Trapping Cardiac Recessive Mutants via Expression-based Insertional Mutagenesis Screening

Abstract

Rationale

Mutagenesis screening is a powerful genetic tool for probing biological mechanisms underlying vertebrate development and human diseases. However, the increased colony management efforts in vertebrates impose a significant challenge for identifying genes affecting a particular organ such as the heart, especially those exhibiting adult phenotypes upon depletion.

Objective

We aim to develop a facile approach that streamlines colony management efforts via enriching cardiac mutants, which enables us to screen for adult phenotypes.

Methods and Results

The transparency of the zebrafish embryos enabled us to score 67 stable transgenic lines generated from an insertional mutagenesis screen using a transposon-based protein trapping vector. Fifteen lines with cardiac monomeric red fluorescent protein (mRFP) reporter expression were identified. We defined the molecular nature for 10 lines and bred them to homozygosity, which led to the identification of one embryonic lethal, one larval lethal, and one adult recessive mutant exhibiting cardiac hypertrophy at one year of age. Further characterization of these mutants uncovered an essential function of methionine adenosyltransferase II, alpha a (mat2aa) in cardiogenesis, an essential function of mitochondrial ribosomal protein S18B (mrps18b) in cardiac mitochondrial homeostasis, as well as a function of DnaJ (Hsp40) homolog, subfamily B, member 6b (dnajb6b) in adult cardiac hypertrophy.

Conclusions

We demonstrate that transposon-based gene trapping is an efficient approach for identifying both embryonic and adult recessive mutants with cardiac expression. The generation of a Zebrafish Insertional Cardiac (ZIC) mutant collection shall facilitate the annotation of a vertebrate cardiac genome, as well as enable heart-based adult screens.

INTRODUCTION

Upon completion of the human genome project, the identification of 20,000+ genes facilitated a systematic approach for characterizing genes that participate in a particular biological process and/or disease.¹ The Cardiovascular Gene Ontology (GO) Annotation Initiative represents such an effort that summarizes the functional knowledge of gene products across all species using structured biological vocabularies² (http://www.geneontology.org/GO.cardio.shtml). More than 4,000 genes implicated in heart development, cardiovascular processes and cardiac diseases have been compiled through the extraction of human genes associated with cardiovascular-related GO processes, genome-wide cardiovascular gene association analyses, and knockout studies of developmentally important genes. However, the current cardiac gene list is incomplete, and the expression and function of each individual gene have not been annotated.

Mutagenesis screening is a powerful genome-wide gene discovery tool used in lower model organisms such as yeast, C. elegans and Drosophila, which has successfully been extended to vertebrates.^{3, 4} An ethylnitrosourea (ENU)-based mutagenesis screen has been conducted to seek mouse mutants that mimic congenital heart diseases⁵ and identify adult modifiers of cardiomyopathy.⁶ The scale of a mutagenesis screen in vertebrates is rather limited because of the significant increase of resources required for colony management compared with lower model organisms. Moreover, most recessive mutants exhibit adult phenotypes that are only detectable either with the use of sophisticated phenotyping techniques or when animals are under certain stresses.⁷ Therefore, most screens in vertebrates have focused on embryonic lethal mutants, which only account for 5–10% of the genome.^{8, 9}

Due to its low-cost maintenance and highly prolific nature, the zebrafish (Danio rerio) has emerged as an increasingly popular vertebrate model for mutagenesis screening. The first two large-scale mutagenesis screens using ENU as a mutagen generated hundreds of mutants, which laid the foundation for the zebrafish to become a premier animal model.^{8, 10} To address the limitations of molecular cloning, mutagens with a sequence tag have been exploited, including either retrovirus- ^{9, 11} or transposon-based vectors.^{12, 13} In mice, the transposon-based gene-trapping approach has been successfully exploited in embryonic stem cells, and this approach has been recently extended to an entire animal.^14–16 Efficient transposon-based gene trapping has been recently established in zebrafish using pGBT-RP2.1 (RP2), which contains both gene reporting and gene-breaking cassettes.¹⁷ When inserted into an intron in an endogenous gene, RP2 hijacks the targeted splicing donor sequence, truncates the encoded protein, and disrupts gene function with high knockdown efficiency.

Here, we leveraged the simplified colony management and high mutagenicity of the RP2 cassette to establish zebrafish as a useful model vertebrate animal for introducing biases in the genes that might be screened for adult cardiac phenotypes. To enrich the cardiac mutants, we integrated an expression-based enriching strategy with mutagenesis screening. Ten insertional lines with embryonic heart expression were bred into homozygosity and then raised to adulthood. Three mutants exhibiting embryonic lethal, larval lethal, and adult recessive phenotypes were identified. Our results demonstrate that transposon-based gene trapping in zebrafish is an efficient approach to identify genes with cardiac expression. The generation of a Zebrafish Insertional Cardiac (ZIC) mutant collection shall facilitate the annotation of the genes in the cardiac genome, and open the door for systematic identifying genetic modifiers of major cardiac diseases.

METHODS

For a detailed description, see the Online Methods in Supplement Material available at http://circres.ahajournals.org.

Ethics statement

Zebrafish (danio rerio) were maintained and handled according to the “Guide for the Care and Use of Laboratory Animals” published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996) and the Mayo Clinic Institutional Animal Care and Use Committee (IACUC). The approved IACUC protocol number is A17610. The IRB protocol that describes the method for extracting RNA from the human heart is 10-000216.

