Search

Summary A variety of mechanosensory neurons are involved in touch, proprioception, and pain. Many molecular components of the mechanotransduction machinery subserving these sensory modalities remain to be discovered. Here, we combine recordings of mechanosensitive (MS) currents in mechanosensory neurons with single-cell RNA sequencing. Transcriptional profiles are mapped onto previously identified sensory neuron types to identify cell-type correlates between datasets. Correlation of current signatures with single-cell transcriptomes provides a one-to-one correspondence between mechanoelectric properties and transcriptomically defined neuronal populations. Moreover, a gene-expression differential comparison provides a set of candidate genes for mechanotransduction complexes. Piezo2 is expectedly found to be enriched in rapidly adapting MS current-expressing neurons, whereas Tmem120a and Tmem150c, thought to mediate slow-type MS currents, are uniformly expressed in all mechanosensory neuron subtypes. Further knockdown experiments disqualify them as mediating MS currents in sensory neurons. This dataset constitutes an open resource to explore further the cell-type-specific determinants of mechanosensory properties., Graphical abstract, Highlights • Patch-seq is performed on mechanically characterized DRG sensory neurons • Transcriptomically defined neuronal classes and current subtypes are correlated • Datasets of specific transcripts of current subtype-expressing neurons are generated • TACAN (Tmem120a) and Tentonin-3 (Tmem150c) do not contribute to DRG mechano-currents, Mechanosensation is the least understood sensory modality at the molecular level. Parpaite et al. combine single-cell RNA sequencing with mechanosensitive current recordings in cultured sensory neurons. One-to-one correspondence between mechanoelectric properties and single-cell transcriptomes provides a valuable resource for the identification of molecular factors involved in mechanosensation.

We previously linked TSHZ3 haploinsufficiency to autism spectrum disorder (ASD) and showed that embryonic or postnatal Tshz3 deletion in mice results in behavioral traits relevant to the two core domains of ASD, namely social interaction deficits and repetitive behaviors. Here, we provide evidence that cortical projection neurons (CPNs) and striatal cholinergic interneurons (SCINs) are two main and complementary players in the TSHZ3-linked ASD syndrome. We show that in the cerebral cortex, TSHZ3 is expressed in CPNs and in a proportion of GABA interneurons, while not in cholinergic interneurons or glial cells. TSHZ3-expressing cells, which are predominantly SCINs in the striatum, represent a low proportion of neurons in the ascending cholinergic projection system. We then characterized two new conditional knockout (cKO) models generated by crossing Tshz3flox/flox with Emx1-Cre (Emx1-cKO) or Chat-Cre (Chat-cKO) mice to decipher the respective role of CPNs and SCINs. Emx1-cKO mice show altered excitatory synaptic transmission onto CPNs and plasticity at corticostriatal synapses, with neither cortical neuron loss nor impaired layer distribution. These animals present social interaction deficits but no repetitive patterns of behavior. Chat-cKO mice exhibit no loss of SCINs but changes in the electrophysiological properties of these interneurons, associated with repetitive patterns of behavior without social interaction deficits. Therefore, dysfunction in either CPNs or SCINs segregates with a distinct ASD behavioral trait. These findings provide novel insights onto the implication of the corticostriatal circuitry in ASD by revealing an unexpected neuronal dichotomy in the biological background of the two core behavioral domains of this disorder.

Flow of cerebrospinal fluid (CSF) may contribute to spine morphogenesis, as mutations affecting both cilia motility and CSF flow lead to scoliosis1. However, the mechanisms underlying detection of the CSF flow in the central canal of the spinal cord remain elusive. Here we used full-field optical coherence tomography (FF-OCT) and bead tracking to demonstrate that CSF flows bidirectionally along the antero-posterior axis in the central canal of zebrafish embryos. In the zebrafish mutantcfap298tm304, previously known askurly, reduction of cilia motility slows transport down the length of central canal. To investigate downstream mechanisms that could transduce CSF flow, we performed calcium imaging in sensory neurons contacting the CSF (CSF-cNs) and found that disruption in cilia motility impaired the activity of CSF-cNs. CSF-cNs across species express the transient receptor potential channel PKD2L1, also known as TRPP3, which contributes to CSF-cN chemosensory properties. Using calcium imaging and whole-cell patch clamp recordings, we found that the loss of the Pkd2l1 channel inpkd2l1mutant embryos also abolished CSF-cN activity. Whole-cell recordings further demonstrated that opening of a single channel is sufficient to trigger action potentials in wild type CSF-cNs. Recording from isolated cellsin vitro,we showed that CSF-cNs are mechanosensory cells that respond to pressure in a Pkd2l1-dependent manner. Interestingly, adultpkd2l1mutant zebrafish develop an exaggerated spine curvature, reminiscent of kyphosis in humans. Our study indicates that CSF-cNs are mechanosensory cells whose spontaneous activity reflects CSF flowin vivo. Furthermore, Pkd2l1 in CSF-cNs contributes to the maintenance of the natural curvature of the spine.

International audience; A gold standard for characterizing mechanically activated (MA) currents is via heterologous expression of candidate channels in naive cells. Two recent studies described MA channels using this paradigm. TMEM150c was proposed to be a component of an MA channel partly based on a heterologous expression approach (Hong et al., 2016). In another study, Piezo1's N-terminal ``propeller'' domain was proposed to constitute an intrinsic mechanosensitive module based on expression of a chimera between a pore-forming domain of the mechanically insensitive ASIC1 channel and Piezo1 (Zhao et al., 2016). When we attempted to replicate these results, we found each construct conferred modest MA currents in a small fraction of naive HEK cells similar to the published work. Strikingly, these MA currents were not detected in cells in which endogenous Piezo1 was CRISPR/Cas9 inactivated. These results highlight the importance of choosing cells lacking endogenous MA channels to assay the mechanotransduction properties of various proteins. This Matters Arising paper is in response to Hong et al. (2016) and Zhao et al. (2016) in Neuron. See also the response papers by Hong et al. (2017) and Zhao et al. (2017) published concurrently with this Matters Arising.