Long-range chemical signalling in vivo is regulated by mechanical signals

Biological processes are regulated by chemical and mechanical signals, yet the interaction between these signalling modalities remains unclear. Using the developing Xenopus laevis brain as a model system, we identified a critical crosstalk between tissue stiffness and chemical signalling in vivo. Targeted knockdown of the mechanosensitive ion channel Piezo1 in retinal ganglion cells (RGCs) led to pathfinding errors in vivo. However, pathfinding errors were also observed in RGCs expressing Piezo1, when Piezo1 was downregulated in the surrounding brain tissue. Depleting Piezo1 in brain parenchyma led to decreases in the expression of the long-range chemical guidance cues, Semaphorin3A and Slit1, and markedly reduced tissue stiffness. While tissue softening was independent of Sema3A depletion, Slit1 and Sema3A expression increased significantly in stiffer environments in vitro. Moreover, stiffening soft brain regions in vivo induced ectopic Sema3A production via a Piezo1-dependent mechanism. Our results demonstrate that brain tissue mechanics modulates the expression of key chemical signals, a likely phenomenon across diverse biological systems.


Introduction
Numerous cellular and tissue-level processes in biology are regulated by long-range concentration gradients of diffusible chemical signals, which act in concert to spatiotemporally control the function of cells within tissues 1 .Morphogen gradients, for example, set positional information during organismal development [2][3][4][5][6] , while growth factor and hormone gradients are essential for organ development and maintenance 7 .Signalling molecules often have multitudinous functions; for instance, the Semaphorin family of proteins regulate cell morphology and motility in the nervous, immune, respiratory, cardiovascular, and musculoskeletal systems during development, homeostasis, and disease 8 .Similarly, Slit proteins play important roles in brain, kidney, heart, and mammary gland development, as well as in various disease states, by regulating cell proliferation, migration, vascularization and more 9 .
Recent work demonstrated that many of these cellular functions are also regulated by tissue mechanics.For instance, mechanical anisotropies in tissues affect growth orientation, while factors such as mechanical stresses, environmental stiffness, and cell or tissue geometry profoundly affect cell proliferation, migration, and fate specification [10][11][12] .In the nervous system, for example, mechanical forces regulate development and various physiological processes, while alterations in tissue mechanics occur in neurodegenerative diseases and other pathophysiological conditions 13 .Thus, besides the well-established role of chemical signals, mechanical cues have recently emerged as equally critical regulators of biological processes.Indeed, cells will always be simultaneously exposed to both chemical and mechanical cues.One example is the extension of neuronal axons along precise paths during nervous system development, termed axon pathfinding.In the Xenopus laevis optic pathway, retinal ganglion cell (RGC) axons exhibit a stereotypical trajectory, exiting the eye via the optic nerve, crossing the midline at the optic chiasm, traversing the contralateral brain surface, and making a characteristic caudal turn at the mid-diencephalon before reaching their target, the optic tectum 14,15 (Fig. 1a).Several membrane-bound and diffusible signalling molecules pattern the surrounding brain tissue, directing Xenopus RGC axons along this specific path 16 .For instance, the caudal turn of RGC axons in the mid-diencephalon is regulated by gradients of the repulsive long-range guidance cues, Slit1 17 and Semaphorin3A (Sema3A) 18 .At the same time, RGC axons are mechanosensitive and encounter an instructive stiffness gradient in the brain, which also contributes to regulating RGC axon pathfinding in that region 19,20 .However, the interaction between chemical and mechanical signalling in vivo remains elusive.

