Re-evaluating functional landscape of the cardiovascular system during development

The cardiovascular system facilitates body-wide distribution of oxygen, a vital process for development and survival of virtually all vertebrates. However, zebrafish, a vertebrate model organism, appears to form organs and survive mid-larval periods without the functional cardiovascular system. Despite such dispensability, it is the first organ to develop. Such enigma prompted us to hypothesize yet other cardiovascular functions that are important for developmental and/or physiological processes. Hence, systematic cellular ablations and functional perturbations are performed on zebrafish cardiovascular system to gain comprehensive and body-wide understanding of such functions and to elucidate underlying mechanisms. This approach identifies a set of organ-specific genes, each implicated for important functions. The study also unveils distinct cardiovascular mechanisms, each differentially regulating their expressions in organ-specific and oxygen-independent manners. Such mechanisms are mediated by organ-vessel interactions, circulation-dependent signals, and circulation-independent beating-heart-derived signals. Hence, a comprehensive and body-wide functional landscape of the cardiovascular system reported herein may provide a clue as to why it is the first organ to develop. Furthermore, the dataset herein could serve as a resource for the study of organ development and function. SUMMARY STATEMENT The body-wide landscape of the cardiovascular functions during development is reported. Such landscape may provide a clue as to why the cardiovascular system is the first organ to develop.


SUMMARY STATEMENT
The body-wide landscape of the cardiovascular functions during development is reported. Such landscape may provide a clue as to why the cardiovascular system is the first organ to develop.

INTRODUCTION
The cardiovascular system has evolved to facilitate oxygen transport throughout the body (Aaronson et al., 2014;Gabella, 1995). Availability of oxygen is required for development and function of virtually all organs. Oxygen deficiency, referred to as hypoxia, results in developmental and functional failure and/or damages of organs (Semenza, 2011(Semenza, , 2014Simon and Keith, 2008). Hence, the cardiovascular system is the first functional organ to develop.
The cardiovascular system is also essential for the delivery of humoral factors and immune cells to various parts of the body (Aaronson et al., 2014;Gabella, 1995). Hormones synthesized and secreted by endocrine organs enter into the circulation and reach to other distant organs, where they control organ growth and homeostasis. Immune cells exploit the vascular system to reach to distant tissues where timely immune responses are required. Such responses could facilitate the repair of tissue damages and/or the elimination of intruders and/or unwanted cells such as dead cells and cancer cells. On the other hand, persistence of such responses could result in chronic diseases.
Contrary to this perceived predominant importance of the canonical function, zebrafish forms various organs without the functional cardiovascular system (Field et al., 2003a;Field et al., 2003b).
Zebrafish is a vertebrate animal, thus with closed circulatory system composed of the heart and the extensive body-wide vascular network. Virtually all vertebrates fail to develop organs without the preceding formation of the cardiovascular system. This is presumably due to the fact that oxygen homeostasis established and maintained by the cardiovascular system is essential for the formation and growth of all functional organs. However, zebrafish appears to form organs even in the absence of the functional cardiovascular system (Field et al., 2003a;Field et al., 2003b). It is assumed that oxygen diffusion through the body wall is sufficient for the initial organogenesis (Field et al., 2003a;Field et al., 2003b). Despite such dispensability for oxygen homeostasis, the cardiovascular system develops first. This observation raises a possibility that the cardiovascular system plays other critical roles in regulating developmental and/or physiological processes. Such cardiovascular function may not be apparent by anatomical observations or the measurements of conventional physiological parameters.
Hence, we re-evaluated the functions of zebrafish cardiovascular system during organogenesis.
Specifically, we addressed two questions: 1) What else other than oxygen homeostasis is regulated by the cardiovascular system?; 2) How is it regulated? Experiments were designed to gain comprehensive and body-wide insights into these questions. This was achieved by selectively eliminating each functional aspect of the cardiovascular system in zebrafish larvae. Genetic and pharmacological manipulations were applied to zebrafish larvae to induce ablation(s) and/or functional perturbations of cardiomyocytes, cardiac contraction, the circulation (hemodynamic force), oxygen supply, and/or vECs (vessels)/hematopoietic cells. Molecular signatures induced by each of such manipulations were characterized by body-wide gene expression patterns.
We find the expression of many organ-specific genes are differentially influenced by each manipulation. Such genes include those which are implicated for their functional importance in metabolism, sterol homeostasis, sensory system development and function, and neural functions.
Surprisingly, body-wide hypoxia has very small, if any, effects on their expressions. Instead, other distinct cardiovascular mechanisms appear to regulate them. Such mechanisms are mediated by 1) local organ-vessel interactions, 2) circulation-dependent signals, and 3) circulation-independent but distantly-acting beating-heart-derived signals. Hence, these functions are acting more dominantly than the regulation of oxygen-homeostasis during organogenesis. Therefore, such functions could explain why the cardiovascular system develops first. These results also suggest such less-appreciated functions of the cardiovascular system are more important than previously perceived. The findings also provide a body-wide landscape depicting organ-specific gene expression patterns that are differentially regulated by the distinct cardiovascular functions and mechanisms. Such landscape may serve as a resource and a platform for the future in-depth analyses of organ development and function.

