Piezo1-mediated spontaneous calcium transients in satellite glia impact dorsal root ganglia development

Spontaneous Ca2+ transients of neural cells is a hallmark of the developing nervous system. It is widely accepted that chemical signals, like neurotransmitters, contribute to spontaneous Ca2+ transients in the nervous system. Here, we reveal an additional mechanism of spontaneous Ca2+ transients that is mechanosensitive in the peripheral nervous system (PNS) using intravital imaging of growing dorsal root ganglia (DRG) in zebrafish embryos. GCaMP6s imaging shows that developing DRG satellite glia contain distinct spontaneous Ca2+ transients, classified into simultaneous, isolated, and microdomains. Longitudinal analysis over days in development demonstrates that as DRG satellite glia become more synchronized, isolated Ca2+ transients remain constant. Using a chemical screen, we identify that Ca2+ transients in DRG glia are dependent on mechanical properties, which we confirmed using an experimental application of mechanical force. We find that isolated spontaneous Ca2+ transients of the glia during development is altered by manipulation of mechanosensitive protein Piezo1, which is expressed in the developing ganglia. In contrast, simultaneous Ca2+ transients of DRG satellite glia is not Piezo1-mediated, thus demonstrating that distinct mechanisms mediate subtypes of spontaneous Ca2+ transients. Activating Piezo1 eventually impacts the cell abundance of DRG cells and behaviors that are driven by DRG neurons. Together, our results reveal mechanistically distinct subtypes of Ca2+ transients in satellite glia and introduce mechanobiology as a critical component of spontaneous Ca2+ transients in the developing PNS.


Introduction
It is widely accepted that spontaneous activity is a critical feature of the developing nervous system [1][2][3].This activity has been visualized by measuring Ca 2+ transients.For years, such spontaneous Ca 2+ transients have been investigated in neurons and are identified as neuronal firing or activity, but recent studies have also revealed an important role for spontaneous Ca 2+ transients in glia.These glial Ca 2+ transients can be in response to neuronal activity or independent of neuronal activity and can be characterized into distinct subtypes [4][5][6][7].For example, glial cells exhibit whole cell and microdomain Ca 2+ transients, which are mechanistically and functionally distinct [8][9][10].Glial cells can also exhibit synchronous Ca 2+ transients in physically connected networks [11,12].Regardless of Ca 2+ transient subtype and unique from neurons, immature glia and their progenitors also proliferate throughout life [13].How glial Ca 2+ transients, proliferation, and physically connected networks are related or regulated, remains largely unexplored.The importance of these concepts is underscored by the prevalence of such processes during normal brain development and in gliomas [4,5,[14][15][16].If glial Ca 2+ transients are critical for nervous system function, we need more investigation into how distinct Ca 2+ transients change over development, whether different molecular components control distinct transient subtypes, and if distinct Ca 2+ transients are linked to proliferation and/or network formation.Lastly, these concepts need to be explored in both the central nervous system and peripheral nervous system (PNS).
What we do know is that spontaneous Ca 2+ transients in the nervous system have largely been characterized as dependent on chemical signals.In neurons, Ca 2+ spontaneous activity is promoted by neurotransmitters and their receptors [17,18].Similarly, glutamate and NMDA drive spontaneous Ca 2+ transients of glial cells like oligodendrocytes and astrocytes [19][20][21][22].We also know chemical signals like ATP can induce purinergic receptors to drive Ca 2+ changes in glia, akin to activity of the glia [23][24][25].Each of these chemical signals causes changes to ion channels that drive spontaneous Ca 2+ transients.However, in addition to ion channels that are induced by chemical signals, mechanosensitive ion channels are also present in the nervous system [26,27].For example, Piezo proteins are mechanosensitive channels that are expressed in the nervous system [26,28,29].These mechanosensitive channels are essential for evoking a subset of peripheral sensory neurons in response to mechanical stimulation [30,31].Peripheral mechanosensitive glia are also present at the skin to ensure response to mechanical stimuli [32].However, the role of mechanosensitive properties in the development of glia is less understood, especially in peripheral glia.This is despite knowledge that mechanical components can have profound effects on cell differentiation and tissue organization and that Trp channels, some of which are at least partially mechanosensitive, are important for Ca 2+ transients in glia like astrocytes [4,8,29,[33][34][35][36].
Here, we use imaging of GCaMP6s in satellite glia of the dorsal root ganglia (DRG) in zebrafish as a model to investigate the role of glial activity in the developing PNS.The DRG is required for somatosensory stimuli in the PNS and contains somatosensory neurons and satellite glia that ensheath those neurons.We identify that satellite glia display at least 3 types (microdomain, isolated, and simultaneous) of spontaneous Ca 2+ transients in early phases of development.By mapping the GCaMP6s events, we identify that the DRG transitioned to synchronized Ca 2+ transients early in development, demonstrating the formation of glial networks within the first 3 days of DRG construction.In a pilot screen and follow-up experimental manipulations, we identify mechanosensitive ion channel Piezo1 as a modulator of the isolated Ca 2+ transients of satellite glia in development and identify that these satellite glia are mechanosensitive.Perturbation of Piezo1 causes not only changes in isolated Ca 2+ transients of DRG satellite glia but also in their expansion and function, demonstrating a potential consequence to altering isolated glial Ca 2+ transients during development.Together, we introduce the role of mechanosensitive ion channels in the spontaneous Ca 2+ transients of the developing PNS.

DRG satellite glia exhibit distinct Ca 2+ transients
To understand if DRG satellite glia display spontaneous Ca 2+ transients, we first explored the Ca 2+ transients of DRG cells in intact ganglia using intravital imaging in zebrafish.To do this, intervals capturing the entire DRG.We defined Ca 2+ microdomains as small regions with significant changes in integrated density of fluorescence of GCaMP6s-caax (Fig 1D and 1E).We quantified the duration of these microdomains and found that they lasted on average for 11.81 +/−9.914s (n = 15 cells, 8 DRG, 7 animals) (Fig 1I).We also quantified the average volume of these microdomains and found that they were on average 30.10+/−18.51μM 3 (n = 15 cells, 8 DRG, 7 animals) (Fig 1J).Together, these results indicate DRG satellite glia exhibit at least 3 distinct Ca 2+ transient events during development: isolated, simultaneous, and microdomains.