Retinal ganglion cell-intrinsic and extrinsic signalling via Piezo1 is required for accurate axon pathfinding in vivo
A key player involved in sensing and responding to brain tissue stiffness 19,21 is the mechanosensitive nonselective cation channel Piezo1 22,23 .Piezo1 is found in a broad range of animal tissues including the bladder, kidneys, lung, and skin 22 , as well as in the developing Xenopus nervous system 19,24 .Downregulating Piezo1 expression in the nervous system in vivo leads to aberrant axon growth and severe pathfinding defects during development 19 .
To disentangle the effect of Piezo1 activity in RGC axons from that of the surrounding neuroepithelial cells on axon pathfinding, we selectively downregulated Piezo1 in the axons while preserving its expression in the surrounding brain tissue.This targeted depletion was achieved by injecting translation-blocking morpholinos into one of the two dorsal blastomeres at the 4-cell stage, resulting in embryos with Piezo1 depletion in one half of the nervous system.As RGC axons cross over completely in the Xenopus optic chiasm to the contralateral brain surface 25 , this approach yielded embryos with Piezo1-depleted RGC axons growing across normal brain tissue and, in the contralateral brain hemisphere, normal RGC axons growing across Piezo1-depleted brain tissue (Fig. 1b).RGC axons were labelled with DiI, and their trajectories analysed at developmental stage 40 26 , when axons had reached their end target, the optic tectum.
Consistent with our previous findings 19 , simultaneous downregulation of Piezo1 in both axons and the surrounding brain tissue led to severe axon pathfinding defects (Fig. 1c,f), compared with age-matched sibling embryos, where both dorsal blastomeres were injected with a control morpholino.The elongation of the RGC axon bundle growing along the brain surface, or optic tract, was significantly lower than in controls (p = 0.002, Kruskal-Wallis with Dunn post-hoc test) (Fig. 1g).Furthermore, only 13% of brains had normal axonal projections, while 61% of brains exhibited stalling and/or misprojection defects (i.e., deviations from the normal trajectory, such as bypassing the caudal turn in the middiencephalon or misrouting dorso-anteriorly into the telencephalon) (Fig. 1h).
Representative images illustrating normal RGC axonal projections and various axon guidance defects are shown in Supplementary Fig. 1.
Nonetheless, only 44% of these brains exhibited normal axonal projections (Fig. 1h).Piezo1 downregulation in the axons resulted in 28% of brains with stalling defects, while 44% of the brains showed misprojecting or both stalling and misprojecting defects (Fig. 1h).These results confirmed that RGC axons probe their mechanical environment via Piezo1, and that this cell-intrinsic mechanosensing is required for accurate pathfinding.However, the overall phenotype was milder than what we had observed in brains with downregulated Piezo1 expression in both RGC axons and the surrounding neuroepithelium 19 (Fig. 1f-h), suggesting that Piezo1 expression in neuroepithelial cells was also required for proper axon pathfinding.
Indeed, when Piezo1 downregulation was limited to the surrounding brain tissue, axon pathfinding of healthy RGCs containing Piezo1 was also perturbed.In fact, the observed guidance defects were even more profound than in the brains with axonal Piezo1 knockdown (Fig. 1e).Optic tract elongation was significantly decreased if compared to controls (p = 0.010, Kruskal-Wallis with Dunn post-hoc test) (Fig. 1g), and only 38% of brains showed normal axonal projections.In contrast, 56% of brains exhibited stalling and/or splaying and misprojecting axons (Fig. 1h).Thus, despite normal Piezo1 levels in RGC axons, axon growth patterns were disrupted, confirming that Piezo1-mediated mechanosensing by neuroepithelial cells was also essential for accurate axon pathfinding.

Piezo1-dependent expression of long-range chemical guidance cues in vivo
We next investigated how perturbations of neuroepithelial mechanosensing may lead to disruptions in growth patterns of healthy axons.At least two RGC-extrinsic factors are crucial for the turning of the optic tract in the mid-diencephalon (Fig. 1a): local tissue stiffness 19 and long-range chemical signalling by the repulsive guidance cues Slit1 17 and Semaphorin3A (Sema3A) 18 , which are both located rostrally to the optic tract's caudal turn (Fig. 2a,e).Chemical perturbations of brain tissue stiffness in vivo mostly lead to a splaying of RGC axons in the diencephalon 19 , while depletion of Slit1 and Sema3A signalling dominantly leads to a stalled axon phenotype 17,27 .Since we found both axon guidance defects after downregulating Piezo1 expression in neuroepithelial cells (Fig. 1e,h), we first tested the effect of Piezo1 downregulation on the expression of these chemical guidance cues using hybridization chain reaction-fluorescence in situ hybridization (HCR-FISH).
In stage 40 control-injected sibling embryos, our HCR-FISH results recapitulate published in situ hybridization data 17,18,28 : Slit1 forms a distinct band between the telencephalon and diencephalon, while Sema3A is highly expressed rostro-ventrally of Slit1 in the telencephalon (Fig. 2b,f).Piezo1 downregulation resulted in a significantly smaller area occupied by Slit1 (p = 0.004, unpaired t-test with Welch's correction) (Fig. 2c,d), and Sema3A mRNA expression was significantly decreased compared to controls (p = 0.025, unpaired t-test with Welch's correction ) (Fig. 2g,h).Standard in situ hybridizations confirmed the dramatic decrease in Sema3A mRNA expression in Piezo1-depleted brain tissue (Fig. 2 i,j).Additionally, Western blot analysis confirmed that Piezo1 downregulation resulted in significant decreases in Sema3A (p = 0.003, ratio paired t-test) (Fig. 2k,l) and Piezo1 protein levels, but not in the housekeeping protein b-actin, (Fig. 2k and Supplementary Fig. 2); Slit1 protein expression could not be assessed due to a lack of suitable antibodies in Xenopus.Hence, our data indicated that the downregulation of Piezo1 -a mechanosensor -in neuroepithelial cells attenuated the expression of key chemical cues, implying a role for mechanical signalling in modulating chemical signalling during development.

