G protein-coupled estrogen receptor regulates heart rate and heart valve thickness in zebrafish

Estrogens act by binding to estrogen receptors alpha and beta (ERα, ERβ), ligand-dependent transcription factors that play crucial roles in sex differentiation, tumor growth and cardiovascular physiology. Estrogens also activate the G protein-coupled estrogen receptor (GPER), however the function of GPER in vivo is less well understood. Here we find that GPER is required for normal heart rate in zebrafish embryos and for normal valve thickness in zebrafish adults. Acute exposure to estrogens increased heart rate in wildtype and in ERα and ERβ mutant embryos but not in GPER mutants. Nuclear estrogen receptor signaling remained normal in GPER mutant embryos. However, GPER mutant embryos exhibited reduced basal heart rate while heart rate was normal in ERα and ERβ mutants. We detected gper transcript in discrete regions of the brain but not in the heart. In the brain, we observed gper expression in cells lacking nuclear estrogen receptor activity, suggesting that GPER acts in the brain to regulate heart rate independently of nuclear estrogen receptor signaling. Additionally, blood flow in embryos has been shown to influence heart valve maturation, suggesting the hypothesis that reduced heart rate during embryonic and juvenile development disrupts heart valve maturation. Consistent with this hypothesis, we find that adult GPER mutants have thinner heart valves than wildtype. Our results demonstrate that estradiol plays a previously unappreciated role in the acute modulation of heart rate during zebrafish embryonic development and that GPER functions as an autonomous estrogen receptor in vivo to regulate basal heart rate and heart valve thickness.


INTRODUCTION
Zebrafish are an established model for human cardiovascular development and function (1) with conserved estrogen signaling (2)(3)(4). While studying the function of ERα (esr1) in zebrafish embryonic heart valves (5,6), we serendipitously observed that estrogen receptor modulators caused acute changes in heart rate.
Estrogens bind two classes of receptors: nuclear hormone receptors (ERa, ERβ) that are ligand-dependent transcription factors (7), and the G protein-coupled estrogen receptor (GPER, also known as GPR30), an integral membrane protein (8,9). It has been difficult to tease apart to what degree ERα and/or ERβ are involved in regulating GPER function in vivo. The observations that ERα can directly activate G proteins in cultured cells (10)(11)(12)(13) and that GPER coimmunoprecipitated with ERα in tumor cells (14) has been used to argue that either GPER is dispensable for estrogen-dependent signaling or that GPER mediates interactions between ERα and G proteins (15). Studies using GPERdeficient mice implicate GPER in ventricular hypertrophy (16), regulation of blood pressure and vascular tone (17,18) and atherosclerosis progression (19), but whether nuclear ER signaling is required for GPER function in these contexts is unknown. Additionally, these studies examined GPER function in adult animals, while the role of GPER during embryonic development is not well understood.
Here we use zebrafish embryos, an established model of human development, to reveal a new function for GPER during cardiovascular development.
Estrogen signaling often differs between males and females. However, zebrafish embryos and larvae are bipotential hermaphrodites that have not begun to sexually differentiate before approximately 10 days post fertilization (dpf) (20), meaning that estrogen levels are uniform between age-matched embryos.
Additionally, zebrafish embryos develop outside of the mother and not within a confined space, such as the uterus. Therefore, zebrafish embryos are not subject to local estrogen concentration gradients, as has been reported to occur in rodents depending upon their position in utero and their proximity to embryos of the same or opposite sex (21,22). These developmental traits make zebrafish a powerful model to study how sex hormone signaling influences the formation and function of non-gonadal tissues. Using complementary genetic and pharmacologic approaches, we sought to characterize how estradiol regulates heart rate and determine to what extent each estrogen receptor mediates estradiol-dependent changes in heart rate in zebrafish embryos.

RESULTS
We exposed 49 hour post fertilization (hpf) embryos to 17β-estradiol (estradiol) and assayed heart rate following one hour exposure. We found that estradiol exposure caused an approximately 20% increase in heart rate ( Fig. 1, mean difference in heart rate between estradiol and vehicle exposed embryos was 26.51 ± 3.36 (standard error) beats per minute (bpm)). Exposure to progesterone, a structurally similar steroid sex hormone, had no effect on heart rate ( Fig. 1, mean difference in heart rate 2.31 ± 6.54), suggesting that the effects on heart rate were specific to estrogens.
