Introduction

Growth of blood vessels during development as well as in the adult organism is a tightly regulated process, which is controlled by endothelial cell (EC) behaviors such as cell migration, proliferation and differentiation. Misregulation of vascular growth not only contributes to ischemic conditions, but overgrowth also directly aggravates diseases such as growth and metastasis of cancers or age-related macular degeneration.

Vascular endothelial growth factors (Vegfs) and their Vegf receptors (VEGFR-1/Flt1, VEGFR-2/Kdrl-Kdr and VEGFR-3/Flt4) are the major regulators of vascular growth processes [1–3].

While VEGFA and VEGFR-2 (Kdrl in zebrafish) mainly regulate angiogenic processes such as sprouting and remodeling of vessels [4, 5], VEGFC and VEGFR-3 have mainly been recognized for their role in regulating development of the lymphatic endothelial system [6–8]. Vegfr3/flt4-deficient zebrafish completely lack lymphatic vessels but show no major defects in blood vessel growth [9]. VEGFR3-deficient mice die of defective vascular development before the lymphatic system becomes established [10]. Vegfc mutant mice as well as zebrafish lack a lymphatic system [11, 12]. Angiogenesis defects observed in vegfc zebrafish mutants include failure in EC migration [during formation of the primordial hindbrain channels (PHBCs)] [11] as well as reduced EC proliferation [in the common cardinal veins (CCVs)] [13].

However, VEGFA as well as VEGFC expression are both upregulated in various tumors and their misregulation is involved in other diseases; therefore, understanding the mechanisms regulating their expression are of clinical relevance [14, 15].

Within cultured fibroblast or cancer cells, VEGFC mRNA expression was shown to be upregulated by cytokines (Interleukin-1α or interleukin-1β, or Tumor necrosis factor-α) [16] and growth factors (Platelet derived growth factor, Epidermal growth factor and Transforming growth factor-β) [17], but not by Hypoxia-inducible factor-1α (HIF1α) [18].

The optical clarity of the externally developing zebrafish embryos is one of the many advantages for using this model for the analysis of vascular development. The growing vasculature can easily be visualized in vivo by endothelial-specific transgenic fluorophore expression [19].

The vascular anatomy of zebrafish embryos has a high structural homology to other vertebrates [20, 21]. Similarly, most signaling pathways regulating vascular development are conserved between zebrafish and mammals [22, 23]. A functional circulatory system including a primitive heart is already established in the zebrafish embryo by 24 h post-fertilization (hpf).

Additionally, recent advances in genome editing using Transcription activator-like effector nucleases (TALENs) or Cas9 nucleases [24, 25] enabled gene-specific targeting in zebrafish.

We performed gene expression analyses to identify novel regulators of angiogenesis in zebrafish embryos and thereby identified metallothionein 2 (mt2) as a candidate.

MTs are low-molecular-weight and cysteine-rich proteins, which are conserved throughout the animal kingdom. There are four classes of mammalian Mt genes, Mt1–4 [26, 27], and two mt genes in zebrafish, mt2 and metallothionein-B-like (mtbl) [28, 29].

The main function of MTs is the regulation of homeostasis, such as the protection against oxidative stress or metals. Both heavy and trace metals such as zinc, copper or iron can be chelated via sulfur-based clusters [30, 31].

However, MTs also display non-canonical functions in angiogenesis and pathological conditions. Mt1 and Mt2 are very similar and the best characterized genes of the MT family, which can act as tumor suppressors [32] and have cardio- and neuroprotective functions [33–35]. Mice deficient for both Mt1 and Mt2 are viable and show beside their greater sensitivity to metals no major developmental defects [36, 37]. When challenged by femoral artery ligation or cortical freeze injury, these Mt1/2 knockout mice show impaired angiogenesis and wound healing [38–40]. MT3 is important for cell growth [41], and its expression is downregulated in a carcinoma cell line [42]. MT3 also has a critical role in the recovery of the brain, since Mt3-deficient mice show increased oxidative stress and apoptosis upon cortical freeze lesion [43]. The non-inducible Mt4 is expressed in epithelial tissues and has only been shown to detoxify of metals [30, 44].

However, how MTs exert their non-canonical functions, such as the regulation of angiogenic processes, is not understood.

Here, we used zebrafish as a model to analyze the role of Mt in angiogenesis. We generated Mt2-deficient zebrafish embryos by performing antisense morpholino-oligonucleotide (MO)-mediated gene knockdown as well as by using TALEN to generate zebrafish mt2 mutants. Using in vivo time-lapse analysis, we show that mt2 deficiency leads to striking angiogenesis defects, especially to defective formation of the PHBCs. Furthermore, we demonstrate that Mt2 acts upstream of vegfc expression in regulating EC migration and proliferation. This regulation of angiogenesis represents a non-canonical function of Mt2, since another Metallothionein family member (Mtbl) cannot regulate vegfc expression.

