Piezo3 is a new member of the PIEZO ion channel family in vertebrates

Abstract

Mechanosensing and mechanotransduction are essential for all living cells. In mammals, Piezo1 and Piezo2 are two mechanically activated cation channels that serve as mechanosensors for a variety of physiological and pathological processes, ranging from touch sensing to sickle cell disease. These two channels are well evolutionarily conserved, and orthologous genes can be traced back to the origin of vertebrates, which underwent whole-genome duplications (WGDs). The number of paralogous genes originating from the vertebrate WGD varies across gene families. Thus, whether there are more PIEZO paralogous genes in vertebrates remains understudied. Here, we identified piezo3, a new paralog of the piezo gene family, and analyzed its evolutionary history using phylogenetic and synteny analyses. The piezo3 gene is present in most vertebrate lineages but absent in birds and most mammals, likely due to nonfunctionalization after WGDs. In addition, we demonstrated that this channel could mediate calcium flux in response to mechanical stimuli in HEK293T cells, suggesting that Piezo3 exhibits PIEZO1/2-like activation and conduction channel functions. Our CRISPR mutation analysis revealed that the zebrafish piezo3 gene is not developmentally essential for skeletons and survival, possibly because its expression overlaps with other PIEZO channels. Mutant zebrafish showed elevated sensitivity to mechanical force and increased locomotor activity under (photopic) light illumination. Our results suggest that this new mechanical-sensing Piezo channel is widespread in vertebrates and may be critical for vertebrate adaptation by modulating mechanical sensing and light responses during evolution.

Graphical Abstract

INTRODUCTION

Mechanosensing and mechanotransduction are fundamental cellular processes that are manifested at many levels of cellular behaviors, including cell division, growth, and the transmission of biochemical signals (Martinac and Kloda, 2003). Ion channels play an important role in transducing mechanical force into electrochemical responses by allowing ions to cross cell membranes along their concentration gradient. MscL (large conductance mechanosensitive ion channels) was the first mechano-gated channel identified as an osmotic stress-release valve in bacteria (Sukharev et al., 1994). Homologous MscL was reported in Archaea but not in mammals (Martinac and Kloda, 2003). It took decades of searching before the PIEZO1 and PIEZO2 channels were identified (Coste et al., 2010). After that, the orthologs of these two genes were extensively investigated in various organisms, especially the mouse (Bagriantsev et al., 2014; Murthy et al., 2017). Piezo channel genes are not limited to vertebrates; invertebrates and plants possess piezo channels (Ergin et al., 2023; Hu et al., 2019; Mousavi et al., 2021; Radin et al., 2021). Now, the mechanotransduction by PIEZO1 and PIEZO2 channels is increasingly being incorporated into many biological research fields (Xiao, 2024). PIEZO1 is broadly expressed across many tissues and organs and plays critical roles in cardiovascular, skeletal, and digestive physiology. PIEZO2 is mainly expressed in the ganglion neurons and is responsible for somatosensory, auditory, respiratory, and urinary sensing (Delmas et al., 2022; Syeda, 2021).

Whole-genome duplications (WGDs) play a significant role in shaping paralogous gene numbers and expanding developmental genetic toolkits that underlie vertebrate morphological novelties and diversity (Dehal and Boore, 2005; Holland et al., 1994; Ohno, 1970). It is well accepted that there are two rounds of WGDs (1R and 2R) around the origin of vertebrates (Dehal and Boore, 2005; Holland et al., 1994; Meyer and Schartl, 1999; Ohno, 1970). All the genes of vertebrate ancestors were duplicated twice, creating four ohnologs through the two rounds of WGDs. One or more of these four ohnologs (paralogs generated by WGD) may be lost in vertebrate lineages. For example, there is one fibrillar collagen and one Hox gene cluster in invertebrates, while there are four fibrillar collagen Hox gene clusters in tetrapods (Zhang and Cohn, 2008). After the vertebrate lineage split, another round of WGD (3R) occurred in the common ancestor of all extant teleosts (Meyer and Van de Peer, 2005). This 3R resulted in more ohnologs in teleost species, although many duplicated genes were lost (Inoue et al., 2015; Semon and Wolfe, 2007). For example, there are multiple potassium channel duplicates, i.e., ohnologs, in zebrafish (Park et al., 2024; Silic et al., 2021; Silic et al., 2022). These duplicated genes play an essential role in embryogenesis and the diversity of vertebrate morphological structures, such as skeletons, through subfunctionalization, neofunctionalization, and gene loss (Force et al., 1999; Meyer and Schartl, 1999; Ohno, 1970; Zhang and Cohn, 2008; Zhang et al., 2009). PIEZO1 and PIEZO2 have been extensively investigated for mechanotransduction in mice and other species, while the evolutionary history of this type of channel in vertebrates remains poorly understood, especially in teleosts, given the three rounds of WGD events in vertebrates.

Zebrafish is one of the most common vertebrate model systems for current biomedical research due to numerous advantages, such as external embryonic development, early transparent embryos, tractable genetics, and vertebrate biology that is homologous to humans (Lieschke and Currie, 2007; Santoriello and Zon, 2012). Moreover, its phylogenetic position and its genome, which underwent 3R, made the zebrafish model attractive for evolutionary functional studies. Here, we report a new vertebrate piezo gene, piezo3, with its evolutionary history. We found that the piezo3 gene is present in most vertebrate lineages but is lost in birds and most mammals. In addition, we examined its expression patterns, which suggest Piezo3 may play a role in embryogenesis and physiology. Our CRISPR mutants support this notion. Although the piezo3 gene is not developmentally essential for skeleton and survival, it modulates zebrafish larval responses to mechanical and photopic stimuli.

RESULTS

Identifying a new piezo channel gene in zebrafish

The two-pore domain potassium (K2P) channels are also involved in mechanosensing and transduction (Brohawn, 2015; Brohawn et al., 2014; Sorum et al., 2024). Recently, we revealed that vertebrate K2P potassium channels were expanded by WGDs (Park et al., 2024). It is reasonable to expect that WGDs also shape PIEZO channels, and we hypothesized that more paralogous genes of piezo1 and piezo2 may exist in teleosts. To test this hypothesis, we chose zebrafish because the functions of piezo genes (piezo1, piezo2a, and piezo2b) have already been reported, and the zebrafish genome is relatively well-annotated (Faucherre et al., 2013; Ramli et al., 2023). First, we searched for piezo genes in the zebrafish genome (GRCz11) using already-known zebrafish piezo1 as bait by BLAT. We identified si:dkey-11f4.7 as a new paralog of the piezo1 gene. In ZFIN, there are six transcripts from this gene (si:dkey-11f4.7-201 to si:dkey-11f4.7-206) that correspond to six transcripts of the gene ENSDARG00000099207 in ENSEMBL. The sizes of these six transcripts (443–972 bps) are much shorter than the piezo1 transcript. In contrast, in NCBI, the si:dkey-11f4.7 gene (Gene ID: 569687) has two much longer transcripts (XM_068223521.1 and XM_009304390.4), and they code two putative proteins (2605aa and 2651aa). This discrepancy likely resulted from the incomplete zebrafish genome annotation. The si:dkey-11f4.7 shared homology with the current three known zebrafish piezo1, piezo2a, and piezo2b (Table 1). It is also worth noting that the piezo2a gene is annotated as two genes (piezo2.a and piezo2a.1) in ZFIN, and both are partial sequences of piezo2a (Table 1).

