ErbB4 tyrosine kinase inhibition impairs neuromuscular development in zebrafish embryos.

Tyrosine kinase inhibitors are widely used in the clinic, but limited information is available about their toxicity in developing organisms. Here, we tested the effect of tyrosine kinase inhibitors targeting the ErbB receptors for their effects on developing zebrafish ( Danio rerio) embryos. Embryos treated with wide-spectrum pan-ErbB inhibitors or erbb4a-targeting antisense oligonucleotides demonstrated reduced locomotion, reduced diameter of skeletal muscle fibers, and reduced expression of muscle-specific genes, as well as reduced motoneuron length. The phenotypes in the skeletal muscle, as well as the defect in motility, were rescued both by microinjection of human ERBB4 mRNA and by transposon-mediated muscle-specific ERBB4 overexpression. The role of ErbB4 in regulating motility was further controlled by targeted mutation of the endogenous erbb4a locus in the zebrafish genome by CRISPR/Cas9. These observations demonstrate a potential for the ErbB tyrosine kinase inhibitors to induce neuromuscular toxicity in a developing organism via a mechanism involving inhibition of ErbB4 function.

INTRODUCTION

Despite the approval of a number of molecularly targeted therapies for treating several different cancer types, only a few protocols have been clinically tested with pediatric patients. Drugs approved for treatment of malignancies in adults are still often used off-label for pediatric patients (Zwaan ). Clinical pharmacology of children, however, differs significantly from that of adults (Kearns ), and the developing organs may be expected to be more vulnerable to toxicity than their mature adult counterparts, with a higher percentage of slowly dividing or quiescent cells. In particular, this is of potential concern in the case of inhibitors of growth-signaling molecules, such as the receptor tyrosine kinases (RTK).

ErbB receptors are a subfamily of RTKs composed of four members: epidermal growth factor receptor (EGFR, ErbB1, HER1), ErbB2 (HER2), ErbB3 (HER3), and ErbB4 (HER4). ErbB1 and ErbB2 have been successfully targeted in different cancer types, such as breast, lung, colorectal, gastric, and head and neck, by both therapeutic antibodies and tyrosine kinase inhibitors (TKI; Arteaga and Engelman, 2014). The recently developed wide-spectrum ErbB TKIs, including afatinib, neratinib, canertinib, and dacomitinib, suppress the catalytic activity of EGFR and ErbB2 as well as ErbB4, while they do not directly target the naturally nearly kinase-inactive ErbB3 (Schwartz ).

During development, the ErbB receptors activate central signaling pathways that regulate vertebrate organogenesis (Holbro and Hynes, 2004). A typical mammalian genome includes four erbb genes, erbb1, erbb2, erbb3, and erbb4 (Stein and Staros, 2000). In the zebrafish (Danio rerio) genome, however, the genes encoding EGFR, ErbB3, and ErbB4 have been suggested to have undergone another round of duplication, producing erbb1a and erbb1b, erbb3a and erbb3b, and erbb4a and erbb4b, while there is only a single erbb2 gene (Reischauer ).

One approach to preclinically addressing the toxicity of RTK TKIs is analyzing experimental models of growing organisms. Zebrafish embryos and larvae are an increasingly used model both for pediatric research (Veldman and Lin, 2008; Lohi ) and for in vivo testing of drug efficacy and toxicology (MacRae and Peterson, 2015). Here, we tested several ErbB-targeting TKIs for toxicity in zebrafish embryos. Our findings indicate that wide-spectrum pan-ErbB TKIs, but not EGFR-specific TKIs (nor TKIs with dual specificity for EGFR and ErbB2), affected the mobility of the embryos. Validation experiments with erbb4a-targeting antisense oligonucleotides and an erbb4a mutant fish line suggested a critical role for ErbB4 in mediating the effects. Mechanistic experiments further indicated that the defect in motility was associated with structural defects in the developing truncal muscle fibers, changed muscle-specific gene expression, and reduced motoneuron length.

RESULTS

Pan-ErbB kinase inhibitors suppress motility of zebrafish embryos

To address the effects of ErbB inhibition on developing tissues, 8–h past fertilization (hpf) zebrafish embryos were treated for 48 h with 10 µM of nine different ErbB TKIs. While there were no statistically significant differences in the survival of the embryos, some of the TKIs were observed to inhibit their movement. To quantitatively analyze the role of ErbB TKIs in regulating motility, touch–response assays were carried out (Granato ), and the movement of the embryos was videorecorded (an example of the procedure is shown in Supplemental Movie S1). The swim tracks for individual embryos are visualized in Figure 1A and Supplemental Figure S1. Inhibitors, such as erlotinib and gefitinib, expected to inhibit EGFR alone, did not affect either the length of the swim bursts (Figure 1B) or the average speed during the bursts (Figure 1C). A moderate and not statistically significant effect was observed with the dual-specific EGFR/ErbB2 inhibitor lapatinib (Figure 1, B and C). However, the irreversible pan-ErbB inhibitors afatinib, neratinib, canertinib, and dacomitinib, capable of suppressing the activity of EGFR, ErbB2, and ErbB4, indicated some reduction in motility, and a statistically significant effect was observed for canertinib (Figure 1). While our analysis was not sufficiently powered to detect relatively small effects, these findings suggested that inhibition of the ErbB4 kinase was associated with the reduced motility of the embryos. Indeed, the TKI AG 1478 shown to block the activity of EGFR as well ErbB4 (Levitzki and Gazit, 1995; Egeblad ) blocked motility in a statistically significant manner (Figure 1).