To explore whether heart rate was influenced by nuclear estrogen receptor or GPER signaling pathways, we employed a pharmacological approach. We exposed embryos to ICI182,780 (fulvestrant), a well-characterized ERα and ERβ antagonist (23) that also acts as a GPER agonist (8). Following one-hour exposure to ICI182,780, heart rate was significantly increased (Fig. 1, mean difference in heart rate 29.81 ± 4.75 bpm). This effect was blocked by coadministration of G36, a specific GPER antagonist (24) (Fig. 1, mean difference in heart rate 4.10 ± 6.23 bpm), suggesting that estradiol increases heart rate via GPER. We also exposed embryos to G1, a specific GPER agonist with no detectable agonist activity against nuclear estrogen receptors (25), and found that heart rate increased significantly (Fig. 1, mean difference in heart rate 40.98 ± 6.35 bpm). Together, our pharmacological results suggest that GPER regulates heart rate acutely.
To definitively test the hypothesis that estradiol regulates heart rate via GPER, we generated GPER mutant embryos, exposed them to estrogen receptor modulators and assayed heart rate. Using CRISPR-Cas technology (26), we generated embryos with a 131 basepair deletion in the gper open reading frame ( Fig. 2A, B). Embryos were viable and grossly normal, allowing us to measure heart rate (Fig. 2C, D). We exposed homozygous maternal zygotic gper mutant embryos (MZgper-/-) to estradiol or to ICI182,780 and found no increase in heart rate compared to embryos exposed to vehicle (Fig. 2E). Our results demonstrate that estradiol increases heart rate in a GPER-dependent manner. Note that zygotic gper mutants exhibited increased heart rate in response to estradiol (Fig. S1, mean difference in heart rate 29.11 ± 3.56 bpm), indicating that GPER is maternally deposited into oocytes and expressed in embryos. This is consistent with previously published results that detected gper transcript in zebrafish embryos at 1 hour post fertilization, suggesting the presence of maternally loaded gper mRNA (27).
To test whether endogenous estrogens regulate heart rate during embryonic development, we examined basal heart rate in GPER mutant embryos reared in untreated water, reasoning that if heart rate was reduced, then that would suggest that endogenous estradiol regulates heart rate via GPER. We compared heart rate in wildtype versus MZgper-/-embryos at 50 hpf and found that MZgper-/-embryos had reduced heart rate compared to wildtype (Fig. 2F, mean difference in heart rate between wildtype and mutant -30.80 ± 7.07 bpm).
These results demonstrate that GPER is required for normal basal heart rate in embryos and strongly suggest that endogenous estrogens influence heart rate via GPER.
Whether GPER acts as an autonomous estrogen receptor in vivo is controversial. Previous reports suggest that GPER activity might require interaction with nuclear estrogen receptors at the membrane or that estrogens activate GPER indirectly, by binding to nuclear receptors in the cytosol that then activate downstream proteins, including GPER (15,28). To determine whether nuclear estrogen receptors influence heart rate, we generated zebrafish with loss-of-function mutations in each nuclear estrogen receptor gene: esr1 (ERa), esr2a (ERβ1) and esr2b (ERβ2) (Fig. S2-S4). All mutant embryos were viable and grossly normal, allowing us to measure heart rate ( Fig. S2-S4). To test whether estradiol increases heart rate via nuclear estrogen receptors, we exposed 49 hpf esr1 -/-, esr2a -/and esr2b -/embryos to estradiol or vehicle for one hour and assayed heart rate. Following estradiol exposure, heart rate was increased in all mutants compared to vehicle control (Fig. 3A, mean difference in heart rate between estradiol and vehicle 25.04 ± 5.68 bpm for esr1 -/-, 37.23 ± 7.66 bpm for esr2a -/-, 32.48 ± 1.92 bpm for esr2b -/-), similar to what we observed when wildtype embryos were exposed to estradiol (Fig. 1). These results demonstrate that nuclear estrogen receptors are not necessary for estradioldependent increase in heart rate.