Materials and methods

Zebrafish maintenance and strains

Zebrafish embryos were maintained at 28.5 °C under standard husbandry conditions [45]. Zebrafish lines used were Tg(kdrl:EGFP)s843 [46], Tg(fli1a:EGFP)y1 [47] and Tg(fli1a:nEGFP)y7 [48]. The vegfc hu6410 allele encodes a stop codon at amino acid position 107 (L107X) [49].

Generation of the mt2 mutant transgenic zebrafish line using transcription activator-like effector nucleases (TALENs)

TALENs were assembled using the Golden Gate method [50]. For targeting the mt2 locus, a 5′ RVD (NH–NG–NH–NH–NI–NG–NI–HD–NG–HD–NG–HD–NG–NH (DNA sequence GTGGATACTCTCTGG)) and a 3′ RVD (NI–HD–NG–HD–NG–NG–NH–NH–HD–NI–HD–NI–NG–NG (DNA sequence ACTCTTGGCACATTC)) were generated with a spacer of 16 bp (AAAAATGGACCCCTGC) to target exon 1. An AvaII (New England BioLabs) restriction site within the spacer region was used for genotyping of putative founders. mRNA was generated using the T3 mMessage mMachine Kit (Ambion) and injected using 100 pg of the TALEN mix.

mRNA and morpholino (MO) injections

MOs blocking either translation (MO) or RNA splicing (spbMO) were obtained from Gene Tools and are as follows:

mt2 MO: 5′-GGTCCATTTTTCCAGAGAGTATCCT (5 ng) and mt2 spbMO: 5′-AGCTGAAACACTTACTCTTGGCACA (7–10 ng), targeting mt2 (BC152694.1); mtbl MO: 5′-CTGGTCCATCTTTACACCGTAGGTC and mtbl spbMO: 5′-AGTTAATCGGCTCACTTTTCTTGTC (both 13 ng) targeting mtbl (NM_001201469.1), upf1 spbMO: 5′-TTTTGGGAGTTTATACCTGGTTGTC (0.1 ng) [51] and smg1 spbMO: 5′-AACCATTGGTTTGTTACCTGGTGCA (12.5 ng) [51] and standard control MO: 5′-CCTCTTACCTCAGTTACAATTTATA.

For overexpression experiments, the mt2 sequence was amplified from 24 hpf cDNA and cloned into the pCS2+ vector for in vitro RNA synthesis using the following primers: mt2fwd 5′-AGACGAATTCGCTCCACCATGGACCCCTGCGAATGTGC and mt2rev 5′-AGACCTCGAGTCATTGACAGCAGCTGGAGC.

Similarly, mtbl was cloned into the pCS2+ vector using mtblfwd 5′-AGACGAATTCGCTCCACCATGGACCAGTGTAACTGCTC and mtblrev 5′- AGACCTCGAGTCATTTGCAGCAGTGTGTGG.

The mRNA was synthesized using SP6 mMessage mMachine Kit (Ambion). For all experiments, the injection was done into the yolk of 1-cell-stage zebrafish embryos, and 0.05 % phenol red (Sigma) was added to the injection solution.

Injection amounts per embryo were as follows: 500 pg mt2 mRNA, 100 pg mtbl mRNA, 100 pg to 500 pg H2B-cherry mRNA [52], 200 pg vegfc mRNA [53] and 200 pg sFLT4 mRNA [54].

RNA and DNA isolation, qPCR analysis and genotyping

RNA from WT, mutants and MO-injected embryos was isolated with Trizol reagent, and cDNA was generated by SuperScript II reverse transcriptase (Invitrogen).

The cDNA was analyzed with real-time quantitative PCR (qPCR) using Power SYBR Green (Applied Biosystems) and the following primers: b-actinfwd 5′-CTGGACTTCGAGCAGGAGAT and b-actinrev 5′-GCAAGATTCCATACCCAGGA (156 bp amplicon); vegfcfwd 5′-GCAGGAACATCAGCACTTCA and vegfcrev 5′-GTGTGGTTGGCGAAGCTTAT (103 bp amplicon); fli1afwd 5′-CTCAGGGAAAGTAGCTCATCG and fli1arev 5′-CTTTTCCGCTGTGCATGTT (139 bp amplicon); myod1fwd 5′-TCTGATGGCATGATGGATTT and myod1rev 5′-TTATTATTCCGTGCGTCAGC (110 bp amplicon). For qPCR at least two different cDNA samples were generated and analyzed. Experiments were performed at least three times.

The knockdown efficiency of the mt2 splice MO was validated with reverse transcription PCR (RT-PCR) and primers mt2fwd 5′-ATGGACCCCTGCGAATGTGC and mt2rev 5′-TCTTCTTGCAGGTAGTACACTG (spliced amplicon 91 bp, non-spliced amplicon 185 bp). The functionality of the mtbl splice MO was analyzed with mtblfwd 5′-GACCAGTGTGACTGCTCCAA and mtblrev 5′-TGCAGGATTTCTCCTTGTCC (spliced amplicon 169 bp, non-spliced amplicon 327 bp).

