ABSTRACT
The family of deleted-in-liver-cancer (dlc) genes encodes RhoGTPases and plays pivotal roles in cardiovascular development, but animal models for studying their functions are sparse due to early embryonic lethality. Gain and loss of function of dlc1 and dlc3 severely altered the growth of intersegmental vessels in the trunk of zebrafish embryos. Additionally, overexpression of dlc1 affected the growth of the common cardinal veins, but could rescue the arrest of angiogenesis induced by Vegfr2 inhibition, placing dlc1 downstream of kdrl signaling. Loss of dlc1 negatively affected the lumenization of the first aortic arch arteries and the lateral dorsal aortae. dlc1 mutants displayed a full obstruction in the early outflow tract during cardiac morphogenesis, which models to alterations in DLC1 detected in congenital heart defects in human patients. This study provides a functional in vivo characterization of dlc1 and dlc3 during vertebrate embryogenesis and places dlc1 as a key gene to control vascular development.
INTRODUCTION
The cardiovascular system is one of the earliest systems formed during embryonic development. Its proper formation requires a multitude of orchestrated processes, including cell proliferation, cell fate determination and migration. Many diseases develop through perturbation of these key physiological processes leading to systemic consequences. Disruptions in organ formation during embryonic development can result in congenital conditions. A well-known example is congenital heart defects (CHD), which can lead to disability or severely impaired organ function and death. Thus, studying the genetic of embryonic development is necessary to understand the mechanisms underlying these pathological conditions, thereby facilitating the development of prenatal screening tests and treatments.
The zebrafish model is suitable for studying many pathological conditions due to their conserved biology, availability of transgenic lines and optical transparency. Moreover, many genes that are embryonic lethal in mammals can be studied during early development in zebrafish larvae. One such gene, deleted in liver cancer 1 (dlc1), has been shown to be a bona fide tumor suppressor gene and is frequently downregulated or lost in many human cancers1-3. The three DLC proteins, DLC1, DLC2 and DLC3, belong to the family of RhoGAP proteins and preferentially target RHOA1 and to a lesser degree, CDC421,4. There is a high degree of similarity between all three family members, each possessing an N-terminal sterile alpha motif (SAM), an enzymatically active GAP domain and a C-terminal steriodogenic acute regulatory protein (StAR)-related lipid transfer (START) domain5. The function of DLC1 as a tumor suppressor is associated with its RhoGAP activity, leading to the inactivation of RHOA after concomitant binding to tensins4, which subsequently affects the formation of stress fibers and focal adhesions 6. Cytoskeletal rearrangements are key requirements in directing endothelial cell migration during blood vessel growth. Vasculogenesis and angiogenesis are thereby tightly regulated processes that depend on the interplay between numerous signaling pathways. Besides the well-known major VEGF/VEGFR signaling pathway7, other guidance factors, such as Semaphorins, Plexins and Neuropilins, also play a crucial role in vascular development and can additionally function as enhancing co-factors in the VEGF/VEGFR signaling axis7,8.
However, not much is known about the role of dlc1 during organogenesis because its homozygous deletion is embryonic lethal in Drosophila, mouse and avian models9-14. Dlc1null mouse embryos showed multiple anatomical changes in the developing vasculature and heart, which were analyzed postmortem due to intrauterine lethality 9. A recent study also linked the occurrence of a rare variant of DLC1 to human CHDs15. To further study the role of these genes in development, a Dlc2-KO mouse model has been established but no phenotype was detected16, whereas there is no current animal model for Dlc316.
In this study, we performed the first in vivo characterization of the functional role of dlc1 and dlc3 (stard8) during embryonic and larval development using the advantages of the zebrafish model. We initially show that both genes are highly expressed in the vascular lineage. Gain and loss-of-function experiments resulted in abnormal development of the vascular network, suggesting an important role for dlc1 and dlc3 in orchestrating endothelial cell migration during angiogenic sprouting in vivo. We show that dlc1 acts directly downstream of the kdrl signaling axis in vivo, which was further confirmed by observing a loss in the integrity of the endothelial cell sheet and an uneven leading edge during common cardinal vein growth, a phenotype reminiscent of dampened Sema3/PlxnD1 signaling and RhoA inactivation 17. Finally, we used CRISPR/Cas9-mediated genome editing to establish dlc1 and dlc3 mutants. dlc1null embryos showed defects in the lumenization of the first aortic arch arteries, in the formation of the lateral dorsal aortae and the development of the immature outflow tract in the embryonic heart. These observations are concordant with a study that recently linked DLC1 mutations to human CHDs, therefore reinforcing our zebrafish dlc1 mutant as an in vivo model to study the etiology of human CHDs.