To test whether endogenous estrogens regulate heart rate via nuclear estrogen receptors, we bred heterozygous fish to generate embryos homozygous for mutations in either esr1, esr2a or esr2b genes and assayed heart rate in 50 hpf embryos. We observed no significant difference in basal heart rate between homozygotes, heterozygotes or wild type siblings within the same clutch (Fig. 3B, mean difference in heart rate between homozygote and wildtype -4.34 ± 1.37 for esr1, -0.46 ± 3.75 for esr2a, -2.37 ± 3.26 for esr2b; between heterozygote and wildtype -3.34 ± 1.02 for esr1, -0.91 ± 1.53 for esr2a, 0.63 ± 1.66 for esr2b).
These results demonstrate that nuclear estrogen receptors are not required for the establishment of normal basal heart rate in embryos.
It is possible that the mutations generated in each nuclear estrogen receptor gene do not cause loss of functional estrogen receptor proteins. To exclude this possibility and show that esr mutants exhibit loss of functional ER proteins, we generated esr mutants on the Tg(5xERE:GFP) c262/c262 transgenic background, where green fluorescent protein (GFP) expression occurs in cells with activated nuclear estrogen receptors (5) (referred to as 5xERE:GFP).
Previous studies using whole mount in situ hybridization demonstrated that esr1 is expressed in embryonic heart valves while esr2b is expressed in the liver (6), therefore we hypothesized that mutants would fail to upregulate GFP in tissues where the relevant receptor is normally expressed. We exposed 2-3 day post fertilization (dpf) 5xERE:GFP, 5xERE:GFP;esr1 -/-, 5xERE:GFP;esr2a -/and 5xERE:GFP;esr2b -/embryos to 100 ng/ml 17β-estradiol overnight and assayed fluorescence. Consistent with esr gene expression patterns, 5xERE:GFP;esr1 -/larvae exhibited fluorescence in the liver but not in the heart (Fig. S2), whereas 5xERE:GFP;esr2b -/larvae exhibited fluorescence in the heart but not in the liver (Fig. S4). esr2a transcript was not detected at these embryonic and larval stages (6) and, as expected, we saw no change in fluorescence between 5xERE:GFP and 5xERE:GFP;esr2a -/- (Fig. S3). We conclude that the zebrafish nuclear estrogen receptor mutants lack estrogen receptor function.
Deleterious mutations can induce genetic compensation (29), however results from the 5xERE:GFP esr mutants suggest that compensatory expression of esr genes is not occurring. For example, it is possible that in the esr1 mutant there is compensatory upregulation of esr2a and/or esr2b that masks a heart rate phenotype. If esr2a or esr2b were upregulated in esr1 mutants, then we would expect to see fluorescence in the heart in 5xERE:GFP;esr1 -/embryos. Instead, we observed no fluorescence in the hearts of 5xERE:GFP;esr1 -/embryos ( Fig.   S2). Similarly, we observed no ectopic fluorescence in 5xERE:GFP;esr2b -/embryos ( Fig. S4), suggesting that esr genes are not compensating for one another in the multiple zebrafish esr mutants.
To further test whether nuclear estrogen receptor signaling is influenced by GPER, we generated gper mutants on the 5xERE:GFP transgenic background and asked whether estradiol exposure reduced nuclear estrogen receptor activity in mutants compared to wildtype. Following overnight exposure to estradiol, 3 dpf 5xERE:GFP and 5xERE:GFP;MZgper-/-larvae exhibited similar fluorescence (Fig. S5). These results demonstrate that nuclear estrogen receptor transcriptional activity does not require GPER and support the hypothesis that GPER acts as an autonomous estrogen receptor in vivo.
Heart rate can be modulated by cardiomyocytes in the heart, or by cells in the central nervous system, which directly innervates the heart to modulate heart rate and also regulates the release of humoral factors, such as thyroid hormone, that bind to receptors in cardiomyocytes and regulate heart rate (30). To determine whether GPER regulates heart rate tissue autonomously, we performed whole mount in situ hybridization to test whether gper transcripts are expressed in 50 hpf zebrafish embryo hearts. We did not detect transcript in the heart or in the vasculature. In contrast, we detected gper mRNA in three discrete anatomic areas of the brain: the preoptic and olfactory areas and in the ventral hypothalamus ( Fig. 4A-C). Thus, gper localization is consistent with the hypothesis that GPER acts in the brain, and not through cells in the heart, to regulate heart rate.