Table 1.

Zebrafish piezo paralogous genes.

Gene name	Other names	NCBI Gene ID Accession ID	ZFIN ID	Ensembl Gene ID	Gene Location	Protein Length AA
piezo1	piezo1, FAM38A, im:7149048	GeneID:567949 XM_691263.9	ZFIN:ZDB-GENE-041111-255	ENSDARG00000076870	Chromosome 7: 55,328,420-55,454,406	2538
piezo2a	Piezo2.a piezo1b; si:ch1073-433p8.1 LOC100334952 piezo2a.1 zmp:0000000946 zmp:0000001249	GeneID:100334952 XM_021470255.2	ZDB-GENE-131119-76 ZDB-GENE-130530-949 (piezo2a.1)	ENSDARG00000076005	Chromosome 24: 854,877-945,921	2752
piezo2b	piezo2 LOC100534973 zmp:0000000807	Gene ID: 100534973 X1: XM_021468270.2 X2: XM_021468271.2 X3: XM_021468277.2	ZDB-GENE-130530-810	ENSDARG00000107086 ENSDARG00000112723 (fragmented)	Chromosome 2: 54,086,436-54,270,366	3005
piezo3	si:dkey-11f4.7	Gene ID: 569687 X1: XM_068223521.2 X2: XM_009304390.4 XM_021478800.1	ZFIN:ZDB-GENE-070912-361	ENSDARG00000099207 (fragmented)	Chromosome 9: 360,252-403,767	2643

Vertebrate piezo genes may have been duplicated through WGDs

This new si:dkey-11f4.7 gene could be unique in zebrafish or conserved in other vertebrates. To address this question, we used this gene as bait and searched NCBI and Ensembl for homologous sequences. This gene is also present in most vertebrate lineages, including chondrichthyans, holostei, teleosts, amphibians, and reptiles, but not in currently sequenced birds and most mammals. To further clarify its relationships with PIEZO1 and PIEZO2, we conducted phylogenetic analyses using maximum likelihood and Bayesian methods. Both analyses yield consistent results. In vertebrates, there are three clusters: PIEZO1, PIEZO2, and Si:dkey-11f4.7. The Si:dkey-11f4.7 clade is closer to the PIEZO2 than the PIEZO1 clade (Fig. 1 and Fig. S1). Thus, we suggest renaming the si:dkey-11f4.7 as PIEZO3 based on our phylogenetic analysis. Most teleost species possess duplicated piezo2 genes (piezo2a and piezo2b), indicating they originated from the third round WGD (3R). Interestingly, the PIEZO3 gene is present in some mammalian genomes, such as those of the armadillo, red deer, and tree shrew, suggesting that the loss of this gene might not have occurred in the ancestor of mammals, but rather independently in each sublineage.

Figure 1. Maximum likelihood phylogeny of animal PIEZO channels.

Some representative species (human, mouse, zebrafish, spotted gar, shark, lancelet, tunicate, fruit fly, and worm) were analyzed. Numbers at each node indicate the supporting values based on 1,000 bootstrap replicates. Branch lengths are proportional to the expected number of replacements per site. The phylogeny was inferred using the JTT model with a gamma distribution in PhyML. All vertebrate PIEZOs were grouped together, and the three types of PIEZO channels formed a distinct clade, which is color-coded and indicated with vertical black lines. The teleost Piezo2 channels were also color-highlighted. All invertebrate PIEZO channels formed a distinct outgroup on the phylogenetic tree.

To further confirm whether the vertebrate piezo genes are true paralogs generated by WGDs, we conducted syntenic analyses across a few representative species. We found that piezo1 is located in an evolutionarily conserved synteny (aprt-cdt1-piezo1-ctu2-rnf166-snai3), and piezo2 is linked with napg and apcdd1 genes (Fig. 2A–B). Similarly, piezo3 is located in a longer synteny (prelid3b-atp5f1e-tubb1-ctsz-nelfcd-gnas-piezo3-npepl1-stx6-apcdd1l-vapb-rab22a). This piezo3 synteny becomes shorter and more variable in teleosts (Fig. 2C). The linkages of piezo2-apcdd1 and piezo3-apcdd1l are consistent with their closer clustering in our phylogeny analysis, suggesting they may have arisen from a common ancestor during the teleost-specific WGD (3R).

Figure 2. Syntenic analyses of the piezo genes across representative vertebrate species.

The Illustration of the genes and their sizes is not proportional to the actual distances between genes. A. Common vertebrate piezo1 synteny. The piezo1 gene was highlighted in red. The synteny (aprt-cdt1-piezo1-ctu2-rnf166) is shared among all the representative species. B. Common vertebrate piezo2 synteny (piezo2-napg-apcdd1). The piezo2 is highlighted in red. The synteny was boxed with blue dashed lines. After duplication, the apcdd1 gene is lost in both the piezo2a and piezo2b synteny blocks. C. Common vertebrate piezo3 synteny. A few more representative species were included for comparison. The piezo3 synteny is boxed with dashed lines. The gene order of this synteny is conserved among sharks, snakes, and mammals, but slightly shifted in spotted gar and teleost. The PIEZO3 genes become pseudogenes (red letters) in humans and mice.

To reveal how the PIEZO3 gene was lost in humans, we further analyzed the sequences of the conserved synteny. According to the current human genome (GRCh38.p14), there is a pseudogene, PIEZO1P2, between NPEPL1 and GNAS (Chromosome 20: 58,740,532–58,783,821). PIEZO1P2 encodes a 2161bp transcript and shares homology with PIEZO1, supporting that this processed PIEZO1P2 pseudogene transcript resulted from PIEZO3 inactivation during evolution. Interestingly, at the downstream of this synteny, there is another pseudogene, PIEZO1P1 (219bp), which is likely a rudimentary part of PIEZO3 (Fig. 2C). Similarly, we found a 178bp conserved region (chromosome 2:174075082:174075795) in the mouse genome (GRCm39). It has 87.72% identity to human PIEZO1P2, indicating that the mouse Piezo3 gene was also inactivated/lost in the same way. According to the GTEX portal, PIEZO1P2 is expressed in the spleen and testis, although its function remains unknown (Fig. S2).