Characterization of "heartless"
To investigate the role of the heart during development, the zebrafish larvae lacking the heart (referred to as "heartless" here) was characterized. The "heartless" was generated by treating Tg(cmlc2:mcherry-NTR) zebrafish larvae with MTZ (Curado et al., 2007;Curado et al., 2008;Dickover et al., 2013) (Fig. 1A, see METHODS). The specific ablation of cardiac muscle, but not endocardium/endothelial cells, was confirmed by the lack of cardiac muscle fluorescent reporter expression (cmlc2:mcherry) (Fig. 1B), and the presence of endothelial reporter gene expression (fli1a:egfp) (Isogai et al., 2003; (Fig. 1B). The lack of circulation was also confirmed (Movies 1, 2). Blood cells were often stuck and accumulated within the heart and between the heart and liver, instead of circulating (Movies 1, 2).
The body-wide gene expression pattern of 4.5 dpf "heartless" was characterized by genome-wide transcriptome analysis. It was compared to two controls, Tg(cmlc2:mcherry-NTR) treated with the vehicle (DMSO) (NTR+DMSO) and wild type treated with MTZ (WT+MTZ). Differentially expressed genes are shown as Volcano plots (Fig. 1D). The sites of their differential expression were determined by whole-mount in situ hybridization (WISH) analyses (Figs. S1, S2). The results are summarized in Fig. 1E, depicting a body-wide landscape of differentially expressed genes in "heartless". Many of such genes are primarily expressed in the liver. Furthermore, their functions are implicated for a broad range of biological processes. For example, fabp10a, rbp2b and gc are involved in metabolism of lipids and vitamins. Several genes, such as c3b, cfb, hamp, angpt2a, il4r.1, are implicated for their roles in tissue remodeling and immune responses. This may suggest that liver development and/or function is more dominantly influenced by the cardiovascular system, as compared to those of other organs, at least during this early-to mid-organogenesis periods. The dominant effect on the liver may also suggest the earlier functional maturation of this organ.

Characterization of cardiomyocyte-specific gene mutations
The MTZ/NTR-mediated cell-ablation system induces cell death of the targeted cells (Curado et al., 2007;Curado et al., 2008;Dickover et al., 2013), thus inflammation could locally occur at the cellablated tissues. In fact, some of the genes influenced in "heartless" are related to tissue inflammation (e.g., hamp, il4r.1, il1b, cxcr4a, etc.). Hence, it is important to examine a possibility of an effect of tissue inflammation. Furthermore, the differential gene expression pattern could be due to the lack of cardiomyocytes or cardiac contraction, as both are absent in "heartless". Thus, these points were addressed by characterizing mutant zebrafish larvae for three cardiomyocyte-specific genes, myl7, cmlc1, tnnc1a, each encoding protein critical for cardiac contraction (Figs. 2A, S4A, S4B, see METHODS). Time-lapse observations of each mutant larva show differential effects on the contraction properties (Movies S3 -S5). The myl7 mutant shows complete lack of contraction (Movie S3). The cmlc1 mutant heart exhibits shuddering movement (Movie S4). The tnnc1a mutant shows significantly reduced ventricular contraction, but relatively normal atrial contraction (Movie S5). These aberrant cardiac contractions also resulted in perturbed circulation (Movies S3 -S5).
The expression analyses of pan-hypoxia indicators show significant upregulation of both phd3 and igfbp1a (Fig. 2B). The expression of vegfaa is significantly upregulated in both cmlc1 and tnnc1a mutants (Fig. 2B), but not in myl7 mutant (Fig. 2B). These results may reflect differences in the degree and/or quality of hypoxic states resulting from the differential cardiac dysfunctions among the mutants.
The gene expression patterns of these mutant larvae and "heartless" were compared and summarized (Fig. 2C). The heatmap shows the similar gene expression patterns among "heartless" and the three mutants, despite the lack of apparent cellular ablations in the mutant hearts and variable hypoxic states in the mutants. These results indicate that defective cardiac contraction and/or circulation, but not the local tissue inflammation or the cardiomyocyte-absence itself or hypoxia, is a primary cause of the differential gene expression patterns in "heartless".
The impact of the myl7 mutation on the gene expression pattern was completely offset (R 2 =0.95501) by the re-expression of myl7 using its own cardiomyocyte-specific promoter (Fig. S4C, Table S1). Such gene expression rescue was accompanied by the complete rescue of the normal contractility and circulation (Movie S6). This result further indicates the importance of cardiac contraction to establish and/or maintain the normal body-wide gene expression landscape. The myl7 promoter-driven cardiomyocyte-specific re-expression of cmlc1 or tnnc1a was not sufficient to rescue the defective cardiac contractions or circulation in each corresponding mutant (Fig. S4C, Movies S7, S8, Table S1). Such incomplete functional rescues resulted in only partial or the lack of gene expression rescues (Fig. S4D). These results further support the indication that the normal cardiac contraction and circulation determines the body-wide gene expression landscape.
A few differences in the gene expression patterns were found among the three mutant larvae (Fig.   2C). Such differences could be explained by the presence of genetic interactions between cardiac and non-cardiac genes (black arrows and lines in Fig. 2D). According to the interactions, each non-cardiac gene is categorized into six groups (I -VI) ( Fig. 2D): (I) Positive regulation by all three cardiacgenes; (II) Positive regulation by myl7 and cmlc1, but not by tnnc1a; (III) Negative regulation by all three cardiac genes; (IV) Negative regulation by cmlc1, but not by myl7 or tnnc1a; (IV) (V) Negative regulation by cmlc1 and tnnc1a, but not by myl7; (VI) Negative regulation by tnnc1a, but not by myl7 or cmlc1.
In addition to such cardiac and non-cardiac gene interactions (black arrows and lines in Fig. 2D), there also appears an intra-cardiac interaction (red lines in Fig. 2D). The expression of cardiac nppa is suppressed by cmlc1 and tnnc1a, as indicated by its upregulation by the mutation of either gene (Fig.   2C, D). The expressions of cmlc1 and tnnc1a require myl7, as they are both downregulated by the myl7 mutation (Fig. 2C, D).
These results further indicate a differential sensitivity of each gene expression to the cardiac contraction and/or circulation properties. It is possible that subtle changes in the local mechanical signals may also influence gene expressions in non-cardiac organs and also within the heart.