Genetic evidence using esr mutants suggests that GPER acts independently of nuclear estrogen receptors to regulate heart rate (Fig. 3). To further test the hypothesis that GPER acts as an autonomous estrogen receptor in vivo, we asked whether GPER and nuclear estrogen receptors are expressed in the same cells in the brain, reasoning that if GPER and nuclear estrogen receptors fail to colocalize, this would support the idea that GPER acts as an autonomous estrogen receptor in vivo. We took 5xERE:GFP embryos at 1 day post fertilization and exposed them overnight to 100 ng/ml estradiol. When the embryos reached 48 hpf, we used two color fluorescent in situ hybridization to detect gfp and gper transcripts simultaneously. Since all three nuclear estrogen receptor genes activate the 5xERE:GFP transgene, detecting gfp allows us to monitor activity of all three estrogen receptors using a single RNA probe. In the olfactory and preoptic areas, we found no colocalization between gfp and gper Because MZgper-/-fish are viable, we asked whether loss of GPER and reduced basal heart rate in embryos might have an effect on adult heart morphology. We dissected hearts from 8-9 month old adult wildtype and MZgper-/-zebrafish. Female mutants had a mean 47% reduction in atrioventricular (AV) valve width compared to wildtype ( Fig. 5E; MZgper-/-33.5 ± 6.35 μm (mean ± standard deviation), n=10; wildtype 62.49 ± 18.53, n=9, p=0.0002 Student's t test), while male mutants had a mean 40% reduction in AV valve width compared to wildtype ( Fig. 5F; MZgper-/-29.39 ± 6.32, n=7; wildtype 49.14 ± 9.68, n=5, p=0.0016 Student's t test). We also analyzed the single thickest part of AV valve leaflets in each heart and found that mutant valves were on average 45% thinner at their thickest point than wildtype valves from fish of the same sex (female These results are consistent with previous observations that eliminating blood flow in embryos impairs heart valve formation (31). Together, these findings demonstrate a correlation between reduced blood flow in embryos and reduced valve thickness in adults and suggest that GPER regulation of heart rate is important for proper heart valve maturation.

DISCUSSION
Our results support the hypothesis that GPER acts as an autonomous estrogen receptor in vivo. Previous reports using cultured cells demonstrated that fluorescently labeled or isotopic estradiol specifically binds membranes from cells expressing GPER (8,9). Additionally, estradiol exposure increased cyclic AMP and calcium levels in HEK293 and COS7 cells in a GPER-dependent manner (8,9), while estradiol exposure increased phosphoinositide 3-kinase activity in SKBR3 breast cancer cell line in a GPER-dependent manner (8). However, because these studies utilized cells that either express artificially high levels of GPER or are tumorigenic, the findings do not address whether GPER acts as an estrogen receptor in vivo under normal physiologic conditions. Our genetic and pharmacologic results strongly suggest that GPER is an estrogen receptor in vivo. If estradiol was binding to ERα or ERβ, and these receptors activated GPER, then we would expect to see no increase in heart rate in esr1, esr2a or esr2b mutants following exposure to estradiol. Instead, all esr mutants responded normally to estradiol exposure (Fig. 3), suggesting that ER and GPER signaling pathways are distinct in this context. Consistent with these results, we found gper transcript expressed in cells in the brain that lack nuclear estrogen receptor activity (Fig. 4), further supporting the hypothesis that GPER responds to estrogens independently of nuclear estrogen receptors in vivo.
A study in mouse hearts also supports the idea that GPER acts as an autonomous estrogen receptor in vivo. In adults, estradiol administration reduces cardiac damage following ischemia (32)(33)(34). To identify the receptor required for estradiol's protective effects, Kabir and colleagues subjected hearts from male mutant mice lacking either GPER, ERα or ERβ to ischemia-reperfusion injury in the presence of estradiol or vehicle. Estradiol treatment protected wildtype and ERα and ERβ mutant mice from injury, but had no effect on GPER mutant mice (35), demonstrating that estradiol exerts its protective effects via GPER, independently of ERα or ERβ. The extent to which ER and GPER signaling pathways interact likely depends on cell type, developmental stage, sex and/or pathology. Studying the influence of estrogens on heart rate in zebrafish embryos is a powerful in vivo system where GPER activity is dissociated from classical nuclear estrogen receptor signaling.