Zebrafish piezo genes’ expression during embryogenesis

As PIEZO3 is a new vertebrate piezo channel, no functional study has been conducted. To gain insight into its functions and validate that piezo3 is an actively expressed gene, we performed whole-mount in situ hybridization during zebrafish embryogenesis. This gene was not turned on until 5-somite stage, and it is mainly expressed in neuronal tissues (Fig.3 A–F). At the 15-somite stage, it was expressed at the ventral side of the brain (Fig. 3G–I). From the 21-somite to 24 hpf (hours post fertilization) stage, piezo3 was mainly expressed in the brain and the eye, including the lens (Fig. 3J–M). The brain and eye expression remained but became less expressed at 48 hpf. Evident expression was detected in pharyngeal arches and pectoral fin buds (Fig. 3N–O). This expression of the pharyngeal arch and pectoral fin buds continued to the 72 hpf stage. In contrast, the head expression was reduced (Fig. 3P–Q), suggesting Piezo3 might play a role in neural, pharyngeal, and pectoral fin development.

Figure 3. Zebrafish piezo3 gene expression during zebrafish embryogenesis.

Whole-mount in situ hybridization of zebrafish embryos at bud stage (A-C), 5 somite stage (D-F), 15 somite stage (G-I), 21 somite stage (J-K), 24 hpf (L-M), 48 hpf (N-O), and 72 hpf (P-Q). The anterior is to the left in all the whole-mount images, and the dorsal is to the top in all transverse sections (L, N, P). The piezo3 gene expression is shown laterally (A, D, G, J, L, N, P). Dorsal view of gene expression of piezo3 (B, E, H, K, M, O, Q). Ventral view of gene expression of piezo3 (C, F, I). The white dashed lines indicate the proximate positions of the sections shown in the insert (L’) of the same panel (L). Magnified pectoral fin buds are highlighted in the inserts (N’, P’) in the corresponding panels (N, P). Scale bars were added to the left of the bottom row. 250 μm for whole-mount images. e, eye; fb, forebrain; hb, hindbrain; mb, midbrain, op, optic vesicles; ov, otic vesicles; pa, pharyngeal arches; pf, pectoral fins; tb, tail bud.

The piezo3 gene is paralogous to piezo1, piezo2a, and piezo2b. To elucidate whether they are coexpressed during zebrafish embryogenesis, we also systematically analyzed the three paralogs. The piezo1 was first detected in the forebrain during the late gastrulation, the bud stage, and continued to the 5-somite stage (Fig. S3A–F). From the 15-somite to the 21-somite stage, it was also expressed at the otic vesicles and ventral somite mesoderm (Fig. S3G–K). This ventral somite and otic vesicle expression continued to 24 hpf (Fig. S3L–M). This expression is consistent with its functions in red blood cells and the vertebral column (Faucherre et al., 2014; Ramli et al., 2023). At 48 hpf, the somite expression retracted, but the piezo1 transcript was detected at the otic vesicle, pharyngeal arch, and pectoral fin buds. These expression domains remained at 72 hpf (Fig. S3P–Q).

The piezo2a gene was first detected in the forebrain at the 5-somite stage and was lightly expressed in the neural tube (Fig. S4A–F). The forebrain domain continued at the 15-somite stage (Fig. S4G–I). From the 21-somite to the 24 hpf, piezo2a was mainly expressed at the ventral brain and ventral mesoderm of somites (Fig. S4J–M). Similar to piezo1, this domain may be related to vertebral column development (Ramli et al., 2023). At 48 hpf, the ventral mesoderm recessed, while the pharyngeal arches, pectoral, and caudal fin buds started to express the piezo2a gene (Fig. S4N–O). These expression patterns persisted to 72 hpf and 5 dpf (Fig. S4P–S). Interestingly, its expression within caudal fin buds is relatively stable at the proximal part. In contrast, piezo2a showed a dynamic expression pattern in the pectoral fin buds, with expression broadly distributed throughout the mesenchyme at 48 hpf, restricted to the basal mesenchyme at 72 hpf, and localized to the epithelial-mesenchymal interface at 5 dpf. This dynamic expression pattern suggests a potential role for piezo2a in pectoral fin development. The piezo2b gene is expressed mainly in Rohon-Beard neurons of the neural tube and the trigeminal ganglia (Fig. S5A–Q), consistent with a previous report (Faucherre et al., 2013).

Zebrafish Piezo3 conducts calcium cations in HEK293T cells

PIEZO1 and PIEZO2 selectively conduct Ca²⁺, Na⁺, K⁺, and Mg²⁺, and both are sensitive to streptomycin (Coste et al., 2010). To examine whether zebrafish Piezo3 has a similar function, we first took advantage of AlphaFold to examine its 3D structure using the available amino acid sequences. Indeed, the zebrafish Piezo3 has a very similar 3D structure to human PIEZO1, PIEZO2, and the other three zebrafish Piezo1, Piezo2a, and Piezo2b (Fig.S6 A–E). Next, we cloned the full-length zebrafish piezo3 into a CMV promoter-driven plasmid construct with an IRES (Internal Ribosome Entry Site)-mediated EGFP marker. We co-transfected it with a Ca²⁺ reporter, jRGECO1a, into HEK293T cells. After transfection, there are four types of cells: 1) Cells with both EGFP and red fluorescence (Piezo3 and jRGECO1a), green-only (Piezo3), red-only (jRGECO1a), and no fluorescence cells (un-transfected). To study the effects of Piezo3, we compared the red fluorescence intensity (Calcium readout) between red-only cells (no Piezo3) and both green and red cells (Piezo3 and jRGECO1a) (Fig. 4A–B). We observed a slightly higher baseline fluctuation in the Piezo3 expression cells (Fig. 4C, H). To test their mechanical response, we applied a gentle force using a blunt-end glass probe mounted on a micromanipulator. A rapid rise in red signal transient and a slower signaling decay were observed (Fig. 4D, H), indicating calcium ions were moving through the Piezo3 channel. To test whether the Piezo3 is sensitive to streptomycin, we added streptomycin at two concentrations to the cell medium and repeated the stimulation experiments. We found that the cells showed significantly smaller or no changes in red signal (Fig. 4E, H). The Piezo3-expressing cells responded to mechanical force (Fig. 4F, H) at a low streptomycin concentration. However, the cells failed to respond to probe poking at the higher streptomycin concentration (Fig. 4G–H). Overall, these results suggest that zebrafish Piezo3 is a functional mechanical channel, similar to Piezo1 and Piezo2.

Figure 4. Piezo3 mediated intracellular calcium signals in response to mechanical force in HEK293T cells.

A. Experimental design for co-transfection and calcium signal measurement of the mechanical force-induced response. B. Representative images of the RFP+/GFP+ and RFP+/GFP− HEK293T cells in untouched or mechanically stimulated conditions. The merged image of the bright-field and GFP channels showed piezo3-expressing GFP+ and GFP− cells, and the RFP channel showed jRGECO1a-expressing RFP+ cells. The red rectangle labeled the cells that were mechanically stimulated. C-D. Calcium signal responses of the RFP+/GFP+ and RFP+/GFP− HEK293T cells in streptomycin-free culture conditions without (C) and with (D) glass probe touch. E-F. Calcium signal responses at a low streptomycin (172μM) condition without (E) and with (F) glass probe touch. G. Calcium signal responses at a high streptomycin (1032μM) condition with glass probe touch. The vertical black arrow indicates the time of glass probe application. RFP was recorded at 0.5 frames/s for 3–4 min. ΔF/F₀ (ΔF=F-F₀) was used to describe the calcium signal response (mean + 95%CI). 10–15 cells were recorded for each group. H. Comparisons of the overall calcium signal responses between PIEZO3-expressing and non-PIEZO3 HEK293T cells. The total area under the curve (AUC) over the ΔF/F₀=0 baseline was calculated to quantify the overall calcium signal response for the record duration. *p<0.05.

