Thus, these data corroborated our notion that activation of SNc-derived dopaminergic neurons is required for EE-induced adult neurogenesis and cognitive enhancement.

Dopamine Mediates EE-Induced Adult Neurogenesis Mainly Through DRD2 Receptor.

Dopamine acts on both D1-like receptors (DRD1 and DRD5) and D2-like receptors (DRD2, DRD3, DRD4) (32). Neural progenitor cells (NPCs) isolated from the adult hippocampus highly expressed Drd2 and Drd3 mRNAs, modestly expressed Drd1 mRNA, and barely expressed Drd4 and Drd5 mRNA (SI Appendix, Fig. S6A). While primary hippocampal neurons isolated from newborn mice predominantly expressed Drd1, Drd2, and Drd3 mRNAs and barely expressed Drd4 and Drd5 mRNAs (SI Appendix, Fig. S6D). Exogenous dopamine increased NPC proliferation (SI Appendix, Fig. S6 B and C) and dendritic development of primary hippocampal neurons (SI Appendix, Fig. S6 E and F), which was consistent with previous studies (33, 34). However, these dopamine-induced effects in NPCs (SI Appendix, Fig. S6 B and C) and primary hippocampal neurons (SI Appendix, Fig. S6 E and F) were completely blocked when treating with DRD2 antagonist (Haloperidol), but not with DRD1 antagonist (SCH-23390) or DRD3 antagonist (SB-277011), suggesting that dopamine regulates neurogenesis and neuronal development mainly through DRD2 receptor. To consolidate these findings, we generated a lentivirus expressing a short hairpin RNA (shRNA) against Drd2 (referred to lenti-shDrd2) and mCherry. In comparison to control shRNA (referred to lenti-shNC), lenti-shDrd2 effectively knocked down DRD2 expression (SI Appendix, Fig. S6 G and H). Notably, dopamine was unable to promote NPC proliferation and dendritic development of primary hippocampal neurons after knockdown of DRD2 expression (SI Appendix, Fig. S6 I and J).

To investigate whether DRD2 is required for EE-induced adult neurogenesis in vivo, we generated a lentivirus carrying a Cre-dependent expression of Drd2-shRNAs and mCherry (lenti-DIO-shDrd2). While a lentivirus carrying a Cre-dependent expression of control shRNA and mCherry (lenti-DIO-shNC) was used as control. We stereotactically injected the lentiviruses into the DG of Nestin-Cre mice and subjected them to EE (SI Appendix, Fig. S7 A and B). By performing a 2-h EdU pulse labeling after EE exposure, we found that Nestin-Cre mice injected with lenti-DIO-shDrd2 exhibited a decrease in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors compared to Nestin-Cre mice injected with lenti-DIO-shNC (SI Appendix, Fig. S7 C–E). Furthermore, the number of DCX⁺ immature neurons was significantly reduced in Nestin-Cre mice injected with lenti-DIO-shDrd2 (SI Appendix, Fig. S7F). Next, we assessed the requirement of DRD2 for EE-induced maturation of adult-born neurons. To this end, we generated a retrovirus carrying a Cre-dependent expression of Drd2-shRNAs and mCherry (RV-DIO-shDrd2). The retrovirus carrying a Cre-dependent expression of control shRNA and mCherry (RV-DIO-shNC) was used as control. We stereotactically injected the retrovirus into the DG of Nestin-Cre mice and subjected them to EE (SI Appendix, Fig. S7G). After 4 wk of retroviral injection, we found that the total dendritic length and dendritic complexity of RV-mCherry-labeled adult-born neurons were decreased in Nestin-Cre mice injected with RV-DIO-shDrd2 compared to Nestin-Cre mice injected with RV-DIO-shNC (SI Appendix, Fig. S7 H and I). To investigate whether DRD2 is required for EE-induced cognitive enhancement, we injected the mice with lentivirus and subjected them into cognition-related behavioral tests after EE exposure (SI Appendix, Fig. S7J), we found that Nestin-Cre mice injected with lenti-DIO-shDrd2 exhibited lower performances than Nestin-Cre mice injected with lenti-DIO-shNC during the NOR test (SI Appendix, Fig. S7K), the NOL test (SI Appendix, Fig. S7L), the contexture fear conditioning test (SI Appendix, Fig. S7M), as well as the Y-maze test (SI Appendix, Fig. S7N).

The SNc-Derived Dopaminergic Neurons Receive the Inputs from the Somatosensory Cortex.