RESULTS
dlc1 and dlc3 are expressed in the lateral plate mesoderm (LPM) and in endothelial cells
Currently, only mouse and chicken models have been established to analyze early dlc expression and function during embryonic development9,10. To investigate the spatial and temporal distribution of dlc1, dlc2 (stard13b) and dlc3 during zebrafish development, we performed in situ hybridization at different embryonic stages. Specific expression of all three genes was detected from 6 somites (ss) onwards. dlc1 expression was first detected in the posterior LPM (PLPM) and in the head of the embryo (Fig. 1A, Supplemental Fig. 1A). At 24 hours post fertilization (hpf), dlc1 expression was broadly distributed throughout embryonic tissues and in the vascular system. This included the mid cerebral vein (MCeV) in the head (Fig. 1B, arrowhead) and the aortic and venous endothelium in the trunk (Fig. 1B, C). dlc3 was expressed in the anterior LPM and the PLPM (Fig. 1D solid arrows, Supplemental Fig. 1B), intersomitic regions, and the tail bud at 6 ss (Fig. 1D, Supplemental Fig. 1C). At 24 hpf, dlc3 expression strongly resembled kdrl expression, marking the axial trunk vasculature and the first sprouting intersegmental vessels (ISVs) (Fig. 1E). Due to the absence of dlc2 expression in the LPM and vascular tissues (Supplemental Fig. 1D-G), the physiological role of dlc2 was not further investigated. The endothelial expression of dlc1 and dlc3 was confirmed in sorted kdr1:eGFP+ cells by qPCR analysis. Both genes were enriched in endothelial cells (ECs) at 26 hpf compared to eGFP-negative sorted cells (Fig. 1F, G). Altogether, dlc1 and dlc3, but not dlc2, are expressed in putative angioblasts in the LPM and in ECs at later stages of development.
Overexpression of dlc1 and dlc3 affects angiogenesis
Overexpression of dlc1 and dlc3 (by injection of full-length mRNA at the 1-cell stage) induced a high lethality by 8 hpf ranging from 75% to 95% of injected embryos. The surviving embryos showed severe ISV defects, which manifested in hyperbranching and misguided vessels (Fig. 2A-C). To overcome this problem, we established a stable transgenic line, hsp70:dlc1, with dlc1 under the control of the heat-shock promoter hsp70 (Fig. 2D and Supplemental Fig. 2). Analysis of ISVs at 2 days post fertilization (dpf) following heat shock at 13ss revealed prominent patterning defects. Thus, temporal restriction of dlc1 overexpression (Supplemental Fig. 2) recapitulated the phenotype observed in mRNA overexpression experiments (Fig. 2D-H), while greatly reducing lethality. Besides the ISV phenotype, heat-shocked embryos also displayed alterations in the growth of the bilateral common cardinal veins (CCVs). The CCV migrates as a collective EC sheet that ultimately connects the venous system and the endocardium of the cardiac inflow tract (IFT). In heat-shocked embryos, the leading front and edges of the migrating CCV cell sheet often appeared uneven and/or small holes were observed in the EC sheet (Fig. 2I, J). These two CCV phenotypes appear to be dependent on the RhoGAP domain of dlc1, as truncated dlc1 mRNAs containing only the GAP and START domains caused a similar phenotype (Fig. 2K, K’), while other truncated constructs of dlc1 did not induce this phenotype (Supplemental Table 1). When we treated heat-shocked embryos with a Rho Activator, we partially restored the growth of the CCVs (Fig. 2L-O), demonstrating that dlc1 alters the CCV growth pattern in a GAP-dependent manner.
dlc1 affects CCV growth via the Sema3d/PlxnD1 and the Nrp1/RhoA axes
The CCV growth pattern induced by dlc1 overexpression (Fig. 3A) is similar to the phenotype observed in sema3d morphants and mutants 17. A recent zebrafish study showed that Sema3d regulates collective EC migration in the CCVs by two distinct mechanisms. The guidance and straight outgrowth of the CCV is dependent on Sema3d signaling through PlexinD1, whereas the cytoskeletal integrity of the EC sheet is maintained by Sema3d signaling through Nrp1. The latter regulates actin network organization through RhoA, which stabilizes the EC sheet morphology17. To investigate whether Dlc1 mediates its effect on CCV development via Sema3d and Nrp1, zebrafish sema3d and rat Nrp1 mRNA17 were injected in one-cell stage hsp70:dlc1 embryos, that were then heat-shocked at 13ss. While sema3d mRNA rescued the uneven growing edge of the CCV, holes in the migrating EC sheet were still present (Fig. 3B, C, F). Conversely, Nrp1 overexpression rescued the EC sheet integrity but not the uneven CCV growth pattern (Fig. 3D, E, G). These results suggest that Dlc1 controls EC migration via the Sema3d/PlxnD1 and EC sheet integrity via the Nrp1/RhoA axis during CCV growth.