Zebrafish piezo3 is not essential for skeletal development or survival and may be responsible for larval mobilization

To investigate the biological functions of the zebrafish Piezo3 channel, we mutated this gene using CRISPR-Cas9 (Kim and Zhang, 2020). Three loss-of-function mutants, pu117 (large knockout, LKO), pu118 (indel and small knockout, SKO), and pu119 (indels) were generated and validated by sequencing (Fig. 5A–B). To eliminate potential off-target mutations, we outcrossed F₁ mutants to wildtype fish in established fish lines. Then, homozygotes were generated by in-crossing F₂ adults for phenotype assessment. Homozygotes are fertile and show no evident morphological alterations. Consistently, we observed an approximate Mendelian ratio with the three mutants (Fig. 5C), suggesting that the piezo3 gene is not developmentally essential for overall survival. To further validate the loss of function of mutant pu117, we performed quantitative RT-PCR on 3 dpf zebrafish embryos (anterior parts due to the piezo gene overlapping expression) and found that the piezo3 mRNA was indeed significantly decreased to around 25% compared to wildtype siblings, while the other 3 piezo genes did not show any evident alteration (Fig. 5D–E). This result suggests that pu117 is indeed a loss-of-function mutant, and there is no transcriptional compensation from other piezo genes.

Figure 5. Zebrafish piezo3 is a developmentally nonessential gene.

A. Illustration of the piezo3 locus based on the genomic fragment sequence, CR753874.9 in NCBI. Exons are shown as black vertical lines. Three CRISPR gRNA target sites are highlighted by blue vertical bars. The large knockout (LKO) truncated region is indicated by an orange bar, and the small knockout (SKO) region is shown as a green bar. Piezo3 quantitative RT-PCR primers are indicated by red triangles below exons 3 and 5. B. Predicted piezo3 proteins of full frame and 3 mutant alleles generated by CRISPR-Cas9: pu117 (LKO, exon 14 – exon 29), pu118 (exon 14 indel mutation + exon 29 SKO), pu119 (exon 14 in-frame indel mutation + exon 29 indel mutation). C. Genotype distribution from a F2 heterozygote in-crosses from three piezo3 mutant lines (pu117, pu118, pu119). Genotypes are close to the expected Mendelian ratios. D. Illustration of quantitative RT-PCR experiment. The anterior part of the larval fish was dissected for total RNA isolation. E. Result of quantitative RT-PCR of 4 piezo genes in pu117 mutants and wildtype siblings. Only piezo3 mRNA decreased significantly. F-I. Hematoxylin and eosin (H&E) staining of adult zebrafish eyes in wild type (WT) and piezo3 LKO homozygous fish (pu117/pu117). CH, choroid; GCL, ganglion cell layer; ILM, inner limiting membrane; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segments; ON, optic nerve; ONL, outer nuclear layer; OPL, outer plexiform layer; OS, outer segments; RPE, retinal pigment epithelium. J-U. Alcian blue and alizarin red staining of 13 dpf larvae (J-M), 28 dpf juveniles (N-Q), and adult (R-U) WT and piezo3 LKO homozygous fish. bh, basihyal; ch, ceratohyal; cl, cleithrum; d, dentary; LL, left lateral; mc, Meckel’s cartilage; mx, maxilla; pm, premaxilla; op, opercle; te, teeth; V, ventral. V-AC. Alizarin red staining of the adult skeleton in WT and pu117 homozygous fish. Whole fish (V, Z) and higher magnification views of vertebrae (W, AA) and pectoral fin (X, Y, AB, AC) are shown. anal, anal fin; caudal, caudal fin; cl, cleithrum; co, coracoid; dorsal, dorsal fin; HA, haemal arches; IBM, intermuscular bone; mr, marginal ray; NA, neural arches; pec, pectoral fin; pel, pelvic fin; pr1–4, proximal radials; SC, scapula; VC, vertebral column. AD. VMR experimental processes. AE. VMR assays from 3 independent experiments on pu117 mutants and wildtype siblings. Each individual zebrafish larval total traveled distance in a 1-hour light-on period is recorded. Data are presented as Tukey box-and-whisker plots; boxes indicate the 25^th-75^th percentiles (interquartile range), center lines indicate the median, and whiskers extend to the most extreme values within 1.5 times the interquartile range. Values outside this range are shown as individual points. AF-AG. pu117 and pu118 larval mechanical tapping assays. The average traveled distance was recorded for each individual upon tapping. ** indicated significance with p< 0.01, and * indicated p<0.05.

We did not detect histological differences in the adult retina (Fig. 5F–I). As piezo3 was also expressed in the pharyngeal arches (Fig. 3N, P), we next examined pu117 larval and adult gill arches and found no cartilage changes (Fig. 5J–U). Since piezo1 plays an essential role in skeletogenesis and is responsible for intramuscular bones in zebrafish and carp (Zhang et al., 2025), we then examined the adult zebrafish pu117 skeletons using alizarin red staining. However, we did not find any evident abnormality in the overall axial and intramuscular bones (Fig. 5V–AC). As piezo3 was expressed in the eyes of 1 dpf fish embryos (Fig. 3L), we first analyzed the vision of 7 dpf pu117 larval zebrafish using the visual-motor response (VMR) assay (Fig. 5AD–AE). We found that the total distance traveled by the mutants in the light-on phase was greater than that of the wildtype sibling control, although there was no difference in their startle response (Fig. 5AE). Moreover, the pu117 and pu118 mutants were more sensitive to mechanical tapping compared to the wildtype sibling control (Fig. 5AF–AG), suggesting that Piezo3 might have a repressive function in larval fish mobilization to mechanical stimuli.

DISCUSSION

Mechanosensing and transduction are essential for cellular and organism physiology. The PIEZO1 and PIEZO2 channels are critical for vertebrate somatosensation, embryogenesis, and the pathogenesis of various diseases. Currently, only Piezo1 and Piezo2 genes have been reported in the literature and annotated in most vertebrate genomes (Bagriantsev et al., 2014; Coste et al., 2010). Here, we report a new member of the vertebrate Piezo channel family, Piezo3.