Generating and cloning GBT lines

The gene-break transposon (GBT) mutagenesis vector GBT-RP2.1 (RP2) plasmid solution(1 nL of a 50 ng/μL) was combined with 100 ng/μL of Tol2 transposase mRNA and injected into zebrafish embryos at the single-cell stage to generate insertional lines as described previously.¹⁷

Three methods including inverse PCR, 5′-RACE, and 3′-RACE were combined to clone and validate the GBT integration sites. Inverse PCR was conducted to determine the exact integration positions of RP2. Genotyping primers flanking the predicted integration site were designed to confirm the integration position using a RP2 vector-specific primer combined with a gene-specific primer. Occasionally, the exact integration site of the transposon within the targeted intron was determined using a series of PCR primers spanning the intron.¹⁷

Cardiac phenotyping

Fraction shortening (FS), ventricle chamber size, and the heart rate of an embryonic zebrafish heart were measured by documenting videos of the beating heart.¹⁸ To quantify the cardiomyocyte cell size, number, PCNA, and apoptosis, dissected embryonic hearts were subjected to immunostaining as described.¹⁹ To determine the adult heart enlargement phenotype in the GBT0411 line, the index of ventricular surface area to body weight (VSA/BW) was used according to the previously described method.^{20, 21} Frozen sections (10–12 μm) of adult heart ventricles were subjected to immunostaining using previously described methods.²¹ The stained images were captured using a Zeiss Axioplan II microscope equipped with ApoTome and AxioVision software (Carl Zeiss).

Statistical analysis

All values are presented as the mean ± standard deviation. Student’s t-test was used for comparisons between special two groups and ANOVA was used for assessment differences among three or more groups. Significance was accepted at P<0.05.

RESULTS

Fifteen GBT lines exhibit cardiac mRFP expression

The RP2 plasmid and transposase mRNA was injected into 4,000 embryos to initiate the mutagenesis screen. Embryos showing a mosaic monomeric red fluorescent protein (mRFP) signal in the heart at 2–4 days post fertilization (dpf) were pre-selected to enrich for lines with cardiac expression. Among the 545 pre-selected embryos, 441 embryos survived to adulthood and were defined as F₀ founders. Subsequently, outcrosses of these F₀ founders were performed, and 67 stable F₁ insertional lines were generated (Online Figure I). The mRFP expression in the heart during the first four days of embryogenesis was detected in 15 F₁ lines using fluorescence microscopy (Figure 1A). Other lines, such as GBT0365, which exhibited mRFP-tagged gene expression in the central nervous system, did not show detectable expression in the embryonic heart.

Figure 1. A collection of 15 GBT lines with cardiac expression of tagged genes.

(A) Lateral views of GBT embryos at 2–4 days post fertilization (dpf) are shown. mRFP expression was observed in the heart (arrowhead) of 15 GBT lines but not in other GBT lines, such as GBT0365 (arrow), which exhibited neuronal expression. Scale bar=200 μm. (B-C) Cardiac expression analysis of tagged genes in either embryonic (B) or adult zebrafish hearts (C) by RT-PCR. (D) Cardiac expression analysis of human orthologs of tagged genes using RT-PCR. * indicates candidate genes and orthologs that have not been included in the present ZIC collection. (E) Whole-mount in situ hybridization was used to determine the expression patterns of the tagged genes. Lateral views are shown in left panels, and ventral views are shown in right panels. Arrowheads indicate heart expression in 3 cardiac lines, while the arrow indicates the heart area in the sncgb control embryo, which showed no detectable signal. Scale bar=200 μm.

To facilitate molecular cloning of the tagged genes, we outcrossed 15 potential cardiac insertional lines to obtain fish with a single copy insertion. Southern blotting was used to identify fish with a lower copy number of insertions (Online Figure II). After 2–4 generations of outcrossing, 13 of the 15 lines contained a single insertion. The remaining insertional lines, GBT0363 and GBT0421, contained multiple insertions that could not be separated using simple crossing (Table 1). We cloned the candidate RP2 integration loci using a combination of 5′-RACE, 3′-RACE, and inverse PCR technologies (Table 1). In 10 of the 13 lines, the loci were integrated within well-annotated genes in the zebrafish genome (Zv9). As expected, 9 of the 10 insertions were located in introns downstream of the ATG start codon, with one exception in which a RP2 element was inserted into the first intron located upstream of the ATG translational start site of the cysteine-serine-rich nuclear protein 1b (csrnp1b) gene in the GBT0416 line. An upstream AUG was located within the reading frame of the mRFP coding sequence, which likely accounts for the mRFP expression in GBT0416. The RP2 insertions in 3 lines, GBT0413, GBT0418, and GBT0423, were within repetitive genomic regions based on the Zv9 genomic database. Further annotation of these genomic regions is required to determine the molecular nature of these 3 lines (Table 1). In this study, we primarily focus on the 10 cardiac GBT lines for which the molecular nature was clear.

Table 1.