dlc1 acts downstream of kdrl signaling
Neuropilins function as co-receptors to enhance the binding activity of Vegfa and Vegfc to Flt1, Kdrl and Flt4, respectively7,8. Since we showed that Dlc1 interacts with the Sema3d/PlxnD1 and the Nrp1/RhoA axes, we next examined its role in Kdrl signaling. To investigate this, hsp70:dlc1 embryos were divided into two groups and only one group was heat-shocked at 13ss prior to treatment with a high dose (1.5µM) of the VEGFR2 inhibitor SU5416 for 5 hours (Fig. 3H). At 3dpf, the non-heat-shocked SU5416-treated embryos showed pronounced vascular defects throughout the whole embryo (Fig. 3I). All vasculogenic and angiogenic sites were affected, which led to a complete absence of blood circulation. In 20% of these embryos, ISVs were able to partially regrow, however, the large majority failed to extend beyond the horizontal myoseptum (Fig. 3I, asterisks). In sharp contrast, temporal-controlled overexpression of dlc1 restored ISV growth in 83% of examined embryos (Fig. 3J, K). Overexpression of Dlc1 not only rescued the initial sprouting defect, but also prevented the ISV extension defects that were observed in control SU4516-treated embryos. In line with our initial gain-of-function phenotype, we still observed some additional hyper-sprouting (Fig. 3J, arrowheads). This finding suggests that Dlc1 acts downstream of the Kdrl signaling axis and plays a crucial role during vascular development.
dlc1 activity is modulated by phosphorylation
In human epithelial cells, DLC1 activity is regulated by a phosphorylation-dephosphorylation cycle downstream of EGFR signaling18. Therefore, we investigated whether inhibiting protein phosphatase 2 A (PP2A) dephosphorylation of Dlc1 could suppress its activity in vivo (Fig. 3L). Treatment with okadaic acid (OA), a PP2A inhibitor, caused only minimal effects on vascular development (Fig. 3M). Hsp70:dlc1 embryos were heat-shocked at 13ss and subsequently treated with OA overnight. OA treatment rescued the hyper-sprouting ISV phenotype induced by dlc1 gain-of-function (Fig. 3N-P), suggesting a conserved regulation of dlc1 in zebrafish in vivo, compared to what was previously observed in vitro for human cells.
The dlc1 mutation impairs the development of the vascular system
By using CRIPSR/Cas9, we generated two different dlc1 mutant alleles: a -3/+2 bp indel mutation and a -4 bp deletion (Supplemental Fig. 3A). Both mutations resulted in premature stop codons after 174 and 173 translated amino acids, respectively; resulting in severely truncated proteins that lacked most of the FAT domain, and the GAP and START domains. Based on our knowledge of the Dlc family of proteins, we considered these as null alleles. The mutant phenotypes were analyzed on a kdrl:eGFP transgenic background from 24hpf until 2.5dpf. Matings of dlc1+/- adults always induced the same two phenotypes in mutant embryos with a Mendelian ratio (20-25%). However, the distribution of these two phenotypes varied considerably and was independent of the mating pairs. The less severe phenotype consisted of the impaired formation of the lateral dorsal aortae (LDAs) and aortic arches 1 (AA1s) (12%-25% of mutant embryos per clutch), which in most cases did not result in embryonic lethality (Fig. 4). In some cases, the AA1s were functionally impaired, resulting either in stenosis (Fig. 4A, B arrow) or failure to connect to the LDA (Fig. 4B right side, arrowhead). In most cases, the bilateral LDA either failed to correctly merge into one vessel (Fig. 4E, F, F’) or failed to complete proper lumenization (Fig. 4G, G’). This was phenocopied by dlc1-MO injection (Supplemental Fig. 3C,D; Fig. 4H, I, I’). Erythrocytes could often only circulate through one LDA, possibly compensating for the stenosis in the other LDA (Fig. 4I arrows). Dlc1 mutant embryos showed stenotic phenotypes throughout the LDA, including the looped lateral region, where the LDA merge into the AA1 (Fig. 4C, D). Surprisingly, dlc3 mutants (Supplemental Fig. 3B) showed only transient mild edema and decelerated blood flow, especially in the head, but these phenotypes resolved completely by 2.5dpf (data not