Possible cross-regulations among liver genes?
Many genes influenced in "heartless" are liver-genes (Fig. 1E). Hence, a possibility of their crossregulations was examined by analyzing the mutants for 9 liver genes (Fig. S5A, B). Nonsensemediated-decay (NMD) was confirmed, except for rbp2b (Fig. S5C). The expression of none of the 54 genes regulated in "heartless" was affected by any of these liver gene mutations (Fig. S5D, Table S1), indicating the existence of very little, if any, one-on-one cross-regulations among these liver-genes.

Characterization of "vesselless"
Oxygen and humoral factors are delivered to tissues and organs through the circulation. Immune cells use the vessels to reach to their target tissues. The circulation requires cardiac contraction and vessels.
Therefore, the absence of either one results in the lack of circulation. Hence, any effects caused by the lack of circulation are likely to be shared between "heartless" and larvae without vessels. In contrast, the gene expression regulated by the local vEC-derived signal(s) could be influenced only in the larvae without vessels, but not in "heartless". To distinguish between these two types of regulations, we generated the larvae without the vessels but with intact contracting cardiac-muscle. This was made possible by using cloche/npas4l mutant (Liao et al., 1997;Reischauer et al., 2016)(referred to as "vesselless" here) (Fig. 3A). The absence of the vessels and endocardium, but the presence of cardiomyocytes ( Fig. 3B) and cardiac contraction (Movie S9), was confirmed.
The upregulation of two hypoxia indicators, phd3 and igfbp1a, was detected in "vesselless", as in "heartless" (Fig. 3C). No significant changes of vegfaa expression was detected in "vesselless", like in "heartless" (Fig. 3C), which may in part reflect a possibility of its expression in hematopoietic cells (Liao et al., 1997;Reischauer et al., 2016). These results indicate that the degree and the quality of body-wide hypoxia in "vesselless" is comparable to that in "heartless". Hence, it is likely that the difference between "vesselless" and "heartless" is the absence/presence of the vessels, but not the states of the cardiac-contraction or hypoxia.
The comprehensive transcriptome analysis of 5 dpf "vesselless" was conducted to identify genes whose expressions are specifically dependent on the presence of the vessels (Figs. 1B, 3B). The GO enrichment analysis unveiled families of genes whose expressions are specifically affected in "vesselless", but not in "heartless" (Fig. 3D, E). They include those related to sensory/neural system (i.e., synaptic transmission, response to light stimulus, sensory perception, phototransduction, detection of light stimulus, neurological system process, neurotransmitter transport, sensory perception of light stimulus, visual perception, regulation of synaptic transmission) and sterol homeostasis (i.e., sterol biosynthetic process, sterol metabolic process) (Fig. 3D, E). They are affected only in "vesselless", but not in "heartless", despite hypoxia and the lack of circulation in both.
Many sensory/neural system genes are specifically downregulated in "vesselless" (Fig. 3E). One such gene is olfactory-marker-protein-a (ompa) (Fig. 3E). WISH analyses show ompa is expressed in the olfactory bulb (Fig. S6). A comparison to a pan-olfactory-bulb marker, ompb (Celik et al., 2002;Yoshida et al., 2002;Yoshida and Mishina, 2003), shows that both ompa and ompb are expressed in the olfactory bulb, but the ompa expression is restricted to a sub-domain of the neuroepithelium (Fig.   S6).
They include a family of genes encoding enzymes critically involved in cholesterol biosynthesis (Lu et al., 2015;Mazein et al., 2013;Paton and Ntambi, 2009) (Fig. S7). WISH analyses indicate that all were expressed in the liver, and five (hmgcra, sqlea, lss, msmo1, fads2) were also expressed in the brain and two (hmgcra, fads2) in the intestine (Fig. S8). Cholesterol in the circulation is taken up by vECs via endocytosis (Anderson et al., 2011;Ho et al., 2004). This vEC mechanism is critical to maintain the cholesterol level in circulation (Anderson et al., 2011;Ho et al., 2004). It is possible that this mechanism operates as a negative feedback, suppressing the expression of the genes encoding cholesterol biosynthesis enzymes -hence, the absence of vECs in "vesselless" could induce their upregulation.
This possibility was further supported by an experiment using atorvastatin, a potent inhibitor of HMG-CoA reductase, an enzyme required for cholesterol biosynthesis (D'Amico et al., 2007;Ho et al., 2004). The atorvastatin treatment of wild type or "heartless" zebrafish larvae induced the upregulation of these genes (Fig. S9, Table S1). In contrast, the atorvastatin treatment of "vesselless" induced only a weak or no upregulation (Fig. S9, Table S1).
We also examined their expression in 2 dpf larvae where major organs, such as liver, are yet to grow, but vECs are already present, thus an indirect pleiotropic effect is minimized (Fig. 3F). The result shows that their expressions are also upregulated in 2 dpf "vesselless" larvae ( Fig. 3F). These results suggest that their expressions are in a negative feedback loop where vECs function as a suppressive interface.
That such differential expressions of the genes were due to the lack of vessels is further supported by the characterization of etv2/etsrp morphant (Fig. S10). Etv2/etsrp morpholino injection was previously shown to reduce the vascular network in a relatively specific manner (Craig et al., 2015;Sumanas and Lin, 2006;Veldman and Lin, 2012) (Fig. S10A). Gene expression correlation analysis between "vesselless" and etv2/etsrp morphant showed a high correlation coefficient Table S1), supporting the indication that the differential gene expressions in "vesselless" is due to the lack of vessels, rather than a pleiotropic effect of the cloche/npas4l gene mutation.
Taken together, the results suggest that vECs themselves function as positive and negative regulators for sensory gene and sterol homeostasis gene expressions, respectively (Fig. 3H). In "vesselless", the absence of vEC-derived signals, such as cell-cell contacts, ECM, secreted paracrine factors, may induce the downregulation of sensory system genes such as ompa and opsin/rhodopsin genes in olfactory and visual systems, respectively ( Fig. 3E, G). The absence of vECs in "vesselless" also induces the lack of cholesterol endocytosis, causing the elimination of the negative feedback loop of cholesterol biosynthesis gene expression and hence the upregulation of their expression ( Fig. 3E, F, G).