DNA was extracted using lysis buffer (10 mM Tris–HCl, 50 mM KCl, 0.3 % Tween 20, 0.3 % Nonidet P-40, pH 8.3) with 0.5 μg/μl proteinase K (Roche) overnight at 55 °C, followed by 10 min denaturation at 95 °C.

The genotype of the mt2 mu290, the mt2 mu292 and the mt2 mu289 mutants was analyzed with primers mt2fwd 5′-TCTTCTTGCAGGTAGTACACTG and mt2rev 5′-TAAAAGCAGAGCACAAACACG and the restriction enzyme AvaII.

The genotype of the vegfc hu6410 zebrafish mutants was analyzed in a multiplex PCR with WT and mutant zebrafish-specific inner and outer primers. As inner primers vegfcfwd 5′-CTTTCATCAATCTTGAACTTTT (WT specific) and vegfcrev 5′-TAAATTAATAGTCACTCACTTTACT (mutant specific with one mismatch) were used and as outer primers vegfcfwd 5′-GATGAACTCATGAGGATAGTTT and vegfcrev 5′-AAACTCTTTCCCCACATCTA.

Whole-mount in situ hybridization

Whole-mount in situ hybridization was performed as described [55]. The mt2 probe was amplified from 24 hpf zebrafish embryo cDNA with the T7 promoter site and the following primers: mt2fwd 5′-GGAACTTTCAAGCTCTTTGTGG and mt2rev 5′-gTAATACgACTCACTATAggGACAAAGGACATGGCAGAAAA. The vegfc probe is already described [53].

Confocal microscopy and in vivo time-lapse analysis

Zebrafish embryos were analyzed with confocal microscopy as previously described, using 1 % agarose embryo moulds [56]. The fluorescent images were acquired using the Sp5 DM 6000 upright confocal microscope (Leica) or the inverse LSM 780 confocal microscope (Zeiss).

BrdU incorporation and immunohistochemistry

Proliferation analysis was performed as described [57] with following modifications: Embryos were grown to 24 hpf and then incubated in 10 mM 5-bromo-2′-deoxyuridine (BrdU) for 30 min on ice. After 8 h of further incubation and BrdU incorporation, embryos were fixed in 4 % paraformaldehyde (PFA) at 32 hpf. After incubation in 2 M HCl for 1 h, permeabilization (phosphate-buffered saline (PBS) with 0.3 % Triton X−100 (Sigma) and 0.1 % Tween 20 (Sigma)) and blocking (PBS with 0.3 % Triton X−100 and 4 % BSA (Roth)), the following antibodies were used for immunostaining: mouse anti-BrdU (1:100, Roche), Alexa 546 anti-mouse (1:500, Invitrogen) and Alexa 488 anti-GFP (1:500, Invitrogen, for ECs of Tg(kdrl:EGFP)s843). After each antibody incubation, extensive washing was performed (PBS with 0.3 % Triton X−100).

Phenotypic analysis, quantifications, statistics and softwares

For evaluation of the rescue experiments, different clutches of at least three different experiments were scored for the existence of the PHBCs. If the PHBCs were not present at all or developed to <50 %, they were considered as missing; if the PHBCs were developed to more than 50 % or fully connected, they were considered as existent. For rescue experiments of vegfc hu6410 zebrafish mutants, only embryos with a strong PHBC phenotype or with fully developed PHBCs were taken for analysis for both Ctr and mt2 mRNA-injected zebrafish, and each embryo was subjected to subsequent genotyping. Cell numbers of fixed Tg(kdrl:EGFP)s843 or Tg(fli1a:nEGFP)y7 zebrafish embryos were evaluated with help of the Spots function of Imaris. Cells in the PHBC, the anterior cluster and the posterior clusters were counted at 32 hpf in confocal stacks. Similarly, ECs of the ACV, PCV and CCV were counted at 32 hpf, while ECs of the Ses were counted at 48 hpf.

For analysis of the amount of proliferating cells in the CCVs, BrdU-positive cells were calculated relative to the total number of ECs in the CCVs. For quantifying the Ses, the region between somites 9 and 14 has been analyzed. The P values for the experiments were calculated with a two-tailed Student’s t-test. The rescue experiments for the PHBC phenotypes were evaluated for significance with the Chi-Square test using Microsoft excel. SDS 2.3 and RQ Manager (Applied Biosystems) were taken for analysis of the real-time data. Primers were designed using Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/primer3/input.htm).

Where possible, the analysis was performed blind to experimental conditions.

Results

Mt2 regulates EC behavior during angiogenesis

To identify regulators of EC migration, we screened for functional involvement of candidate genes using morpholino antisense oligonucelotides (MO) to knockdown protein expression in zebrafish embryos and analyzed their vascular development using endothelial-specific GFP expression (Tg(kdrl:GFP)s843).