Satellite glia cell networks are established during early DRG construction
To understand how these types of activity may change over development, we quantified the average amount of isolated and simultaneous Ca 2+ transients events in the same animal at 2, 3, and 4 dpf.While we did not see a significant change in isolated Ca 2+ transients over this developmental period ( Current research proposes that DRG satellite glia form networks in vitro [42][43][44].To further test if this occurs in vivo and to determine when in development it arises, we measured synchronized networks in sox10 + cells.To identify a synchronized network of cells, we compared the Ca 2+ transient profiles of individual cells by computing the correlation between 2 Ca 2+ transient profiles.To determine how this changed in development, we quantified the percent of high correlation coefficients (>0.5) per cell in each DRG at 2, 3, and 4 dpf.By creating network maps of individual DRG that show how the activity of each cell is related (Fig 1K -1M), we found that sox10 + cells at 2 dpf had an average of 37.48% ± 32.08% high correlation coefficients (n = 50 cells, 10 DRG, 6 animals).At 3 dpf, we measured that sox10 + cells had an average of 39.33% ± 30.61% high correlation coefficients (n = 27 cells, 6 DRG, 4 animals) and by 4 dpf, sox10 + cells had a significant increase in the percent of high correlation coefficients, with an average of 58.26% ± 32.If glial networks are forming, we hypothesized that gap junctions may also increase during the time when synchronized Ca 2+ transients are present.Cxn43 is known to be present in satellite glia and contribute to gap junctions in synchronized neural networks [45][46][47].Therefore, we stained for Cxn43 at 2, 3, and 4 dpf in animals expressing Tg(sox10:meGFP), which labels satellite glia in the DRG with membrane-localized GFP.The 2 dpf DRG had an average of 0.500 ± 0.707 Cxn43 puncta (n = 18 DRG, 6 animals).This increased to an average of 1.000 ± 0.845 Cxn43 puncta per DRG at 3 dpf (n = 29 DRG, 10 animals) and by 4 dpf, there was a significant increase in the number of Cxn43 puncta present in the DRG with an average of 2.208 ± 1.062 Cxn43 puncta per DRG (n = 24 DRG, 8 animals) (2 dpf versus 4 dpf: p < 0.0001, 3 dpf versus 4 dpf: p < 0.0001 post hoc Tukey test) (Fig 1O and 1P).These results support the hypothesis that DRG cells begin forming glial connections during its earliest construction.
To determine if there are functional gap junction connections, we treated animals expressing Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s); Tg(neurod:tagRFP) with either carbenoxolone (CBX), a gap junction inhibitor, or a control treatment of DMSO.Animals treated with CBX at 3 dpf demonstrated a significant decrease in the percent of high correlation coefficients with an average percent of 22.00% ± 22.34% (n = 34 cells, 5 DRG, 3 animals) compared to an average percent of high correlation coefficients of 36.58% ± 31.22% when treated with a DMSO control (n = 26 cells, 7 DRG, 4 animals) (DMSO versus CBX: p = 0.0391 unpaired t test) (Fig 1Q).These results strongly support the idea that functional gap junctions are present in glial networks in DRG during its early construction.

Satellite glia Ca 2+ transients are impacted by altering mechanobiology
Our measurements indicated that DRG satellite glia cells demonstrate distinct Ca 2+ transients.To identify potential molecular components involved in these Ca 2+ transients, we performed a chemical screen targeting various chemical signals shown to affect Ca 2+ transients using transgenic animals expressing Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s); Tg(neurod:tagRFP) (Fig 1R and 1S).Additionally, we included a broad-mechanosensitive ion channel antagonist, GsMTx4, because of the underappreciated role that mechanobiology has during neurodevelopment.We hypothesized that GsMTx4 would reduce the amount of observed Ca 2+ transients if mechanobiology had an important role during early development.Each animal was exposed to the pharmacological agent 30 min prior and during the imaging window and then GCaMP6s intensity was measured for 1 h with a 15-s imaging interval.We reasoned that an overall change in the abundance of Ca 2+ transients could help us identify molecules that are important for either isolated or simultaneous spontaneous Ca 2+ transients.We found that GsMTx4 significantly reduced the amount of Ca 2+ transients observed compared to DMSO (Fig 1S) (DMSO versus GsMTx4: p < 0.0001 post hoc Dunnett test).We also measured a significant change following treatment with Thapsigargin (Thaps) (Fig 1S) (DMSO versus Thaps: p = 0.0010 post hoc Dunnett test).While chemical signaling has been widely described in spontaneous Ca 2+ transients, the role of mechanobiology in the process is less known, which led us to investigate the potential role of mechanobiology in spontaneous Ca 2+ transients in the DRG.
To first explore the possibility that mechanical features impact spontaneous Ca 2+ transients in the DRG, we tested if the cells in the developing DRG are sensitive to mechanical perturbation.To do this, we imaged the DRG of transgenic zebrafish expressing GCaMP6s in satellite glial Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s) cells during tissue compression (Fig 2A).Tissue compression was administered by bending the animal with a microneedle as they were imaged on the confocal microscope (Fig 2B).At 2 dpf, 57% of DRG expressing Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s) responded to tissue compression (n = 7 DRG, 7 animals) (Fig 2C and 2D).In order to better understand how much compression was needed, we measured the distance that the animal was compressed.This compression impacted not just the DRG itself, but also all surrounding tissue.We found that the average amount of compression needed to elicit a response in the DRG was 207.8 μM (n = 16 DRG, 16 fish) (Fig 2H).There was notable (P) Quantification of the average number of Ca 2+ events per sox10 + cell following genetic manipulation via injection of uas:cas9mkate-u6:piezo1gRNA or uas:cas9mkate-u6:emptygRNA into animals expressing Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s) at 3 dpf.Additionally, a group of Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s) injected with uas:cas9mkate-u6:piezo1gRNA and treated with 40 μM Jedi2 treatment was also variability in this compression assay.This was due to variability in the amount of force, size of the tip of the needle, positioning of the animal, and angle that the microneedle was positioned.It is possible that this response of sox10 + cells was secondary to neuronal firing.We, therefore, tested if neurons fired in response to compression at 2 dpf in Tg(neurod:gal4+myl7); Tg(uas: GCaMP6s) animals but could not detect Ca 2+ transients in neurons after compression (n = 5 DRG, 5 animals) (Fig 2C).Examining DRG axonal projections in Tg(ngn1:GFP) animals also showed that neurons at 2 dpf did not have peripheral axons at their final targets in the periphery (Fig 2G).It therefore seems unlikely that such Ca 2+ transients in sox10 + satellite glia after tissue compression are secondary to neuronal activity.To understand if sox10 + satellite glia continued to be sensitive to mechanical compression, we repeated this assay at 3 dpf.By 3 dpf, 82% (n = 11 DRG, 11 animals) of DRG expressing Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s) responded to tissue compression (Fig 2C).At 3 dpf, 100% (n = 5 DRG, 5 animals) of DRG expressing Tg(neurod:gal4+myl7); Tg(uas:GCaMP6s) also demonstrated Ca 2+ transients after tissue compression (Fig 2C).While the neuronal population of the DRG does respond to tissue compression at a later age, our data suggests satellite glia respond to the mechanical tissue compression at early ages without neuronal activation.