Cloning of cardiac insertional mutants

Line	Allele	ZIC	Chr.	Gene	Position	mRNA disruption	Human ortholog	CV GO list	Phenotype
GBT0410	xu0410Gt	ZIC1	2	vapal	20	99%	VAPA	New
GBT0411	xu0411Gt	ZIC2	7	dnajb6b	227	99%	DNAJB6	Existing	Adult cardiac hypertrophy
GBT0412	xu0412Gt	ZIC3	10	xpo7	9	99%	XPO7	New
GBT0415	xu0415Gt	ZIC4	21	arrdc1b	40	100%	ARRDC1	New
GBT0416	xu0416Gt	ZIC5	24	csrnp1b	−7a	100%	CSRNP1	New
GBT0419	xu0419Gt	ZIC6	21	rxraa	10	100%	RXRA	Existing
GBT0364	xu0364Gt	ZIC7	10	mat2aa	31	100%	MAT2A	Existing	Embryonic lethal
GBT0422	xu0422Gt	ZIC8	7	insrb	479	100%	INSR	Existing
GBT0424	xu0424Gt	ZIC9	18	v2rl1	163	100%	VMN2R1	New
GBT0425	xu0425Gt	ZIC10	19	mrps18b	75	99%	MRPS18B	Existing	Larval lethal
GBT0413	xu0413Gt		UGC	ENSDARG00000094054
GBT0418	xu0418Gt		3	zgc:113030	24		ZNF595b
GBT0423	xu0423Gt		4	ENSDARG00000091251	29
GBT0363c	xu0363Gt		8	atp5f1&others	9	100%d	ATP5F1
GBT0421e	xu0421Gt		2	fam78b&others			FAM78B
GBT0365	xu0365Gt		12	sncgb	41	100%	SNCG

^aThe RP2 insertion is upstream of the ATG start codon;

^bCandidate human ortholog with the highest amino acid sequence similarity;

^cThis line contains 5 RP2 insertions;

^dPercent mRNA disruption for the atp5f1 gene;

^eThis line contains 8 RP2 insertions; UGC, unplaced genomic scaffold.

It was expected that the splice acceptor in the RP2 vector would hijack a splice donor of each tagged gene, which would truncate the encoded protein and lead to functional disruption.¹⁷ We assessed the efficiency of gene disruption at the level of RNA splicing. Homozygous embryos were identified using genotyping, and the expression level of tagged transcripts was subsequently measured via quantitative RT-PCR using a primer pair targeted to the flanking exons. Compared with wild-type (WT) siblings, native transcriptional splicing events in all 10 cardiac mutant lines were reduced by ≥99% (Table 1). Similarly, the targeted splicing event in synuclein, gamma b (sncgb) was completely disrupted in the homozygous GBT0365 non-cardiac control line. These data demonstrate efficient knockdown of the tagged genes in the individual GBT lines.

Assessment of the cardiac expression of 10 tagged genes

Among the 10 RP2-tagged genes, only 5 have been listed in the current Cardiovascular GO Annotation Initiative (Table 1). To verify that the 10-tagged genes were indeed expressed in the heart, we examined their expression via RT-PCR using RNA extracted from zebrafish embryonic hearts. All of the 10 tagged genes showed positive signals in RT-PCR assays, whereas the sncgb control did not (Figure 1B). Positive signals were also detected for the 3 candidate genes in GBT0418, GBT0363, and GBT0421, including zgc:113030, ATP synthase, H+ transporting, mitochondrial F0 complex, subunit b, isoform 1(atp5f1), and the family with sequence similarity 78, member B(fam78b). We also validated their expression in adult zebrafish hearts using both RT-PCR (Figure 1C) and fluorescence microscopy in dissected hearts (Online Figure III). Then, we assessed the corresponding human orthologs for these 13 genes and sncgb as a control. Based on the RT-PCR analysis using RNA extracted from normal human heart tissue, we detected cardiac expression for all 13 orthologous genes (Figure 1D). Taken together, these findings suggest that all 10 tagged genes exhibited cardiac expression in both embryonic and adult zebrafish, as well as in adult humans.

According to the published whole-mount in situ hybridization (ISH) data, three genes, insulin receptor b (insrb/GBT0422), vomeronasal 2 receptor l1 (v2rl1/GBT0424), and mitochondrial ribosomal protein S18B (mrps18b/GBT0425), exhibited ubiquitous expression and 4 genes, DnaJ (Hsp40) homolog, subfamily B, member 6b (dnajb6b/GBT0411), csrnp1b/GBT0416, retinoid X receptor, alpha a (rxraa/GBT0419), and methionine adenosyltransferase II, alpha a (mat2aa/GBT0364), showed restricted expression but with no reported heart signal at the embryonic stage. No ISH expression data have been reported for the remaining 3 genes, vesicle-associated membrane protein-associated protein A like (vapal/GBT0410), exportin 7 (xpo7/GBT0412), and arrestin domain-containing 1b (arrdc1b/GBT0415). We hypothesized that the lack of documented cardiac expression could reflect the relatively low sensitivity of ISH assays, low abundance of the transcripts, or dynamic expression patterns. To test these possibilities, we performed ISH assays for the 3 tagged genes, dnajb6b, csrnp1b, and mrps18b, and sncgb as a control. A short color developing time revealed RNA expression patterns similar to those that have been previously reported²² (and www.zfin.org) (Figure 1E), whereas a prolonged color reaction showed cardiac signals for all 3 genes but not for the sncgb non-cardiac control (Figure 1A and 1E).

We documented the dynamic expression of mRFP in the heart from these insertional lines. As exemplified by GBT419, mRFP signal was observed in the posterior hindbrain and somite at 20 hours post fertilization (hpf), which gradually expanded to the tectum, pharyngeal arch at 36 hpf (Online Figure IV). As the embryo developed, clear cardiac signals started to emerge at 72 hpf, whereas the pharyngeal arch signal faded at 96 hpf. The dynamic mRFP pattern reflected the endogenous expression profile of the tagged gene rxraa, as revealed by ISH using an rxraa riboprobe (Online Figure IV).^{22, 23} In addition, Tg(fli1a:EGFP), a transgenic reporter strain that labels the entire endocardium layer was crossed with GBT lines to determine the cell type-specific expression in the hearts of zebrafish from these cardiac lines. Based on the comparison with the EGFP signal, we identified tagged fusion proteins that were expressed in both layers (GBT0416), only in the endocardium (GBT0422), or only in the myocardium (GBT0411) (Online Figure IV). The myocardium expression of the tagged Dnajb6b-mRFP fusion in the GBT0411 line was further confirmed by crossing with Tg(titin:actn2-EGFP), a myocardium marker transgenic fish (Online Figure IV).²⁴ Thus, these analyses demonstrated that GBT lines with affected cardiac genes can be reliably identified, and their cardiac expression can be faithfully revealed using the mRFP reporter.