We identified Mt2 as a potential regulator of EC migration. For a detailed analysis, we injected MOs either inhibiting mRNA translation (using MO covering the ATG) or blocking mRNA splicing (spbMO). Embryos injected with mt2 MO or mt2 spbMO, showed brain necrosis but no other major morphological defects (Fig. 1). Of the affected vessels, the PHBCs are the first to develop. They grow by angiogenic sprouting out of an anterior cluster and a posterior cluster of ECs, which start at 18 hpf to migrate toward each other and connect around 22–23 hpf to form a functional vessel (Fig. 1d, e; movie 1). At 24–25 hpf, circulation starts and blood flow can be observed going through the PHBCs. However, we observed not only defective growth of the PHBCs, but also of the CCVs and the Ses at different time points of development (Fig. 1, Fig. S1).

Fig. 1
figure 1

mt2 morphants fail to form the PHBCs. a Schematic illustration of the vasculature of a 24 hpf old zebrafish embryo with the location of the PHBC and the adjacent PHBC forming clusters. b, c Brightfield images of Ctr (b) and mt2 (c) MO-injected zebrafish embryos at 24 hpf showed no morphological defects, apart from a mild necrosis in the head. d Schematic illustration of the development of the PHBC (dark blue) between 17 and 30 hpf. e, f Confocal micrographs from time-lapse movies showing the development of the PHBC in zebrafish embryos between 18 and 30 hpf. The vasculature was visualized by transgenic GFP expression using Tg(kdrl:EGFP)s843 embryos. In embryos injected with Ctr MO, ECs migrate from the anterior and the posterior cluster and connect to form the PHBC before 24 hpf (at around 23 hpf; e). In embryos injected with mt2 MO ECs fail to migrate and therefore do not form the PHBCs (f). White arrows indicate the anterior and posterior migration front of the PHBC. White arrowheads indicate filopodia in mt2 morphants. g Quantification of EC numbers in Ctr MO-injected (black bars) or mt2 MO-injected (white bars) and mt2 spbMO-injected (gray bars) embryos counted from vascular-specific nuclear GFP expression (Tg(fli1a:nEGFP)y7). While the total EC numbers were not affected, mt2 MO-injected embryos showed fewer ECs in the PHBC and more ECs in the clusters. n = 20, *P < 0.05; n.s., not significant; error bars indicate standard error of the mean (SEM). h Analysis of mt2 splicing efficiency in embryos injected with Ctr MO or mt2 spbMO. RT-PCR analysis showed a 185 bp amplicon in embryos injected with mt2 spbMO, while functional splicing led to a 91 bp amplicon in Ctr MO-injected embryos. (Color figure online)

Full size image

We used in vivo time-lapse imaging to further analyze PHBC formation in control MO (Ctr)- or mt2 MO-injected Tg(kdrl:EGFP)s843 embryos. The Ctr and mt2 MO-injected (morphant) embryos were indistinguishable from each other until 18 hpf (Fig. 1e, f), with both displaying normal development of the lateral dorsal aorta (LDA). In Ctr morphants, the ECs migrated, connected and thereby formed the PHBCs (Fig. 1e, movie 1), whereas mt2 morphant ECs failed to migrate out of the clusters and did not connect to form the PHBC (Fig. 1f, movie 2). However, the ECs were motile and formed filopodia, but the directed migration required for the connection of the PHBCs was perturbed (movie 2). Some mt2-deficient embryos extended sprouts from the anterior and posterior cluster to develop the PHBCs, but no proper connection was established. To determine, whether this defect is caused by defective migration or reduced EC numbers, we counted the number of ECs in the PHBCs as well as in the anterior and posterior cluster at 32 hpf, long after PHBC formation should have been completed. While the total EC number in PHBCs and clusters was similar, Ctr morphants had an average of 21 cells in the PHBC and 12 cells in the clusters, whereas mt2 morphants had an average of 13 cells in the PHBC and 22 cells in the cluster (Fig. 1g). Therefore, our results indicate that Mt2 regulates EC migration during PHBC angiogenesis.

Additionally, we analyzed CCV formation in Ctr and mt2 morphants in more detail (Fig. S1). The CCVs grow at a 90 °C angle out of the trunk ACV and posterior PCV, by a combination of collective EC migration and proliferation [13]. At 32 hpf the total cell number in ACV, PCV and CCV was reduced in mt2 morphants compared with Ctr morphants (Fig. S1j). However, the percentage of cells in the CCV was significantly reduced from 35 % CCV cells in Ctr embryos to 26 % in mt2-deficient embryos (Fig. S1i). To test whether Mt2 regulates EC migration or proliferation in the CCV, we performed proliferation analysis by BrdU incorporation in mt2 morphants. Proliferation was strongly decreased in mt2 morphants (Fig. S2).

In sum, our results indicate that Mt2 regulates angiogenesis, by regulating EC migration in the PHBCs and EC proliferation during CCV formation.

mt2 zebrafish mutants phenocopy the mt2 morphants

Despite performing extensive control experiments, MOs have been shown to exhibit off-target effects [58–60]. To verify the phenotype of the mt2 morphants, we used TALENs [50] to induce double-strand breaks in the mt2 gene. As expected, errors made by the repair machinery of the cell then led to mutations in the double-strand break area [50]. Since the mt2 sequence is very short, we targeted exon 1, which consists of 25 base pairs (bp) only (Fig. 2a). We identified several different alleles of mt2 mutations and further analyzed three of them.