FIGURE 1:

ErbB kinase inhibitors reduce motility of zebrafish embryos. Touch–response assays measuring motility of embryos treated from 8 to 48 hpf with the buffer control DMSO or with 10 µM of the indicated ErbB TKIs. (A) The colored lines demonstrate videorecorded tracks of swim bursts of individual embryos stimulated to move by a physical touch to the tail. (B, C) Quantification of the length of the entire swim burst (B) and the average speed during the burst (C) in the touch–response assays. Mean values are indicated above each column. DMSO, n = 18; erlotinib, n = 14; gefitinib, n = 11; osimertinib, n = 9; lapatinib, n = 7, afatinib, n = 7; neratinib, n = 6; canertinib, n = 7; dacomitinib, n = 9; AG 1478, n = 11. *p < 0.05.

Movie S1

Movie S1

erbb4a is the predominant erbb4 gene expressed during zebrafish organogenesis

To address the relative expression of the two erbb4 genes in zebrafish, whole embryos were lysed and subjected to real-time reverse transcription-PCR (RT-PCR) analysis at different time points after fertilization. While both loci were expressed during early development, there was a switch of the predominant transcript at 24 hpf (Supplemental Figure S2A). Expression of erbb4b peaked at 6 hpf, the earliest time point analyzed, and declined thereafter. In contrast, erbb4a was expressed at a low level at 6 hpf, reached a similar level to erbb4b at 24 hpf, and further increased to clearly dominate over erbb4b at 48 hpf (Supplemental Figure S2A). Western blot analysis with anti-ErbB4 E200 against the intracellular domain of ErbB4 confirmed the expression of ErbB4 protein, and demonstrated significantly increased protein expression from 48 hpf onward (Supplemental Figure S2B). Whole mount immunofluorescence analysis with the ErbB4 antibody HFR-1 demonstrated the strongest ErbB4 expression in the brain and somites of 48-hpf embryos (Supplemental Figure S2, C–F).

erbb4a is necessary for motility of zebrafish embryos

As an independent approach to addressing the function of erbb4a, morpholino antisense oligonucleotides were injected into one- to eight-cell stage embryos. A morpholino-targeting erbb4a mRNA reduced the amount of total ErbB4 protein present in lysate of whole embryos by 99% (Figure 2A, lanes 1 and 2), further supporting the significance of erbb4a as the major ERBB4 ortholog and also validating the functionality of anti-ErbB4 E200 antibody in zebrafish.

FIGURE 2:

erbb4a is necessary for normal motility of zebrafish embryos. (A) Western blot analysis of ErbB4 expression (anti-ErbB4 E200) in 48-hpf embryos injected at the one- to eight-cell stage with erbb4a-targeting or control morpholino (lanes 1 and 2). β-tubulin (antibody E7) expression was analyzed from the same lysates to control loading. (B) Touch–response assays measuring motility of 48-hpf morphants and embryos treated with AG 1478 or untreated. The colored lines demonstrate videorecorded tracks of swim bursts of 10 randomly selected embryos stimulated to move by a physical touch to the tail. (C, D) Quantification of the length of the swim burst (C) and the average speed during the burst (D) in the touch–response assays. Control morpholino, n = 32; erbb4a morpholino, n = 17; DMSO, n = 17; AG 1478, n = 18. (E) Western blot analysis of ErbB4 expression (anti-ErbB4 E200) in 48-hpf embryos coinjected at the one- to eight-cell stage with erbb4a targeting or control morpholino together with in vitro transcribed mRNA encoding EGFP or a fusion protein of ErbB4 and EGFP (ERBB4-EGFP). β-tubulin (clone E7) expression was analyzed from the same lysates to control loading. (F) Ten randomly selected swim tracks of 48-hpf morphants subjected to touch–response assays. (G, H) Quantification of the length of the swim burst (G) and the average speed during the burst (H) in the touch–response assays. Control morpholino + EGFP, n = 27; erbb4a morpholino + EGFP, n = 18; erbb4a morpholino + ERBB4-EGFP, n = 12.

Validation experiments with AG 1478 confirmed the observations of the ErbB kinase inhibitor screen. Similarly to treatment with AG 1478, the 48-hpf erbb4a morphant embryos moved for shorter distances in response to a contact, as compared with respective controls. Figure 2B demonstrates the swim bursts of 10 randomly selected embryos treated or not with either erbb4-targeting morpholino. Both the burst length (Figure 2C) and the average velocity during the movement (Figure 2D) were reduced in embryos with reduced ErbB4 signaling. Both ErbB4 protein expression (Figure 2E) and the effects on embryonic motility (Figure 2, F–H) were rescued by coinjecting the embryos with human ERBB4 mRNA simultaneously with the erbb4a-targeting morpholino oligonucleotide, supporting the specificity of the antisense morpholino approach.