WGDs play a crucial role in the evolution of modern vertebrates and flowering plants (Ohno, 1970; Van de Peer et al., 2017). It is well accepted that there were two consecutive rounds of WGD (2Rs) around the origin of vertebrates (Dehal and Boore, 2005; Meyer and Schartl, 1999; Meyer and Van de Peer, 2005; Ohno, 1970). These 2Rs generated four copies of all the genes in the vertebrate ancestor, and some of these ohnologs may have been lost during evolution. This could lead to 1–4 genes in most tetrapods. For example, there is only one hedgehog gene, while mice and humans have SHH (sonic hedgehog), IHH (Indian hedgehog), and DHH (desert hedgehog) (Pereira et al., 2014). Our phylogenetic and syntenic analyses revealed that Piezo3 most likely resulted from two rounds of WGDs in the vertebrate lineage, because the Piezo3 cluster is parallel to that of Piezo1 and Piezo2. Piezo3 was retained in teleosts, amphibians, and reptiles, but lost in birds and most mammals. Only a few mammals (platypus, echidna, Tasmanian devil, armadillo, red deer, and tree shrew) still carry this gene. It would be interesting to know whether the Piezo3 has a unique function or is just retained as a redundant gene in these species. The WGD also occurred within vertebrate lineages. The teleost-specific WGD, 3R, occurred in the ancestor of modern teleosts around 300 million years ago (Brunet et al., 2006; Glasauer and Neuhauss, 2014; Meyer and Schartl, 1999; Ravi and Venkatesh, 2018; Woods et al., 2000). This is the reason why most teleosts, including zebrafish, have 6–8 Hox gene clusters (Amores et al., 1998; Wagner et al., 2003). According to our phylogenetic and syntenic analysis, the piezo2 gene was further duplicated into piezo2a and piezo2b in this lineage, as two paralogs that were clustered with tetrapod PIEZO2. Based on our results, we propose a model for the presence of the piezo3 gene in vertebrates (Fig. 6A). The zebrafish piezo3 originated from the 2R, and piezo2a and piezo2b originated from the 3R. The teleost piezo1 and piezo3 genes were most likely duplicated, but one ortholog was lost immediately after 3R, leaving only one gene in this lineage (Fig. 6B).

Figure 6. Model of vertebrate piezo gene evolution and their gene expression summary in zebrafish embryogenesis.

A. The piezo3 gene is present in most vertebrate lineages but is lost in birds and most mammals (hexagons). Blue and green bars on the tree indicated WGD events. B. Based on our phylogenetic analysis, we proposed that there is one piezo gene in chordates. After two rounds of WGDs, four piezo genes were created in the ancestor of vertebrates. The piezo4 gene was lost soon after the vertebrate lineage split. In the teleost lineage, the 3^rd WGD further created two ohnologs for piezo1, piezo2, and piezo3. However, one of piezo1 and piezo3 was lost, whereas piezo2a and piezo2b were retained in almost all teleost species. The gray colored boxes indicate gene loss. C-E. Schematic summary of piezo gene expressions in early zebrafish embryos (1–3 dpf). The gene expression domains are color-coded. Multiple piezo genes are expressed at the pharyngeal arches and pectoral fin buds.

Gene expression typically corresponds to its function during embryonic development. Both regulatory and coding regions of the ancestral piezo gene can be duplicated through WGDs. Coding regions generally evolve more slowly than regulatory regions due to functional constraints (True and Carroll, 2002). Thus, duplicates have three fates: loss, sub-functionalization, and neo-functionalization based on the DDC model (Force et al., 1999). Consistently, we observed some overlaps among them (Fig. 6C–E). For example, piezo1, piezo2a, and piezo3 are all expressed in the pectoral fin buds, while only piezo2a is expressed in the caudal fin buds, suggesting a sub-functionalization. Among the four piezo genes, piezo2b has distinct expression in Rohon-Beard neurons. This could be a case of neo-functionalization, but it needs to be confirmed by examining the piezo gene expression in cephalochordates and urochordates in the future.

Calcium signaling in transfected human HEK293T cells suggests that zebrafish Piezo3 is a cation channel, like human PIEZO channels. It is also sensitive to mechanical stimulation and can be inhibited by streptomycin, suggesting that it has evolutionarily conserved functions. Future piezoelectrophysiology using a patch clamp should further clarify and confirm the biophysical properties of piezo3 mechanosensing and mechanoconduting. Interestingly, the piezo3 mutant larvae were more sensitive to tapping mechanical force and showed greater traveled distance in our VMR assays. So, losing Piezo3 made the fish more sensitive to tapping force and more active during locomotion in the light phase, but not in the dark phase. Overall, these results suggest that Piezo3 may have a repressive role in force sensing and photopic response. Its absence in birds and most mammals may facilitate their adaptation to their environment compared to other lower vertebrates.

Piezo channels also play an important role in a variety of vertebrate morphogenesis. Piezo1 is expressed in the mouse cardiovascular system. Mice lacking Piezo1 exhibited embryonic lethality due to the absence of major blood vessels, disorganized embryonic vasculature, and pericardial effusion (Li et al., 2014; Ranade et al., 2014). Similarly, piezo1 is also required for aortic and outflow tract valve development and large arteries (Abello et al., 2025; Duchemin et al., 2019; Faucherre et al., 2020). Piezo1 and Piezo2 are also expressed in skeletal tissues, and are critical for long bones in the limbs (Wang et al., 2020; Zhou et al., 2020) and jawbone development of mice (Nie et al., 2024). In contrast to the mouse, loss of piezo1 is not embryonic lethal, but results in short intermuscular bones in zebrafish and carp (Zhang et al., 2025). Like piezo1, piezo2a null mutant is also viable and fertile. However, the double null of piezo1 and piezo2a in fish larvae only survived to 8 days with scoliosis (Ramli et al., 2023). In contrast to the null mutant, an in-frame deletion mutant fish with 11 amino acid deletions also developed scoliosis (Ramli et al., 2023). This difference could be caused by a dominant negative effect, in which likely other piezo proteins, such as piezo2a, were also inhibited by the in-frame mutant.

Our 3 piezo3 CRISPR mutant zebrafish are viable as adults and show no obvious morphological defects, even in regions where piezo3 is expressed during embryogenesis, such as the eyes, pharyngeal arches, and fin buds. Thus, our results support the conclusion that piezo3 is not a developmentally essential gene for skeletons and survival, given the presence of mRNA and the expected Mendelian ratios of each genotype in our in-cross experiment. One explanation is that zebrafish have four piezo genes, each with redundant functions, as reflected in their overlapping gene expression domains (Figs. 3, S3–S5). This overlap may not be limited to early embryogenesis but could also occur during later organogenesis. For example, the piezo1, piezo2a, and piezo3 are all expressed in the 2–3 dpf pectoral fin buds and pharyngeal arches. The mutation of any single piezo gene may not be sufficient to generate a clear morphological phenotype. No such alterations in these two structures are found in the above-mentioned piezo1 and piezo2a mutants, supporting this idea (Ramli et al., 2023). Indeed, our quantitative RT-PCR results suggest that this redundancy could be the main reason, instead of transcriptional compensation. Future functional and anatomical analyses of double and triple mutants with piezo1 and piezo2a are needed to elucidate Piezo3’s physiological functions. In addition, we recently found that the gene knockin into the promoter region of the zebrafish pax9 gene was effective in reporting gene expression and simultaneously inactivating gene activity (Dong and Zhang, 2026). Such a piezo3 knockin-knockout fish line with a fluorescent protein or a genetically encoded piezo-dependent biosensor (e.g., GenEPi) (Yaganoglu et al., 2023) will be useful for future functional investigations.