Distinguishing differential roles of the cardiovascular system in regulating the body-wide gene expression landscape
The "heartless" lacks cardiomyocytes, resulting in the absence of cardiac contraction and circulation 1C, 2B, 3C). Hence, some, if not all, of the differential gene expressions detected among these models could be caused by one or the combinations of differential conditions of each model, but not by the common ones such as hypoxia. To determine which of these conditions are contributing to the differential gene expression patterns, we examined additional models.
One such is a cardiotoxin-treated model (Fig. 4A). Haloperidol is known to disturb the normal cardiac contraction in zebrafish (Milan et al., 2006). While cardiomyocytes and vasculature were present as in normal larvae (Fig. 4B), the zebrafish larvae treated with this drug for 5 hrs exhibited arrested heart-beat and no circulation (Movie S10). Hypoxia of the haloperidol-treated larvae was characterized by the expression of pan-hypoxia indicators (Fig. 4C). The haloperidol-treatment caused significant upregulated expression of all three pan-hypoxia indictors (Fig. 4C).
Contributions of hypoxia were also examined (Fig. 5). The body-wide hypoxia was induced by two methods (Fig. 5A): DMOG-treatment and hypoxia-chamber (Gerri et al., 2017). The larvae were treated with 100 µM DMOG for 6 hrs, 10 hrs, or with 125 µM for 24 hrs (Fig. 5A, see also METHODS). The heart-beat and the circulation appeared relatively normal in the DMOG-treated larvae (Movie S11). Body-wide hypoxia was also induced by incubating the larvae in hypoxiachamber for 24 hrs (3.5 dpf -4.5 dpf) (Fig. 5A, see also METHODS). The 24-hrs treatment/incubation period in the hypoxia models are comparable to the duration of the cardiacablation in "heartless" (see METHODS). Hypoxia was evaluated by the expressions of pan-hypoxia indicators ( Fig. 5B). In both models, the expression of phd3 was significantly upregulated (Fig. 5B).
The expressions of both ifgbp1a and vegfaa expression were upregulated in the larvae treated by DMOG for 10 hrs and 24 hrs, and also in those incubated in hypoxia-chamber for 24 hrs (Fig. 5B).
The differential gene expression patterns were analyzed in these additional models and compared to those of "heartless" and "vesselless" (Fig. 6). We also introduced the vessel-ablation to "heartless" (i.e., "heartless+vesselless") to determine the dependence of the differential gene expression in "heartless" on the presence of vessels (Fig. 6). The structural and functional characteristics of each model are summarized in Fig. 6A. The differential gene expressions among the models are summarized and presented as heatmap ( Fig. 6B) with the original raw data in Table S1.
The heatmap shows that many of the differential gene expressions found in both or either "heartless" and/or "vesselless" are unaffected in the hypoxia models (i.e., DMOG and hypoxiachamber). This result indicates that the hypoxic condition induced up to 24 hrs of DMOG or hypoxiachamber incubation is less influential on the expressions of many of these differentially regulated genes in "heartless" and/or "vesselless". Based on the degree of the upregulation of phd3, a panhypoxia marker, a comparable degree of hypoxia appears to be achieved by "heartless", "vesselless", "heartless+vesselless", DMOG and hypoxia-chamber models (Fig. 6B).
Such circulation-dependent signals could be mechanical, humoral and/or cellular signals. The result also suggests that oxygen homeostasis contributes to much lesser extent, if any, to their expressions.
No influences were detected in either of the hypoxia models ( Fig. 6B. Table S1). These results suggest that the local-presence of vECs themselves in olfactory bulb is a main contributor to the regulation of the ompa expression. The circulation appears to influence to much lesser extent, and the hypoxia impose little, if any, influence on its expression.
Such results suggest that their expressions are mainly regulated by the local presence/absence of vECs, rather than the circulation or hypoxia.
Retinol binding protein 2b (rbp2b) exhibits a distinct expression pattern (Fig. 6B). Rbp2b is implicated for vitamin A and lipid homeostasis (Kanai et al., 1968;Li and Norris, 1996) and primarily expressed in the liver (Figs. S1, S3). The expression of rbp2b is significantly downregulated in "heartless", but not in "vesselless" (Fig. 6B). Such downregulation in "heartless" is maintained in the "heartless+veseelless" combination model (Fig. 6B, Table S1). Furthermore, the significant downregulation was also detected in the haloperidol model (Fig. 6B). These results suggest that the liver expression of rbp2b requires cardiac contraction, but the vessels are dispensable.
What could be such a mediator relayed from the beating heart to the liver in the absence of the vessels? A possible candidate is a nervous-system-derived mediator. The innervation of the cardiomyocytes is critical for the cardiac function. Dopaminergic (DA) neurons are known to regulate a number of hemostatic processes involving cardiovascular functions (Aaronson et al., 2014;Myers and Olson, 2012;Noble, 2002). Hence, we examined a role of DA neurons in regulating the liver  S11C). Such dispensability of otpa + /otpb + DA neurons suggests an existence of other type(s) of beating-heart derived vessel-independent signals in regulating the liver rbp2b expression.