If this response to mechanical force is mediated by mechanosensitive ion channels, we would hypothesize that it would be reduced upon treatment of GsMTx4, which broadly blocks mechanosensitive ion channels.To test this hypothesis, we imaged animals expressing either Tg(sox10:gal4+myl7); Tg(uas:GCaMP6s); Tg(neurod:tagRFP), or Tg(neurod:gal4+myl7); Tg (uas:GCaMP6s) that were treated with GsMTx4.We found that treatment with GsMTx4 reduced the response to mechanical stimuli to 20% of animals (n = 5 DRG, 5 animals) expressing sox10 + GCaMP6s at 2 dpf (Fig 2C and 2E).At 3 dpf when treated with GsMTx4, there was a significant reduction in response with 0% of animals (n = 4 DRG, 4 animals) expressing sox10 + GCaMP6s responded to mechanical force and 25% of animals (n = 4 DRG, 4 animals) expressing neuronal GCaMP6s responded to mechanical force (sox10 3 dpf DMSO versus GsMTx4: p = 0.0110, neurod 3 dpf DMSO versus GsMTx4: p = 0.0476 Fisher's exact test) (Fig 2C).Additionally, we investigated the effect of GsMTx4 treatment on 3 phases of this assay.We assessed the change in fluorescence in DRG expressing Tg(sox10:gal4); Tg(uas: GCaMP6s) during the resting state, compression state, and decompression state (Fig 2E).We define these phases as follows: resting phase is the initial fluorescence before compression with the needle, compression phase is when the needle is actively putting force on the animal, and decompression phase is when the needle has been released.We compared the average change in fluorescence during these phases between DMSO and GsMTx4-treated animals at 3 dpf expressing Tg(sox10:gal4); Tg(uas:GCaMP6s).We found that animals treated with DMSO had an average change in fluorescence of 0.003+/−0.006during resting phase, an average change in fluorescence of 0.010+/−0.010during compression, and an average change in fluorescence of 0.029+/−0.019during decompression (n = 9 DRG, 9 animals).We found that animals treated with GsMTx4 had an average change in fluorescence of 0.006+/−0.Decompression DMSO versus Decompression GsMTx4: p = 0.0025, multiple unpaired t tests) (Fig 2F).These data support the hypothesis that DRG are responsive to mechanical forces and identify that sox10 + cells are mechanosensitive, at least partially independent of neuronal activity.

Satellite glia Ca 2+ transients can be altered by manipulating Piezo1
We next explored the potential molecular determinant of this mechanical component.The mature DRG is known to express mechanosensitive channels Piezo1 and Piezo2; however, Piezo2 is restricted to neurons while Piezo1 is expressed in neurons and satellite glia in mice [31].To investigate this in zebrafish, we utilized RNAscope to determine spatiotemporal distribution of piezo1 RNA in animals expressing Tg(sox10:meGFP).We found that 79% of DRG (n = 24 DRG, 8 animals) at 3 dpf contained piezo1 RNAscope puncta within sox10 + satellite glia (Fig 2I and 2J).Additionally, we utilized Whole-mount HCR-FISH targeting piezo1 RNA 3 dpf animals expressing Tg(sox10:meGFP) and found similar expression of piezo1 (S4 Fig).
One possible explanation for an increase in Ca 2+ transients in sox10 + satellite glia is that the sox10 + satellite glia are active in response to neuronal activity.To investigate whether the observed change in Ca 2+ transients in sox10 + satellite glia was a consequence of altered neuronal activity, we treated animals expressing Tg(neurod:gal4+myl7); Tg(uas:GCaMP6s) with Piezo1 agonists and quantified the amount of Ca 2+ transients per DRG neuron.We found following DMSO treatment that DRG neurons exhibited an average of 2.42 ± 1.71 Ca 2+ transient events per hour.When animals were treated with either Yoda1 or Jedi2, an average of 3.50 ± 2.39 or 2.57 ± 2.00 Ca 2+ transient events per hour, respectively, could be detected.When animals were treated with GsMTx4, there was an observed 3.75 ± 1.67 average number of Ca 2+ transient events (DMSO n = 33 neurons, 24 DRG, 10 animals, Yoda1 n = 8 neurons, 7 DRG, 4 animals, Jedi2 n = 35 neurons, 18 DRG, 5 animals, GsMTx4 n = 8 neurons, 8 DRG, 4 animals) (Fig 2O).Overall, we found that Piezo1 agonists did not contribute to an increase in Ca 2+ transients in the neurod + population.These data are most consistent with the hypothesis that sox10 + satellite glia display Ca 2+ transients in response to Piezo1 agonists independent of an increase in neuronal activity.
We also quantified the number of Ca 2+ microdomains following manipulations of Piezo1 in Tg(sox10:gal4+myl7) animals injected with uas:GCaMP6s-caax.In DMSO-treated animals, sox10 + cells displayed an average number of 0.14 ± 0.38 Ca 2+ microdomains at 3 dpf.Animals treated with Jedi2 displayed an average number of 1.38 ± 0.92 Ca 2+ microdomains at 3 dpf.When animals were treated with GsMTx4, we observed an average of 0.14 ± 0.38 Ca 2+ microdomains at 3 dpf (DMSO n = 7 DRG, 7 animals, Jedi2 n = 8 DRG, 6 animals, GsMTx4 n = 7 DRG, 7 animals) (Fig 3G).Similar to isolated Ca 2+ transients, we found a significant increase in the average number of Ca 2+ microdomains per DRG when animals were treated with a Piezo1 agonist (DMSO versus Jedi2: p = 0.0025 post hoc Dunnett test) (Fig 3G).We also quantified the average duration of the identified Ca 2+ microdomains and did not find a significant difference (Fig 3H).These additional findings suggest that Piezo1-mediated mechanical forces contribute to the number of observable Ca 2+ microdomain events in addition to isolated Ca 2+ transient events.