Piezo1 depletion, but not Sema3A downregulation, leads to brain tissue softening
The splaying of axons seen in Piezo1-downregulated brain parenchyma (Fig. 1e,f) recapitulated the phenotype of RGC axons growing through brain tissue that was chemically softened in vivo 19 .In order to test if, in addition to the chemical landscape, tissue mechanics was also altered in Piezo1 knockdown brains, we employed atomic force microscopy (AFM)-based stiffness mapping 19,20 to measure the stiffness of live developing Xenopus brains at stage 40.Tissue stiffness was quantified by the reduced apparent elastic modulus, K, whereby a larger K value indicates stiffer tissue.Measurements were performed in vivo in a rectangular grid on the exposed brain surface (Fig. 3a,b) of stagematched sibling embryos for control injected embryos and embryos in which Piezo1 was downregulated either exclusively in the axons or in the surrounding brain tissue, or in both.
The overall mechanical landscape of brains with Piezo1-downregulation in solely the RGC axons was similar to that of control embryos (median K ctrl = 432 Pa, K Axon k/d = 467 Pa; p > 0.999, Kruskal-Wallis with Dunn post-hoc test ) (Fig. 3c,d), suggesting that the observed axon pathfinding defects shown in Fig. 1d were not a consequence of alterations in tissue stiffness but rather originated in the axons' reduced ability to sense their mechanical environment.Conversely, downregulating Piezo1 either exclusively in the surrounding brain tissue or simultaneously in the axons and surrounding brain tissue resulted in a close to two-fold decrease in brain stiffness (median K Brain k/d = 231 Pa, K Double k/d = 240 Pa; p < 0.0001, Kruskal-Wallis with Dunn post-hoc test) (Fig. 3c,d).Hence, the knockdown of Piezo1 in the developing neuroepithelium led to considerable changes in both the chemical and mechanical landscape encountered by growing RGC axons.
In order to test whether the observed tissue softening following knockdown of Piezo1 was specific to the central nervous system or a more general phenomenon, we also measured the stiffness of skin in Xenopus embryos at stages 28-31 using AFM (see Methods for details).Also here, Piezo1 knockdown led to a significant softening of the tissue (Supplementary Fig. 3), indicating that Piezo1 may not only sense but also regulate tissue stiffness in other organ systems.
We then wanted to know whether there was a causal relationship between the decrease in guidance cue expression (Fig. 2) and the decrease in tissue stiffness (Fig. 3c,d) in Piezo1 knockdown brains.In particular, we asked if tissue softening was a consequence of the decrease in Sema3A and Slit1, or if, vice versa, the decrease in guidance cue expression was a consequence of the change in tissue stiffness.Given the spatially restricted expression of Slit1 mRNA and the uncertainty if protein levels had changed after Piezo1 knockdown, we here focused on the effect of Sema3A depletion on tissue stiffness.
To downregulate endogenous Sema3A protein in the brain without altering the early developmental expression of Piezo1 levels, we electroporated the forebrain of stage 29/30 embryos with either a Sema3A or a control fluorescein-tagged morpholino 27 .The electroporated embryos were allowed to develop to stage 40, at which point tissue stiffness was measured using AFM (Fig. 3e).Downregulating Sema3A protein expression in the forebrain did not result in brain tissue softening.In fact, the stiffness of the Sema3Adepleted forebrain region increased compared to controls, with a median K = 455 Pa in the Sema3A knockdown brain and K = 350 Pa in the control (p < 0.0001, Wilcoxon rank-sum test) (Fig. 3f,g), while brain tissue mechanics in adjacent regions was unaltered in these embryos (median K Sema3A k/d = 332 Pa, K ctrl = 341 Pa; p = 0.67, Wilcoxon rank-sum test) (Fig 3h).These data showed that the observed stiffening was specific to the localized Sema3A depletion, and suggested that the decrease in guidance cues in Piezo1 knockdown brains is likely not responsible for the decrease in tissue stiffness.