DISCUSSION
It has been assumed that the cardiovascular system in zebrafish is dispensable for oxygen homeostasis at least during early-to mid-larval periods. This notion suggests an existence of other important cardiovascular function(s) during development. Herein we show several evidences suggesting that vascular-organ interactions, circulation-dependent signals, and circulation-independent but distantly acting beating-heart-derived signals are important mediators of such non-oxygen regulating functions ( Fig. 7).
What could be the identities of these signals? For the sensory system (e.g., the ompa expression in the olfactory bulb), vEC-derived signals such as cell-cell contacts, ECM and/or secreted paracrine factors are potential candidates (Crivellato et al., 2007). Although such vEC-derived signals are the main regulators, the circulation also appears to influence the ompa expression (Fig. 6B, Table S1).
Therefore, it is possible that the strength/levels of such vEC-derived signals may be partially dependent on the circulation. Alternatively, the mechanical signals mediated by the circulation could function together with the vEC-derived signals to regulate the expression of the sensory system genes.
In the case of locally regulating the sterol homeostasis genes by vECs, a negative-feedback loop system, such as cholesterol endocytosis, installed within vECs is a potential candidate (Anderson et al., 2011;Ho et al., 2004).
We also found 18 genes that are regulated by circulation-dependent signals, but not much by hypoxia (Fig. 6B). Mechanical stimuli, such as shear-stress, could be such circulation-dependent signals (Sato, 2013) (Fig. 7). Humoral factors (Sato, 2013) and immune cells and/or factors presented by them are also candidates.
A beating-heart-derived long-distance acting signal is another one. An example regulated by such a signal is rbp2b (Fig. 7). We show that otpa + /otpb + DA neurons are dispensable for this regulation ( Fig. S11C). What could be such a signal? A possibility of diffusible small chemicals and/or peptides regulated and released by the contracting cardiac-muscle (Fig. 7) would be worth exploring in the future.
We also show that each of such signals and cardiovascular functions are highly selective for specific organs (Fig. 7). In particular, the local vascular-organ interactions appear to be preferentially exploited by sensory organs (such as olfactory and visual systems) (Figs. 3G, 7). Our results also indicate a critical importance of the vascular interface for maintaining cholesterol homeostasis in brain, liver and intestine. In this case, the vasculature appears to function as suppressive interface to prevent hyper-activation of the cholesterol biosynthesis pathways (Figs. 3G, 7). It is possible that the usage of such local cardiovascular functions is more effective for certain organs (e.g., sensory organs) than others. Alternatively, such special organs may lack a system that can utilize circulationdependent signals or may be less efficient in using them.
Our finding also has an evolutionary implication. We identified a possibility that the heart, via contraction, sends signal(s) to the liver in a circulation-independent manner, as indicated by the differential expression patterns of rbp2b (Figs. 6B, 7). This mechanism could be related to a primordial function of the cardiovascular system. Like in vertebrates, the heart is the first functional organ system to develop also in invertebrates. However, invertebrates lack extensive vascular network.
The beating-heart alone is sufficient to facilitate the body-wide transport of oxygen and humoral factors. Hence, it is possible that such signaling system by the beating-heart without the use of the vessel has survived natural selection and has remained in some vertebrates such as zebrafish.
Do our findings apply to other vertebrates such as mammals? In mice and frog, the critical importance of local interactions between vECs and developing organs have been shown for the development of liver and pancreas, respectively (Cleaver and Dor, 2012;Lammert et al., 2001Lammert et al., , 2003Talavera-Adame and Dafoe, 2015). Recently, an importance of vEC-tissue interactions is also implicated in liver organoid formation in vitro (Camp et al., 2017). Neurovascular interactions play critical roles in development and disease in mice and human (Crivellato et al., 2007;Mukouyama et al., 2002;Okabe et al., 2014;Visconti et al., 2002;Wang et al., 1998;Weinstein, 2005). Several paracrine factors, collectively referred to as angiocrine factors, have been discovered (Butler et al., 2010a;Ding et al., 2010;Ding et al., 2011). They are secreted from tissue-specific vECs and facilitate organ regeneration (Butler et al., 2010b;Rafii et al., 2016). These findings illustrate an importance of such local-functions of the vasculature in other vertebrates including mammals and human.
Our results also indicate that the zebrafish larvae without the functional cardiovascular system (i.e., "heartless" and "vesselless") are under hypoxia at least based on the upregulated expression of two pan-hypoxia indicators, phd3 and igfbp1a (Fig. 6B). Previously, it was assumed that zebrafish without the functional cardiovascular system can form organs as oxygen diffusion through body wall is sufficient at least up to mid-larval period. However, our results indicate that the functional cardiovascular system is indeed necessary for oxygen homeostasis and zebrafish organs form under hypoxic microenvironment.
Experimental manipulations of the cardiovascular system induce changes in oxygen-homeostasis, making it challenging to study the roles of the cardiovascular system in regulating oxygenindependent developmental and/or physiological processes. Here, we identified a set of noncardiovascular genes (e.g., ompa, fabp10a, dpys, ugt1a5, pfkfb3, etc.) that are regulated by the formation and functions of the cardiovascular system, but are only little, if any, influenced by oxygen homeostasis. Hence, these non-cardiovascular genes could be utilized as the targets of manipulations to determine biological processes that are independent of oxygen homeostasis. No direct manipulations of the cardiovascular system are required in this approach, thus it is applicable to mammalian models. For example, the inhibitions of such gene functions could identify biological processes that are independent of oxygen homeostasis, but are mediated by the cardiovascular system and are critical for organismal development and homeostasis, in mammalian models such as mice.
The dataset presented herein also provides a list of marker genes for distinct functional aspects of the cardiovascular system. These markers are useful for evaluating functional manipulations of the cardiovascular system in the future experiments. Furthermore, many diseases are caused by the changes of the functional states of the cardiovascular system (Noble, 2002). Hence, it is also possible that the differential gene expression patterns reported here could be exploited to evaluate the effects of therapeutic treatments on the cardiovascular system in diseases and/or disease models.
In addition, the dataset could also serve as a useful resource to design experiments to gain further in-depth insights into the roles of the cardiovascular system in regulating organ development and function. The olfactory vasculature could function as a guide to induce the differentiation of a ompapositive subset of neuroepithelial cells. Such vasculature-guided ompa-positive neuroepitheial cells could possess a unique physiological function. In each organ, only subsets of genes are responsive to the manipulations of the cardiovascular system. These cardiovascular-sensitive genes may collectively assume unique developmental and/or physiological functions during organ maturation. Such questions could be systematically addressed by using the annotations of the genes reported in this paper.