Altering Piezo1 has functional consequences to DRG development
The specific function of isolated Ca 2+ transients is relatively unknown.We therefore used Piezo1 manipulations to test the potential functional consequence of increasing isolated Ca 2+ transients.We first hypothesized that Piezo1-sensitive isolated Ca 2+ transients were important for the formation of synchronized glial networks.These glial networks form between 2 and 4 dpf (Fig 1N).Therefore, to first test this hypothesis, we treated animals expressing Tg(sox10: gal4+myl7); Tg(uas:GCaMP6s); Tg(neurod:tagRFP) with either DMSO, GsMTx4, or Jedi2 for 30 min daily at 2 and 3 dpf.We performed our Ca 2+ imaging paradigm on these animals at 4 dpf and first assessed the amount of isolated and spontaneous Ca 2+ transient events following consecutive days of treatment.Following consecutive days treated with DMSO, sox10 + cells showed an average of 2.06 ± 1.85 simultaneous Ca 2+ transients and an average of 0.63 ± 0.75 isolated Ca 2+ transients.Following treatment on consecutive days with Jedi2, sox10 + cells showed an average of 1.56 ± 1.19 simultaneous Ca 2+ transients and an average of 1.52 ± 1.85 isolated Ca 2+ transients.Taken together, these data confirm that following consecutive days of treatment, Jedi2 significantly increases the amount of isolated Ca 2+ transients (DMSO versus Jedi2: p = 0.0132 post hoc Dunnett test).Interestingly, after treatment with the broad-mechanosensitive antagonist, GsMTx4, on consecutive days in development, sox10 + cells displayed an average of 0.86 ± 0.56 simultaneous Ca 2+ transient events and an average of 0.68 ± 0.72 isolated Ca 2+ transient events, significantly reducing the amount of simultaneous Ca 2+ transients (DMSO versus GsMTx4: p = 0.0050 post hoc Dunnett test) (DMSO n = 32 cells, 6 DRG, 4 animals, Jedi2 n = 25 cells, 5 DRG, 4 animals, GsMTx4 n = 22 cells, 4 DRG, 3 animals) (Fig 4D and 4E).To answer whether these changes ultimately impacted synchrony, we quantified the average percent of high correlation coefficients per sox10 + cell following these treatment paradigms.We found that when treated with DMSO, sox10 + cells had an average of 35.88 ± 28.78% of high correlation coefficients.When treated with either GsMTx4 or Jedi2, sox10 + cells had an average of 29.09 ± 31.65% high correlation coefficients or an average of 24.48 ± 27.16% high correlation coefficients, respectively, (DMSO n = 32 cells, 6 DRG, 4 animals, GsMTx4 n = 22 cells, 4 DRG, 3 animals, Jedi2 n = 25 cells, 5 DRG, 5 animals) (Fig 4F).There was no significant change in the average percent of high correlation coefficients when treated with Jedi2 or GsMTx4 and thereby inconsistent with the idea that Piezo1-sensitive isolated Ca 2+ transients or Ca 2+ microdomains impact the formation of a presumptive glial network, at least as identified by synchronous activity.
In addition to the synchronous glial networks that we identify form by 4 dpf, it is also known that the DRG rapidly expands during this developmental time [38].We, therefore, tested the hypothesis that Piezo1-sensitive events like isolated Ca 2+ transients could be important for DRG expansion.To do this, we increased isolated Ca 2+ transients via Piezo1 through Jedi2 or Yoda1 treatment and assessed the number of cells present in the DRG at 5 dpf.To assay both neuronal and glial expansion, we treated zebrafish expressing Tg(neurod:tagRFP) with Piezo1 agonists or a mechanosensitive antagonist for 30 min each day from 2 to 4 dpf, and then used immunohistochemistry against Sox10, thereby identifying TagRFP + neurons and Sox10 + satellite glia at 5 dpf.Animals treated with DMSO on consecutive days displayed an average number of 3.85 ± 1.05 neurod + cells and an average number of 5.09 ± 1.33 Sox10 + cells.When animals were treated with GsMTx4, DRG also contained an average of 4.18 ± 1.22 neurod + cells and 4.91 ± 0.99 Sox10 + cells.However, treatment with Jedi2, resulted in DRG with an average of 2.23 ± 0.81 neurod + cells and average number of 3.36 ± 0.81 Sox10 + cells (Fig 4I and 4J), showing a reduction in the amount of DRG cells (DMSO versus Jedi2: p < 0.0001 post hoc Dunnett test) (DMSO n = 53 DRG, 15 animals, GsMTx4 n = 34 DRG, 10 animals, Jedi2 n = 39 DRG, 10 animals).With treatment with Yoda1, DRG (n = 18 DRG, 7 animals) also displayed a significant reduction in DRG cells, with an average number of 2.50 ± 1.04 neurod + cells (DMSO versus Yoda1: p < 0.0001, post hoc Dunnett test).To rule out the possibility that the change in DRG cell number after pharmacological treatments was from a nonspecific change in the size of the entire animal, we repeated this assay and measured the length of animals at 5 dpf and found no significant difference (S6 Fig) .Taken together, these findings identify that altering Piezo1-sensitive isolated or microdomain Ca 2+ transients impacts DRG development through a reduction in the number of cells present.