Our zebrafish larval behavior assays reveal that piezo3 is involved in larval responses to mechanical and photopic stimuli. However, the roles of piezo3 in the visual and mechanical sensing-response loop remain unknown, particularly with respect to the embryonic development of the organs or tissues involved. Future detailed examination of these structures, such as the lateral line, Mauthner cells, and relevant sensing and motor neurons, will provide critical information.

Overall, we identified a new member of the vertebrate Piezo channel family, piezo3, which originated during vertebrate WGDs and is present in lower vertebrates but has been lost in most birds and mammals. We also found that the zebrafish PIEZO3 channel responded to mechanical stimuli in HEK293T cells, and that PIEZO3 modulated zebrafish larval responses to mechanical and photopic stimuli. Since piezo3 is a new member of the PIEZO channel family, further studies are needed to confirm its functions. Our present data provide an essential foundation for such inquiry.

MATERIALS AND METHODS

Zebrafish strains, husbandry, and CRISPR mutation

Zebrafish were raised and maintained in accordance with AAALAC-approved standards at the Purdue animal housing facility, following protocols approved by the Purdue Animal Care and Use Committee (PACUC, #1210000750). All the zebrafish experiments were carried out in a wildtype TAB (AB/Tübingen, RRID: ZIRC_ZL1) background (Allende et al., 1996; Amsterdam et al., 1999). Zebrafish were maintained according to the Zebrafish Book, and all zebrafish embryos were staged based on the Kimmel staging guide (Kimmel et al., 1995).

The piezo3 loss-of-function mutants were generated using our established CRISPR method (Kim and Zhang, 2020). Briefly, gRNAs were designed against exons 14 and 29, and DNA templates were synthesized by IDT DNA. The gRNAs were prepared using HiScribe^® T7 High Yield RNA Synthesis Kit (NEB E2040L). Then, gRNAs were co-injected with the Cas9 protein (PNA Bio, CP01) into the single-cell stage of fish embryos. The T7E1 assay was used to determine mutations before the fish embryos were raised to adulthood as F0 founders, so that their offspring, F1 fish, will be further screened using the T7E1 assay. Positive F1 adult fish will be established as founder fish lines after sequencing validation.

Bioinformatics and phylogenetic analysis

Zebrafish Piezo channel protein sequences were first identified by searching the (RRID: SCR_006472) and Ensembl (RRID: SCR_002344) databases using human gene symbols. Any zebrafish Piezo channels with partial sequences in Ensembl were further searched for full-length sequences using BLASTp in NCBI. All the retrieved zebrafish piezo genes were also validated in ZFIN (RRID: SCR_002560). Orthologous and paralogous genes were retrieved from Ensembl using the comparative genomics function in GRCz11. For all the orthologous protein sequences, representative protein sequences from the major metazoan taxa were selected based on their taxonomic positions and the quality of available protein sequences. When multiple isoforms were present, the longest sequence was chosen for analysis. Piezo3 orthologs were identified by BLAST search in both NCBI and Ensembl. The final protein sequences selected for the final analysis are listed in Supplementary Table S1.

Phylogenetic and syntenic analyses were performed as we described previously (Hensley et al., 2016a; Hensley et al., 2016b; Silic et al., 2021; Zhang and Cohn, 2006; Zhang et al., 2006). Briefly, multiple protein sequences were aligned using the MUSCLE program with default settings (Edgar, 2004). To identify the best evolutionary model for phylogenetic analysis, we conducted a model selection test using maximum likelihood (ML) with default parameters in MEGA (RRID: SCR_000667) (Kumar et al., 2024). The models with the lowest B.I.C. (Bayesian Information Criterion) scores were considered to describe the substitution pattern the best, and JTT (Jones-Taylor-Thornton model) +G (gamma distribution) was chosen. Then, we constructed phylogenetic trees using PhyML3.1 (Felsenstein, 2004). ML phylogenetic analysis was performed using JTT + G with 1000 bootstrap replicates (Zhang and Cohn, 2006; Zhang et al., 2006). The final phylogenetic trees were viewed and generated with FigTree (RRID: SCR_008515) V1.4.2 (http://tree.bio.ed.ac.uk/software/figtree). The syntenic analysis was first carried out in the Genomicus browser (version 100.01). Individual gene position was then verified in Ensembl, UCSC genome browser (RRID: SCR_011791), and the Synteny Database (Catchen et al., 2009).

For the PIEZO1P2 RNA (ENSG00000237121.1), gene expression data were retrieved from the GTEx Portal: https://www.gtexportal.org. Protein tertiary structures of zebrafish Piezo1, Piezo2a, Piezo2b, Piezo3, and human PIEZO1 and PIEZO2 were generated using ColabFold: AlphaFold2 (https://www.alphafold.ebi.ac.uk, RRID: SCR_023662) with MMseqs2 (v1.5.5) (Kim et al., 2025; Varadi et al., 2022). Full-length amino acid sequences were submitted to the ColabFold notebook using default parameters.

Zebrafish gene cloning

Partial sequences of zebrafish piezo genes were cloned from a mixed cDNA pool. Briefly, About 100 embryos (1–3 dpf) were dechorionated, pooled, and used for total RNA extraction using TRIzol reagent (Thermo Fisher, 15596026) according to the manufacturer’s guidance. The total RNA amount and quality were checked using a Nanodrop spectrophotometer (RRID: SCR_018042). Reverse transcriptions were performed with the SuperScript^® III First-Strand Synthesis System (Thermo Fisher, 18080051) following manual instructions. Phusion^® High-Fidelity DNA Polymerase master mix (New England Biolabs, M0531) or PrimeSTAR^® GXL DNA Polymerase (Takara Bio, R050A) was chosen for PCR amplification. Right-sized PCR products were examined on DNA agarose gels and purified with NucleoSpin Gel and PCR Clean-up Kit (Takara Bio, 740609.250). Purified PCR products of each gene were then cloned into the pJet1.2 vector (Thermo Fisher, K1232) according to the manual instructions. Correct clones were verified for correct insertional gene orientation by endonuclease digestion and sequencing. The full length of the zebrafish piezo3 ORF was synthesized and cloned into the pCDNA3.1 vector by Gene Script to reduce random mutations of long-range PCR. Finally, the piezo3 ORF was subcloned into the pBac vector carrying an IRES-EGFP tag to make a plasmid construct, pBAC-piezo3-IRES-EGFP.

Whole mount in situ hybridization, cryosection, and imaging

Riboprobe DNA templates were prepared either by PCR amplification or plasmid linearization with an endonuclease at the 5’ end of the protein-coding region in the pENTR-D or pDONR221 vector. DNA templates were purified before in vitro transcription. Anti-sense riboprobes were synthesized by in vitro transcription using T7 RNA polymerase (Thermo Scientific, EP0111) and DIG RNA Labeling Mix (Millipore-Sigma, 11277073910) according to the manufacturers’ manuals. All synthesized riboprobes were purified using Sigma Spin post-reaction clean-up columns (Sigma, S5059) and stored at −80°C in a freezer before use.