Between 2 and 5 dpf, zebrafish heart rate normally increases (36,37). Our results support the hypothesis that endogenous estradiol regulates this increase in heart rate. The finding that GPER mutant embryos have lower basal heart rate compared to wildtype embryos implicates endogenous estradiol. Additionally, a recent HPLC analysis of endogenous estradiol concentration in zebrafish embryos found that estradiol concentrations increased from 137 pg/embryo at 48 hpf to 170 pg/embryo at 72 hpf (38). Taken together, these results support the hypothesis that endogenous estradiol regulates heart rate in zebrafish embryos and larvae. The source of embryonic estradiol, whether synthesized by the embryo or maternally deposited in the yolk, is not known.
Blood flow in embryos is a critical determinant of cardiovascular development (39). Zebrafish embryos in which blood flow is drastically reduced or eliminated exhibited a valve dysgenesis phenotype (31). MZgper-/-embryos exhibit a ~20% reduced heart rate compared to wildtype, yet are viable, allowing us to examine the adult heart following reduced basal heart rate in embryos.
Based on the heart rate phenotype in MZgper-/-embryos and the detection of gper transcripts in the brain but not the heart, we propose that the reduced heart rate in GPER mutant embryos causes a reduction in blood flow that prevents heart valves from maturing properly and manifests in adults as thinner atrioventricular valves compared to wildtype. While gper transcript is expressed in the adult zebrafish brain (2), it is not known whether gper is expressed in the juvenile or adult zebrafish heart, where it could directly influence valve cell division and/or valve cell survival separately from influencing heart rate and blood flow.
There are several mechanisms by which GPER activity in the brain could regulate heart rate. Neurons that express GPER could be part of a chain of neurons that ultimately terminates in motor neurons that innervate the heart and regulate contractions. It is not known whether the zebrafish heart is innervated by 50 hpf and thus it is unclear whether GPER is influencing sympathetic or parasympathetic neuron activity. Parasympathetic neurons terminate at the heart and activate muscarinic acetylcholine receptors to reduce heart rate. In contrast, sympathetic neurons terminate at the heart and activate beta-adrenergic receptors to increase heart rate. GPER could be activating sympathetic activity or reducing parasympathetic activity. However, there is conflicting evidence as to whether autonomic control of the heart occurs in two-day old zebrafish embryos.
Milan and colleagues reported that propranolol, a beta-adrenergic receptor antagonist, reduced heart rate in 2 dpf embryos, while isoproterenol, a betaadrenergic receptor agonist, and atropine, a muscarinic acetylcholine receptor antagonist, both increased heart rate (40). These results suggest that at 2 dpf, the zebrafish embryo heart can respond to sympathetic and parasympathetic activity. In contrast, Schwerte and colleagues reported that atropine and propranolol did not influence heart rate in embryos younger than 5 dpf, while zebrafish failed to respond to adrenergic stimulation before 4 dpf (41). Future studies are required to determine when functional autonomic innervation of the zebrafish heart initially occurs.
Even if the autonomic nervous system does not regulate heart rate in 2 dpf zebrafish embryos, GPER can still function in the brain to regulate heart rate.
Therefore, GPER activity could trigger neuronal activity that leads to systemic release of humoral factors, such as thyroid hormone, known to regulate heart rate (30). Expression of gper transcript in the ventral hypothalamus / pituitary ( Figure 4) is consistent with this hypothesis.
While our results illuminate GPER signaling in the context of embryonic heart rate, it is not clear to what extent GPER influences heart rate at later stages of development. At larval, juvenile and adult stages it is impossible to assess heart rate without immobilizing or anesthetizing zebrafish, manipulations that themselves influence heart rate. In adult mice with mutations in GPER, there was no significant difference in basal heart rate between mutant and wildtype mice of either sex (16,17,43). It is possible that GPER regulates heart rate in embryos but not in adults. Additionally, heart rate in GPER mutant mice was assayed using general anesthesia, which is known to depress heart rate compared to conscious mice (44). Anesthesia may mask the effect of GPER on basal heart that we observe in conscious animals. We also cannot exclude the possibility that the effects of GPER on heart rate are specific for zebrafish.