To investigate whether this change in cell abundance was a consequence of increased Ca 2+ transients nonspecific to Piezo1-mediated activity, we utilized an optogenetic approach to increase Ca 2+ transients in the DRG.We injected uas:Chr2-tdTomato into animals expressing Tg(sox10:gal4); Tg(uas:GCaMP6s) and exposed the animals to 488 nm light for 30 min each day from 2 dpf until 4 dpf (Fig 4K) [51].Prior to this experiment, we verified that the uas: Chr2-tdTomato injected construct would increase Ca 2+ transients by exposing animals over a short period of time to 488 nm light and quantifying the change in fluorescent intensity (S7 Fig) .Following the 488 nm exposure from 2 to 4 dpf, we fixed the animals at 5 dpf and stained for the neuronal marker HuCD (Fig 4L).We then quantified the number of HuCD + cells and sox10 + only cells to compare the number of both populations to a group of injected animals that were not exposed to 488 nm light during development.In animals exposed to 488 nm light, we found an average of 2.556+/−0.527HuCD + neurons and an average of 3.778 +/−0.667sox10 + satellite glia (n = 9 DRG, 4 animals).The animals that were not exposed to 488 nm light during development had an average number of 3.000 HuCD + neurons and an average of 3.500+/−1.000sox10 + satellite glia (n = 4 DRG, 4 animals).We found no significant difference in the number of neurons or satellite glia present in the DRG following after Chr2 manipulation (Fig 4K -4N).These findings support the hypothesis that increasing Piezo1-mediated isolated Ca 2+ transients impacts the DRG via lowering cell abundance.
It is possible that this decrease in cell abundance was caused by a decrease in cell divisions or from an increase in cell death.In order to understand the cause of the decrease in cell abundance, we assessed the number of cell divisions and cell death occurring following consecutive treatment with Piezo1 agonists.To do this, we treated animals expressing Tg(sox10:meGFP) from 2 to 4 dpf with Jedi2 or a DMSO control.We then utilized overnight time lapse imaging to assay cell divisions or cell death.When treated with DMSO at 2 dpf, 88.24% of DRG had cell divisions (n = 17 DRG, 6 animals).At 3 dpf following DMSO treatment, cell divisions occurred in 70.83% of DRG (n = 24 DRG, 8 animals).Following treatment of Jedi2 at 2 dpf, 67.74% of DRG had cell divisions (n = 31 DRG, 8 animals).At 3 dpf following consecutive treatment with Jedi2, 33.33% of DRG had cell divisions (n = 24 DRG, 8 animals) (Fig 4O).We found a significant decrease in the number of cell divisions following consecutive treatment of Jedi2 at 2 and 3 dpf (DMSO versus Jedi2 3 dpf: p = 0.0199 Fisher's exact test) (Fig 4O).However, we did not observe a significant change in the number of observed cell deaths (Fig 4P).The most likely explanation for this data is that the decrease in cell abundance in Piezo1-manipulated animals is from a reduction in cell divisions.
Lastly, we questioned whether this Piezo1-mediated decrease in cell abundance had functional consequences to the animal's physiology.To answer this question, we treated animals with Piezo1 agonists and mechanical-channel antagonists during development and then tested the animal's response to sensory stimuli.We previously demonstrated that larval zebrafish DRG neurons are active after zebrafish larvae are submerged in 4˚C water [52,53].This submersion causes a shivering phenotype that is at least partially dependent on intact DRG neurons and axons [52,53].We treated animals with Jedi2 or GsMTx4 for 30 min each day from 2 to 4 dpf and then assayed sensory responses at 5 dpf.Following consecutive days of treatment, DMSO control-treated animals had a shivering phenotype average duration of 11.60 ± 5.9 s.When treated on consecutive days with GsMTx4, animals had an average length of shivering of 6.55 ± 6.97 s.Following consecutive days of treatment with Jedi2, animals had an average length of shivering of 6.47 ± 6.01 s.There was a significant decrease in the length of shivering following consecutive days of Jedi2 treatment suggesting that overactivation of Piezo1 during development impacts DRG response to sensory stimulus (DMSO versus Jedi2: p = 0.0427 unpaired t test) (DMSO n = 12 animals, GsMTx4 n = 11 animals, Jedi2 n = 13 animals) (Fig 4Q).The noted nonsignificant decrease to the average length of shivering following GsMTx4 may be due in part from unidentified pathways targeted with this broad mechanosensitive ion channel antagonist.Overall, these results support the hypothesis that impacting DRG development via Piezo1-mediated isolated Ca 2+ transients results in a functional consequence to the animal's physiology.

Discussion
Activity of neural cells during development is well documented.We define this activity in satellite glia as significant changes in Ca 2+ transients, which is a separate and different process from known neuronal firing activity.This can be broadly categorized into evoked and spontaneous Ca 2+ transients.In glia, spontaneous Ca 2+ transients are further divided into subtypes characterized as whole cell and microdomain Ca 2+ transients.However, the developmental, molecular, and functional features of these glial Ca 2+ transients merits more investigation.Here, we demonstrate that satellite glia in the DRG exhibit distinct subtypes of spontaneous Ca 2+ transients during early developmental times.We further reveal that distinct subtypes of Ca 2+ transients are sensitive to manipulation of Piezo1.The functional consequence of disrupting such Piezo1-sensitive events is supported by data that shows cell abundance and sensory behavior is impacted by Piezo1-manipulations. Overall, we reveal developmental, molecular, and functional characteristics of glial Ca 2+ transients in the DRG.
Despite clear roles of neural Ca 2+ transients in development and homeostasis of the nervous system in the animal, it is unclear when and which cells in the DRG show spontaneous Ca 2+ transients.It is well appreciated that cultured DRG neurons exhibit spontaneous Ca 2+ transients [54,55].Ca 2+ reporters have also demonstrated that satellite glia demonstrate Ca 2+ transients in culture [56,57].These cultured satellite glia also exhibit synchronized Ca 2+ transients.Recent work has also shown Ca 2+ transients in the vertebrate DRG neurons in vivo [58,59].However, such work was restricted to mature animals.Our work reveals that both DRG neurons and glia in the animal display Ca 2+ transients during the earliest stages of DRG construction.Even on the first day of genesis, DRG cells demonstrate Ca 2+ transients.We also identify the molecular mechanisms that mediate some of these Ca 2+ transients.What remains unknown is how these Ca 2+ transients change as the animal approaches adulthood or in neuropathologies.These are important topics to study because we know altered activity of both glia and neurons has been implicated in neuropathologies.Further, in addition to the spontaneous Ca 2+ transients we focus on, neurons in the DRG also are evoked by specific stimuli.How glia respond to evoked stimulation in the animal is almost entirely unknown.