Fig. 2
figure 2

TALEN-generated mt2 zebrafish mutants fail to form the PHBC and have different levels of NMD of mt2 transcripts. a 5′ and 3′ TALEN arms were designed to target exon 1 of the mt2 gene to induce mutations in the genome. b TALEN injection resulted in various genomic mutations. Illustrated is the comparison of the amino acid sequence in WT and different Mt2 mutant alleles, red color indicates mutated amino acids. In mt2 mu289 mutants a 6 bp deletion resulted in deletion of two amino acids, while in mt2 mu290 and mt2 mu292 frameshift mutations resulted in complete changes of the amino acid sequence. c–f Confocal images of the PHBCs at 24 hpf. WT (c) and MZmt2 mu289 mutant zebrafish embryos (d) form a PHBC, while MZmt2 mu290 and MZmt2 mu292 mutant embryos fail to connect the PHBCs. The vasculature was visualized by transgenic GFP expression from Tg(fli1a:EGFP)y1 for MZmt2 mu289 and MZmt2 mu290 mutant embryos and from Tg(kdrl:EGFP)s843 for MZmt2 mu292 mutant embryos. White arrows indicate the anterior and posterior migration front of the PHBC. mt2 expression was analyzed by in situ hybridization in 24 hpf-old embryos. While WT siblings (g) and MZmt2 mu289 mutants (h) had similar mt2 expression levels, nonsense-mediated decay led to degradation of mt2 mRNA transcript in MZmt2 mu290 (i) and MZmt2 mu292 (j) mutant embryos. Black arrows indicate mt2 expression in cells of the yolk extension, black arrowheads label the region of the PHBCs. k–n Brightfield images of WT, MZmt2 mu289, MZmt2 mu290 and MZmt2 mu292 mutant embryos at 24 hpf. WT siblings and MZmt2 mu289 displayed no morphological defects. MZmt2 mu290 and MZmt2 mu292 mutant embryos are smaller in size and display necrosis in the head

Full size image

In the mt2 mu289 mutant allele only 6 bp were deleted, which resulted in deletion of the second and third amino acid of the Mt2 protein (Fig. 2b, S3). The mt2 mu290 sequence has two-point mutations and an insertion of 8 bp, which led to a frame shift and an early stop codon. The mt2 mu292 sequence has a deletion of 15 bp, which lead to a frame shift and an early stop codon (Fig. 2b, S3). Since the mutations of mt2 mu290 and mt2 mu292 are located next to the start codon and there is no downstream start codon in frame, the original Mt2 protein sequence should be completely lost, supposedly resulting in null mutants.

By mating F1 heterozygous carriers of each mt2 allele, we obtained homozygous F2 embryos. To our surprise, we only detected very weak phenotypes (data not shown). Since mt2 is maternally provided [61], we hypothesized that the maternal mRNA is sufficient to rescue mt2 deficiency during the early developmental stages analyzed. Therefore, we raised homozygous F2 embryos to adulthood. When mating homozygous mt2 mutant fish to obtain maternal and zygotic mutant (MZ) F3 offspring, we observed strong morphological and angiogenesis phenotypes (Fig. 2), phenocopying the mt2 morphants. Both MZmt2 mu290 and MZmt2 mu292 mutants failed to connect the PHBCs, had reduced cell numbers in the CCVs and defective Ses formation. (Fig. 2e, f; Fig. S4). Additionally, in a subset of mt2 mutant embryos the morphology of the PHBC clusters was affected, with the ECs forming ectopic sprouts (arrowhead, Fig. 2e). The MZmt2 mu289 zebrafish mutants, which lack only two amino acids, displayed only a very weak phenotype. The PHBCs (Fig. 2d) and the Ses (Fig. S4g) developed normally in those mutants, while a mild phenotype could be observed in the CCVs (Fig. S4b).

The penetrance and severity of the phenotype for both null mutants were variable within the clutch and between clutches. MZmt2 mu290 zebrafish mutants showed severe phenotypes at higher rates than MZmt2 mu292 zebrafish mutants (compare Table 1), although both should not retain any amino acid sequence of Mt2. In order to investigate whether the mutations were causing strong alleles, we examined the level of gene transcription. One mechanism potentially interfering with mRNA transcript stability in mutants is nonsense-mediated decay (NMD), whereas MO-mediated blocking of translation would rather stabilize the transcript.