Fish husbandry
Zebrafish were maintained in circulation-type aquarium system (Iwaki) with 14 h/day and 10 h/night cycle at around 27 o C. The fertilized eggs were collected and raised at 28.5 o C in egg water (0.06% artificial marine salt supplemented with 0.0002% methylene blue) until around epiboly stage and subsequently in 1/3 Ringer's medium (1.67 mM HEPES, 38.7 mM NaCl, 0.97 mM KCl, 0.6 mM CaCl 2 , pH 7.2) containing 0.001% phenylthiourea (PTU) (Sigma) to prevent pigmentation. Embryos and larvae were staged to days post fertilization (dpf) according to Kimmel et al (Kimmel et al., 1995).
Zebrafish maintenance and experiments were conducted in accordance with animal protocols approved by the Animal Care and Use Committee of Advanced Telecommunications Research Institute International (A1403, A1503, A1603).
For the liver-specific ablation, Tg(fabp10a:CFP-NTR) homozygous fish were used. Tg (fabp10a:CFP-NTR) embryos/larvae were treated with 7 mM MTZ prepared as above from 2.5 dpf to 5.5 dpf MTZ-containing media was changed daily during the treatment.

RNA extraction
To obtain total RNA, embryos and larvae were harvested in a 1.5 -2 ml tube at appropriate stage and frozen in liquid N 2 to be stored in -80 o C. To prepare total RNA for RNAseq analysis of embryonic and larval stages, 10 -20 embryos and larvae were pooled in a 1.5 ml tube, and total RNA was isolated using RNeasy Mini Kit (QIAGEN). The pooled embryos/larvae were homogenized in Buffer RLT included in the kit using 5 ml syringe and 24G needle by passing through the needle for 20 times.
Alternatively, they were crashed in 700 μl Buffer RLT using approximately 50 zirconia balls with 1.5 mm diameter (YTZ balls) (NIKKATO) by centrifuging at 4260 rpm for 60 sec in Cell Destroyer PS1000 (BMS). After homogenization, the manufacture instruction was followed. To prepare RNA for real-time PCR analysis, embryos/larvae were individually harvested in a 1.5 ml or 2 ml tube and total RNA from each individual embryo/larva was isolated by AllPrep DNA/RNA Mini Kit (QIAGEN).
Individual Embryo/larva was either homogenized using a syringe as described above or crashed in 700 μl Buffer RLT using approximately 50 zirconia balls with 1.5 mm diameter (YTZ balls) (NIKKATO) by centrifuging at 4260 rpm for 60 sec in Cell Destroyer PS1000 (BMS). After the homogenization or crushing, the manufacture instructions were followed. Subsequently, the isolated genomic DNA and total RNA were subjected to genotyping and reverse transcription reaction.

RNA sequencing
Total RNA was prepared from two biological replicate pools of 4.5 dpf wild type and cardiomyocyte-  (Afgan et al., 2016). The obtained bam file was used to calculate fragments-per-kilobase-of-exon-per-million (FPKM) of transcripts and the differential gene expression data using Cuffdiff (Trapnell et al., 2010). To perform Gene ontology enrichment analysis, the enriched genes were defined as those with log 2 fold ≥1 and with p <0.05. A gene ontology enrichment analysis is performed by R package "topGO" using a root category "BP" and a reference database "org.Dr.eg.db". To prepare volcano plot graph from RNAseq data, p-value and fold-change was calculated using DESeq2 (Love et al., 2014) with default settings.