The embryonic lethal GBT0364 line uncovers an essential function of Mat2aa in cardiogenesis

We conducted incrosses for all 10 characterized cardiac lines and identified GBT0364 as an embryonic lethal line with ~25% offspring exhibiting a pericardiac edema phenotype at 3 dpf (Figure 2C). The RP2 insertion was present in the first intron of mat2aa, which encodes a highly conserved protein showing 89% amino acid identity with the human ortholog (Figure 2A and Online Figure V). While a strong maternal mRFP signal was detected at the one-cell stage in the offspring of a female (Online Figure VI), the mRFP signals emerged in the offspring of a male parent at 50% epiboly, indicating the onset of zygotic mat2aa expression (Online Figure VI). The mRFP expression was restricted to particular tissues including the eye, somites, pharyngeal arch, brain, pineal gland, and heart at later stages (Online Figure VI). A similar tissue-restricted and dynamic RNA expression pattern was detected using ISH with either a mRFP or mat2aa riboprobe (Online Figure VI), confirming that the mRFP tag in the RP2 vector faithfully reports the expression of the endogenous gene.

Figure 2. mat2aa is disrupted in the embryonic lethal GBT0364 line.

(A) A RP2 element was inserted into the first intron of the mat2aa gene in the GBT0364 line. (B) mat2aa transcripts were dramatically abolished in the GBT0364 homozygous mutant embryos, as indicated using RT-PCR. (C) GBT0364 homozygous mutants presented a cardiac edema (arrow) phenotype at 3 dpf. (D) Mat2aa ATG morpholino injection resulted in cardiac edema phenotype (arrow) in WT Wik embryos similar to that observed in GBT0364. Scale bars=0.5 mm. (E-F) Western blotting demonstrating that GBT0364 homozygous mutant embryos exhibited significantly reduced methylation levels on histone proteins at 4 dpf. n=3. (G) GBT0364 homozygous mutant embryos also exhibited a reduced global DNA methylation level at 4 dpf. The values are presented as the mean±standard deviation obtained from three independent biological repeats performed in triplicate. *P<0.05.

The RP2 insertion dramatically abolished the expression of mat2aa transcripts in GBT0364 homozygous mutants, which resulted in pericardiac edema at 3 dpf and eventual death at 8 dpf (Figure 2B and 2C, Online Figure VII and data not shown). The results from the genotyping PCR analysis revealed that this insertion was tightly associated with embryonic lethal phenotypes (Online Figure VII). The linkage between the aberrant RNA splicing event and the phenotypes was also validated using a rescue experiment using morpholinos that blocked the RP2-disrupting splicing event¹⁷ (Online Figure VII). Additionally, the cardiac phenotypes in the GBT0364 were faithfully recapitulated by knocking down Mat2aa expression via the injection of an ATG translational morpholino (Figure 2D and Online Figure VII). We assessed the functional disruption of the Mat2aa, which is an enzyme that catalyzes the synthesis of S-Adenosylmethionine (SAMe), a principal methyl donor for both DNA and protein methylation.²⁵ We detected a significant reduction of the methylation levels of the histone proteins H3k9 me3 and H3k27 me3 (Figure 2E and 2F) and a compromised global DNA methylation level in GBT0364 homozygous embryos (Figure 2G).

The expression of Mat2aa in the heart is restricted to the myocardium, as demonstrated by crossing GBT0364 line with either Tg(fli1a:EGFP) or Tg(titin:actn2-EGFP) lines (Figure 3A and 3B). We sought to perform a cardiac-specific rescue assay by exploiting the loxP sites integrated into the RP2 vector, and generated the Tg(cmlc2:YFP-Cre) transgenic fish line in which Cre recombinase expression was driven specifically in cardiomyocytes (Online Figure VIII). We further characterized the cardiac phenotypes in GBT0364 and revealed a reduced ventricular chamber size, decreased cardiac myosin light chain 2 (cmlc2) expression, and compromised cardiac functions, as indicated by reduced fraction shortening and heart rates (Figure 3C through 3G). Consistent with the multiple-tissue expression of mat2aa, non-cardiac defects were also observed in GBT0364 homozygous mutants, including reduced body length and pineal gland mRFP signal (Online Figure VII). In Tg(cmlc2:YFP-Cre); GBT0364/GBT0364 fish, the cardiac-specific Cre recombinase could effectively excise the mutagenic core of the RP2 vector in GBT0364/GBT0364 fish, as indicated by a switch in fluorescence from mRFP to YFP in cardiomyocytes (Online Figure VIII). Consequently, the cardiac phenotypes, including pericardial edema, reduced fraction shortening, and reduced heart rates, were rescued (Figure 3C through 3G). The decreases in cmlc2 expression levels and ventricular chamber size were also restored, whereas non-cardiac phenotypes such as the decreased mRFP expression level in the pineal gland were not rescued (Online Figure VII). Taken together, our data established a direct causality between the myocardial expression of Mat2aa and its role in regulating cardiogenesis and cardiac functions.