Whole-mount in situ hybridizations were performed according to our established method, with some modifications (Park et al., 2024; Silic et al., 2021; Silic et al., 2022). Briefly, chorions were removed using pronase (Sigma, PRON-RO) treatment before fixation for 0.5–3 dpf fish embryos. All fish embryos were staged and then fixed with 4% PFA (paraformaldehyde) for 1–2 days at 4°C, followed by dehydration in serial methanol (25%, 50%, 75%, and 100%) with PBST (Phosphate-buffered saline solution with 0.1% tween-20). Dehydrated embryos were stored in 100% methanol at −20 °C until use. We rehydrated fish embryos using a methanol gradient in PBST (100%, 75%, 50%, and 25%) for in situ hybridization experiments. Embryos (12–24 hpf) were then bleached with 6% H₂O₂ in PBST and permeabilized with proteinase K (10 μg/ml in PBT). After permeabilization, embryos were washed with PBT and fixed in 4% PFA with 0.2% glutaraldehyde for at least 3 hrs at room temperature to inactivate proteinase K. Next, fish embryos were washed with PBST (2 × 10 minutes) to remove fixatives. Then, fish embryos were incubated in pre-hybridization solution [50% formamide, 5XSSC (0.75M NaCl, 75mM sodium citrate, pH 7.0.), 2% Roche blocking powder, 0.1% Triton-X, 50 mg/ml heparin, 1 mg/ml Torula yeast RNA, 1 mM EDTA, 0.1% CHAPS, DEPC-treated ddH₂O] overnight at 65°C in hybridization oven with gentle shaking (40 rpm). Riboprobes (1–2 μl) were added on the second day in the morning. Then, embryos were further hybridized for 48–72 hours before washing unbound riboprobes with 2X SSC and 0.2X SSC (3 × 30 minutes for each solution). Next, fish embryos were washed with KTBT 2x for 10 minutes each (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 10 mM KCl, 1.0% Tween-20) before antigen blocking using 1–2% sheep serum (Sigma, S2263) and 2 mg/ml BSA in KTBT for at least 3 hours. Anti-digoxigenin antibody conjugated to alkaline phosphatase (Roche, 11093274910) was added to the blocking solution at a 1:5000 dilution, and embryos were incubated overnight at 4°C with gentle shaking. After 6 times 1 hr washing with KTBT and overnight KTBT incubation at 4°C with gentle shaking, the following day, color reaction was carried out in NTMT solution (100 mM Tris-HCl, pH 9.5, 50 mM MgCl2, 100 mM NaCl, 0.5% Tween-20) with 75 mg/ml NBT, 50 mg/ml BCIP, and 10% DMF (N, N-dimethylformamide). For 2–3 dpf fish embryos, 1mM levamisole was added to reduce the effect of endogenous alkaline phosphatase. Color development was carried out in the dark with gentle rocking. We closely monitor the color density of each reaction. Once the embryos developed suitable color densities, the reaction was stopped with NTMT washing. The finished samples were imaged immediately or stored in 4% PFA for later imaging. For histological analysis, post-hybridization embryos were equilibrated overnight in 15% sucrose, then in 30% sucrose with 20% gelatin, after which they were embedded in 20% gelatin for cryosectioning (10–25 μm) on a cryotome. Images were acquired using an Axiocam 305 color camera on a Zeiss Stereo Discovery V12 (RRID: SCR_027509) and an Axio Imager 2 (RRID: SCR_018876) compound microscope. Whole-mount embryos were imaged in 3% methylcellulose.

Cell Culture, Transfection, and Calcium Signal Measurement

HEK293T (RRID: CVCL_0063) cells were cultured in DMEM (Sigma-Aldrich, #D6429) supplemented with 10% FBS (Gibco, #A3840001) in a cell culture incubator at 37 °C with 5% CO₂. The full-length zebrafish piezo3, pBAC-piezo3-IRES-EGFP (10 μg) and pGP-jRGECO1a (1.5 μg, RRID: Addgene #61563) were co-transfected into HEK293T cells at the 50% cell confluence in the 10-cm cell culture plate (Corning, #353003) using 20μl TurboFect (Thermo Scientific, #R0533) according to the manufacturer’s manual. Transfection efficiency was assessed by measuring GFP and RFP signals around 48h after transfection. The isolated single cells of RFP+/GFP+ and RFP+/GFP− were selected using a Nikon Eclipse Ti2 inverted microscope (RRID: SCR_021068) with a polygon 1000 pattern illuminator and an ORCA-Fusion digital camera. The videos and images were captured at 488 × 488 pixels (0.5 frame/s) using a Crestoptics CICERO spinning disk confocal microscope and NIS-Elements Advanced Research (RRID: SCR_027181) under the RFP (Excitation: 568 nm; Emission: 593–668 nm) and GFP (Excitation: 488 nm; Emission: 500–550 nm) channels.

To assess the basal calcium transient level without stimulation, the RFP signal from selected cells was recorded for 3 minutes, 48h after transfection. For mechanical stimulation, a borosilicate glass capillary (1.0 mm outer diameter) was generated using a Flaming/Brown micropipette puller (Model P-1000, Sutter Instrument, RRID: SCR_016842), and the tip was blunt-ended by brief flaming. The blunt-ended capillary was then mounted on a manual micromanipulator (WPU, M3301) and used to apply transient mechanical force to selected cells. To ensure comparable responses, adjacent RFP+/GFP+ and RFP+/GFP− cells were simultaneously stimulated using a single poking event. RFP calcium signals from the paired adjacent RFP+/GFP+ and RFP+/GFP-cells were continuously recorded for 4 min with 2-second intervals following the poking event. The inhibitory effect of streptomycin was evaluated at low (172 μM) and high (1032 μM) concentrations on the calcium signal after 2 hours’ administration in cell culture dishes. Streptomycin was supplied from a Penicillin–Streptomycin stock solution, corresponding to approximately 17200 μM streptomycin (10,000 μg/mL, Gibco, #15140122). To quantify calcium signals, cytoplasmic red fluorescence was manually defined and measured using ImageJ (RRID: SCR_003070) (Schneider et al., 2012). Briefly, the red channel was extracted from each frame’s RFP images and converted to 8-bit grayscale for fluorescence intensity analysis. Regions of interest (ROIs) corresponding to the cytoplasm of the paired RFP+/GFP+ and RFP+/GFP− cells were manually outlined and kept consistent throughout the analysis. For each frame, the mean pixel intensity within the ROI was calculated and used to represent the RFP signal intensity at the corresponding time point, with fluorescence intensities ranging from 0 to 255 in 8-bit grayscale images. Time-course fluorescence changes were subsequently analyzed to evaluate calcium responses following the mechanical poking events. Calcium responses for each cell were quantified as the change in fluorescence relative to baseline fluorescence (ΔF/F0), where ΔF=F-F0. F represents the RFP fluorescence intensity at a given time point, and F0 represents the baseline RFP fluorescence intensity at the beginning of the recording. To evaluate the overall calcium response during the recording period, the total area under the curve (AUC) of the ΔF/F0 traces was calculated, including both positive and negative deflections relative to the ΔF/F0=0 baseline. The student t-test was performed in GraphPad Prism 9.0 (RRID: SCR_000306) for statistical analysis.

