Generation and validation of a myoglobin knockout zebrafish model

Previous studies using myoglobin (Mb) knockout mice and knockdown zebrafish have presented conflicting results about in vivo phenotypes resulting from the loss of this conserved and highly expressed protein, and therefore a new well-characterized knockout model is warranted. We here describe the generation of three distinct zebrafish mb knockout lines using the CRISPR/Cas system. None of the three lines exhibited any morphological phenotypes, changes in length, or lethality during embryonic and larval development. The adult homozygous knockout mb(Auzf13.2) zebrafish line were absent of Mb protein, had an almost complete degradation of mb mRNA, and showed no changes in viability, length, or heart size. Furthermore, transcriptomic analysis of adult heart tissue showed that mb knockout did not cause altered expression of other genes. Lastly, no off-targeting was observed in 36 screened loci. In conclusion, we have generated three mb knockout lines with indistinguishable phenotypes during embryonic and larval development and validated one of these lines, mb(Auzf13.2), to have no signs of genetic compensation or off-target effects in the adult heart. These findings suggests that the mb(Auzf13.2) shows promise as a candidate for investigating the biological role of Mb in zebrafish. Supplementary Information The online version contains supplementary material available at 10.1007/s11248-023-00369-3.


Introduction
Myoglobin (Mb) functions as an intracellular O 2 carrier and storage protein, highly expressed in vertebrate heart and skeletal muscle (Wittenberg and Wittenberg 2003).The in vivo function of Mb has been the topic of many studies using several different mouse knockout (KO) models (Garry et al. 1998;Gödecke et al. 1999;Park et al. 2019;Ono-Moore et al. 2021), and one zebrafish knockdown (KD) model (Vlecken et al. 2009) (Supp.Figure 1).However, the in vivo findings have been contradictory (Supp.Table 1).Initial studies in mice showed no apparent effects of the absence of Mb, except for reduced coloring of the heart and the soleus muscle (Garry et al. 1998).Later, it was reported that most Mb (−/−) mice die in utero with signs of cardiac failure (Meeson et al. 2001), whereas a recent study found normal survival rate of Mb (−/−) mice (Ono-Moore et al. 2021).In assessments of Vol:.( 1234567890) endurance exercise with high O 2 demand, one study found a decline (Merx et al. 2005), while another study observed no changes in endurance of Mb (−/−) mice (Garry et al. 1998).Reports on heart size in adult Mb (−/−) mice are also conflicting, with studies indicating no change (Gödedke et al. 2003;Merx et al. 2005;Hendgen-Cotta et al. 2008;Ono-Moore et al. 2021), and one reporting a decreased heart size (Hendgen-Cotta et al. 2017).A consequence of Mb loss in mice may involve reducing O 2 consumption in heart and skeletal muscle by shifting substrate utilization from fatty acids to carbohydrates.In the hearts of Mb (−/−) mice, increased glucose (Flögel et al. 2005) and lactate (Meeson et al. 2001) utilization has been reported.Flögel et al. (2005) observed a significant decrease in fatty acid utilization, while Meeson et al. (2001) only noted a non-significant trend.This shift from fatty acids is expected to result in an increased respiratory exchange ratio (RER).However, conflicting findings have emerged regarding changes in the RER in Mb (−/−) mice, with Merx et al. (2005) reporting an increased RER, whereas Ono-Moore et al. ( 2021) and Christen et al., (2022) did not observe this.Lastly, the recent study by Ono-Moore et al. (2021) did not report a transition from oxidative to non-oxidative muscle types as reported earlier in mice (Grange et al. 2001).One finding that is consistent between studies of Mb (−/−) mice is the increase in heart vascularity (Gödecke et al. 1999;Meeson et al. 2001;Mammen et al. 2003).In zebrafish, a morpholino mb knockdown model showed increased lethality, body curvature and heart defects (Vlecken et al. 2009).Taken together, these findings have failed to produce a consistent picture of how Mb functions in vivo, and whether compensation at various levels may occur or not.
Several factors might explain these contradictory findings.Different species and strains may show variable adaptation in the absence of Mb.Additionally, the method used to generate KO and KD models can result in unintended off-target effects.For instance, the Mb KO models (Garry et al. 1998;Gödecke et al. 1999) were made by replacing exon 2 with a neo selection cassette (Supp.Figure 1), which potentially interferes with expression of other genes (Pham et al. 1996;Scacheri et al. 2001;Meier et al. 2010).The risk of morpholino off-target effects has also been documented (Kok et al. 2015), and earlier studies, e.g.Vlecken et al. 2009, may not follow the best practice of today (Stainier et al. 2017).To the best of our knowledge, neither of the Mb KO models have considered the recent concept of genetic compensation (GC) (El-Brolosy et al. 2019;Ma et al. 2019).In short, GC is a process where degraded mRNA may result in the upregulation of similar genes.The common strategy of introducing an early frame shift to generate a KO model may trigger GC.Two other approaches can be used to avoid GC: creating RNAless mutants by deleting the promoter region or the entire locus, or generating in-frame deletions that lead to the expression of a non-functional protein (Sztal and Stainier 2020).However, larger deletions may cause the removal of regulatory elements, and translated non-functional protein may give rise to unwanted side effects.
Here, we generated three genetically distinct mb KO zebrafish lines using two different CRISPR/Cas approaches.All lines showed identical embryonic and larval development, indicating that none of the lines show any GC or off-target effects.One of the established lines was further validated using transcriptomics to confirm the lack of GC and off-target effects.We are confident that these lines will facilitate future studies of the in vivo function of Mb in zebrafish.