Inhibition of erbb4a induces structural defects in developing muscle fibers

To address a potential general effect of erbb4a down-regulation on development, the head–trunk angle (HTA) was analyzed as a gross measure of developmental stage (Kimmel ). No significant difference (p = 0.99) in HTA between embryos injected with control morpholino (average = 152 ± 13°, n = 13) or morpholino targeting erbb4a (average = 152 ± 12°, n = 13) was observed. However, the number of somites was significantly (p < 0.001) reduced in 48-hpf erbb4a morphants (29.6 ± 0.3 somites/embryo, n = 18) in comparison with the embryos treated with control morpholinos (31.2 ± 0.2 somites/embryo, n = 17). A similarly significant (p < 0.001) decrease in the number of somites was observed in AG 1478–treated embryos (29.0 ± 0.3 somites/embryo, n = 8) in comparison with embryos treated with the dimethyl sulfoxide (DMSO) buffer alone (30.8 ± 0.3 somites/embryo, n = 9).

To address the role of erbb4a at the level of skeletal muscle fibers, whole-mount 48-hpf embryos were stained with an anti-myosin antibody and imaged by confocal microscopy (Figure 3A). Both the AG 1478 treatment and the erbb4a-targeting morpholino resulted in a significantly reduced diameter of muscle fibers (Figure 3, A and B), suggesting a role for ErbB4a signaling in maintaining normal myocyte morphology.

FIGURE 3:

Effect of erbb4a knockdown on myocyte and motoneuron morphology. (A) Confocal immunofluorescence images of 48-hpf embryos treated from 8 to 48 hpf with 10 µM of the ErbB kinase inhibitor AG 1478 or the DMSO buffer alone. Whole-mount embryos were stained with the anti-myosin F59 (red) and the DNA stain 4′,6-diamidine-2-phenylindole (blue). (B) Quantification of the thickness of individual F59-positive myocytes from confocal immunofluorescence images. The number of embryos analyzed was as follows: control morpholino, n = 14; erbb4a morpholino, n = 15; DMSO, n = 16; AG 1478, n = 15. Ten myocytes was analyzed from each embryo. (C) Confocal immunofluorescence images of 48-hpf embryos injected at the one- to eight-cell stage with erbb4a-targeting or control morpholino (top panels) or treated from 8 to 48 hpf with 10 µM of the ErbB kinase inhibitor AG 1478 or the DMSO buffer alone (bottom panels). Whole-mount embryos were stained with anti–synaptotagmin-2 (znp-1) to visualize motoneurons. Scale bar, 40 µm. (D) Quantification of the length of znp-1-immunoreactive motoneuron axons from confocal immunofluorescence images. The numbers of neurons (embryos) analyzed were as follows: control morpholino, n = 39 (11); erbb4a morpholino, n = 38 (10); DMSO, n = 33 (11); AG 1478, n = 38 (11).

erbb4a knockdown is associated with reduced motoneuron length but not with reduced AchR expression in the skeletal muscle

Defects in developing muscle could be secondary to impaired innervation. Indeed, ErbB4 signaling has been suggested to regulate AchR expression and function in mouse models (Rimer ; Schmidt ). However, knockdown of erbb4a did not result in significant changes in AchR expression in 48-hpf zebrafish embryos when measured either by real-time RT-PCR analysis of AchR subunits α1 (chrna1), β1 (chrnb1), and δ (chrnd) (p = 0.08, 0.25, and 0.86, respectively; n = 10) or by AchR staining with fluorescent α-bungarotoxin (p = 0.91 for the area covered by fluorescence signal; n = 6). In contrast, confocal immunofluorescence analyses of motoneurons by anti–synaptotagmin-2 (znp-1) antibody demonstrated that the motoneuronal axons of erbb4a morphants were shorter than in control morpholino-injected embryos (Figure 3, C and D). Similar findings were observed with ErbB kinase inhibitor AG 1478–treated embryos (Figure 3, C and D).

Inhibition of ErbB4a alters expression of several muscle-cell specific genes

To characterize erbb4a-dependent signaling in the muscle fibers at the molecular level, expression of the muscle-specific myosin genes cmcl2, myhc4, smyhc1, and tpm1 was analyzed from total lysates of 48-hpf embryos by real-time RT-PCR. Significantly reduced expression of smyhc1 and tpm1 was observed after erbb4a morpholino injection (Figure 4A) and AG 1478 treatment (Figure 4B). The erbb4a-targeting morpholino also significantly reduced the expression of cmcl2 (Figure 4A).

FIGURE 4:

erbb4a is necessary for expression of several muscle-specific genes. (A, B) Real-time RT-PCR analysis of the expression of muscle-specific genes in 48-hpf embryos injected at the one- to eight-cell stage with erbb4a-targeting or control morpholino (A) or treated from 8 to 48 hpf with 10 µM of the ErbB kinase inhibitor AG 1478 or the DMSO buffer alone (B). Control morpholino, n = 10; erbb4a morpholino, n = 10; DMSO, n = 11; AG 1478, n = 12. (C) Western blot analyses of the expression of the muscle-specific proteins tropomyosin (clone CH-1), myosin light chain (clone F310), and myosin heavy chain (clone F59) in the 48-hpf embryos. (D) Western blot analysis of the expression of tropomyosin in embryos treated with 10 µM of AG 1478 for the indicated periods of time postfertilization. All other time intervals included incubation with DMSO. β-actin (sc-1616) expression was analyzed from the same lysates to control loading.