In summary, this study identified a role for GPER in the regulation of embryonic heart rate. The zebrafish estrogen receptor mutants we developed enable experiments to rapidly and conclusively identify the causative estrogen receptor associated with any estrogen signaling phenotype, as demonstrated with the estradiol-dependent increase in heart rate reported here. This has significant implications for studies of estrogenic environmental endocrine disruptors, which are frequently tested on zebrafish to identify effects on embryonic development, organ formation and function (45). Zebrafish estrogen receptor mutants can now be used to determine whether such effects are specific for estrogen receptors and to identify the precise receptor target. Our results also establish a need to consider the impact on cardiac function when considering the toxicity of estrogenic environmental endocrine disruptors.

Zebrafish
Zebrafish were raised at 28.5°C on a 14-h light, 10-h dark cycle in the UAB Zebrafish Research Facility in an Aquaneering recirculating water system (Aquaneering, Inc., San Diego, CA). Wildtype zebrafish were AB strain (46) and all mutant and transgenic lines were generated on the AB strain. To visualize nuclear estrogen receptor activity, transgenic line Tg(5xERE:GFP) c262/c262 was used for all studies unless otherwise mentioned (5). All procedures were approved by the UAB Institutional Animal Care and Use Committee.

Embryo collection
Embryos were collected during 10 minute intervals to ensure precise developmental timing within a group.

Measurement of heart rates
All embryos were reared at 28.5 o C and heart rate was measured at room temperature. Following one hour incubation in treatment compounds at 28.5 o C, heart rate (beats per minute, bpm) was calculated by counting the number of heart beats in fifteen seconds and multiplying that number by four. Prior to measurements, each dish was removed from the incubator and placed under the microscope light for 4 minutes at room temperature, allowing embryos to acclimate to the light and eliminate any effect of the startle response. Control groups were counted first and last to ensure that the overall heart rate did not increase during the duration of counting due to natural increases in heart rate during development. All heart rates were measured on a Zeiss Stemi 2000 dissecting microscope with a halogen transmitted light base (Carl Zeiss Microimaging, Thornwood, NJ).

Generation of guide RNA and Cas9 mRNA
Plasmids pT7-gRNA and pT3T3-nCas9n were obtained from Addgene (numbers 46759, 46757) (26). pT7-gRNA was digested simultaneously with BsmBI, BglII and SalI for one hour at 37 o C followed by one hour at 55 o C. To generate esr2a, esr2b and gper gRNAs, oligonucletides containing target site sequences (see

Live imaging
Live zebrafish embryos and larvae were visualized using a Nikon MULTIZOOM AZ100 equipped with epi-fluorescence and an Andor Clara digital camera unless otherwise noted. To validate mutants with 5xERE reporter activity, larvae were treated overnight with 100 ng/mL estradiol beginning at 2-3 dpf. Following overnight treatment, larvae were washed in E3B, anesthetized with 0.04% tricaine and imaged in Petri dish containing E3B. For Fig. S1 H-K, larvae were mounted in bridged coverslips in E3B with 0.04% tricaine (46). Images were captured on a Zeiss Axio Observer.Z1 fluorescent microscope equipped with an Axio HRm camera and Zen Blue 2011 software (Carl Zeiss Microscopy, Oberkochen, Germany). Adjustments, cropping and layout were performed using Photoshop CS6 and InDesign CS6 (Adobe Systems Inc., San Jose, CA).