By imaging GCaMP6s in sox10 + cells and probing gap junction components, we reveal that synchronized cellular Ca 2+ transients indicative of glial networks form within 3 days of DRG genesis.We know that glial networks are essential in the central nervous system for circuit formation, neuronal health, and signal transduction.However, glial networks in the PNS are less understood.Because of this, whether glial networks exist in developing DRG was not known.Our work identifies that DRG satellite glia are not synchronized initially, but by 4 days of DRG construction, become synchronized.One interesting aspect of this increased synchronization is that it occurs while the population is simultaneously expanding via cell divisions.Further investigation in additional PNS populations will provide an understanding if this is a unique process to the DRG or if it is found in additional areas of the PNS.It will also be important to probe the plasticity of the synchronized glial network and how it could be altered.
While most currently published research has focused on identifying molecules involved in synchronous Ca 2+ transients of cells that make up neural circuits, we have identified a mechanism that contributes specifically to what we define as isolated Ca 2+ transients within the DRG.We found that mechanobiology via Piezo1 contributes to an increase in isolated Ca 2+ transients without altering simultaneous Ca 2+ transients.Our findings provide insights into the importance of understanding mechanical forces on the cellular level during development.Recent work has brought the ability to visualize Piezo1 activity into fruition using GenEPi [60].Using this new toolset would provide great insight into the roles of Piezo, while identifying when and where it is active.GenEPi will allow direct inquisition into the activity of native Piezo1.It is possible that these mechanical forces provide insight to satellite glia regarding whether proliferation is needed.For example, if the mechanical forces acting on a satellite glia are high, this may cause signaling via Piezo1 to halt or promote proliferation [29,35,61].Alternatively, the ability to sense larger mechanical forces on the level of tissues may be important for DRG expansion.If this were true, an increase in mechanical forces may signal to the DRG that there is no room for further proliferation.One potential signaling cascade that may be involved is YAP/Taz.YAP/Taz is a well-known controller of cellular proliferation and has also been shown to modulate DRG development [13].So, it is possible that over activating Piezo1 is altering localization of Yap/Taz [62].Another area of research that would impact these ideas is the utility of Ca 2+ microdomains observed in the DRG.If these microdomains are indicative of mechanical forces on the subcellular level, we may hypothesize either an increase or decrease in the amount of these microdomains which would further inform proliferative decisions in DRG satellite glia.Further investigation into the mechanical forces acting on the microenvironment of the DRG needs to be completed.
Our findings highlight the importance of understanding the role of mechanical signals in PNS development and the function of distinct Ca 2+ transients in that process.

Limitations of findings
We found a transition from an asynchronous population to a synchronous population during early DRG development.We attribute this transition to an increase in correlation coefficients between cells present in the DRG, which we hypothesize is a result from increased gap junction formation.But, this transition may also be partially explained by other processes.This work only investigates Ca 2+ transients in the first 3 days of DRG genesis, so whether the observed synchrony remains through later stages of development is unknown and merits further study.We currently hypothesize that the decrease in cell proliferation is a result of Piezo1 overactivation.Whether Piezo1 controls proliferation only through isolated Ca 2+ transients cannot be determined with our experiments.Our pharmacological manipulations also do not distinguish between cell-autonomous and non-autonomous roles of Piezo1 in satellite glia.Because of this, we cannot rule out the possibility that some observed phenotypes are a result of nonautonomous signaling.However, the piezo1 expression in DRG sox10 + cells and response to Piezo1 agonists, as well as our genetic manipulations, suggests a cell-autonomous role.But it remains a possibility that over activating Piezo1 could result in a change in a cell's ability to communicate unidentified signals to surrounding cells.If this is the case, then non-autonomous roles of Piezo1 could contribute to the reduction in cell abundance.We use an established cell-specific gRNA approach and validate that the gRNA creates indels; however, a caveat is that the nature of those lesions or the portion of cells that have homozygous mutations is not known.Further probing the role of Piezo1 in germline mutants would validate those findings, although, in mice global germline mutations to Piezo1 are embryonic lethal [63].Throughout this work, we utilize GsMTx4 as a Piezo1 antagonist alongside our other pharmacological treatments.Because GsMTx4 is more broadly inhibitive of mechanosensitive ion channels, further work is needed to test antagonistic effects on Piezo1.We hypothesize that further work utilizing more specific Piezo1 antagonists will complement the data shown here.Our findings support the hypothesis that DRG satellite glia are mechanosensitive, but to further understand the amount of force needed to elicit a response in these cells and the true dynamics of the cells during a response will need additional studies targeting mechanical properties.Lastly, we identify satellite glia as sox10 + neurod -cells residing in the DRG with an ensheathing morphology.It is likely that a subset of progenitor cells exist in the DRG during this developmental time period and potentially into adulthood [37].If this is the case, our findings would suggest that altering Piezo1-mediated Ca 2+ transients in both satellite glia and DRG progenitors likely impact DRG development.

Ethics statement
Experimental procedures adhered to the NIH guide for the care and use of laboratory animals.All experiments were approved by the University of Notre Dame Institutional Animal Care and Use Committee (IACUC) (protocol 19-08-5464) which is guided by the United States Department of Agriculture, the Animal Welfare Act (USA) and the Assessment and Accreditation of Laboratory Animal Care International.
In vivo overnight imaging.Animals were anesthetized using veterinary grade 3-aminobenzoic acid ester (Tricaine) for mounting purposes only.Animals were then mounted laterally on their right side in glass-bottomed 35-mm petri dishes [38] and covered in 0.8% low melt agarose.For overnight time lapse imaging, a mixture of egg water and tricaine was added to the dish.Images were acquired on spinning disk confocal microscopes custom built by 3i technology (Denver, Colorado) that contains: Zeiss Axio Observer Z1 Advanced Mariana Microscope, X-cite 120LED White Light LED System, filter cubes for GFP and mRFP, a motorized X,Y stage, piezo Z stage, 20× Air (0.50 NA), 63× (1.15NA), 40× (1.1NA) objectives, CSU-W1 T2 Spinning Disk Confocal Head (50 μM) with 1× camera adapter, and an iXon3 1Kx1K EMCCD camera or Prime 95B back illuminated CMOS camera, dichroic mirrors for 446, 515, 561, 405, 488, 561, 640 excitation, laser stack with 405 nm, 445 nm, 488 nm, 561 nm, and 637 nm.Overnight time-lapse images were collected every 5 min for 24 h capturing a 40 μM z stack.Adobe Illustrator and ImageJ were used to process images.Only brightness and contrast were adjusted and enhanced for images represented in this study.