Western blot analyses indicated that the steady-state expression levels of myosin proteins recognized by antibodies raised against chicken tropomyosin or chicken fast-type myosin light chain were clearly reduced in response to either erbb4a-targeting morpholino or AG 1478 (Figure 4C). In contrast, the myosin species recognized by antibodies raised against chicken slow-type myosin heavy chain remained unchanged after both genetic and chemical down-regulation of ErbB4 function (Figure 4C).

To elucidate the temporal requirement of ErbB activity, AG 1478 was added to the embryos at 8, 24, or 48 hpf and washed away at 24, 48, or 72 hpf. The effect on muscular tissue was assessed by Western blot analysis of tropomyosin expression at the 72-hpf time point for all treatment periods. A robust decrease in tropomyosin level was already detected when AG 1478 was administered at 8 hpf regardless of the washout time point at 24, 48, or 72 hpf (Figure 4D, lanes 2–4 vs. lane 1). An attenuated, but still clear, response was also observed at the later treatment period from 24 or 48 hpf to 72 hpf (Figure 4D, lane 5 or 6 vs. lane 1).

Rescue of reduced motility of erbb4a morphants by muscle-specific ERBB4 expression

To more directly address whether the reduced motility of erbb4a-deficient embryos was caused by a defect of developing myocytes, a construct to overexpress human ERBB4 under the muscle-specific acta1 promoter (Jacoby ) was cloned, as described in Materials and Methods. This construct, encoding both ErbB4 and EGFP, or a control construct encoding EGFP alone, was coinjected into the embryos with the erbb4a-targeting or control morpholino. Interestingly, the mosaic muscle cell–specific expression of ErbB4 in the 48-hpf G0 embryos was sufficient to at least partially rescue the reduction in the tropomyosin expression (Figure 5A) and the reduction of muscle fiber diameter (Figure 5, B and D), as well as the shortening of motoneuron axons (Figure 5, C and E). Moreover, the expression of the ErbB4 construct with the muscle-specific promoter was sufficient to partially rescue the compromised motility of the erbb4a morphants in the touch–response assays (Figure 5, F–H).

FIGURE 5:

Muscle-specific expression of ERBB4 can rescue neuromuscular phenotypes in erbb4a morphants. Analyses of 48-hpf embryos coinjected at the one- to eight-cell stage with erbb4a-targeting or control morpholino together with constructs encoding EGFP, or ErbB4 and EGFP (ERBB4/EGFP), under the muscle-specific acta1 promoter. (A) Western blot analysis of ErbB4 expression (anti-ErbB4 E200). Tropomyosin (clone CH1) expression was analyzed from the same lysates. All the samples were run on the same gels, but some nonrelevant lanes were omitted between samples 2 and 3, as indicated by white line. (B) Confocal whole-mount immunofluorescence analysis of GFP-positive mosaic muscle cells. GFP = green; anti-ErbB4 (E200) = red. Scale bar, 20 µm. (C) Confocal whole-mount immunofluorescence analysis of motoneurons with anti–synaptotagmin-2 (znp-1) antibody. Scale bar, 40 µm. (D) Quantification of the thickness of individual GFP-positive myocytes from confocal immunofluorescence images (representative images shown in B). The numbers of GFP-expressing myocytes (embryos) analyzed were as follows: control morpholino + EGFP, n = 28 (5); erbb4a morpholino + EGFP, n = 65 (5); erbb4a morpholino + ERBB4/EGFP, n = 26 (5). (E) Quantification of the length of motoneuron axons from confocal immunofluorescence images (representative images shown in C). The numbers of neurons (embryos) analyzed were as follows: control morpholino + EGFP, n = 24 (8); erbb4a morpholino + EGFP, n = 29 (9); erbb4a morpholino + ERBB4/EGFP, n = 18 (6). (F) Ten randomly selected swim tracks of touch–response assays with 48-hpf morphants expressing the rescue constructs encoding EGFP, or ErbB4 and EGFP (ERBB4/EGFP), under the muscle-specific acta1 promoter. (G, H) Quantification of the length of the swim burst (G) and the average speed during the burst (H) in the touch–response assays. Control morpholino + EGFP, n = 32; erbb4a morpholino + EGFP, n = 15; erbb4a morpholino + ERBB4/EGFP, n = 20.

Spontaneous motility is reduced in erbb4a mutant embryos

To control for the reported lack of specificity of the morpholino approach (Kok ; Rossi ), as well as potential promiscuity of chemical inhibition, an erbb4a mutant fish line was generated using CRISPR/Cas9-mediated targeted mutagenesis. We successfully isolated a 22 base pair–deletion allele of erbb4a (erbb4a) (Figure 6A) that causes a frame shift and premature stop codon in the DNA sequence encoding the extracellular domain of ErbB4a. Based on in silico translation, this results in a truncated ErbB4a protein lacking transmembrane, kinase, and intracellular domains (Figure 6B). Indeed, Western blot analysis of a pool of 96-hpf mutants using the anti-ErbB4 E200 recognizing the intracellular domain of ErbB4 demonstrated protein expression levels reduced to 65% for the heterozygous erbb4a embryos and to 19% for the homozygous erbb4a embryos (Figure 6C). The remaining 19% were comparable to the relative amount of the erbb4b transcript in the homozygous erbb4a embryos (16% when compared with the total number of erbb4a + erbb4b transcripts in the wild-type embryos by real-time RT-PCR), suggesting that the remaining ErbB4-immunoreactive band in the mutants derived, not from ErbB4a, but from ErbB4b protein.