Environmental stiffness regulates chemical guidance cue expression
To test if, instead, tissue stiffness may modulate chemical guidance cue expression, independently of the decrease in Piezo1 expression, we first exposed brain tissue to different mechanical environments and measured their effect on the expression of Slit1 and Sema3A mRNA using HCR-FISH.Brain tissues of wild type embryos at stage 37/38, at which point axons already respond to Sema3A 18 and Slit1, was harvested from stiff brain regions 19 that produce the guidance cues (telencephalon) and from the adjacent soft brain region 19 that typically does not (hypothalamus) (Fig. 4a).Both tissue types were separately embedded in soft (40 Pa) and stiff (450 Pa) 3D hydrogels (Fig. 4a) 29 .Cells at the edges of the tissue regions embedded in the substrates pulled on the surrounding hydrogel with forces on the order of ~50 nN, visibly deforming the material and thus probing its mechanical properties (Supplementary Fig. 4; Supplementary movie 1).After 24 hours in culture, Slit1 and Sema3A expression in telencephalic tissue was independent of the environmental stiffness (p Slit1 = 0.824, p Sema3A = 0.542, ratio paired t-test) (Supplementary Fig. 5).However, in hypothalamic tissue, we found a significant increase in both Slit1 and Sema3A mRNA expression in stiff compared to soft substrates (p Slit1 = 0.025, p Sema3A = 0.050, ratio paired ttest) (Fig. 4b-e), suggesting that altering environmental mechanics may modulate the expression of chemical guidance cues in soft brain tissues in vitro.
To corroborate the effect of tissue stiffness on the expression of long-range chemical signals in vivo, we used AFM-based compression stiffening 19,30 to locally stiffen stage 35/36 soft hypothalamic brain regions, for > 6 hours (until embryos reached stage 40) in vivo and subsequently assessed mRNA expression with HCR-FISH (Fig. 4f).Slit1 expression was not significantly different from controls in compression-stiffened brains (p = 0.169, unpaired ttest with Welch's correction) (Supplementary Fig. 6).However, in line with our in vitro data (Fig. 4e), stiffening the tissue resulted in the ectopic production of the chemical guidance cue Sema3A at the hypothalamus (Fig. 4,h).In compression-stiffened brains, there was a significant increase in mRNA levels compared to controls (p = 0.011, unpaired t-test with Welch's correction) (Fig. 4g-i), indicating that in vivo tissue stiffness may indeed regulate the expression of the long-range chemical signal, Sema3A.Since Piezo1-depleted brains had decreased Sema3A expression (Fig. 2) and were softer than control brains (Fig. 3), and as an increase in tissue stiffness led to an increase in Sema3A expression in vivo (Fig. 4g-i), we finally tested if stiffening telencephalic regions, where Sema3A is normally produced, is sufficient to rescue Sema3A expression in Piezo1 knockdown brains (Fig. 4j).Compression-stiffening Piezo1-depleted stage 35/36 brains for > 6 hours did not restore Sema3A levels (p = 0.170, unpaired t-test with Welch's correction) (Fig 4k-m), suggesting that both tissue stiffness and adequate levels of Piezo1, which not only regulates but also detects tissue stiffness, are required to lay down the appropriate chemical landscape for supporting axon pathfinding in the developing brain.