Quantitative RT-PCR

Total RNAs were extracted from the anterior part of 3 dpf fish embryos (Fig. 5D) using TRIzol reagent (Thermo Fisher Scientific, 15596026) following the manufacturer’s guide. About 1 μg of RNA was reverse-transcribed using the Invitrogen^™ SuperScript^™ III First-Strand Synthesis System (Thermo Fisher Scientific, 18080–051) following the manufacturer’s instructions. The resulting cDNAs were used as templates directly. Quantitative RT-PCRs were performed using 2× SYBR Green qPCR Master Mix (Cat# B21202, Selleck Chemicals LLC) in 20 μL reactions containing 1× SYBR mix, 0.5 μM forward and reverse primers (Supplementary Table S2), and 1μL cDNA as template. PCR amplifications were performed on a QuantStudio^™ 3 Real-Time PCR System, 96-well, 0.2 mL format (Thermo Fisher Scientific, A28136) under the following cycling conditions: 95 °C for 3 min, followed by 40 cycles of 95 °C for 10 s and 60 °C for 30 s, with a melt-curve stage to confirm amplification specificity. Relative gene expression was calculated using the 2⁻ΔΔCt method, using ACTB (β-actin) as the internal normalization control (Cui et al., 2017).

Bone and cartilage staining

Fish were euthanized with an overdose of MS-222 and fixed in a freshly prepared solution containing 5% formalin, 5% Triton X-100, and 1% KOH. Specimens were rocked at 42 °C for 48h until pigmentation was removed. For adult fish bone staining, after a brief rinse with deionized water, calcified bone was stained with 0.02% Alizarin Red in 1% KOH for 2–3 days with gentle rocking (Sakata-Haga et al., 2018). During staining, scales were removed under a binocular microscope using fine tweezers to facilitate visualization of the internal skeleton. For larvae and juvenile cartilage and bone staining, a modified double staining protocol was used (Walker and Kimmel, 2007). A double-stain solution was prepared by mixing 10 ml of Alizarin red solution (0.5% Alizarin Red in H2O) and 1ml of Alcian blue solution (0.02% Alcian blue, 200 mM MgCl2, 70% Ethanol). Specimens were incubated in the double-stain solution with gentle rocking for 1–2 days, depending on the staining stage. For both bone and cartilage staining, specimens were then washed in a clearing solution (20% Tween-20, 1% KOH) at 42 °C for 6 h until background coloration was eliminated. Samples were subsequently transferred through a graded glycerol series (25%, 50%, 75%) and stored in 100% glycerol for immediate imaging or long-term preservation. Images were acquired using an Axiocam 305 color camera on a Zeiss Stereo Discovery V12 microscope.

Visual-Motor Response (VMR) and Tapping Assay

At 6 dpf, wildtype and homozygous mutant zebrafish larvae were transferred to Whatman UNIPlate square 96-well plates (VWR, Radnor, PA) along with their respective E3 treatment medium, with one larva per well and 24 larvae per treatment group. Three independent experiments were performed. The plates were placed in lightproof boxes inside the incubator for overnight dark habituation. At 7 dpf, they were transferred to the Zebrabox (Viewpoint Life Sciences, Montréal, QC, Canada) to measure the visual-motor response (VMR) displayed by the larvae. The total irradiance of the LED stimulus across the visible spectrum (380–780 nm) was 310.45 μW/cm^2 (Venkatraman et al., 2020). Larval movement was recorded in tracking mode, which binned activity every second. The data collection protocol consists of a 30-minute dark period, a 1-hour light period, and a 5-minute dark period. All VMR experiments were conducted between 9 a.m. and 6 p.m. to minimize circadian rhythm effects on vision (Emran et al., 2008; Ganzen et al., 2021). The VMR data were first normalized to remove confounding factors, including light-intensity variation across the 96-plate and batch effects across biological replications using an established procedure (Xie et al., 2019). We first analyzed the distance traveled by the larvae during the first second after light onset and offset, a typical approach for detecting visual startle responses in larvae, using a linear mixed-effects model with genotype as the fixed effect and biological replicate as the random effect (Wang et al., 2025). The results showed no significant effect of the genotype on distance traveled (on: β = −0.01341, p = 0.606; off: β = 0.07686, p = 0.0974). We also noticed a substantial difference in locomotor activities during the 1-hour light period. We analyzed this difference using the total distance traveled during the 1-hour period in a linear mixed-effects model, with genotype as the fixed effect and biological replicate as the random effect. The results showed a non-significant biological replicate effect and a significant increase in distance traveled by the piezo3 mutant (β= 63.2, p = 0.0065). Since the biological replicate effect was non-significant, we combined the total distance traveled for each genotype across all replicates and visualized the combined data in a boxplot (Fig. 5AE).

A tapping assay was conducted to evaluate the total distance traveled by larvae in response to a mechanosensory stimulus, i.e., a physical tap. These larvae were transferred to 96-well plates and placed in the Zebrabox as described in the VMR assay section. To induce mechanosensory stimulation, a metal rod was manually tapped against the side of the 96-well plate under constant ambient light. 16 mutants (pu117 or pu118) and 16 wildtype sibling controls were measured in each experiment repeat. The pu117 mutant experiment was repeated once. Each experiment duration consists of 12 taps, with a 10-second interval between taps. Larval responses were recorded by the Zebrabox in tracking mode. Each plate was tested twice. The tapping assay data were analyzed in two steps: 1) filtering and 2) hypothesis testing. First, the total distance data were filtered to remove unsuccessful taps. A tap was considered unsuccessful if the average total distance traveled at that tap was less than two standard deviations above the average total distance traveled during periods between taps. The average total distance traveled during the in-between periods was calculated across all genotypes and all in-between time intervals. Second, the filtered data from all taps and plates were pooled within each genotype. The Wilcoxon rank-sum test was used to test the difference in average total distance traveled between genotypes. The resulting p-values were adjusted using Holm’s method to correct for multiple comparisons (Holm, 1979).

1. Piezo3 is a new member of PIEZO channel family in vertebrates.

2. Piezo1–3 most likely originated through vertebrate whole genome duplications.

3. Piezo3 is present in vertebrates except birds and most mammals.

4. Zebrafish piezo genes show distinct and overlapping expression during embryogenesis

5. Zebrafish Piezo3 conducts calcium ions

6. The piezo3 CRISPR zebrafish mutants do not exhibit evident morphological changes, but zebrafish larvae respond actively to mechanical and photopic stimuli.