FIGURE 6:

Defective motility in erbb4a mutant embryos. (A) Sanger sequencing data of wild-type and mutant erbb4a alleles. Red and blue lines indicate 5′ and 3′ sequences flanking the mutation site, respectively. The black bracket in the wild-type sequence indicates the deletion site and sequence missing from the ubs35 mutant allele of erbb4a. (B) Schematic presentation of wild-type and mutant ErbB4a proteins. (C) Western blot analysis of ErbB4 expression in 96-hpf wild-type and heterozygote (erbb4a) and homozygote (erbb4a) mutant embryos. Five to six embryos per genotype were pooled for Western analysis. (D–F) Spontaneous motility assay. Ninety-six-hpf zebrafish embryos were analyzed for spontaneous motility in 24-well plates (1 embryo/well) using a DanioVision instrument. (D) Tracks of swimming over a period of 10 min. (E, F) Quantification of the length of the swim tracks (E) and the average speed (F). Wild-type erbb4a, n = 13; heterozygote erbb4a, n = 18; homozygote erbb4a, n = 13.

To measure the effect of the erbb4a mutation on motility, an assay recording spontaneous motility for a period of 10 min was carried out (Supplemental Movie 2). Indeed, the embryos homozygous for the erbb4a mutation (erbb4a) moved for a reduced distance, as well as at lower speed, in comparison with heterozygous erbb4a embryos (Figure 6, D–F).

Movie S2

Movie S2

DISCUSSION

Treatment with pan-ErbB TKIs was observed to reduce the motility of developing zebrafish embryos. Several lines of evidence indicate that these effects were mediated, at least partially, by ErbB4. Chemical inhibition of the ErbB kinase activity, as well as genetic targeting of erbb4a, resulted in decreased motility of embryos, reduction of the diameter of muscle fibers, and reduced expression of the myosin genes cmcl2, smyhc1, and tpm1. The defects in the motility of the embryos were partially rescued by muscle-specific mosaic expression of human ERBB4, implying a requirement for ErbB4 expressed by the myofibers for the development of skeletal muscle. Consistently, the motility of embryos with targeted erbb4a was reduced both in touch–response assays and in assays measuring spontaneous swimming.

Therapeutic TKIs may promote off-target effects by inhibition of other structurally homologous kinases. Here, we observed similar defects in motility using different ErbB kinase inhibitors, as well as by antisense oligonucleotide and mutational approaches targeting erbb4a, indicating that the effects on motility were indeed on-target and specific for blocking ErbB4 signaling.

To confirm the expression of erbb4a (and erbb4b) in zebrafish independent of the use of antibodies, we analyzed relevant RNA-seq data sets available in repositories. In the embryonic brain (whole dissected head), the erbb4a expression was relatively high (erbb4a: 2.86 [range 2.59–3.13] FPKM; erbb4b: 1.53 [1.28–1.78] FPKM; Borck ). In adult muscle, erbb4a was also clearly expressed (erbb4a: 0.72 [0.58–0.87] FPKM; erbb4b: 0.76 [0.56–0.97] FPKM; Louie ). In FACS-purified spinal cord neurons, erbb4a mRNA was expressed at very low level, if at all (erbb4a: 0.016 [0–0.07] FPKM; erbb4b: 0.11 [0–0.23] FPKM; Friedmann ). Taken together, the expression pattern from our immunofluorescence analysis was in line with published RNA-seq expression data.

ErbB4 has previously been demonstrated to be expressed in human (Plowman ; Junttila ), mouse (Elenius ), and rat (Lebrasseur ) skeletal muscle tissue. In the rodent myocytes, ErbB4 protein is concentrated in the neuromuscular junctions, where it has been suggested to act as a postsynaptic neuregulin receptor (Zhu ; Trinidad ). The neuromuscular junctions are superficially normal and functional in mice lacking both Erbb2 and Erbb4 in the skeletal muscle (Escher ). However, detailed analysis has demonstrated altered AchR recycling and signal transmission in the neuromuscular junctions in the muscles of mice lacking both Erbb2 and Erbb4 (Schmidt ). ErbB4 has also been shown to activate AchR transcription in cultured rodent myotubes (Rimer ). Although our real-time RT-PCR and bungarotoxin labeling analyses failed to demonstrate significant quantitative changes in AchR expression, it remains to be elucidated whether AchR recycling or signaling is affected in the erbb4a-deficient zebrafish model.

ErbB-targeting TKIs are being tested in several early clinical trials for pediatric malignancies, such as leukemias, brain malignancies, and osteosarcomas (www.clinicaltrials.gov). While the extent to which ErbB TKIs provoke adverse neuromuscular effects in pediatric patients remains to be addressed, muscle spasms and musculoskeletal pain are among the common side effects associated with the ErbB4-targeting pan-ErbB inhibitors, such as afatinib, neratinib, and dacomitinib, in adult patients (Yang ; Chan ; Wu ). Interestingly, inhibitory mutations in the ERBB4 gene have been linked to familial and sporadic amyotrophic lateral sclerosis (ALS; Takahashi ), a disease characterized by both motoneuron loss and muscle atrophy (Al-Chalabi ). These findings suggest a role of ErbB signaling, and specifically ErbB4, in the regulation of neuromuscular function also in humans.