Table 1 mt2 zebrafish morphants, MZmt2 zebrafish mutants and vegfc hu6410 zebrafish mutants display many common phenotypes
Full size table

We therefore subjected 24 hpf-old MZmt2 mutant embryos to in situ hybridization to analyze the presence of the mt2 transcript. While WT and MZmt2 mu289 mutant embryos expressed mt2 as published [61], almost no expression could be observed in MZmt2 mu290 and MZmt2 mu292 mutants (Fig. 2g–j). Interestingly, the efficiency of NMD was not the same for both null mutants. While the great majority of MZmt2 mu292 embryos completely lacked mt2 expression (Fig. 2j), some MZmt2 mu290 mutant embryos retained mt2 message partially (Fig. 2i), which correlated with the different frequencies of angiogenesis defects (Table 1). We hypothesized that the more efficient the NMD was for the mt2 zebrafish mutant, the more compensation mechanisms might take place to attenuate the phenotype. To analyze, whether mRNA stability could indeed influence the phenotypic severity, we partially ablated two subunits of the NMD mediating complex (smg1 and upf1) by injecting smg1/upf1 MOs in WT and in MZmt2 mu290 mutant embryos. We could indeed observe an increase in the number of affected embryos, when message degradation by NMD was reduced (Fig. S5). This indicates that indeed the correlation of the stronger phenotype with the reduced mRNA degradation is functionally relevant. The sum of this data implies that different levels of mRNA degradation can lead to differences in the phenotypes of generated zebrafish mutants and morphants, potentially by regulating unknown compensatory mechanisms.

Mt2 acts upstream of Vegfc in regulating angiogenesis

Mt2 deficiency resulted in angiogenic defects during PHBC and CCV formation. Both of these processes have been described to be regulated by Vegfc during zebrafish embryonic development. Vegfc mutants or morphants fail to connect the PHBCs and have reduced proliferation in their CCVs [11, 13].

We therefore carried out different rescue experiments to analyze whether there is an interaction of Mt2 and Vegfc signaling. We ubiquitously overexpressed Vegfc in WT or mt2 morphant embryos by injection of vegfc mRNA into 1-cell-stage embryos. Overexpression of vegfc in WT embryos did not alter EC migration to form the PHBCs (Fig. 3a), but significantly reduced the number of embryos with PHBC connection defects from 44 % affected mt2 morphants to 25 % affected vegfc-injected mt2 morphants (Fig. 3c, j, n). Furthermore, we overexpressed the Vegfc ligand trap sflt4 [54], which is a soluble form of the Vegfr3, that titrates away Vegfc and therefore results in the same phenotypes as the genetic vegfc mutation (Fig. 3e, k). By combining sflt4 with high amounts of mt2 mRNA injection, we could compensate the PHBC formation failure (Fig. 3f, k, o). Injection of vegfc mRNA rescued the sflt4 mRNA injection to a similar extent (data not shown), indicating that Mt2 overexpression could indeed compensate for Vegfc ligand depletion. Interestingly, when we repeated the same experiment of rescuing Vegfc deficiency by mt2 overexpression in vegfc hu6410 mutant embryos, Mt2 failed to rescue (Fig. 3i, l, p), suggesting that vegfc is the only relevant target of mt2. We confirmed the upregulation of vegfc transcripts after mt2 mRNA injection in vegfc hu6410 mutant embryos with qPCR (Fig. S6). Taken together our data showed that Mt2 deficiency can be overcome by Vegfc addition and that Mt2 overexpression can outcompete Vegfc protein depletion, but not Vegfc mutation. These results are consistent with a mechanism, in which Mt2 regulates vegfc RNA expression (Fig. 3m).

Fig. 3
figure 3

Mt2 acts upstream of Vegfc in PHBC formation. a–c, j Injection of vegfc mRNA rescued mt2 deficiency in mt2 morphants. Overexpression of vegfc mRNA does not disturb PHBC formation (a). Upon mt2 MO injection (b), 44 % of the embryos lack the PHBC, while upon co-injection with vegfc mRNA (c) PHBC formation becomes rescued in half of the affected embryos (quantification of different experiments shown in j). d–f, k mt2 overexpression rescued PHBC formation defects induced by overexpression of a Vegfc ligand trap (sflt4 overexpression). Injection of mt2 mRNA resulted in normal PHBC development (d). Depletion of Vegfc through injection of sflt4 mRNA led to a failure in PHBC formation in 68 % of the embryos (e). Co-injection of both mt2 and sflt4 mRNA rescued PHBC formation and left only 23 % of embryos showing no PHBC (quantifications of different experiments shown in k). g–i, l Overexpression of mt2 mRNA in embryos with a genetic null mutation in the vegfc gene (vegfc hu6410) could not rescue the PHBC phenotype. Embryos were scored for their phenotype and subsequently genotyped for the vegfc mutation (quantifications of different experiments shown in l). The analysis was performed using Tg(kdrl:EGFP)s843 (a–f) and Tg(fli1a:EGFP)y1 (g–i) embryos. j–k Quantifications of the phenotypes observed after injection of indicated reagents: Black bars label percent of embryos with the PHBC formed, white bars label percent of embryos lacking the PHBC. Statistical significance was calculated with the Chi-square test, n = 228 (j), n = 277 (k), n = 124 (l), **P < 0.01; ***P < 0.001; n.s., not significant. m–p Schematics representing the proposed mechanisms of angiogenesis regulation in the experiments shown in a–l