shown). Overall, there were no specific changes to the vasculature in dlc3 mutants that could explain this transient phenotype. Only in rare cases did dlc3 mutant embryos display a comparable LDA phenotype to dlc1 mutants (Supplemental Fig. 4A, B). Surprisingly, the dlc1 or dlc3 mutants did not show any modified ISV growth, whereas both MOs considerably affected ISV growth in the trunk (Supplemental Fig. 3C-E, Supplemental Fig. 5). In vivo time-lapse microscopy of dlc1 and dlc3 morphants showed growing ISV stalling at the horizontal myoseptum, as they failed to properly elongate dorsally (Supplemental Fig. 5A-F and Supplemental Movies 1-3). To determine if the dlc1 morphants showed an alteration in tip and/or stalk cell identity, we purified ECs from their trunks and tails (Supplemental Fig. 5G). There were no changes in the expression of genes involved in tip or stalk cell identity between dlc1 morphants and control embryos (Supplemental Fig. 5H). This indicated that dlc1 likely controls the directed EC migration of sprouting ISVs but does not affect the identity of tip and stalk cells in the angiogenic sprout. Taken together this data suggests that dlc1 controls vasculogenesis in the LDA and AA1s and EC migration during angiogenic sprouting in zebrafish.
dlc1 mutant embryos as a model of congenital heart defects (CHDs)
A second phenotype was found in 0%-13% of embryos obtained from dlc1+/- incrosses, which did not survive past 5dpf (Fig. 5). This severe phenotype was observed as early as 28-30 hpf due to the lack of blood circulation and the development of pericardial edemas (Fig. 5A, B). Microangiographies confirmed that no blood or liquid could leave the heart and enter circulation in these mutants (Fig. 5C, D). The absence of blood circulation was mainly due to a defective immature outflow tract (OFT; 80%), resulting in backflow of blood from the ventricle to the atrium (Fig. 5B, Supplemental Movies 4 and 5), preventing blood from leaving the heart. Furthermore, the defective OFT formation resulted in enlargement and dilation of the heart but did not affect atrium and ventricle specification (Fig. 5E, F). qPCR analysis on pools of isolated mutant hearts revealed significantly higher expression of plxnd1 and a trend towards elevated kdrl expression (not significant) compared to control hearts, whereas expression of the cardiomyocyte marker myl7 was similar (Fig. 5G). Due to the cardiac enlargement, the observed transcriptional changes could either be attributed to more endocardial cells or to an overall higher expression of plxnd1 and kdrl. Moreover, mutant hearts displayed a change in bmp4 expression that normally marks the OFT, the atrio-ventricular (AV) canal, and to a much lesser extent, the IFT (Fig. 5H, M). In dlc1 mutants, bmp4 staining was highly present in the OFT and the IFT, but only remotely in the AV canal (Fig. 5I, M), even though no changes in overall bmp4 expression could be detected by qPCR (Fig. 5J). This rather suggests a change in bmp4 distribution within the heart than a change at the transcriptional level (Fig. 5M). dlc1 morphants phenocopied the abnormal bmp4 distribution in the heart (Fig. 5K, L), but not the obstructed OFT phenotype. This indicates that the changes in bmp4 distribution are probably not a result of the morphological alterations of the heart due to the lack of blood flow, but rather depended on the proper migration of bmp4-expressing cells within the heart of dlc1 mutants. Decreased dlc3 levels did not cause any cardiac phenotypes, but dlc1/dlc3 double mutants exhibited higher occurrence of the cardiac phenotype. Concordant with this hypothesis, double mutant fish never survived to adulthood. Moreover, injection of low doses of the dlc3-MO (4 ng) into dlc1 mutants favored the appearance of the severe cardiac phenotype by 2-fold (Fig. 5N). These results point towards a certain degree of redundancy between dlc1 and dlc3 during cardiovascular development, although dlc1 seems to be the major player in this process. Overall this suggests that dlc1 is indispensable for initial early OFT formation and proper cardiogenesis in the zebrafish embryo.