Taken together, our observations indicate that ErbB4 inhibition promotes neuromuscular toxicity in developing zebrafish. These findings underline the need to characterize the toxicological properties of novel molecularly targeted therapeutics used for pediatric patients.

MATERIALS AND METHODS

Fish culture

Wild-type zebrafish (D. rerio) were maintained according to standard procedures (Nüsslein-Volhard and Dahm, 2002) in the aquatic facilities of the Laboratory of Animal Physiology, University of Turku; Laboratory of Aquatic Pathobiology, Åbo Akademi University; and Biozentrum, University of Basel. The embryos were cultured at 28.5°C in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 0.01% methylene blue). In samples analyzed by fluorescence microscopy, pigmentation was inhibited by the addition of 0.2 mM phenylthiourea (Sigma) into the E3 medium. Housing of adult zebrafish and experiments with zebrafish embryos were performed according to the European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes, Statutes 1076/85 and 62/2006 of The Animal Protection Law in Finland, EU Directive 86/609, and Swiss national guidelines of animal experimentation (TSchV) under licenses MMM/465/712-93 (issued by the Ministry of Agriculture and Forestry, Finland) and 1014H and 1014G1 (issued by the Veterinäramt-Basel-Stadt).

ErbB tyrosine kinase inhibitors

To block ErbB kinase activity in the embryos, 10 µM of tyrosine kinase inhibitors AG 1478 (Calbiochem), erlotinib, gefitinib, lapatinib, afatinib, neratinib, canertinib (all from Santa-Cruz Biotechnologies), osimertinib, and dacomitinib (both from Cayman Chemical), diluted in DMSO, or DMSO alone was applied in E3 medium at 8 hpf, except as indicated for experimentation shown in Figure 4D. The final concentration of DMSO was 0.1% of the culture medium.

Morpholino antisense oligonucleotides

To block endogenous expression of erbb4a in zebrafish embryos, a morpholino antisense oligonucleotide targeting the translation initiation site of the erbb4a mRNA (5′-AGAAGGAAGCGAACCGGCCACATTT-3′) and a standard control morpholino targeting a mutated splice site of human β-globin mRNA (5′-CCTCTTACCTC AGTTACAATTTATA-3′) were purchased from Gene Tools LLC. The 25 nucleotide–long antisense morpholino sequence contained 15 mismatches with erbb4b and 12 mismatches with human ERBB4.

DNA constructs and mRNA production

To generate the fusion expression plasmid pcDNA3.1ERBB4EGFP, a BglII/NotI fragment of pERBB4-EGFP (Williams ) including an insert encoding human ErbB4 (nucleotides 99-3974; NM_001042599.1) fused to EGFP was blunted and ligated into the EcoRV site of pcDNA3.1. A NotI/XbaI fragment of pCS2FA-transposase (Kwan ) encoding the Tol2 transposon was inserted into pcDNA3.1 (Invitrogen) to generate pcDNA3.1Tol2. RNA was transcribed in vitro from template plasmids pcDNA3.1EGFP (kind gift from Chris Cartwright, Stanford University, Stanford, CA), pcDNA3.1ERBB4EGFP (human ERBB4), or pcDNA3.1Tol2 using T7 RNA polymerase and mScript mRNA production system (Epicentre). Poly A tails and 5′ cap structures were synthesized using mScript mRNA production system (Epicentre).

Plasmids for zebrafish transgenesis were cloned using the Tol2kit system (Kwan ). The expression constructs pDestTol2A2-acta1-EGFPpA for expression of the EGFP under the muscle-specific acta1-promoter was generated in a Gateway LR recombination reaction including pDestTol2A2 (Kwan ), p5E-acta1 (Jacoby ), pME-EGFP (Kwan ), and p3E-EGFPpA, using previously described methods (Kwan ). The plasmid p3EP2AEGFPpA encoding the self-cleaving peptide P2A from porcine teschovirus fused to EGFP was generated from p3E-EGFPpA (Kwan ) by PCR using the primers 5′-CAGGCTGGAGACGTGGAGGAGAACCCTGGACCTGTGAGCAAGGGCGAG-GAGCTGTTCAC-3′ and 5′-CTTCAGCAGGCTGAAGTTAGTA GCTCCGCTTCCCA-TGGTGGCCACTTTGTACAAGAAAG-3′. The P2A peptide allows the expression of two proteins under the same promoter, and is functional in zebrafish (Kim ). The expression constructs pDestTol2A2-acta1-ERBB4-P2A-EGFPpA for expression of human ErbB4 together with EGFP under the muscle-specific acta1-promoter was generated in a recombination reaction including pDestTol2A2, p5E-acta1, pDONR223-ERBB4 (Addgene plasmid 23875; Johannessen ), and p3EP2AEGFPpA.

Microinjection of morpholino oligonucleotides, mRNAs, and DNA constructs

Embryos at the one- to eight-cell stage were microinjected using Femtotip glass capillary needles (Eppendorf) and a Femtojet pressure system (Eppendorf) equipped with an Injectman2 micromanipulator (Eppendorf) attached to a StereoLumarV.12 stereomicroscope (Zeiss). Injection volume was calculated by injecting a small droplet into halocarbon oil. The diameter of the droplet was measured and the volume calculated.

