Cardiac neural crest contributes to cardiomyocytes in amniotes and heart regeneration in zebrafish

  1. Weiyi Tang
  2. Megan L Martik
  3. Yuwei Li
  4. Marianne E Bronner  Is a corresponding author
  1. California Institute of Technology, United States

Abstract

Cardiac neural crest cells contribute to important portions of the cardiovascular system including the aorticopulmonary septum and cardiac ganglion. Using replication incompetent avian retroviruses for precise high-resolution lineage analysis, we uncover a previously undescribed neural crest contribution to cardiomyocytes of the ventricles in Gallus gallus, supported by Wnt1-Cre lineage analysis in Mus musculus. To test the intriguing possibility that neural crest cells contribute to heart repair, we examined Danio rerio adult heart regeneration in the neural crest transgenic line, Tg(−4.9sox10:eGFP). Whereas the adult heart has few sox10+ cells in the apex, sox10 and other neural crest regulatory network genes are upregulated in the regenerating myocardium after resection. The results suggest that neural crest cells contribute to many cardiovascular structures including cardiomyocytes across vertebrates and to the regenerating heart of teleost fish. Thus, understanding molecular mechanisms that control the normal development of the neural crest into cardiomyocytes and reactivation of the neural crest program upon regeneration may open potential therapeutic approaches to repair heart damage in amniotes.

https://doi.org/10.7554/eLife.47929.001

eLife digest

Before birth, unspecialized stem cells go through a process called differentiation to form the many types of cells found in the adult. Neural crest cells are a group of these stem cells found in all animals with backbones (i.e. vertebrates) including humans. These cells migrate extensively during development to form many different parts of the body. Due to their contributions to diverse organs and tissues, neural crest cells are very important for healthy development.

The heart ventricle is one of the tissues to which neural crest cells contribute during embryonic development in fish and amphibians. However, it was unclear whether this is also the case for birds or mammals or whether neural crest cells have any roles in the regeneration of the adult heart after injury in fish and amphibians.

To address these questions, Tang, Martik et al. used cell biology techniques to track neural crest cells in living animals. The experiments revealed that neural crest cells contribute to heart tissue in developing birds and mammals and help repair the heart in adult zebrafish. Further results showed that the contribution of neural crest cells to the heart is controlled by the same genes during both the growth of the embryonic heart and the repair of the adult heart.

These results provide new insights into the repair and healing of damaged heart muscle in fish. They also show that similar processes could exist in mammals, including humans, suggesting that activating neural crest cells in the heart could treat damage caused by heart attacks and related conditions.

https://doi.org/10.7554/eLife.47929.002

Introduction

The neural crest is an important stem cell population characterized by its multipotency, migratory behavior, and broad ability to differentiate into derivatives as diverse as elements of the cardiovascular system, craniofacial skeleton, and peripheral nervous system. However, not all neural crest cells are alike, with distinct populations existing along the body axis. One of the most unique neural crest populations is the ‘cardiac neural crest’ that contributes to the outflow septum and smooth muscle of the outflow tract of the heart. Ablation studies in chick embryos show that removal of the cardiac crest results in a broad range of defects, including persistent truncus arteriosus, abnormal myocardium function, and misalignment of the arch arteries (Kirby et al., 1983Waldo et al., 1999; Bockman et al., 1987). These defects are highly reminiscent of some of the most common human congenital heart defects. Importantly, other neural crest populations cannot rescue the effects of cardiac neural crest ablation even when grafted in its place, exemplifying the uniqueness of this population (Kirby, 1989).

Classically, quail-chick transplantation experiments have been used to uncover contributions of the cardiac neural crest to the heart, with some more recent attempts using antibody staining of migratory neural crest cells or LacZ retroviral lineage analysis as well as transgenic lines such as Wnt1-Cre driven β-galactosidase in mammals (Kirby et al., 1983; Kuratani and Kirby, 1991; Boot et al., 2003; Jiang et al., 2000). The results suggest that the cardiac neural crest contributes to smooth muscle cells lining the great arteries, outflow tract septum and valves, mesenchyme that remodels pharyngeal arch arteries, and parasympathetic innervation of the heart, such as the cardiac ganglion. However, inconsistencies remain between different lineage approaches, most of which suffer from high background and low cellular resolution.