The primary somatosensory cortex (S1) contains distinct groups of neurons that receive input from tactile processing. To determine whether the S1 innervates the dopaminergic neurons in the SNc, we took advantage of a herpes simplex virus (HSV)-based approach for anterior tracing of monosynaptic outputs (35). We simultaneously injected an AAV carrying a hSyn-driven thymidine kinase (TK) and GFP (AAV-hSyn-TK-GFP) and a genetically modified version of anterograde HSV encoding a membrane-targeted tdTomato (tdT) reporter (H129ΔTK-tdT) into the S1 (Fig. 5A). This strategy restricted the expression of H129ΔTK-tdT in the neurons of S1 and transduces the expression of tdT reporter to their direct synaptic targets. Notably, the distributions of postsynaptic outputs from the S1 were observed in the SNC, in which the majority of tdT⁺ cells were immunopositive for TH (Fig. 5 A and B). These results were supported by the previous notion (36), in which dopaminergic neurons in the SNc, but not the VTA, mainly receive excitatory inputs from the S1.

Fig. 5.

The S1 to SNc afferent mediated EE-induced dopaminergic neuron activation and dopamine release. (A) Schematic illustration of the stereotactical injection of viruses used for the anterograde tracing. Confocal image (Left) represents GFP⁺ cells in the primary somatosensory cortex (S1). (Scale bar, 500 μm.) Confocal images (Right) represent the anterograde tracing tdT⁺TH⁺ cells in the SNc. [Scale bar, 100 μm (Left), 40 μm (Right).] (B) Percentage of TH⁺ cell among all tdT⁺ anterograde tracing cells in the SNc (n = 3 mice). (C) The experimental design and illustration for viral grafting, CNO administration, and the c-Fos analysis in the mice housed either in SH or in EE. (D) Confocal images of c-Fos⁺TH⁺ cells in the SNc of the mice housed either in SH or in EE. (Scale bar, 20 μm.) (E) Percentage of c-Fos⁺TH⁺ cells among total TH⁺ cells in the SNc (n = 5 mice). (F) The experimental design and illustration of viral grafting, CNO administration, and the fiber photometry recording of neuronal activity in the SNc of the mice housed either in SH or in EE. (G) Representative traces of population activity for dopaminergic neurons in the SNc of the mice housed either in SH or in EE after vehicle or CNO treatment. (H–J) Quantification of events (H), average peak ΔF/F (I), and percentage of above threshold activity (J) in dopaminergic neurons of the SNc (n = 8 mice). (K) The experimental design and illustration of viral grafting, CNO administration, and the fiber photometry recording of dopamine levels in the DG. (L) Representative traces of dopamine levels in the DG of the mice housed either in SH or in EE after vehicle or CNO treatment. (M–O) Quantification of events (M), average peak ΔF/F (N), and AUC (O) for dopamine levels in the DG (n = 8 mice). Values represent mean ± SEM. Two-sided paired t test or two-way ANOVA with Tukey’s post hoc test; **P < 0.01, ***P < 0.001, ns; not significant.

The S1 to SNc Afferent Is Required for EE-Induced Activation of Dopaminergic Neurons.

To investigate whether EE activate dopaminergic neurons in the SNc via the S1, we injected a retroAAV2 carrying a hSyn-driven flippase (Flp) (retroAAV2-hSyn-Flp) into the SNc of the mice, which allowed us to selectively expressed Flp in neurons of the S1 that project to the SNc (Fig. 5C). Simultaneously, an AAV carrying a Flp-dependent hSyn-driven inhibitory DREADDs (AAV-hSyn-fDIO-hM4Di-mcherry) was injected into the S1 of the same mice (Fig. 5C), which allowed us to inhibit the S1 to SNc afferent upon CNO administration. The mice were then subjected either in SH or in EE. In contrast to vehicle-treated mice, CNO-treated mice failed to increase the number of c-Fos⁺TH⁺ cells in the SNc when switching the housing from SH to EE (Fig. 5 D and E). To consolidate these findings, we used the above strategy to inhibit the S1 to SNc afferent, and then injected an AAV-EF1α-DIO-GCaMP6f into the SNc of DAT-Cre mice to record the activity of dopaminergic neurons in freely moving mice (Fig. 5F). Fiber photometry recording showed that the GCaMP6f fluorescent signals are not able to induce in CNO-treated mice when switching the housing condition from SH to EE (Fig. 5 G–J). Furthermore, we injected AAV-hSyn-gDA3m into the DG of WT mice to record dopamine levels in freely moving mice (Fig. 5K). Fiber photometry recording showed that the gDA3m fluorescent signals were unable to elicit in CNO-treated mice when switching the housing from SH to EE (Fig. 5 L–O). Therefore, these data suggested that EE induces dopaminergic neuron activation through the S1 to SNc afferent.