Volumes of 2–4 nl were used in injections. Morpholinos and in vitro transcribed mRNAs were injected into the yolk at final concentrations of 7.5 ng/embryo (lowest dose to cause robust erbb4a knockdown at 48 hpf) and 500 pg/embryo, respectively. DNA constructs were injected into the yolk at a final concentration of 125 pg/embryo together with 120 pg/embryo of in vitro transcribed Tol2 transposon mRNA.

CRISPR/Cas9-mediated targeting of endogenous erbb4a locus

The guide RNA specific for the erbb4a locus was designed using ChopChop online software (Montague ; Labun ; (http://chopchop.cbu.uib.no/). The preparation of single-guide RNA (sgRNA) and its injection into one- to four-cell stage embryos were performed as previously described (Gagnon ). In brief, an erbb4a-targeting DNA oligonucleotiude (5´-TTCTAATACGACTC­ACTATAGAAGCAGAGCTGGCTGTTGTGTTTTAGAGCTAGA-3´) was annealed with the DNA oligonucleotide (5´-AAAAGCA CCGACTC­GGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTT­AACTTGCTATTTCTAGCTCTAAAAC-3´). The annealed oligonucleotides were filled in using Phusion DNA polymerase. The gel-purified DNA fragment was then used as a substrate for the in vitro RNA transcription reaction with T7 RNA polymerase (MAXIscript T7 Transcription Kit; Ambion). The resulting sgRNA was precipitated with ethanol and ammonium acetate and dissolved in RNase-free water. Purified sgRNA was complexed with recombinant Cas9 protein (Gagnon ) and injected into one- to four-cell stage embryos of the Tu zebrafish strain. The injected embryos were grown into adulthood and outcrossed with wild-type fish. The founder fish (G0) were identified by screening the offspring for the presence of mutations in erbb4a locus using PCR amplification and digestion with T7 endonuclease or AluI (both from New England Biolabs). A 22 base pair–deletion allele (erbb4a) was successfully isolated and F3 generation was used in the experiments. The erbb4a genotyping PCR was carried out as nested PCR with outer primer pair (5´-ACTCTGGTTCATTGAAACGTCA-3´; 5´-CAT CATGCATCAGAGTGCCAAG-3´) and inner primer pair (5´-ATCAACACATGCTGTATCTGCTCT-3´; 5´-CTTGAT­AGGAA AGTGCAGCGAG-3´) using Taq DNA polymerase (New England Biolabs). To obtain DNA, RNA, and protein samples from the same individual embryos to allow first genotyping and then pooling samples, a NucleoSpin TriPrep kit (Macherey-Nagel) was used.

Protein extraction and Western blotting

Five to twenty embryos were homogenized using a handheld motorized pellet pestle (Sigma) in lysis buffer (1% Triton X-100, 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 2 mM phenyl methyl sulfonyl fluoride, 10 μg/ml aprotinin, and 10 μg/ml leupeptin) or directly in sample buffer (2% SDS, 10% glycerol, 60 mM Tris-HCl, pH 6.8, 100 mM dithiothreitol, and 0.01% bromophenol blue). Homogenates were centrifuged for 15 min at 12,000 × g at 4°C. Supernatants were analyzed by Western blotting, as previously described (Kainulainen ), using the following primary and secondary antibodies: anti-ErbB4 (E200; Abcam), anti–β-tubulin (clone E7), anti–slow muscle myosin heavy chain (clone F59), anti–fast muscle myosin light chain (clone F310), anti-tropomyosin (clone CH1), anti–β-actin (sc-1616; Santa Cruz Biotechnology), horseradish peroxidase (HRP)-conjugated goat anti-rabbit (sc-2004; Santa Cruz Biotechnology), HRP-conjugated goat anti-mouse (sc-2005; Santa Cruz Biotechnology), and HRP-conjugated rabbit anti-goat (sc-2768; Santa Cruz Biotechnology). Anti–β-tubulin, anti-myosin, anti-tropomyosin, and anti–synaptotagmin-2 (znp-1) monoclonal antibodies developed by M. Klymkowsky, F. E. Stockdale, J.J.-C. Lin, and B. Trevarrow, respectively, were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biology, University of Iowa, Iowa City.

RNA extraction and real-time RT-PCR

RNA was extracted from one to four embryos homogenized into Trisure RNA extraction reagent (Bioline Reagents) supplemented with 10 µg/ml glycogen (Roche), as recommended by the manufacturer. cDNA was synthesized using the RevertAid H-Reverse Transcription kit (Fermentas). Real-time RT-PCR analyses of diluted (1:5) cDNA samples were carried out using ABI 7900HT (Applied Biosystems) and the following primers and probes: erbb4a, 5′-AAACCGCAACTTGTC­TTTCC-3′, 5′-CCAGAGGAAGATAGTCAAACTGG-3′, probe #21; erbb4b, 5′-ATGTGCATCCCCTGCACT-3′, 5′-CGTCTGAAGGCTGGCAGT-3′, probe #43; rpl13a, 5′-GCGGACCGAT­TCAATAAGGG-3′, 5′-GAAAGACGACCGAGGTGAGA-3′, probe #147; cmcl2, 5′-CAGG­AGCCCAGACCAACA-3′, 5′-AGCAGTT­TTCCCCCTCTTG-3′, probe #112; myhc4, 5′-CAAGCA GAAG­CAGCGTGA-3′, 5′-GGGTAGCACAAAGCCTTCAG-3′, probe #82; smyhc1, 5′-TGCCAAGACCATCAGAAATG-3′, 5′-CACACCAAAGTGAATTCGGATA-3′, probe #52; tpm1, 5′-GAACGCCTTGGACAGAGC-3′, 5′-TTCCAACTGAATTA­GTTCGTCTTCT-3′, probe #30; chrna1, 5′-TGTGTTTTACCTGCCCACAG-3′, 5′-GATCAACTCGACGATCACCA-3′, probe #15; chrnb1, 5′-CGAGTACTTCATCCGCAAGC-3′, 5′-ACCGTCAGGCTGGTATC­TGT-3′, probe #103; chrnd, 5′-TCATTGTGTTGAACCTGCATT-3′, 5′-GGAAGACGCTCCAAAA AGAA-3′, probe #107. All primers were obtained from Oligomer (Helsinki, Finland), and probes were Universal ProbeLibrary probes from Roche Applied Sciences.