Discussion
During numerous processes in diverse biological systems, including brain development, cells encounter a plethora of chemical and mechanical cues in their environment, which they detect, integrate, and interpret.We here discovered an intricate interplay between mechanical and long-range chemical signalling in vivo, with both influencing each other, showing that neither cue can be fully understood in isolation.
Collectively, our results underscore a crucial role of Piezo1 in both RGC-intrinsic and cellextrinsic regulations of axon pathfinding in the developing Xenopus retino-tectal system.
We found that the mechanosensitive ion channel, Piezo1, has at least a dual function in the developing brain.On the one hand, it is involved in sensing tissue stiffness 19 (Fig. 1d,h).On the other hand, it inherently contributes to regulating the mechanical properties of the tissue (Fig. 3a-d), and in turn, the appropriate tissue stiffness is required for the expression of long-range chemical guidance cues (Fig. 4).Downregulation of Piezo1 in brain tissues led to a decrease in tissue stiffness (Fig. 3) and a subsequent decrease of the expression of Sema3A and Slit1 (Figs. 2, 4).In contrast, the downregulation of Sema3A was followed by an increase in tissue stiffness (Fig. 3e-h), indicating that feedback mechanisms between tissue stiffness and gene expression levels do not always act reciprocally.
Recent work has shown that softer substrates lead to a decrease in Piezo1 expression levels in vitro 31,32 .We here found the converse to also hold true; a decrease in Piezo1 expression leads to tissue softening in vivo (Fig. 3), suggesting a positive feedback loop.In line with our data, Piezo1 levels in microglial cells correlate with the stiffness of Aβ plaqueassociated brain tissues 33 , and in macrophages with the stiffness of ischemic tissues 34 .Furthermore, in Drosophila, dPiezo depletion led to softening of gliomas driven by different oncogenic mutations but not of non-transformed brains 35 .How Piezo1 contributes to the regulation of tissue mechanics remains to be studied.Potential contributions could arise from the impact of its activity on cell proliferation 36,37 , on the mechanical properties of cells 38 , or on the extracellular matrix composition 39 .
In vitro, environmental stiffness affected Sema3A and Slit1 mRNA expression in the hypothalamus but not in telencephalic regions (Supplementary Fig. 5).At stage 37/38, the telencephalon is already stiffer than the adjacent hypothalamus 20 and chemical cues (Slit1 and Sema3A) are already expressed in this region 28 .Thus, exposing this stiff tissue, which is already poised to produce chemical cues, to a stiffer environment in vitro may not lead to further alterations in their expression profile.
While increasing tissue stiffness in vivo was sufficient to induce ectopic Sema3A expression in soft brain regions of wild type embryos, in Piezo-downregulated brains tissue stiffening did not rescue Sema3A expression (Fig. 4), indicating that Piezo1 in neuroepithelial cells was also required to detect mechanical signals and induce Sema3A expression (Fig. 4).A dependence of Sema3A expression on tissue stiffness is consistent with expression patterns found in healthy Xenopus brains.Here, Sema3A is expressed in the telencephalon, which is significantly stiffer than the diencephalon and hypothalamus 19 , where Sema3A is not expressed (Fig. 2).However, Sema3A is already expressed in the telencephalon region at stage 32 28 -before the telencephalon stiffens 20 -indicating that other factors contribute to regulating Sema3A.
The expression of both Sema3A and Slit1 is known to be regulated by FGF signalling 17,28 .FGF receptors, however, can be activated in the absence of ligands through mechanical forces 40 .As cellular traction forces are higher in stiffer environments, stiffer tissues could lead to the enhanced activation of FGF receptors, thus facilitating guidance cue expression in the developing brain even in the absence of FGF.If, for example, FGF levels decrease over time or distance during development, higher Sema3A and Slit1 expression levels could still be maintained via Piezo1-mediated responses to enhanced tissue stiffness (Fig. 4).This would also explain why the expression of Sema3A can be triggered in the brain ectopically through a stiffer environment (Fig. 4c,e) or the external application of a mechanical force (Fig. 4h,i).Depleting Piezo1 early in development may not only affect tissue stiffness (Fig. 3) but potentially also neural stem/progenitor cell fate determination 21,41 or FGF signalling 42 .Thus, there are several potential pathways how the mechanosensitive ion channel Piezo1 might regulate long-range chemical signalling.Future work will reveal molecular details on how Piezo1 activity is coupled to gene expression profiles of key signalling molecules.
In the developing Xenopus brain, Piezo1 is critical for setting up the mechanical and chemical landscape required for proper RGC axon guidance.Our findings raise the intriguing possibility that tissue mechanics may regulate the transcription of many other biochemical factors 43 in various other organ systems -and vice versa.Understanding the complex interplay between chemical and mechanical signalling is a challenge but has great potential to provide new insights into the regulation of development, physiology, ageing, and disease.

Methods
Detailed methods are available in the online version of the paper, including the associated code and references.

Figure 2 :
Figure 2: Piezo1 downregulation attenuates the expression of long-range chemical

Figure 4 :
Figure 4: Altering stiffness in vitro and in vivo affects chemical guidance cue