Zebrafish husbandry and ethics
Zebrafish were breed in-house using AB founders from the European Zebrafish Resource Center.Fish were fed four times daily and maintained on a 14-h-light-10-h-dark cycle on recirculating housing systems at 28 °C.Embryos were obtained by natural crosses, reared in E3 buffer [5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO 4 , 10 −5 % (w/w) methylene blue, 2 mM HEPES pH 7.2] at 28 °C.All experiments were carried out according to Danish legislation and approved by The Danish Animal Ethics Council (permit number 2017-15-0201-01380/2023-15-0201-01448).
STAR aligned reads were visualized in the Integrative Genomics Viewer (Robinson et al. 2011).CRIS-PRoff (Anthon et al. 2022) was used to generate a list of potential off-targets of gRNA4, 7, 12, and 21 in the danRer11 genome.The list was filtered for off-targets in transcribed regions with a read coverage of at least 10x.The filtered lists were manually screened for INDELs 1 kb up-and downstream from potential offtargets if coverage would allow.

Data analysis
The galaxy platform (v 23.0) was utilized for alignment, counting, and differential gene expression analysis (Community 2022).One-way Anova, X 2 , and exact Fisher's test were performed in R using stats (v 4.3.1).All graphs were produced in R using the tidyverse package (v 2.0.0)(Wickham et al. 2019).

Results and discussion
Nineteen gRNAs targeting the mb gene in zebrafish were designed in silico and co-injected individually with Cas9 mRNA into 1-cell zebrafish embryos.At 24 hpf, the mb gene was amplified using PCR and Sanger sequenced to estimate in vivo efficiency by using TIDE and ICE (Brinkman et al. 2014;Tim et al. 2018) (Supp.Table 4).

Vol.: (0123456789)
To generate potential KO founders (F0), two distinct approaches were employed.Line mb Auzf13.2 and mb Auzf13.6 were generated using gRNAs 4, 7, 12, and 21 together, whereas line mb Auzf13.3 was generated using gRNA 13 (Fig. 1a, Supp.Table 2), as detailed in Methods.The first line, mb Auzf13.2 , has a 1 bp deletion (C) and a 6 bp insertion (GGT GGT ), resulting in a frameshift and a premature termination codon (Fig. 1a, Supp.Table 5-6), expected to lead to nonsense mediated decay (NMD), and potentially GC.The frameshift in mb Auzf13.2occurs before His-89 (positioned in helix F8), involved in heme binding, causing any translated protein to be dysfunctional (Supp.Table . 7).The second line, mb Auzf13.3 , carries a large deletion (438 bp) spanning the acceptor splice site of exon 2 in the mb gene (Fig. 1a, Supp.Table 5), most likely resulting in transcription of mRNA lacking exon 2 (Supp.Table 6).The splicing of exon 1 and exon 3 will lead to a frameshift in exon 3. The absence of exon 2 will result in non-functional protein, as His-89 is located within this exon (Supp.Table 7).Because exon 3 contains the native termination codon, it is not expected that the introduction of a premature termination codon will result in NMD, thus decreasing the risk of GC.The third line, mb Auzf13.6 , carries a large (856 bp) in-frame deletion, resulting in the splicing of exon 1 and exon 2 (Fig. 1a, Supp.Table 5-6).Similarly, any translated protein will not be functional due to the lack of His-89 (Supp.Table . 7).In the absence of NMD, it is unlikely that the modification of the mb Auzf13.6 line will result in GC.
To characterize the three mb KO lines, we incrossed heterozygous F2 or F3 to allow blinded screening for viability, length, and any morphological/developmental defects during the first six days of development before genotyping.We observed the expected Mendelian ratios at 6 dpf in all three lines (Fig. 1b).Furthermore, we found no significant difference in length, according to genotype for any of the three lines (Fig. 1c).Lastly, we observed no genotype specific changes in the frequency of enlarged pericardium (Fig. 1d), or increased body curvature (Fig. 1e).In contrast, an earlier study based on mb morpholino knockdown in zebrafish, reported an enlarged pericardium, increased body curvature, and increased lethality (Vlecken et al. 2009).
For further characterization, we chose the mb Auzf13.2line as its limited nucleotide change makes it least likely to disrupt hypothetical genetic regulatory elements.F2 or F3 heterozygous mb Auzf13.2 were in-crossed and allowed to grow to adulthood.We observed no genotype specific changes in Mendelian distribution (Fig. 2a), length (Fig. 2b), or ventricle size (Fig. 2c-d) in the adult mb Auzf13.2zebrafish.Altogether, these findings indicate no change in development of the mb Auzf13.2(−/−)fish compared to wild-type fish.We did not observe a change in coloring of the heart by visual inspection (Fig. 2c) as seen in mice (Garry et al. 1998;Gödecke et al. 1999), possibly due to the presence of residual blood in the spongious zebrafish heart, as compared to the more compact mammalian heart.Nevertheless, the absence of Mb in heart muscle of mb Auzf13.2(−/−)fish was confirmed by western blotting (Fig. 2e).
To assess possible unwanted genomic changes in the mb Auzf13.2(−/−)line, we screened 36 in silico predicted potential off-target sites located in transcribed regions with sufficient read coverage (Table 1).Except for one deletion, located ~ 350 bp from a potential off-target, we did not observe any deviation compared to the reference genome (Supp.Figure 3).Based on the large distance between the predicted offtarget site and the observed deletion, it is most likely not a consequence of the editing procedure.The observed deletion is located in the untranslated region (UTR) of the igf1rb gene, thus unlikely to have any effect.The deletion is located on a different chromosome than mb, and is probably carried by our wildtype strain (AB).