DISCUSSION
Our work represents the first detailed in vivo characterization of dlc1 and dlc3 in a vertebrate model that can survive several days despite severe cardiovascular phenotypes. While many excellent cell culture studies provided topical and cancer-related pathophysiological data on Dlc1’s functions, they ultimately lack the anatomical context of a whole animal. This makes it difficult to fully understand the physiological role of the Dlc1 protein, in particular when neighboring tissues and cell complexes are missing to provide a contextual scaffold. This in turn may influence the occurrence and consequential manifestation of multifaceted and/or spatially distinct phenotypes.
We showed that dlc1 and dlc3 are initially expressed in the LPM and become enriched in ECs during zebrafish development. The overexpression of DLC1 in human cells results in enhanced velocity of migrating cells and a loss of their directionality due to dephosphorylation of focal adhesion proteins19. We found that those alterations of cell migration translated into differential phenotypes at two distinct anatomical sites in the zebrafish. While gain-of-function greatly affected angiogenic sprouting in the zebrafish trunk, causing mis- and hyperbranching ISV growth, we also observed an uneven leading edge of ECs, as well as a disrupted EC layer during CCV growth. The specificity of the trunk phenotype was confirmed by rescuing the aberrant ISV sprouting phenotype through inhibiting the final activation step of Dlc1 using a specific PP2A inhibitor. This mechanism was previously described by Ravi and colleagues18 in epithelial cell culture studies and appears to be globally conserved in vivo.
In zebrafish, the loss of directional vessel growth is associated with deficient Sema3a/PlxnD1 signaling during trunk angiogenesis20 and this pathway directly interacts with the Vegfa/Kdrl/Nrp1 axis21,22. We found that dlc1 overexpression can prevent the arrest of ISV growth induced by Vegfr2 pathway inhibition, suggesting that Dlc1 acts directly downstream of the Kdrl axis in vivo. In addition to ISV hyperbranching following dlc1 overexpression, we observed altered migration patterns of ECs during CCV growth. This phenotype has been described in sema3d-deficient zebrafish embryos17, as Sema3d regulates CCV growth in a plxnd1-dependent manner17. Furthermore, Sema3d undergoes EC-specific autocrine signaling through Nrp1 to modulate actin network organization via RhoA and ROCK17, which is important to preserve the EC sheet integrity. In line with this study, we showed that sema3d mRNA injection exclusively rescued the uneven growth of the front and edges of the CCV, while rat Nrp1 mRNA solely rescued the integrity of the endothelial layer in temporally controlled dlc1 overexpression experiments. During angiogenesis, attractive and repulsive signaling cues need to be translated into cytoskeletal rearrangements. NRP1-knockdown in vitro disrupts F-actin network organization in HUVECs23, suggesting a direct interaction of NRP1 and RHOA-mediated actin cytoskeletal rearrangements. Consequently, overexpression of truncated dlc1 mRNA encoding the GAP and START domains was sufficient to induce a similar, but milder CCV phenotype, compared to full-length dlc1 overexpression. This indicates that the disruption of the straight growing edge and lesions in the EC sheet are a result of Dlc1 GAP-dependent inactivation of RhoA, which could be partially rescued by a RhoA activator. These results support the role of Dlc1 in orchestrating EC migration via the Sema3d/Plxnd1, as well as the Nrp1/Rhoa axis during CCV growth in zebrafish. We propose that dlc RhoGAPs may represent a functional link in translating pro-(VEGF) and anti-angiogenic (Sema3/Nrp1) guidance cues into cytoskeletal rearrangements in vivo. However, it is still unclear how dlc1 precisely affects single components of the actin cytoskeletal machinery in vivo during angiogenesis, and how it interacts with the Sema3/Nrp1 axis.
Our results confirm that Dlc1 activation requires dephosphorylation in vivo, which is consistent with in vitro results that showed ERK-mediated phosphorylation-dephosphorylation cycles are essential for DLC1 to become fully functional18. ERK inactivation of FAK triggers the release of the activated FAK binding complex. This enables PP2A-mediated dephosphorylation of the Ser-(P)308 and Thr-(P)301 residues in DLC1, leading to its activation18. FAK represents an important component in this signaling cascade and it can also be inactivated following dephosphorylation by Sema3E in vitro, thereby modulating anti-angiogenic signaling24. We propose that Dlc1 acts downstream of the Vegfa/Kdrl/Nrp1 signaling cascade (Fig. 6), while undergoing the phosphorylation-dephosphorylation cycle and subsequently inactivating RhoA, which was previously described in vitro18 (Fig. 6). Moreover, our data also suggests that dlc1 acts downstream of the Sema3/PlxnD1 axis (Fig. 6), thereby providing a functional link between guidance cues and cytoskeletal rearrangements to control directed EC migration in vivo.