Immunofluorescence staining

Forty-eight-hpf embryos were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) overnight at 4°C, or with 100% methanol for 20 min at –20°C. The samples were rehydrated in successive 5-min incubations in solutions containing 100%, 75%, 50%, 25%, and 0% methanol in PBST (0.1% Tween-20 in PBS). Nonspecific binding was blocked in 5% goat serum (Life Technologies) in PBST for 3 h at 4°C. The samples were incubated overnight at 4°C with the primary antibodies anti-myosin heavy chain (F59), anti-ErbB4 (HFR-1; Abcam), or anti–synaptotagmin-2 (znp-1; Developmental Studies Hybridoma Bank) diluted in 5% goat serum. After four washes with PBST, the secondary antibodies goat anti-mouse Alexa 488 and goat anti-mouse Alexa 555 (Life Technologies) in 5% goat serum were applied to embryos for 3 h at room temperature. After four washes with PBST, the embryos were mounted in 87% glycerol for imaging with an LSM510 META confocal microscope (Zeiss) or with a StereoLumar V.12 stereomicroscope (Zeiss).

To stain for acetylcholine receptors (AchR), embryos were fixed with 4% PFA for 2 h at room temperature, permeabilized with water for 4 h at 4°C, and treated with 1 mg/ml collagenase for 45 min at room temperature. After three washes with PBS, nonspecific binding was blocked with PBDT (PBS, 1% bovine serum albumin, 1% DMSO, 0.5% Triton X-100) supplemented with 5% goat serum for 1 h. To visualize AchRs, Alexa-555 labeled α-bungarotoxin (Life Technologies) was added to blocking solution and the incubation was continued for 3 h at room temperature. After four washes with PBDT, the embryos were mounted in glycerol and imaged with a Zeiss LSM 510 confocal microscope.

Image processing and analysis were performed using FIJI software. Noise in all confocal images was reduced using a median filter (2 × 2), and the contrast and brightness of images were adjusted. Bungarotoxin-alexa-555 stainings were quantified using median 2 × 2 filtering, local thresholding (radius 7), and the particle analysis tool in FIJI (particle size > 4).

Touch–response assay and digital video motion analysis

To analyze the motility of the zebrafish embryos, a touch–response assay was used (Granato ). Dechorionated 48-hpf embryos were placed in E3 medium and allowed to accommodate to room temperature. The embryos were gently touched at the tail with a plastic inoculation loop, and the subsequent swim burst was recorded with a digital video camera. Files were converted into AVI format, and image analysis was performed using ImageJ and the ParticleTracker plug-in (written by Fabrice Cordelieres, Institute Curie, Orsay, France; http://rsb.info.nih.gov/ij/plugins/track/track.html). Both the distance of the swim burst as well as the average velocity during the burst were recorded. Embryos that did not show any response to touch were excluded from further analyses.

Assay measuring spontaneous swimming and digital video motion analysis

To analyze the motility of erbb4a mutant embryos in the absence of physical touch, a DanioVision (Noldus IT) instrument equipped with controlled lighting was utilized. The 96-hpf embryos were assayed in 24-well plates (1 embryo/well). The plate was subjected to 5 min of light, followed by 5 min of darkness. The movement of the embryos was recorded throughout the experiment. The embryos were automatically tracked using Ethovision XT software, and the moved distance and velocity were measured. After experimentation, the embryos were genotyped as described above.

Statistical analyses

Two sample comparisons were performed using Student’s t test and three sample comparisons using analysis of variance (ANOVA) followed by Dunnet’s t post hoc test (IBM SPSS Statistics 20 and GraphPad Prism version 6.05). To improve the fit of data into the normal distribution, the touch response data presented in Figure 6, G and H, were subjected to transformation by natural logarithm prior to ANOVA and Dunnet’s t post hoc test. Columns in figures indicate average + standard error.

Bioinformatic analysis of RNA-seq data

Preexisting RNA-seq data sets were analyzed using Chipster software v3.12 (Kallio ). In brief, the FASTQ files (SRR5115715, SRR1648855, and SRR1616929) were retrieved from Sequence Read Archive (www.ncbi.nlm.nih.gov/sra), and the reads were aligned against the Ensembl genome assembly of the D. rerio genome (Danio_rerio.GRCz10.90) using TopHat2 (Kim ). This was followed by a reference-based assembly of transcripts using Cufflinks (Trapnell ). FPKM (fragments per kilobase of exon model per million reads mapped) values were used to report the expression levels.

Click here for additional data file.