Inhibition of the S1 to SNc Afferent Prevents EE-Induced Adult Neurogenesis and Cognition Enhancement.

To assess whether the neuronal activity in the S1 is affected in the mice with defective touch perception, we examined the expression of c-Fos in the S1 of the mice housed either in SH or in EE (SI Appendix, Fig. S8A). In comparison to WT mice housed in SH, the number of c-Fos⁺ cells was significantly increased in the S1 and the motor cortex of WT mice housed in EE (SI Appendix, Fig. S8 B–E). While P2-cKO mice housed in EE showed an increase in the number of c-Fos⁺ cells in the motor cortex, but not in the S1, compared to P2-cKO mice housed in SH (SI Appendix, Fig. S8 B–E). Thus, these data suggested that exposure to EE failed to activate the S1 in the mice with defective tactile perception.

To investigate the requirement of the S1 to SNc afferent for EE-induced adult neurogenesis and cognitive enhancement, we injected a retroAAV2-hSyn-Flp into the SNc of WT mice and an AAV-hSyn-fDIO-hM4Di-mcherry into the S1 of the same mice to inhibit the S1 to SNc afferent upon CNO administration (SI Appendix, Fig. S8F). The mice were then subjected to EE. By performing a 2-h EdU pulse labeling after EE exposure, we found that CNO-treated mice exhibited a decrease in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors compared to vehicle-treated mice (SI Appendix, Fig. S8 G–I). Furthermore, CNO-treated mice generated fewer DCX⁺ immature neurons than vehicle-treated mice (SI Appendix, Fig. S8J). By injecting RV-GFP in the DG to label newborn neurons, we found that the total dendritic length and dendritic complexity of adult-born neurons were significantly reduced in CNO-treated mice compared to vehicle-treated mice (SI Appendix, Fig. S8 K–M). By subjecting the mice into cognition-related behavioral tests after EE exposure (SI Appendix, Fig. S8N), we found that CNO-treated mice performed worse than vehicle-treated mice during the NOR test (SI Appendix, Fig. S8O), the NOL test (SI Appendix, Fig. S8P), the contexture fear conditioning test (SI Appendix, Fig.S8Q), as well as the Y-maze test (SI Appendix, Fig. S8R). Thus, these data suggested the activation of the S1 to SNc afferent is required for EE-induced adult neurogenesis and cognition enhancement.

Stimulation of the S1 to SNc Afferent Enhances Adult Neurogenesis and Cognition in P2-cKO Mice Housed in EE.

Since Merkel cell–mediated mechanotransduction regulates EE-induced dopaminergic neuron activation through the S1 to SNc afferent, we hypothesized that stimulation of the S1 to SNc afferent is sufficient to promote the activation the SNc-derived dopaminergic neurons in the mice with defective tactile processing when exposed to EE. To address this, we injected a retroAAV2-hSyn-Flp into the SNc of P2-cKO mice and simultaneously injected an AAV carrying a Flp-dependent hSyn-driven excitatory DREADDs (AAV-hSyn-fDIO-hM3Dq-mcherry) into the S1 of the same mice (SI Appendix, Fig. S9A), which allowed us to activate the S1 to SNc afferent upon CNO administration. The mice were then subjected into EE. We found that the number of c-Fos⁺TH⁺ cells in the SNc were significantly increased in CNO-treated P2-cKO mice housed in EE (SI Appendix, Fig. S9 B and C). To consolidate these findings, we used the above strategy to stimulate the S1 to SNc afferent and injected an AAV-EF1α-DIO-GCaMP6f into the SNc of P2-cKO::DAT-Cre mice to record the activity of dopaminergic neurons in freely moving mice (SI Appendix, Fig. S9D). Fiber photometry recording showed that the GCaMP6f fluorescent signals were increased in CNO-treated P2-cKO::DAT-Cre mice when exposed to EE (SI Appendix, Fig. S9 E–H). Furthermore, we injected AAV-hSyn-gDA3m into the DG of P2-cKO mice to record dopamine levels in freely moving mice (SI Appendix, Fig. S9I). Fiber photometry recording showed that the gDA3m fluorescent signals were elevated in CNO-treated P2-cKO mice housed in EE (SI Appendix, Fig. S9 J–M). Therefore, these data suggested that stimulation of the S1 to SNc afferent is able to increase dopaminergic neuron activity and promote dopamine release in the mice with defective touch perception when exposed to EE.