Figure 3. Myocardium-specific rescue of cardiac phenotypes in GBT0364 via cre-loxP system.

(A-B) Myocardium-specific expression of Mat2aa is demonstrated by non-overlapping expression of mRFP with EGFP from Tg(fli1a:EGFP) (A) and overlapping expression with Tg(titin:actn2-EGFP) (B). Insets are images of higher magnification. Scale bar=20 μm. (C) Ventricular chamber size is reduced in GBT0364 homozygous mutant embryos, which can be rescued by the Tg(cmlc2:YFP-Cre) transgene. Lateral views of DIC images are shown. (D) cmlc2 expression is reduced in the hearts of GBT0364 homozygous mutant embryos, which can be rescued using the Tg(cmlc2:YFP-Cre) transgene. Ventral views of embryos following whole-mount in situ hybridization are shown. Scale bars=20 μm. (E-G) Quantification of the ventricular volume (E), percent fraction shortening (F), and heart rate (G). n=6, *P<0.05.

To gain further mechanistic insight into the function of mat2aa in the control of ventricular chamber size, we assessed the cardiomyocyte (CM) cell size and cell number by immunostaining dissected embryonic hearts with anti-β-catenin (to quantify individual CM cell size) and anti-Mef2 (to define CM identity) antibodies (Figure 4A). We detected significant reductions of the CM cell size and cell number in homozygous embryos at 3 dpf (Figure 4A through 4E). The results of PCNA co-staining and TUNEL assays revealed that reduced CM proliferation and increased CM apoptosis, respectively, could potentially explain the decreased CM cell number (Figure 4D through 4H). Previous studies have suggested that S-Adenosylmethionine affects apoptosis in hepatocytes;^{25, 26} therefore, we injected a p53 morpholino (MO) into GBT0364 homozygous mutant embryos to inhibit apoptosis. The injection of the p53 MO delayed the occurrence of pericardial edema at 3–4 dpf. The cellular analysis indicated that the injection of the p53 MO effectively rescued the CM cell number but not the CM size, and affected CM apoptosis but not CM proliferation (Figure 4). Taken together, the results suggested a role for Mat2aa in regulating CM cell size and cell number during cardiogenesis. Consistent with its function in the liver, the function of Mat2aa in regulating CM number is partially ascribed to p53-regulated apoptosis.

Figure 4. Mat2aa regulates cardiomyocyte cell size and cell number.

(A) Representative images of embryonic hearts at 3 dpf after co-staining with anti-β-catenin (green) and anti-Mef2 (red) antibodies from a GBT0364 WT sibling and homozygous mutant with or without injection of 1 nL of p53_MO. Scale bar=20 μm. (B) Quantification of cardiomyocyte (CM) cell size. Injection of p53_MO did not alter the reduced CM size in GBT0364 homozygous mutants. (C) Quantification of CM cell circularity. No significant changes of circularity were observed. (D) Representative images of embryonic hearts at 3 dpf after co-staining with anti-TUNEL (green) and anti-Mef2 (red) antibodies from a GBT0364 WT sibling and homozygous mutant with or without injection of 1 nL p53_MO. (E) Quantification of CM cell number in the ventricle. Injection of p53_MO partially rescued the reduced CM number in GBT0364 homozygous mutant embryos. (F) Quantification of the CM cell apoptosis index in the ventricle. Injection of p53_MO partially rescued the increased TUNEL index of CMs in GBT0364 homozygous mutant embryos. (G) Representative images of embryonic hearts at 3 dpf after co-staining with anti-PCNA (green) and anti-Mef2 (red) antibodies from a GBT0364 WT sibling and homozygous mutants with or without injection of 1 nL p53_MO. Arrows indicate PCNA+/Mef2+ cells. (H) Quantification of the CM cell proliferation index in the ventricle. Injection of p53_MO did not alter the reduced PCNA index of CMs in GBT0364 homozygous mutant embryos. n=6, *P<0.05.

The larval lethal GBT0425 line uncovers an essential function of Mrps18b in cardiac mitochondrial homeostasis

During our inbreeding efforts for the remaining 9 cardiac lines, we found that the GBT0425 line failed to yield any viable homozygous adults. A detailed analysis of the homozygous GBT0425 larvae revealed an apparent developmental delay concomitant with a significantly compromised cardiac contractility at 10 dpf, and ultimately resulted in death at 12 dpf (Figure 5C and 5D). This mutant line showed no visible phenotypes during early embryogenesis and was therefore not detected in our initial phenotype-based screen. A single RP2 insertion was identified in the 2nd intron of the mrps18b gene (Figure 5A). The mrps18b gene encodes one of the 29 components of the conserved small subunit of the mitochondrial ribosome²⁷ (Online Figure IX). The occurrence of the normal splicing event for the full length transcript was dramatically reduced in GBT0425 homozygous mutant embryos (Figure 5B and Table 1). The genetic linkage between the mrps18b gene and the observed phenotype was validated by genotyping the 16 larvae exhibiting these phenotypes, which were all confirmed at the molecular level to be homozygous mutants (Online Figure IX).

Figure 5. GBT0425 line uncovers essential functions of Mrps18b as a mitochondrial protein.