Whole-mount in situ hybridization (WISH)
To synthesize an antisense RNA probe, the template DNA was amplified by PCR using KOD-Plus-Neo (TOYOBO) from cDNA synthesized from zebrafish total RNA of appropriate stage of WT or cloche mutant (for hmgcra, sqlea, nsdhl, cbp, fads2). For lss, cyp51, msmo1, scd, ompa and ompb, the cDNA sequences were chemically synthesized for use as templates of PCR. The primers used for PCR are listed in Supplementary Table 1. Primer sequence was designed using Blast primer (NCBI/ Primer-BLAST: http://www.ncbi.nlm.nih.gov/tools/primer-blast/) and, T3 and T7 sequence were added at 5' end of forward and reverse primer, respectively. PCR product was purified by QIAquick PCR Purification Kit (Qiagen). The sequence was confirmed using T3 and T7 primers. For synthesizing antisense RNA probe, the following mixture was used: DIG RNA labeling mix (Roche Diagnostics), Transcription buffer (Roche Diagnostics), RNase inhibitor (Roche Diagnostics), T7 RNA polymerase (Roche Diagnostics), and 200 ng of template DNA. Then it was incubated for 1.5 -3 hour at 37 o C, followed by precipitation with lithium-chloride precipitation solution. Precipitated DIG-labeled RNA was re-suspended in nuclease free water and mixed with equal volume of formamide to be stored at -80 o C.
To prepare embryos for whole mount in situ hybridization, anesthetized zebrafish embryos/larvae were fixed in 4% Paraformaldehyde Phosphate Buffer Solution (PFA) (Nacalai Tesque) overnight at 4 o C. The fixed embryos were dehydrated three times for 5 minutes in 100% methanol at room temperature and were stored at -30 o C for at least 2 days. Before hybridization, the embryos in methanol were dehydrated five times for 5 minutes in PBST (phosphate buffered saline containing 0.1% Tween-20), and then permeabilized in 10 μ g/ml proteinase K in PBST for 30 minutes at room temperature. After quick wash with PBST, the embryos were post-fixed for 20 minutes in 4% PFA at room temperature, and then washed 5 x 5 minutes in PBST. Then, the embryos were incubated in hybridization solution (50% formamide, 5×SSC, 50 μ g/ml Heparin, 500 μ g/ml tRNA, 0.1% Tween-20, 9.2 mM Citric acid, pH 6.0) without DIG-RNA probe at 68 o C for at least one hour. Hybridization was performed in hybridization solution containing DIG-labeled probe (1:200 dilution) at 68 o C for 16 hours. Following to hybridization, the embryos were washed with 50% formamide /50% 2×SSC once

Two-color fluorescence in situ hybridization
DNP-labeled RNA probe was synthesized using T7 RNA polymerase (Roche) by incubating template DNA with 0.35 mM DNP-11-UTP (Perkin Elmer), 1 mM of ATP, GTP and CTP, 0.65 mM UTP (Invitrogen) for 3 hrs at 37 o C. The synthesized DNP-probe was purified by LiCl precipitation.
Embryos were prepared and hybridized with DIG-labeled and DNP-labeled riboprobe as described in WISH using AP-system, except for addition of 5% dextran sulfate (Sigma) to the hybridization buffer. The hybridized embryos were washed and blocked as WISH method of APsystem. After blocking, to detect DIG-labeled probe, the embryos were incubated with anti-digoxigenin-POD, Fab fragments (1:1000; Roche) in PBST containing 2% sheep serum and 2 mg/ml BSA for overnight at 4 o C. After the incubation, the embryos were washed 6 x 15 min in PBST and then 2 x 5 min in 1x amplification diluent, followed by the incubation with TSA Plus Cyanine 5 solution (1:50 dilution in amplification diluent buffer) (Perkin Elmer) for 1 hr at R.T. After the incubation, embryos were washed 2 x 5 min with PBST and then the first TSA reaction was quenched in 2% H 2 O 2 in PBST for 60 min at R.T. The embryos were then washed 4 x 5 min in PBST, followed by the incubation with anti-DNP-HRP (1:500; Perkin Elmer) in PBST containing 2% sheep serum and 2 mg/ml BSA for overnight at 4 o C. The antibody was washed out in PBST, and then the embryos were incubated in TSA Plus Cyanine 3 (1:50 dilution in amplification diluent buffer) (Perkin Elmer) for 1 hr at R.T. After the incubation, the embryos were washed 6 x 5 min in PBST and mounted in Prolong Diamond (Molecular probes) for imaging under confocal microscope.