Proper ISV growth was severely impaired in dlc1 and dlc3 morphants and diametrically opposite to the gain-of-function phenotype. Time-lapse analysis showed that dlc1-deficient embryos displayed abnormal ISV growth including stalled sprouts, extension defects, and in severe cases, the formation of vascular plexi instead of elongated vessels. As tip and stalk cell identities appeared normal in dlc1 morphants, an explanation for this phenotype could be the aberrant function of repellent guidance cues. A stalling phenotype for growing ISVs has also been described in global sema3d-KO experiments17. However, further investigation is needed to identify which Class 3 Semaphorin is involved in conjunction with PlxnD1/Nrp1 in dlc morphants. While ISV development was impaired in dlc1 morphants but not in mutants, we believe several factors could potentially be responsible for this variance. One explanation could be a partial genetic compensation by other dlc family members or even other genes. Another reason could be the fact that proper angiogenesis and lumenization of vessels depend on sufficient blood flow and shear forces, both of which are impaired in several circumstances of the observed dlc1 mutant phenotypes. Those alterations might mask the emergence and penetrance of an ISV phenotype in dlc1null embryos.
Besides its role in angiogenesis, knockout of dlc1 impaired proper AA1/LDA patterning as well as OFT development. AA1/LDA stenosis could be linked to a defective epithelial-to-mesenchymal transition (EMT), a process that converts immobilized ECs into highly motile cells that are able to move through the extracellular matrix. EMT is crucial in the development of most adult tissues and organs25, especially the heart26. It involves tight regulation of differential expression and/or activation of RhoGTPases27, among others. Effects on proper EMT has been associated with ectopic DLC1 expression in nasopharyngeal carcinoma cells 28, it is plausible that a similar mechanism could contribute to proper AA1/LDA patterning and OFT development in the zebrafish. Defects in OFT development are the cause in 30% of congenital heart defects (CHD) in humans29. Similarly, we showed that the formation of the immature cardiac OFT is defective in dlc1 and dlc1/dlc3 double zebrafish mutants resulting in embryonic lethality around 5dpf. Cardiac morphogenesis has been shown to be dependent on RhoGTPase signaling and proper EMT in chick and mice30,31. Furthermore, mutations in PLXND1 are associated with truncus arteriosus in humans, which was recapitulated in PlxnD1null mice that die shortly after birth22,32. These mice exhibit defective aortic arch arteries, persistent truncus arteriosus and septal defects33. Additionally, this cardiac phenotype is dependent on the GAP domain of PlxnD1 and similar defects have been observed in Sema3C and Nrp1null mice34-35,36. Plexins have previously been shown to bind RhoGTPases37, it is possible that RhoA and its regulators, such as Dlc1, could be present nearby to transduce signals to the actin cytoskeleton to direct EC migration and EMT.
Isolated hearts from dlc1null zebrafish mutants showed increased plxnd1 and kdrl expression compared to WT hearts, whereas myl7 remained largely unchanged. The overall expression level of bmp4 was unaltered in mutant compared to WT hearts, however there was a striking change in its expression pattern. The alterations in the bmp4 expression pattern in dlc1null embryos could point towards an impaired migration of bmp4-expressing cells in the dlc1null zebrafish heart. It is unlikely that bmp4 distribution was affected by the abnormal heart shape as dlc1 morphants displayed the same altered bmp4 expression pattern, without showing the OFT defects seen in mutants. It is very likely that the observed AA1 phenotypes and the disrupted OFT development are interdependent, as a recent study showed that ECs from the aortic arches, in particular the AA1, migrate and contribute to OFT formation38. However, attempts to image OFT and AA1/LDA formation using time-lapse microscopy were prevented by the delayed folding speed of the eGFP fluorophore in our transgenic lines and the depth of the immature OFT within the tissue at 20-22 hpf. Future studies of the earliest events in OFT formation would benefit from the generation of transgenic dlc mutants with fast superfolding enhanced fluorophores39 and the use of specialized microscopy equipment, such as the light sheet fluorescence microscope or SPIM.
The role of dlc3 during cardiovascular development remains elusive as no cardiac phenotypes were observed in mutants. However, our data suggest a partial redundancy with dlc1, while dlc2 seemed not involved in this process. Future experiments should address potential compensatory mechanisms between the three family members. It is noteworthy that the zebrafish dlc1 mutant displays a causative phenotype of CHD, which is congruent with the enrichment of rare DLC1 variants in human CHD patients15. As genetic models of sporadic CHD are rare and the Dlc1 mutation causes early intrauterine embryonic lethality in the mouse model9, the zebrafish dlc1 mutant will provide a valuable tool to study the underlying mechanisms involved in early OFT development.