To investigate whether stimulation of S1 to SNc afferent is able to enhance adult neurogenesis and cognition in the mice with defective touch perception after EE exposure, we again used the above strategy to stimulate the S1 to SNc afferent in P2-cKO mice and subjected the mice in EE (SI Appendix, Fig. S10A). By performing 2-h EdU pulse labeling after EE exposure, we found that CNO-treated P2-cKO mice exhibited an increase in the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors compared to vehicle-treated P2-cKO::DAT-Cre mice (SI Appendix, Fig. S10 B–D). Furthermore, CNO-treated P2-cKO mice generated more DCX⁺ immature neurons than vehicle-treated P2-cKO mice (SI Appendix, Fig. S10E). With the aid of RV-GFP to label adult-born neurons in the DG, we found that CNO-treated P2-cKO mice displayed an increase in the total dendritic length and dendritic complexity of RV-GFP-labeled adult-born neurons compared to vehicle-treated P2-cKO mice after EE exposure (SI Appendix, Fig. S10 F–H). By subjecting the mice into cognition-related behavioral tests after EE exposure (SI Appendix, Fig. S10I), we found that CNO-treated P2-cKO mice performed better than vehicle-treated P2-cKO mice during the NOR test (SI Appendix, Fig. S10J), the NOL test (SI Appendix, Fig. S10K), the contexture fear conditioning test (SI Appendix, Fig. S10L), as well as the Y-maze test (SI Appendix, Fig. S10M). Collectively, these data supported our conclusion that activation of the S1 to SNc afferent is essential for EE-induced adult neurogenesis and cognitive enhancement.

To investigate whether enhancing adult neurogenesis could rescue the cognitive impairment in the mice with defective touch perception after EE exposure, we infused recombinant Wnt3a into the DG and subjected the mice in EE (SI Appendix, Fig. S11 A and E). Wnt signaling is required for NSC proliferation and adult hippocampal neurogenesis (37). Administration of Wnt3a was able to increase the number of EdU⁺ cells, EdU⁺Nestin⁺Sox2⁺ NSCs, and EdU⁺Nestin⁻Sox2⁺ intermediate progenitors in P2-cKO mice after exposure to EE (SI Appendix, Fig. S11 B–D). Notably, we found that, in the NOR test (SI Appendix, Fig. S11F), the NOL test (SI Appendix, Fig. S11G), the contexture fear conditioning test (SI Appendix, Fig. S11H), and the Y-maze test (SI Appendix, Fig. S11I), Wnt3a-treated P2-cKO mice outperformed vehicle-treated P2-cKO mice after exposure to EE. Therefore, these data corroborate that adult hippocampal neurogenesis is involved in EE-induced cognitive enhancement.

Discussion

It is worth noting that there is no standardization of EE with regard to procedure and methodology, which varies drastically in terms of environment size, stimulation used (various tunnels and toys), as well as with or without running wheels. Exercise is often studied independently of EE. Both EE and exercise enhance adult hippocampal neurogenesis and cognitive function (19, 38), but the total locomotion of the mice does not correlate with neurogenesis in EE (22). In this study, by subjecting the mice into EE without a running wheel, we found that exposure to EE failed to promote adult hippocampal neurogenesis and improve cognitive function in the P2-cKO mice, in which the tactile sensitivity was impaired due to a specific ablation of Piezo2 from Merkel cells. Meanwhile, by subjecting the mice to voluntary running, we found that P2-cKO mice were still responsive to vulnerary running-induced adult neurogenesis and cognitive improvement, suggesting unlikely involvement of tactile processing in the neural consequences of exercise. Therefore, our study indicated that tactile perception is required for promoting hippocampal neurogenesis and cognitive function in EE but not in voluntary exercise.

Tactile sensation serves as a fundamental sense for mice. Through this sense, mice form a representation of the world, which includes EE. The barrel cortex houses the somatosensory representation of the whiskers on the rodent’s face and constitutes an early stage of cortical processing for tactile information (39). It occupies roughly 70% of the primary somatosensory cortex. Even though damage to the barrel cortex impacts the function of the hippocampus, EE significantly alleviates this effect (40–42). The positive influence of EE might be attributed to the modulation of BDNF, which is essential for neuronal survival and promotes learning and memory (12). Although Piezo2 in Merkel cells of whisker hair follicles responds to tactile transduction (4), our study demonstrated that exposure to EE increases BDNF expression in P2-cKO mice. These findings suggest that tactile input does not affect BDNF expression in the brain in the same way.