Microscopy and image process
To observe and record the heart beating and the circulation, embryos/larvae were anesthetized using 0.012% MS-222 and were mounted either laterally or ventrally in 1.0% NuSive GTG Agarose (Lonza) on glass-bottomed 35 mm dish. Imaging was performed using a 10x dry objective lens (Plan Apo, NA0.45) and 20x dry objective lens (Plan Apo, NA 0.75) mounted on Nikon A1R confocal microscope. Time-lapse image was recorded with a resonant scanner for 15 -30 f/s imaging, and converted to Quick time movie using IMARIS software (BITPLANE) or to AVI movie using Figi software. These movies were converted into mp4 movies using iMovie software.
To take images of whole mount in situ hybridization, specimens were mounted in 75 -80% glycerol and imaged using 4 x (Plan Apo/NA0.20) or 10 x (Plan Apo/NA 0.45) (Nikon) objective lens mounted on Nikon eclipse inverted microscope and 1 x objective lens (Plan Apo) mounted on Leica M165 FC microscope. Images of two-color fluorescence WISH were taken using a 20x dry objective lens (Plan Apo, NA 0.75) and 40x water immersion objective lens (Apo LWD, NA 1.15) mounted on Nikon A1R confocal microscope.
For preparing gRNA, we followed either plasmid-based method, where the template sequence for gRNA was cloned in plasmid, or cloning-free method. For the plasmid based method (Jao et al., 2013 Kit (QIAGEN) from several colonies, the successful cloning was confirmed by sequencing with M13Forward primer. The plasmid with gRNA target sequence was linearized by BamHI and used as a template of in vitro transcription reaction. gRNA was transcribed using MEGAshortscript kit (Ambion). The cloning-free method was also used to generate templates for gRNA synthesis (Gagnon et al., 2014). The 1 μl of 100 μM gene-specific oligonucleotides containing T7 or SP6 sequence, 20 base target sequence without PAM, and a complementary region to constant oligonucleotide were mixed with 1 μl of 100 μM oligonucleotide encoding the reverse-complement of the tracrRNA tail with 1x NEBuffer2 in a total volume of 10 μl to anneal by the following procedure: denaturation for 5 min at 95 o C, cooling to 85 o C at -2 o C /sec and then cooling from 85 o C to 25 o C at -0.1 o C /sec. The single strand DNA overhangs were filled with T4 DNA polymerase by adding 2.5 μl 10 mM dNTPs mix, 1 μl 10x NEBuffer2, 0.2 μl 100x NEB BSA and 0.5 μl T4 DNA polymerase (NEB) and then incubated at 12 o C for 20 min. The resulting double strand DNA was purified using QIAquick PCR purification kit (Qiagen). The gRNAs were transcribed using MEGAshortscript kit (Ambion). gRNAs were treated with DNase, which is included in the kit, and precipitated using lithium-chloride precipitation solution (Ambion). For making Cas9 mRNA to co-inject with gRNAs to zebrafish eggs, we used pCS2-nls-zCas9-nls (Jao et al., 2013) (Addgene) as a template DNA. The template DNA was linearized by NotI (NEB) and purified using QIAquick PCR purification kit. Capped nls-zCas9-nls mRNA was synthesized using mMESSAGE mMACHINE SP6 transcription kit (Ambion) in a volume of 20 μl. The synthesized mRNA was treated with DNase and precipitated with lithium-chloride precipitation solution (Ambion).
To assay and determine the indel mutation by gRNAs, the genome obtained from individual embryos of 1 to 4 dpf or tail clipping for direct sequencing and/or high resolution melt (HRM) analysis was used (Thomas et al., 2014). Embryo or tail fin clips were transferred into 25 to 50 μl of lysis buffer (10 mM Tris-HCl (pH 8.0), 50 mM KCl, 0.3% Tween20, 0.3% NP40, 1 mM EDTA, 0.2 mg/ml Proteinase K (Invitrogen)), and incubated at 55 o C for 2 hrs to overnight, followed by the incubation at 98 o C for 10 min. For direct sequence, the sequence spanning the gRNA target site was amplified by PCR using 1 μl of the lysed sample as a template, and purified the PCR product. Primers used for the PCR and direct sequencing are listed in Supplementary Table 2 Table S2.

Rescue experiment
The pTol2 Embryos were harvested on 2 dpf for the analyses. embryos/larvae at appropriate stage were treated in 6 cm petri dish.

Hypoxia chamber experiment
Hypoxia Incubator Chamber (Billups-Rothenberg) was used to induce body-wide hypoxia. Sibling embryos/larvae in a dish were separated into two groups each in 6 cm dish, containing 10 mL of 1/3 Ringer's solution containing PTU. One dish was set in hypoxia chamber to flow 5% O 2 /95% N 2 gas mixture for 5 min with a flow rate of 20 L/min following to the manufacturer's instruction. Hypoxia chamber was placed in 28.5°C incubator for 24 hrs and then the larvae were harvested.

Cardiotoxin treatment
Haloperidol (Sigma) was dissolved in DMSO to make the stock solution of 50 mM and 25 mM, respectively. Haloperidol was diluted to 50 µM in 1/3 Ringer's solution containing PTU.
Embryos/larvae were treated with haloperidol for 5 hrs at 28.5°C to 4.5 dpf. Before harvesting, the larvae were assessed under dissection microscope to select those with no heart-beating.

Atorvastatin treatment
Atorvastatin Calcium Trihydrate (Wako) was dissolved in 100% DMSO at a concentration of 10 mM.
Drug was diluted in 0.001% PTU to make the working solution of 2 μ M with 0.2% DMSO.
Atorvastatin treatment was initiated at 24 hpf and replaced with fresh drug every 24 hrs to 4.5 dpf. To combine atorvastatin treatment and heart ablation, embryos were first treated with atorvastatin to 3 dpf. Then the embryos were soaked in the mixture of 2 μ M atorvastatin and 10 mM MTZ for 6 hrs at 3 dpf, followed by the incubation in atorvastatin solution again to 4.5 dpf.

Data analyses and statistics
For data collection and analysis of qRT-PCR, no statistical methods were used to predetermine sample size. Embryos/larvae subjected to qRT-PCR analysis were blindly collected from the clutch mates.
The etv2/etsrp morphants were identified by the reduced expression of Tg(fli1a:egfp) reporter and processed for the analyses. For collecting larvae in Figure S4, those expressing GFP widely in heart (more than 70% in heart in appearance under fluorescence stereo microscope) were collected for qRT-PCR analysis. For qRT-PCR data analysis of mutant fish of myl7, cmlc1 and tnnc1a in Fig. 2, values of 2 -dCt obtained from two different mutant alleles of each gene were combined to calculate statistics.
Because the expression level of lepb gene in wild type embryo/larva was extremely low, a signal of SYBR Green was not detected in our experimental design of qRT-PCR in most of WT. Therefore, in Figs 2 and 6, and Fig. S5, we considered the Ct of lepb as 46, if the signal was not detected. For drug treatment in Fig. S9, embryos/larvae were randomly selected and processed for each treatment group.
For the WISH experiments, at least 8 individual fish were processed and examined. Student t-test was performed for statistical analysis. Benjamini-Hochberg procedure was also applied to all multiple sample comparisons (i.e., heatmaps) to correct errors for the multiple tests. A p-value less than 0.05 was considered to be statistically significant (* p<0.05, **p<0.01 and ***p<0.001). The horizontal line represents the mean.