(A) A RP2 element was inserted into the second intron of the mrps18b gene in the GBT0425 line. (B) The mrps18b transcript was dramatically reduced in GBT0425 homozygous embryos by RT-PCR using mrps18b-c-F/R primers for the entire transcripts amplification. (C) GBT0425 homozygotes appeared smaller at 10 dpf. Scale bar=200 μm. (D) Quantification of fraction shortening (FS%). FS is significantly reduced in GBT0425 homozygous mutant embryos at 10 dpf, but not at 6 dpf. n=6. (E-L) Fluorescent images of individual cells dissociated from either heterozygous (E-H) or homozygous (I-L) GBT0425 embryonic hearts are shown. Mitochondria are stained with MitoTracker Green FM, which appears to co-localize with the mRFP signal. E-L, scale bars=20 μm. (M-P) Transmission electronic micrographs indicating an apparently reduced mitochondrial number in the GBT0425 homozygous embryonic heart (M,N) and disrupted mitochondrial morphology in the remaining mitochondria (O,P) are shown. Arrowheads indicate representative mitochondria; N, nuclei. M-N, scale bars=2 μm; O-P, scale bars=200 nm. (Q) Quantification of mitochondrial number indicated a significantly reduction in the GBT0425 homozygous embryos at 8 dpf. n=3, *P<0.05.

Consistent with the identity of MRPS18B as a nuclear-encoded mitochondrial protein,²⁷ we detected strong and long-lasting maternal mRFP expression in the GBT0425 embryos (Online Figure IX). Because the RP2 insertion generates a truncated Mrps18b-mRFP fusion protein containing 75 N-terminal amino acid residues (Online Figure IX and Table 1), with the first 35 residues encoding the predicted mitochondrial targeting sequence (http://ihg.gsf.de/ihg/mitoprot.html), it was expected that mRFP would report mitochondria-specific expression. Indeed, we observed punctate perinuclear expression of mRFP in a dissected GBT0425 heart, and the mRFP was co-localized with the mitochondrial stain MitoTracker Green FM (Figure 5E through 5L and Online Figure IX). The truncation of the C-terminal 167 residues in Mrps18b will most likely disrupt the function of the mitochondria ribosome.

In GBT0425 homozygous mutant hearts, reduced number of mitochondria in the dissociated cells was revealed by either MitoTracker Green FM signal or mRFP fluorescence (Figure 5E through 5L). As expected, the mitochondrial number appeared to be also reduced in the somite cells (Online Figure IX). The mitochondrial defects in the heart were confirmed by transmission electron microscopy analysis (Figure 5M through 5Q). In addition, the morphologies of the individual mitochondria were disturbed, appearing degenerated with undefined cristae borders (Figure 5O and 5P). These observed phenotypes in GBT0425 homozygous fish are unlikely caused by non-specific effects of fluorescence proteins, since no detectable phenotypes were found in other ZIC lines such as GBT0410 with comparable level of fluorescence protein expression (Online Figure IX). Taken together, our data revealed that Mrps18b is an essential component of the nuclear-encoded small subunit of the mitochondrial ribosome. The prolonged maternal expression of Mrps18b might protect homozygous GBT0425 mutants from early embryonic lethality, resulting in larval phenotypes when maternal Mrps18b expression is diminished at approximately 8 dpf.

The adult recessive GBT 0411 line links Dnajb6b to cardiac hypertrophy

We continued to examine the offspring of the remaining 8 cardiac lines, but we did not detect any fish with visually detectable phenotypes. We reasoned that invasive phenotyping methods were required to identify the adult mutants. We were particularly interested in GBT0411, which displayed a rather specific mRFP expression in the heart at the embryonic stage (Figure 1A and 1E). A RP2 insertion was identified in the intron between exon 7 and 8 of dnajb6b, which encodes a protein with 53% amino acid identity to human DNAJB6 (Figure 6A and Online Figure V). In humans, exons 1–8 encode the short isoform of the cytosolic DNAJB6(S), while exons 1–10 encode the long isoform, DNAJB6(L), which harbors a nuclear localization signal encoded by the last two exons.²⁸ This RP2 insertion in the GBT0411 line dramatically reduced the normal splicing event between exon 6 and 7 (Figure 6B and Table 1), and switched most Dnajb6b(L) to Dnajb6b(S)-like isoform, as revealed by northern blotting (Figure 6C). In the heart, Dnajb6b was expressed only in the myocardium in both embryonic and adult stages (Figure 6D through 6F and Online Figure IV). At the subcellular level, the mRFP fluorescence exhibited a striated expression pattern in the sarcomere that co-localized with the Actn2-EGFP, which was consistent with the Z-disc localization reported for DNAJB6(S) in human muscles.²⁹ Interestingly, mutations in human DNAJB6 caused limb-girdle muscular dystrophy type 1D (LGMD1D).^{29, 30}

Figure 6. Sarcomere localized Dnajb6b is disrupted in the GBT0411 line.

(A) A RP2 element was inserted into the 6th intron of the dnajb6b gene in the GBT0411 line. (B) The dnajb6b entire transcript was dramatically ablated in GBT0411 homozygous embryos. Shown are results of RT-PCR. (C) Northern blotting hybridization results using either mRFP or the short isoform dnajb6b as a probe. The predominant dnajb6b long isoform mRNA in an adult wild type heart is disrupted, which results in truncated dnajb6b mRNA encoded by exon 1–6 that is similar to the dnajb6b short isoform. (D) Shown are lateral views of a live embryo to indicate heart and eye-specific expression of mRFP at 3 dpf. Scale bar=200 μm. (E-F) Fluorescent images of adult heart sections from GBT0411 line after crossing with either Tg(titin:actn2-EGFP) (E) or Tg(fli1a:EGFP) (F). Scale bars=20 μm.