Full size image

Mt2 regulates transcript levels of vegfc

Given the results above, we used qPCR to analyze vegfc transcript levels in mt2 morphant, MZmt2 mu290 mutant and mt2-overexpressing embryos. qPCR analysis revealed a 20 % decrease in vegfc RNA in mt2 morphants and a 31 % decrease in MZmt2 mu290 mutant embryos (Fig. 4a). mt2 overexpression on the other hand led to a 27 % increase in vegfc RNA levels in zebrafish embryos (Fig. 4a). To test whether vegfc transcripts are specifically affected, we analyzed further genes in mt2-deficient and mt2-overexpressing embryos. We chose fli1a as an EC-specific gene and myod1, a muscle-specific marker to represent other tissues [62]. We observed no significant changes in either fli1a or myod1 transcript levels, irrespective of the mt2 expression level. In contrast the significant changes in vegfc transcripts correlated with the changes in mt2 expression as vegfc levels were decreased in mt2-deficient and increased in mt2-overexpressing cells (Fig S6). The analysis of vegfc RNA transcript levels via in situ hybridization showed similar results in some domains of vegfc expression (Fig. 4b–d). mt2 morphants show reduced vegfc staining, especially in the region, where the PHBCs develop (Fig. 4c, arrowheads). Interestingly, the increase in vegfc RNA expression in mt2-injected embryos was also confined to specific domains, including the region of PHBC development (Fig. 4b–d, arrowheads), but not ubiquitously distributed (Fig. 4d). Therefore, our results show that Mt2 is required for regulating vegfc, e.g., during PHBC formation, but it is not sufficient to induce general vegfc expression ectopically. This can be further substantiated when comparing the WT expression patterns of vegfc and mt2, showing that some domains of vegfc expression are in the same region as mt2 expression, while there are also vegfc expression domains in areas not expressing high amounts of mt2 (Fig. S6). We claim that the regulation of vegfc via Mt2 is specifically confined to specific vascular niches, such as the region of PHBC formation.

Fig. 4
figure 4

Mt2 regulates transcript levels of vegfc. a qPCR analysis of vegfc transcript levels (a) in mt2 morphants, MZmt2 mu290 mutants and mt2 mRNA-injected embryos compared to Ctr embryos. Both morphants and mutants showed a significant decrease in vegfc transcript levels, while overexpression of mt2 led to an increase in vegfc transcripts. n = 3, *P < 0.05; **P < 0.01; error bars represent SEM; b–d in situ hybridization for vegfc in hpf embryos injected with mt2 MO or with mt2 mRNA showed a reduction in vegfc expression upon knockdown of mt2 (c) and an increase after mt2 mRNA injection (d)

Full size image

Other metallothioneins cannot regulate vegfc expression

To get more mechanistic insight how Mt2 could regulate vegfc expression, we questioned whether vegfc expression regulation could be a consequence of a cellular stress and hence would require the detoxifying features characteristic to all Metallothioneins (Mts). Therefore, we performed knockdown and overexpression experiments using another Metallothionein family member, metallothionein-B-like (mtbl; Fig. 5). To analyze mtbl-deficient embryos, we used again both translation and splice blocking MOs for our analysis and validated the functionality of the spbMO using RT-PCR (Fig. 5c). Even though mtbl deficiency led to defective development of the CCVs and Ses (Fig. S7), mtbl morphants showed normal PHBC development (Fig. 5b), indicating that during normal embryonic development Mt2 is specifically required for regulating vegfc expression. We next analyzed whether, as shown for Mt2, excess amounts of ectopic Mtbl could compensate for Vegfc ligand depletion by the ligand trap sflt4. While injection of sflt4 mRNA again provoked defective PHBC development (Fig. 5e), co-injection with mtbl mRNA did not rescue this phenotype (Fig. 5f, g). Furthermore, vegfc transcripts were not significantly changed upon knockdown of mtbl (Fig. 5h). These results indicate that the regulation of vegfc transcription by Mt2 is not based on its Metallothionein characteristics and therefore not part of a cellular stress response, but rather represents an additional specific function of Mt2.

Fig. 5
figure 5

The mt2 paralogue metallothionein-B-like (mtbl) does not regulate PHBC formation. mtbl MO-mediated deficiency does not affect PHBC formation. Tg(kdrl:EGFP)s843 embryos showed normal PHBC development at 24 hpf after injection of Ctr MO (a) or mtbl translation blocking MO (b). c Analysis of mtbl splicing efficiency in embryos injected with Ctr MO or mtbl spbMO. RT-PCR analysis showed a 327 bp amplicon in embryos injected with mtbl spbMO, while functional splicing led to a 169 bp amplicon in Ctr MO-injected embryos. mtbl overexpression failed to rescue PHBC formation defects induced by overexpression of a Vegfc ligand trap (sflt4 overexpression). Overexpression of mtbl via mRNA injection (d) resulted in normal development of the PHBC, while sflt4 mRNA injection resulted in PHBC formation failure (e). This phenotype could not be rescued through co-injection of mtbl mRNA (f). Quantification comparing the percentages of embryos lacking the PHBC, with n = 350, ***P < 0.001; n.s., not significant (g). The analysis was done with Tg(kdrl:EGFP)s843 embryos. h qPCR analysis of mtbl morphants showed no significant increase in the vegfc transcript. Error bars show SEM; n.s., not significant