In vivo calcium imaging.Animals were anesthetized using veterinary grade 3-aminobenzoic acid ester (Tricaine) for mounting purposes only.Animals were then mounted laterally on their right side in glass-bottomed 35-mm petri dishes [38] and covered in 0.8% low melt agarose.For Ca 2+ imaging, egg water was added to the dish with no Tricaine.Images were acquired on a spinning disk confocal microscope custom built by 3i technology (Denver, Colorado) microscopes.Ca 2+ time-lapse imaging consisted of image collection every 15 s for 1 h capturing a 40 μM z stack.For imaging of Ca 2+ microdomains, images were either taken every 5 s for 10 min capturing a 20 μM z stack (Fig 1 ) or taken every 200 ms in a single plane (Fig 3).Adobe Illustrator and ImageJ were used to process images.Only brightness and contrast were adjusted and enhanced for images represented in this study.
Consecutive treatments of pharmacological agents.For consecutive days of treatment with pharmacological treatments, animals were bathed in a mixture of either 40 μM Jedi2, 1 μM GsMTx4, or 100 μM Yoda1 in egg water with 2% DMSO for 30 min each day.For  and 4L).Animals were then washed for 5 min in a series of PBSt, DWt (dH 2 O, 0.1% TritonX-100), and acetone.Following these washes, animals were placed in acetone at −20˚C for 10 min.This was then followed by 3 washes of PBSt for 5 min each.Animals were then placed in 5% goat serum in PBSt for a 1 h incubation period.Animals were then incubated in 5% goat serum in PBSt with primary antibody for 1 h at room temperature followed by an overnight incubation at 4˚C.This was followed by 3 consecutive 30 min washes of PBSt and one 1 h wash in PBSt.Animals were then placed in 5% goat serum in PBSt with the secondary antibody for 1 h at room temperature followed by an overnight incubation at 4˚C.This was then followed by 3 consecutive 1 h washes of PBSt.Animals were then stored in 50% glycerol 50% PBS at 4˚C until imaging.
Whole-mount RNAscope.Animals expressing Tg(sox10:meGFP) were fixed at 2 dpf with 4% PFA in PBS for 30 min.Following fixation animals were placed in new Eppendorf tubes and washed with 25%, 50%, 100% methanol for 10 min each.Animals were then kept in 100% methanol at −20˚C overnight.This was followed by a 5-min wash of 50% methanol in PBStw (PBS, 0.1% Tween-20) and an additional 5-min wash of 25% methanol in PBStw.Liquid was removed from the Eppendorf tubes and the animals were air dried for 30 min.This was followed by two 5-min washes with PBStw.Animals were permeabilized with 10 μg/mL proteinase K in PBStw at room temperature for 6 min.This was then followed by 4 consecutive 10-min washes of PBStw.Following removal of PBStw, 2 drops of ACD probes targeting piezo1 were then added to the sample, which was then incubated at 40˚C for 15 h (1:50, 80 μL, C1, ACD).In the case of positive and negative controls that were used, Probe-Dr-polr2 was used as a ubiquitous positive control and Probe-Dr-dapB was used as a bacterial negative control.After this incubation period animals were then washed with SSCtw (5× saline-sodium citrate buffer, 0.1% Tween-20) for 10 min at room temperature twice.An additional fixation was then done in 4% PFA in PBS at room temperature for 10 min.Animals were again transferred to a new Eppendorf tube and washed 3 times in SSCtw.Animals were then incubated in a series of 2 drops Amp1 at 40˚C for 30 min, 2 drops Amp2 at 40˚C for 30 min, 2 drops Amp3 at 40˚C for 15 min, 2 drops HRP-C1 at 40˚C for 30 min, opal fluorophore 650 (1:500) in PBStw at 40˚C for 30 min, and 2 drops of Multiplex FLv2 HRP blocker at 40˚C for 30 min.Between each of these incubation periods, animals were washed twice with SSCtw.Following this protocol, animals were then processed following the Whole-mount immunohistochemistry protocol to target GFP.
Whole-mount HCR-FISH.Animals expressing Tg(sox10:megfp) were fixed at 3 dpf in 4% PFA in PBS for 24 h at 4˚C.Fixed larvae were then washed in PBS 3 times for 5 min each.To dehydrate and permeabilize the tissue, samples were then washed in a series of 100% methanol 4 times for 10 min each.Samples were then stored in 100% methanol at −20˚C for 24 h.To rehydrate samples, a series of methanol/PBStw washes were utilized.Samples were washed for 5 min in 75% methanol in PBStw, 50% methanol in PBStw, 25% methanol in PBStw, and 100% PBStw.Samples were then treated with 10 μg/mL proteinase K for 30 min.Samples were then washed in PBStw twice for 5 min.A postfix was done on samples in 4% PFA for an additional 20 min.Samples were then washed 5 times for 5 min each with PBStw.Samples were then washed in probe hybridization buffer (Molecular Instruments) for 30 min at 37˚C.Samples were then incubated at 37˚C overnight in probe hybridization buffer containing HCR-FISH probes targeting piezo1 (Molecular Instruments).Samples were washed following incubation with probe wash buffer (Molecular Instruments) at 37˚C 4 times for 15 min each.Samples were then washed twice with SSCtw for 5 min each.Following these washes, samples were then incubated in amplification buffer (Molecular Instruments) for 30 min.Following this incubation, samples were then incubated overnight in amplification buffer containing hairpin b1h1 and hairpin b1h2 (647) overnight.Samples were then washed 5 times in PBStw for 20 min each.Following this protocol, samples then underwent the Whole-mount immunohistochemistry protocol targeting GFP.
Animal behavior in cold stimulus.Animals were treated at 2, 3, and 4 dpf with 1 μM GsMTx4 in 2% DMSO, 40 μM Jedi2 in 2% DMSO, or 2% DMSO in egg water for 30 min each day.Aside from this treatment protocol, animals were raised under normal procedures in 28˚C egg water.Each animal was then taken at 5 dpf and placed in 4˚C egg water for 30 s. Video recordings were recorded with 40 ms exposure to bright field white light.The initial 3 s of the movies were not quantified to allow the animal to be placed into the stimulus and for any adjustments to occur.The duration of the shivering was quantified per second starting from the initial shivering phenotype to the cold water stimulus until there was at least 1 s of no shivering.