We sacrificed some of adult cardiac GBT lines and assessed the morphology of their hearts. We observed significantly enlarged ventricles in homozygous GBT0411 fish at 1 year old (Figure 7A and 7B). Immunostaining using anti-β-catenin and anti-PCNA antibodies revealed an increase of cardiomyocyte cell size, but not cell proliferation in the homozygous GBT0411 mutants (Figure 7C and 7D and Online Figure X). Hallmarks of cardiac hypertrophy, including muscular disarray, were also detected by α-actinin antibody staining (Figure 7E), and re-activation of fetal gene atrial natriuretic factor (anf) was detected at 1 year but not at 4 months old (Figure 7F). The cardiac hypertrophy phenotypes in GBT0411 is less likely caused by side effects of fluorescence proteins, as other 7 ZIC lines do not exhibit any noticeable cardiac phenotypes at ≥1 year old, despite stronger mRFP expression in some ZIC lines (Online Figure X). We also did not notice any significant heart size change in either the GBT0031 heterozygous fish¹⁷ or Tg(titin:actn2-EGFP) transgenic fish²⁴ at >1 year old, despite their stronger mRFP or EGFP expression level than GBT0411. It is noteworthy to point out that promoter-specific toxicity of fluorescent proteins in cardiomyocytes might still exist in zebrafish and need to be evaluated by generating transgenic lines containing fluorescent proteins only. Nevertheless, our data strongly suggest that GBT0411 is a recessive mutant for dnajb6b, with an adult phenotype that implicates a role for dnajb6b in cardiac hypertrophy.

Figure 7. GBT0411 is a recessive adult mutant that exhibits cardiac hypertrophy phenotype.

(A) Representative images of dissected hearts from GBT0411 WT sibling, heterozygous and homozygous fish at 12 months old. Scale bar=1 mm. (B) Quantification of the ventricular surface area to body weight index (VSA/BW) showed significant heart enlargement in the GBT0411 homozygous fish compared to that in WT sibling control at 12 months but not at 4 months old. n=6. (C) Representative images of adult heart sections stained with anti-β-catenin antibodies to indicate cardiomyocyte cell size at 12 months old stage. Scale bar=10 μm. (D) Quantification of cardiomyocyte cell size from (C), which is significantly increased in GBT0411 homozygous compared with that in WT fish at 12 months old. 20–30 cardiomyocytes were measured. (E) Representative images of adult heart sections at 12 months old stained with anti-αactinin antibodies. Muscular disarray was detected in the GBT0411 homozygous fish. Insets are images of higher magnification. Scale bar=20 μm. (F) The quantitative RT-PCR results indicating the re-activated expression of atrial natriuretic factor (anf) in GBT0411 homozygous fish heart at 12 months, but not 4 months old, are shown. V, ventricle; A, atrium; OFT, outflow tract. n=3, *P<0.05.

DISCUSSION

Expression-based insertional mutagenesis screen strategy can be used to annotate the cardiac genome

The current study presents an efficient system to identify cardiac genes and to annotate their expression and functions via a transposon-based insertional mutagenesis screening strategy. Our data demonstrated that RP2-based gene trapping is a sensitive and reliable method for identifying genes with cardiac expression. Five of the ten tagged genes identified in our fluorescence-based approach are novel cardiac genes that had not previously been detected by ISH and/or other methods, and have not been included in the Cardiac GO database. The cardiac expression of all 10 genes has been validated using RT-PCR in both zebrafish and humans. Our data suggested that the compromised sensitivity of the ISH technology can be partially ascribed to difficulties in distinguishing specific signals from background. Moreover, the dynamic expression profile and subcellular localization of the tagged genes can be easily revealed by following the mRFP reporter.

Each GBT line with both cardiac expression and cardiac phenotypes served as an opportunity for elucidating the molecular mechanisms of heart development and/or cardiac diseases. The characterization of the GBT0364 line revealed the myocardium-specific expression of Mat2aa in the heart and suggested essential roles for this protein in cardiogenesis and cardiac functions. Both cardiac expression and functions for mat2a have not previously been reported, despite a number of in vitro studies concerning the role of Mat2a in the apoptosis of leukemic T-cells³¹ and as a transcriptional co-repressor of the Maf oncoprotein³² Our observations prompted further investigations to elucidate functions involving methylation in cardiomyocyte differentiation. The characterization of the GBT0425 line demonstrated the mitochondrial expression of Mrps18b in the heart and revealed an essential function for Mrps18b protein in mitochondrial homeostasis and larval survival. The characterization of the GBT0411 line revealed the cardiomyocyte-specific expression of dnajb6b and suggested its function in adult cardiac hypertrophy. Based on the interaction of DNAJB6 with the hsp70 complex and its implication in regulating protein recycling,³³ defective toxic protein aggregation warrants further investigation as a candidate molecular mechanism.

The expression of many cardiac genes is not restricted only to the heart. Therefore, although the cardiac phenotypes can still be specific, it is desirable to separate the primary phenotypes due to gene disruption within the heart from secondary consequence of gene disruption in the other tissues. Because each GBT line represented a potential revertible mutant that enables tissue-specific rescue experiments, the causality between the observed cardiac phenotypes and their cardiac expression can be conveniently established, as our experimentation with GBT0364 demonstrated. Moreover, our data prompted the generation of additional tissue-specific Cre transgenic lines that can be used to determine the primary functions of genes expressed in other cardiac cell types, including those of the epicardium, endocardium, cardiac cushion, and cardiac conduction system.