Full size image

Discussion

In this study, we showed that Mt2 regulates developmental angiogenesis in zebrafish by regulating vegfc mRNA expression. Vegfc regulates EC migration as a chemoattractant, e.g., by guiding ECs in the PHBCs [9, 11], and indeed, we show that correct migration of the PHBCs was perturbed by deletion of mt2. Additionally, Vegfc regulates EC proliferation [13], which was also perturbed in mt2-deficient zebrafish embryos.

We analyzed the role of Mt2 in zebrafish angiogenesis using MO-mediated Mt2 ablation as well as by using TALENs to introduce mutations in the zebrafish mt2 gene. While we observed the same phenotypes in morphants as well as mutants, the phenotypes occurred at different frequencies between morphants and even between different hypothetical null mutants of mt2 (see Table 1). Multiple mechanisms have been discussed to explain differences between mutant and morphant phenotypes: reinitiation at a downstream AUG or at an alternative start codon, exon skipping or the upregulation of other compensatory genes [63]. From the mt2 sequence we can exclude reinitiation at a downstream AUG or exon skipping as potential mechanisms. We cannot predict whether there would be reinitiation at non-AUG start codons. However, we here provided a detailed analysis demonstrating that differences in mRNA stability, caused by NMD-mediated decay of the transcript, might account for the variability of the observed phenotypes. Our experiments show that a stronger efficiency of NMD led to a weaker penetrance of the phenotype, which might indicate transcript-level-based regulation of compensatory mechanisms in the embryo.

Additionally, even vegfc null mutants do not show full penetrance in failing to form the PHBC (Table 1, supplementary material [11]); therefore, embryos with a reduction in vegfc expression through mt2 deficiency are not likely to display higher phenotypic frequencies.

In our study we identified a role for Mt2 in regulating angiogenesis upstream of transcriptional regulation of vegfc expression.

While in the zebrafish only two mt genes exist, in mammals there are at least four different gene families with differentially expressed isoforms [64]. Analysis of the amino acid sequence via UniProt revealed highest identity of the zebrafish Mt2 to the human and mouse MT1, closely followed by the human and mouse MT2. Mammalian Mt1 and Mt2 are supposedly very similar in their function [26] and have previously been implicated to be involved in angiogenic processes. The MZmt2 zebrafish knockout led to impaired development of major vessels, such as the PHBCs, the CCVs and the Ses and MZmt2-deficient embryos died during larval stages. The murine Mt1/2 double knockout in contrast was viable [36] and only displayed angiogenesis defects when challenged, e.g., by cortical freeze injury or femoral artery ligation [38–40]. As in zebrafish maternal message was capable of compensating mt2 deficiency during embryonic angiogenesis, most likely in mammals other MT family might be able to compensate Mt1/2 deficiency during development. However, a link to angiogenesis has also been established for human Mts in vitro [65].

We demonstrated that Mt2 but not Mtbl regulates angiogenesis upstream of vegfc transcription. Mt family members are involved in regulating a large number of developmental processes, including cell survival, cell proliferation, migration, scavenging of reactive oxygen species, and modulating the immune response. Most of these capabilities have been attributed to the metal-binding capabilities, resulting, e.g., in removal of cofactor ions such as zinc [26, 30, 66]. The zebrafish Mtbl is capable of fulfilling these MT family member functions, but does not rescue PHBC development in mt2 morphants or vegfc ligand reduced embryos. We present here the first evidence for an additional role of zebrafish Mt2 in regulating vegfc expression independent of Mt function. Interestingly, upregulation of different human MT isoforms was observed comparing the responses to physiological or hypoxic conditions [65]. This could be taken as an indication for differential regulatory functions of some human MT family proteins, independent of the functions common to all MTs.

We analyzed whether other transcript levels were regulated by zebrafish Mt2 in addition to vegfc. Neither the Vegfc regulator ccbe1 expression, nor the Pdgf/Vegf family member c-fos-induced growth factor (figf) expression was altered. In contrast, vegfa expression seemed also regulated downstream of MT2 (data not shown). Reduced Vegfa RNA [40] and VEGFA protein levels [38] were reported in Mt1/2-deficient mice. While changes in Vegfc expression explained the PHBC and CCV phenotypes, reduction in Vegfa expression would account for the failures in Se formation, as deficiency in either Vegfa or its receptor Kdrl result in severe Se phenotypes [9, 67].

In sum, we have identified a novel role of MT2 in regulating angiogenesis by regulating vegfc transcription, which might be conserved in mammals.

We for the first time show that this regulatory role is specific to zebrafish Mt2 and represents a novel, non-canonical function of MT2, most likely not attributed to metal-binding capabilities of MT proteins.