RNA in situ hybridization
For synthesis of RNA probes, full-length gper open reading frame was amplified by PCR from genomic DNA extracted from 3 dpf larvae (gper is a single exon gene and therefore the open reading frame sequence is identical in genomic and cDNA) using primers 5'-ATGGAGGAGCAGACTACCAATGTG-3' and 5'-CTACACCTCAGACTCACTCCTGACAG-3' and TA cloned into pCR2.1 vector (Invitrogen). amhc and gfp probes were used as described (5,49). All clones were verified by sequencing. Digoxigenin-labeled antisense RNA and FITC-labeled antisense RNA were transcribed using T7 and T3 polymerase, respectively, as previously described (5). Colorimetric whole-mount in situ hybridization was performed on zebrafish embryos and larvae as described previously, using 5% dextran in the hybridization buffer (50,51). Following colorimetric in situ hybridization, embryos were sequentially cleared in glycerol (25%, 50%, 75% in phosphate buffered saline), mounted in 4% low-melting temperature agarose, and imaged using a Zeiss Axio Observer.Z1 microscope with Zeiss Axio MRc5 camera and Zen Blue 2011 software. Fluorescent in situ hybridization (FISH) was performed as previously described (51) with the following modifications: After rehydration, Proteinase K treatment was extended to 35 minutes. Following hybridization, embryos were washed in 2xSSC prior to being placed in PBT. Embryos were blocked in 2% Roche blocking reagent in 100 mM Maleic acid, 150 mM NaCl, pH 7.5 (52). For double labeling, following development of anti-DIG-POD antibody, reaction was inactivated in 100 mM glycine pH 2 for 10 minutes then incubated in anti-FITC antibody. Following florescent in situ hybridization, embryos were cleared in 50% glycerol, mounted on a bridged coverslip and imaged using a Nikon A1/R scanning confocal microscope with Nikon Advanced Elements software.

Histology on adult zebrafish hearts
Adult wild-type (n=15) and MZgper -/-(n=17) zebrafish at 8-9 months of age were used for heart dissections. Zebrafish were anesthetized in 0.2 mg/mL tricaine, measured with a digital caliper to obtain standard length (SL) (53), and decapitated dorsal to the pectoral fin, then hearts were dissected with forceps in phosphate buffered saline (PBS). Hearts were examined for structural integrity and fixed in 1 mL of 10% formaldehyde in PBS at 4°C for 16-20 hours overnight. Whole hearts were washed three times in PBS.
To confirm heart integrity following dissection and fixation, whole hearts were imaged on a Nikon SMZ1500 stereomicroscope equipped with a Nikon DS-Qi1MC camera. Intact, properly dissected hearts were embedded in 10-15 µL of Histogel (Thermo Scientific) on 0.8 µm AA Millipore filter paper and oriented with atrium and ventricle in the same horizontal plane to ensure downstream collection of coronal sections. Histogelembedded hearts were placed in tissue cassettes (Fisher Scientific #22-272420) and allowed to set on ice for 5 minutes. Excess Histogel was trimmed without disturbing the heart. Cassettes were closed and stored in 70% ethanol until further processing. Hearts were processed routinely into paraffin, embedded, sectioned at 5 µm and stained with hematoxylin and eosin (H&E). Bright-field images of sections were obtained using a Zeiss Axio Observer.Z1 microscope with a Zeiss Axio MRc5 camera and 20x objective (NA 0.8). Tiled images were captured and fused using the stitching algorithm of Zeiss ZEN 2 blue edition software. Cross-sectional length measurements were obtained by taking the mean of 6 sections from each heart (on average) determined to be in the middle of the heart by AV valve visibility. In each section, the thickness of the largest valve leaflet was measured perpendicular to the long axis of the valve. Zebrafish growth and size vary within groups of similarly aged adult fish (53). Therefore, measurements for each sex were normalized to standard length (SL), an established measure of postembryonic zebrafish development (53), by dividing the individual measurements for each fish into the mean SL for that sex. Comparisons between genotypes for each sex were made with two-tailed, unpaired Student's t test. Statistical significance was accepted at a P value of <0.05.

Experimental design and data analysis
Hart rate assays were conducted in separate experiments. Each experiment included comparing groups (treated vs untreated or mutant vs wildtype) using at least 3 embryos per group with all embryos from the same clutch. All experiments were replicated for at least 3 times (n≥3) using different clutches on different days. This is essentially a complete block design with clutch/day as block. Mean heart rate of individual embryos from a clutch was used for comparing treatment groups (or mutant groups) within experiments using two-way ANOVA controlling for clutch/day effect. The overall treatment effect (or the genotype effect in some experiments) was tested using F test. If it was significant, Dunnett's test was then used to compare each treatment group with the vehicle group or mutant group with the wildtype group. For some special individual pairs of comparisons, paired t test was used. Significance level is 0.05. All the analyses were conducted using R (version 3.0.2). Graphs were produced using GraphPad Prism 7.0a software.