Expression-based insertional mutagenesis screen strategy facilitates the identification of recessive cardiac mutants in vertebrates

The present study also demonstrates that the in vivo protein trapping methodology effectively reduces colony management efforts, the major bottleneck for recessive screens performed in vertebrates, thereby offering an appealing alternative method complementary to other mutagenesis approaches. Because fish with GBT insertions can be easily identified using an mRFP tag, the tagged genes can be cloned using PCR-based approaches. The majority of the generated GBT lines represent a near-null mutant allele for the tagged gene, as indicated by the ≥99% knockdown efficiency observed in the GBT lines generated here and the ≥97% knockdown efficiency reported in a previous study.¹⁷ Integration of an expression-based enriching strategy further simplifies colony management. Because of our expertise in the heart, we selectively worked on 15 candidate lines with cardiac expression. De-prioritizing the other 52 lines at an early mutagenesis screen step reduced our colony management efforts by more than 75%. The current preselection strategy was conducted during embryogenesis, based on the general belief that many fetal genes play important roles during the pathogenesis of adult diseases. At the expense of experimental convenience, this approach can be limited by its intrinsic biases towards genes with embryonic cardiac expression. Interestingly, all 10 ZIC genes identified here based on their embryonic expression also express in both adult zebrafish and adult human hearts. If needed, this pre-selection strategy can be complemented by the selection of GBT lines with mRFP expression in dissected adult hearts.

A plausible solution for conducting recessive adult screens in vertebrates is to establish an international consortium with an accessible central location for generating mutants. Different categories of the mutants shall be distributed to individual labs with related expertise for downstream phenotyping. The EUMODIC (European Mouse Disease Clinic) program represents such an ongoing collaborative effort,⁷ with the aim of breeding 500 mutant mice lines into homozygosity. Our expression-based pre-selection offers an efficient decision-making strategy to disseminate the proper mutant lines to downstream laboratories. Currently, thousands of stable GBT lines are being generated at the Mayo Clinic. Images of the tagged genes are being documented and deposited in the website (www.zfishbook.org).³⁴ Therefore, insertional lines with particular expression patterns can be digitally screened, and each category of insertional lines can be disseminated to laboratories with matching expertise for downstream breeding and phenotyping. Indeed, our initial digital screen of 322 GBT lines identified 18 candidate cardiac lines (Online Figure XI). Because each downstream laboratory only needs to handle a limited number of fish lines, it is anticipated that the recessive adult screen can be conducted on a much larger scale.

Besides increased colony management efforts, the focus of the present study, the other major bottleneck for an adult screen is phenotyping. It is likely that cardiac phenotypes will be detected in many or all of the remaining 7 homozygous GBT lines with the use of adequate phenotyping assays or stressing tools. To test this hypothesis, we plan to maintain and expand a Zebrafish Insertional Cardiac (ZIC) mutant collection, starting from the 10 cardiac GBT lines reported here (Table 1). Meanwhile, we are developing cardiac phenotyping assays and stressing tools, as represented by the generation of the first two adult zebrafish models of cardiomyopathies.^{20, 21} An exciting future direction is to test whether the ZIC lines can be screened to identify genetic modifiers for adult cardiomyopathy, as well as other major cardiovascular diseases.

Novelty and Significance.

What Is Known?

• Parts of the human genome are implicated in the development of the heart and cardiovascular diseases; however, the functions of these parts needs to be annotated.

• Mutagenesis screening in model organisms is a powerful tool for genome-wide gene discovery and functional annotation.

• A transposon-based insertional mutagenesis system has been established in zebrafish to generate mutants and to annotate gene functions.

What New Information Does This Article Contribute?

• Integration of an expression-based pre-selection step into insertional mutagenesis screening provides an effective strategy to enrich mutants affecting a particular organ such as the heart.

• The reduce colony management efforts enable the identification of adult recessive mutants with cardiac expression.

• Methionine adenosyltransferase II, alpha a (mat2aa) is essential for cardiogenesis; ribosomal protein S18B (mrps18b) is needed for cardiac mitochondrial homeostasis at the larva stage; and DnaJ (Hsp40) homolog, subfamily B, member 6b (dnajb6b) is implicated in adult cardiac hypertrophy.

Mutagenesis screening in model organisms is a powerful tool for genome-wide gene discovery and functional annotation; however, it is inefficient in identifying genes that affect a particular organ such as the heart and is limited in identifying vertebrate recessive adult mutants due to challenges in colony management. Here, we address these two bottlenecks through the integration of an expression-based selection strategy with insertional mutagenesis screening. We isolated 10 Zebrafish Insertional Cardiac (ZIC) mutant lines with clear molecular nature and identified both known and novel genes presenting cardiac expression and/or functions. In addition to an embryonic lethal mutant, we isolated a mutant exhibiting larval phenotypes, a mutant exhibiting recessive juvenile phenotype, and a mutant with adult recessive phenotypes at 1 year of age. These cardiac phenotypes can be potentially reverted by the deletion of the transposon insert using a cardiomyocyte-specific Cre transgene, enabling the determination of the causality between myocardial expression of the tagged gene and the observed cardiac phenotype. In summary, our data demonstrate an expression-based mutagenesis screening strategy for efficiently generating both embryonic and adult recessive revertible mutants in a particular organ; suggesting an approach that will facilitate the annotation of vertebrate genomes.

Non-standard Abbreviations

ANF

atrial natriuretic factor

CM

cardiomyocyte

ENU

ethylnitrosourea

GBT

gene break transposon

GO

gene ontology

ISH

in situ hybridization

MO

morpholino

mRFP

monomeric red fluorescent protein

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

ZIC

zebrafish insertional cardiac