MATERIALS AND METHODS
Zebrafish husbandry and maintenance
Zebrafish procedures were approved by the Geneva Veterinary Office and were performed according to the guidelines from Directive 2010/63/EU of the European Parliament. All experiments were performed using the AB* wildtype strain and the following transgenic animals: Tg(kdr1:eGFP)s843 40, Tg(gata1:dsRed)sd2 41, Tg(fli1a:lifeact-GFP)mu240 17 and Tg(myl7:dsRed)s879 42. Embryos were dechorionated at 24 hpf and incubated with 0.002% PTU to prevent pigment formation. Embryos and larvae were anaesthetized with 0.2 mg per mL of Tricaine methanesulfonate (MS-222) in E3 medium by immersion. For fixation, the embryos and larvae (older than 24 hpf) were first anaesthetized with Tricaine and subsequently fixed in 4% cold paraformaldehyde (PFA).
mRNA and Morpholino injections
mRNA was transcribed using the mMessage mMachine Kit SP6 (Ambion) according to the manufacturer protocol from a linearized pCS2 vector containing the open reading frame of the polymerase chain reaction amplified product. After transcription, the mRNA was purified by phenol-chloroform extraction and resuspended in RNAse-free water. 1 nL containing 100-200 pg of full length mRNA or the stated concentration of the respective Morpholinos (MO, purchased from Gene Tools) were injected into 1-4 cell stage embryos. dlc1-MO 5’-AGTCTTCACTCTGAAACGATATGGA-3’, dlc3-MO 5’-CTCTGCTGATGGTAACAGACACAGA-3’, standard control MO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′.
Generation of transgenic lines (Tol2) and mutant lines (CRISPR/Cas9)
For Tg(hsp70l:dlc1-p2a-TFP) fish generation, the sequence for the zebrafish dlc1 (without the STOP codon), was inserted in a Tol2 vector containing the hsp70l:xxxp2a-TFP sequence (kind gift from Prof. B. Martin, Stony Brooks University, NY) by In Fusion (Clontech) cloning. For transgenesis, 1-cell stage embryos were injected with 30 pg of the final Tol2 vector, along with 30 pg Tol2 mRNA.
For CRISPR/Cas9-mediated generation of dlc1 and dlc3 mutants, single-guide RNA was generated by annealing oligonucleotides at 95°C for 5 min, and then 22°C for 45 min. The samples were diluted 20-fold and ligated into the pDR274 plasmid (kind gift from Keith Joung; Addgene plasmid 42250), which was previously linearized with BsaI. Guide RNA was generated using the MEGAshortscript T7 kit (Ambion). 500 pg recombinant Cas9 protein (PNAbio) and 250 pg of the guide RNA were co-injected per embryo. The guide RNA sequences for dlc1 were: 5’-GGACCTTAGCGAGCAGCCAG-3’ and for dlc3: 5-CAGGCAACCGCTCCTCAACC-3’. Embryos were screened using primers that flanked the target region (Supplemental Table 2), and mutant alleles identified by sequencing.
Whole-mount in situ hybridization
Whole-mount in situ hybridization (WISH) was performed on 4% PFA-fixed embryos as previously described43. In brief, RNA probes were generated by linearization of TOPO-TA vectors (Invitrogen) containing the PCR-amplified cDNA sequence. Digoxigenin- or FITC-labeled antisense probes were synthesized using an RNA Labeling Kit (SP6/T7; Roche). The staining was revealed with NBT/BCIP or INT/BCIP substrate (Roche). Embryos were imaged in 100% glycerol, using an Olympus MVX10 microscope. Oligonucleotide primers used to amplify and clone cDNA for the production of WISH probes are listed in Supplemental Table 2.
Cryosections
Fixed embryos were subjected to WISH staining and washed afterwards in methanol, followed by PBS. Embryos were mounted in OCT and frozen at -20°C. 5 µm thick sections were obtained using a Cryostat, the consecutive sections were then transferred to Superfrost Microscope slides and stored at -20°C. Sections were defrosted and mounted in Glycerol shortly before imaging on a Zeiss Axioskop 2 plus.