Mechanical compression assay.To apply mechanical force to the DRG, we dorsally mounted animals in 0.8% low melt agarose and placed them on the stage of a spinning disk confocal.On either side of the stage 2 vertical stainless steel rods were mounted onto the air table.A horizontal rod was mounted above the stage with a micromanipulator attached.A glass needle filled with dextran was mounted in the micromanipulator above the animals with the needle pointing toward the animal.Prior to bringing into contact with the animal, the needle was calibrated in the X and Y positions in relation to the middle of the imaging window.The needle was slowly brought into contact with the skin of the animals and was then tapped to apply pressure to the DRG and surrounding tissue.To understand if DRG were active in response to the mechanical forces, we quantified GCaMP6s transients as previously described.As a proxy to understand how much force was needed to activate DRG satellite glia, we quantified the response of DRG to a gradient of force measured by the distance (μM) that the tissue was displaced with animal bending in the XY direction.
Global CRISPR Cas9.In order to validate that our chosen piezo1 gRNA (ACACCCTGA GGATCTTCCAG) was sufficient at creating indels, we injected the chosen piezo1 gRNA with Cas9 (AltR-Cas9 nuclease V3) into embryos at the 1-cell stage.The gRNA was annealed with tracrRNA, and the annealed product was incubated with Cas9 (25 μM) at 37˚C for 5 min prior to injection.The injection mix also included phenol red for visualization.After 24 h, animals were collected for DNA isolation and Sanger sequencing.
Assessment of animal length.Following consecutive days of pharmacological treatment (see method above), animals were imaged at 10.4× magnification with 1 ms of exposure in brightfield to assess the size of the animal.Length in millimeters was used to quantify the size of the animal following consecutive days of pharmacological treatment.

Quantifications and statistical tests
Quantification of GCaMP6s transients.Before quantifying changes in GCaMP6s intensity, we corrected for motion drift by utilizing the Template Matching plugin in ImageJ.We then traced individual cells in the DRG to create regions of interest (ROI).The integrated density of fluorescence was quantified at every time point for each ROI.We then quantified the z score for each time point for each ROI.Time points with a z score of 2.58 or greater were considered active time points.A single activity event was identified by time points with a z score of 2.58 or greater.If consecutive time points were a z score of 2.58 or greater, this was still considered 1 activity event.This process was used to create activity profiles for each ROI in a given DRG.The average number of active time points was calculated and compared to controls via t tests to determine significance.
Quantification of GCaMP6s transients in pilot screen.The number of GCaMP6s transients in the pilot screen were quantified in ImageJ.The number of visual changes in GCaMP6s intensity were quantified for each movie.These quantifications were done per whole DRG identified in the imaging window.
Generation of line graphs.Line graphs were generated using the data obtained from the quantification of GCaMP6s transients.Each cell in the DRG had GCaMP6s transients quantified.The X axis of the line graph is the 1-h period and the Y axis is the Z score.Line graphs were generated in Prism.
Generation of heatmaps.Heatmaps were generated in Prism.Each row of the heatmap corresponds to an individual cell's GCaMP6s transience during the 1 h of imaging.Blue colors indicate low z scores and yellow colors indicate high z scores (>2.58).
Quantification of correlation coefficients.Activity profiles from individual cells were converted to a binary system.Active points where the z score was greater than 2.58 were listed as 1 and inactive time points where the z score was less than 2.58 were listed as 0. These binary activity profiles were then used to quantify the correlation between individual ROIs found in the same DRG; we utilized the cor() function in R to complete this analysis.The results were then used to create a correlation table in R.These functions were part of the Hmisc package in R. Correlation coefficients greater than 0.5 were considered high correlation coefficients.The percent of high correlation coefficients were then quantified per cell and compared with controls via t tests to determine significance.
Quantification of IHC, RNAscope, HCR-FISH.All quantifications of IHC were completed in ImageJ.The number of Cxn43 puncta that coincided with sox10:megfp expression were quantified per DRG (Fig 1O and 1P).The number of DRG that contained piezo1 within the stained GFP expressed was quantified (Figs 2H, 2I, and S4).This was done by going through each individual optical slice to identify piezo1 puncta within sox10 + cells located in the DRG.In addition to location, puncta were at least 0.45 μM 2 in size for all quantifications.The number of Sox10 + cells were quantified per DRG (Fig 4H and 4I).The same approximate location was imaged on multiple fish, capturing between 2 and 4 DRG per fish.The number of HuCD + cells was quantified per DRG (Fig 4M and 4N).
Quantification of mechanical compression assay.Response to mechanical compression was quantified in ImageJ.DRG were traced and the integrated density of fluorescence was quantified at each time point.To attempt quantifying only time points in the same z position, the beginning and end time points quantified were in the same z position that were not manually being changed using the microscope.Due to the nature of this experiment, tissue compression would still alter the z position of the DRG being quantified.To further account for this alteration, the change in integrated density of fluorescence was quantified by subtracting the initial time point from each time point being analyzed (Δf = f−fi).This value was then divided by the time point being analyzed (Δf/f).Large increases in this value following tissue compression were then scored as active responses to mechanical compression.Little to no change in this value following mechanical compression were scored as not active in response to tissue compression.

Statistical analysis
Statistical analysis was completed with Prism.No statistical methods were used to predetermine sample sizes but sample sizes are similar to previous publications.Statistical tests were completed with biological replicates, not technical replicates.No data points were excluded from the analysis.Healthy animals were randomly selected for all experiments.Each experiment was repeated at least with similar results.All data collected and analyzed are presented in the study.

Software
Slidebook, Prism, ImageJ, R, and Adobe Illustrator were used to acquire, analyze and compile figures.
13% high correlation coefficients (n = 34 cells, 7 DRG, 4 animals) (2 dpf versus 4 dpf: p = 0.0044 post hoc Tukey test, 2 dpf n = 46 cells, 3 dpf n = 27 cells, 4 dpf n = 34 cells) (Fig 1N).Additionally, we observed that the percent of cells displaying Ca 2+ transients together increased by 4 dpf (S3A-S3C Fig).These data are consistent with the hypothesis that DRG satellite glial networks are present in vivo and form by at least the third day of DRG construction in zebrafish.