Conclusion
We successfully generated three distinct KO zebrafish lines for mb using the CRISPR/Cas system.Although these lines carried different mutations, all were expected to result in disruption of Mb function, either by the disruption of translation, or by the translation of a non-functional Mb protein.There were no significant changes in viability, embryo length, or other morphological phenotypes during the initial six days of development in all lines.Among the three lines, we selected the mb Auzf13.2line, which carried a mutation least likely to interfere with regulatory elements, for further validation.We confirmed the absence of Table 1 Assessment of modification at predicted off-target sites The total number of critical off-targets as predicted by CRISPR-off (Anthon et al. 2022) for the four gRNAs used for generation of the mb Auzf13.2zebrafish line is shown in the All row.The following two rows show the number of off-targets in genes and exons, respectively.The last row shows the number of INDELs found within 1 kb of the predicted off-target sites

Fig. 1
Fig. 1 Generated myoglobin (mb) knockout lines, embryonic body length and genotype distribution.a Schematic representation of the zebrafish mb gene (top) and the three variants generated in this study.Exons are shown as grey bars, introns as black lines, and stop codon and UTR are white bars.The positions of guide RNAs and His-89 are indicated by numbered arrows and text, respectively.For generation of the mb Auzf13.2 and the mb Auzf13.6 lines, gRNAs 4, 7, 12, and 21 were used.For generation of the mb Auzf13.2gRNA 13 was used.Insertions are shown as a black bar in mbAuzf13.2 .Deletions are shown as dotted lines in mb Auzf13.3 and mbAuzf13.6 .b Genotype frequency at 6 dpf from F2 or F3 mb(±) in-crosses.Using an X 2 -test, no statistically significant deviation from the expected Mendelian distribution in mb Auzf13.2(X 2 = 3.372, df = 2, p = 0.1853), mb Auzf13.3(X 2 = 0.84921, df = 2, p = 0.654) or mbAuzf13.6

Fig. 2
Fig.2Characterization of adult (3-6 months post fertilization) mb Auzf13.2zebrafish resulting from in-crossed mbAuzf13.2(±) .a Genotype frequency distribution.Using an X 2 -test, no statistically significant deviation from the expected Mendelian distribution (X 2 = 0.18367, df = 2, p = 0.9123) was found.b Zebrafish length relative to mean length of mb Auzf13.2(+/+)siblings.Using a one-way ANOVA test, no statistically significant difference in length according to genotype (F(2) = .474,p = 0.624) was found.c Adult hearts from wild-type and knockout (mb Auzf13.2(−/−) ) fish.d Ventricle size measured by its cross section.Using a one-way ANOVA test, no statistically significant difference in ventricle cross section mb mRNA and of Mb protein in heart ventricles of the mb Auzf13.2(−/−)line.Additionally, we screened 36 potential CRISPR/Cas off-targets and found no mutations at these sites.Moreover, there were no alterations in viability, standard body length, or heart size in adult mbAuzf13.2.The removal of Mb did not lead to any appreciable changes in the expression of other genes, indicating no GC mechanisms in the mb Auzf13.2line.Based on these findings, we are confident that the mb Auzf13.2KO line has undergone sufficient validation to warrant further functional analysis.This line holds potential for elucidating the in vivo function of the mb gene in zebrafish. functional