Heat shock and chemical inhibitor treatments
Heat shock was performed at the 13-somite stage (ss) of embryonic development for 45 min in a 39°C water bath. Embryos were returned to 1xE3 (room temperature) and incubated at 28°C. Subsequent chemical inhibitor treatments were performed 0.5-1 hour after the heat shock and incubated for the stated times at 28°C in the dark. The chemicals were diluted in 1xE3 in the following concentrations: 1.5µM of the VEGFR2 inhibitor SU5416 (Selleckchem) and 15nM Okadaic acid (Enzo).
Confocal microscopy
Live (anaesthetized) embryos were embedded in 0.5% agarose supplied with 0.07 mg per mL Tricaine in E3 in a glass-bottom dish. Confocal imaging was performed using a Nikon inverted A1r spectral. For time-lapse experiments, embryos were imaged in a heated chamber at 28°C overnight. The larvae were kept anaesthetized for the duration of the experiment (several hours to overnight). Images were analyzed and the contrast and brightness equally adjusted using the FIJI (ImageJ) software.
Microangiography
3-4 nl of 2 mg per mL Rhodamine B isothiocyanate-Dextran (70kDa, Sigma) dissolved in PBS and 5 mM HEPES were injected into the sinus venosus of 30 hpf embryos. For this, the embryos were anesthetized with Tricaine, placed dorsally into an injection mold and fixed in place by applying 3% Methylcellulose. Embryos were immediately imaged after the injection (≤ 30 min).
FACS
Embryos were dissociated in 0.9x ice-cold PBS using 0.5mg per mL Liberase (Roche) then resuspended in 0.9x PBS/1% FCS. Dead cells were excluded by SYTOX-red (Life Technologies) staining. Cell sorting was performed using an Aria II (BD Biosciences). Sorted cells were collected in RLT buffer (Qiagen) and frozen at -80°C.
Heart isolation
Isolation of embryonic hearts was performed in general as previously described44 with slight adjustments. Batches of not more than 20 embryos were anaesthetized in Tricaine/E3 and washed three times with ice cold Leibovitz’ L15 medium (Invitrogen) supplemented with 10% FCS. The embryos were then mechanically dissociated (20-30x) in L15 medium/10% FCS using a syringe and 19G needle mounted to a ring stand and directly transferred to a petri dish containing the same medium. Intact isolated hearts were manually collected with a P20 pipette (5-10 hearts per pool) and immediately placed on ice. Identification of the hearts was achieved by a combination of DIC microscopy and fluorescently marked hearts in the transgenic lines kdrl:eGFP (endocardium) or myl7:dsRed (cardiomyocytes). After mechanical dissociation of the hearts in 0.9x PBS using a 27G needle, heart cells were recovered and resuspended in RLT buffer.
RNA extraction, cDNA synthesis and quantitative real-time PCR
RNA was extracted using the RNeasy Minikit (Qiagen) according to the manufacturer protocol and reverse transcribed into cDNA using the Superscript III kit (Invitrogen) or the qScript Kit (Quantabio). Quantitative real-time PCR (qPCR) was performed using the KAPA SYBR FAST Universal qPCR Kit (KAPA BIOSYSTEMS) in a QuantStudio3 system (Life Technology). All primers used are listed in Supplemental Table 2.
Statistical analysis and Reproducibility
Each experiment has been performed at least 3 times on different experimental days (biological replicates; N). Embryos were individually assessed on the day of the experiment and the data pooled for graphical and statistical analysis. The total and pooled numbers of embryos within a full set of experiments (N = 3) was defined as n. Specific n values are presented above bar graphs per condition where applicable. All data in graph bars represent the mean ± SD. Stacked bar graphs containing multiple groups display only the mean for clarity. Statistical significance of differences between 2 groups was determined using an unpaired, 2-tailed t test, while the significance of multiple groups was determined using a 1-way ANOVA with Dunnett’s post hoc correction. To analyze the independence of the occurrence of the dlc1 mutant phenotype after dlc3 MO injection, the Fisher’s exact test was used. P values of less than 0.05 were considered statistically significant. All statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software).
Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.
Author contributions
TL executed the experimental work. JYB designed the experiments, provided funding and supervised the research. TL and JYB jointly wrote the manuscript.
Conflict of interest
The authors declare no conflict of interest.
Acknowledgements
JYB was founded by the Swiss National Fund (#31003A_166515), and was endorsed by a Chair in Life Sciences from the Giorgi-Cavaglieri Foundation. We would like to thank Chantal Combepine and Corentin Pasche for excellent technical assistance, as well as Dr. W. Herzog and Dr. Lucia Du for critical reading of the manuscript.