Dopamine controls various physiological functions in the brain. The dopaminergic projections originating from the SNc and the VTA directly innervate the DG, thus directly influencing the microenvironment of this neurogenic niche to regulate NSC dynamics (33, 43, 44). Alteration in midbrain dopamine neurons by 6-OHDA or MPTP toxin decreased proliferation and differentiation of NSCs in the adult hippocampus (33, 45). In this study, we found that EE specifically activated dopaminergic neurons in the SNc, but not the VTA, thereby promoting dopamine release in the DG. However, EE-induced dopaminergic neuron activation in the SNc was completely blocked in P2-cKO mice. Furthermore, using viral-based neural circuit tracing approaches, we showed that the SNc-derived dopaminergic neurons mainly received the inputs from the S1, which is consistent with previous study (36). Meanwhile, we showed that EE was able to activate the S1 to the SNc afferent in WT mice, but not in P2-cKO mice. Importantly, inhibition of the S1 to SNc afferent impeded EE-induced adult neurogenesis and cognitive improvement in WT mice, while stimulation of the S1 to SNc afferent was able to enhance adult neurogenesis and cognitive function in P2-cKO mice after EE exposure. Collectively, we had identified an EE-activated neuraxis, in which the DG receives the tactile input through a circuit that originates from the S1 and relays via dopaminergic neurons in the SNc. Our data indicated that this neuraxis is required for EE-induced adult neurogenesis and cognitive enhancement. However, we could not completely rule out the possibility that somatosensory signals from other pathways (such as the entorhinal cortex, the thalamus, and the subiculum) might play an important role in hippocampal function. Moreover, dopamine has been found to suppress the transmission of the medial performant path to granular cells in the DG (46), suggesting a potential indirect effect of SNc-derived dopamine on EE-mediated improvements in cognition and neurogenesis.

Dopamine functions by acting on its receptors, which are classified as either D1-like receptors (DRD1 and DRD5) or D2-like receptors (DRD2, DRD3, DRD4) (32). In this study, we found that Drd2 and Drd3 were highly expressed in NSCs isolated from the adult hippocampus, while Drd1, Drd2, and Drd3 were extremely expressed in primary hippocampal neurons. Dopamine has a lower affinity for D1-like receptors than D2-like receptors (32). Using dopamine receptor antagonists, we found that inhibition of DRD2, but not DRD1 and DRD3, blocked the NSC proliferation and neuronal development induced by exogenous dopamine, which is consistent with previous study (47). Moreover, D2-like receptor agonist has been shown to promote the proliferation of NSCs (47, 48), supporting that dopamine induces the proliferation of NSC via DRD2 receptor. Interestingly, treatment with haloperidol, a D2-like receptor antagonist, did not affect the proliferation of NSCs (49), while others showed an increase in NSC proliferation (50). These discrepancies might be due to difference in experiment conditions including experimental design, drug dosage, and species because haloperidol is not a selective D2-like receptor antagonist (51). Future studies are required to evaluate the effect of each dopamine receptor subtype in NSCs using more selective dopamine receptor agonists/antagonists or genetic loss of function.

EE has been known to delay the onset and progression of pathological symptoms and ameliorate the brain function in various neurodegenerative disorders (14), such as Alzheimer’s disease and Parkinson’s disease. Moreover, aging is accompanied by a steady decline in touch sensitivity and acuity (52). For example, wrinkles and loss of firmness and elasticity in the skin are accompanied with aging, which may negatively influence the skin-neural coupling (53). Our study demonstrated that tactile perception mediates EE-induced adult neurogenesis and cognitive enhancement, which highlights the need for more evidence regarding age-related changes in peripheral nerve function in the skin, as well as the impact of the skin-neural coupling of touch in the central nervous system.

Methods

Detailed methods are provided in SI Appendix and include the following items: Mice; Viruses information; Stereotactical injection of viruses; Fiber photometry recording; EdU and drug administration; Osmotic pump grafting; Tissue preparation and immunohistochemistry; Cell quantification of the brain sections; Dendritic morphology analyses; Isolation and culture of adult dentate NPCs; Immunocytochemistry analysis; RNA isolation and real-time PCR; Western blot assay; Measurement of neurotransmitters; Open-field test; Novel object recognition test; Novel objective location test; Fear conditioning test; Elevated plus maze test; Y-Maze test and Statistical analysis.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.