To reconcile these differences, here, we use a multi-organismal approach to examine the lineage contributions of cardiac neural crest to the heart. Using a novel retroviral labeling approach in chick and confirmed by Wnt1-Cre reporter lines in mouse, we reveal a previously undetected contribution of the amniote cardiac neural crest to the trabecular myocardium of the ventricles, a derivative previously thought to be confined to non-amniotic vertebrates (Sato and Yost, 2003; Li et al., 2003; Cavanaugh et al., 2015).

The homologous cardiac neural crest contribution to cardiomyocytes across diverse species raised the intriguing possibility that these cells may contribute to cardiac repair. As the adult zebrafish heart exhibits extensive regenerative capacity, we turned to this model to test whether the neural crest may contribute to heart regeneration (Poss et al., 2002). Intriguingly, we show that resected adult zebrafish hearts reactivate many genes of a neural crest gene regulatory program during the regeneration process. Taken together, these results demonstrate an evolutionarily conserved contribution of neural crest cells to cardiomyocytes across vertebrates and a previously unappreciated role during heart regeneration.

Results

Labeling the chick cardiac neural crest using Replication Incompetent Avian retrovirus

To specifically label cardiac neural crest cells prior to their emigration from the neural tube and identify novel progeny of chick cardiac crest, we use a replication-incompetent avian retrovirus (RIA) that indelibly and precisely marks neural crest progenitors for long term lineage analysis at single cell resolution and without the need for tissue grafting. To this end, the post-otic neural tube of the hindbrain adjacent to somites 1–3 was injected at Hamburger and Hamilton (HH) stage 9–10 with high-titer (1 × 107 ifu/mL) RIA (Figure 1A), which drives expression of nuclear localized H2B-YFP under control of a constitutive RSV promoter (Li et al., 2017; Tang et al., 2019; Fields-Berry et al., 1992; Chen et al., 1999; Hamburger and Hamilton, 1951). At this stage in the development, premigratory cardiac neural crest cells are positioned within the dorsal neural tube and about to emigrate. Accordingly, this labeling approach solely marks hindbrain neural tube cells including premigratory cardiac neural crest cells that subsequently delaminate from the dorsal neural tube during a two-hour time window when the virus remains active.

Figure 1 with 1 supplement see all
Retrovirally mediated fate mapping of cardiac neural crest reveals novel derivatives.

(A) Schematic diagram of the approach: Replication Incompetent Avian (RIA) retrovirus encoding nuclear H2B-YFP was injected into the lumen of the hindbrain from which cardiac neural crest arises. (B) One day post-infection (HH14), whole mount image (lateral view) showing virally labeled progeny (green) in the cardiac migration stream en route to pharyngeal arch 3. (B’) Brightfield image to show anatomical information. A, anterior; P, posterior; D, dorsal; V, ventral. (C) Two days post-infection (HH18), virally labeled cardiac crest has populated pharyngeal arches 3, 4 and 6, highlighted with dashed line. (D) Transverse section showing that labeled cardiac crest expresses neural crest marker HNK-1 (red). D, dorsal; V, ventral; L, left; R, right. (E) Low magnification transverse section of an E6 embryo (DAPI, blue). Dashed boxes show relative positions of cardiac crest-derived populations. (F–H) High magnification image of selected regions in E: pharyngeal arch arteries (F); aorticopulmonary septum (G); Neural crest derivatives located in the outflow tract express Troponin T (magenta), a myocardium marker (H, H’). (I) Temporal map of the establishment of distinct cardiac neural crest derivatives. Labeled cells initially are in the migration stream, cranial nerve IX (CN-IX) and mesenchyme around pharyngeal arch arteries (PAA). Subsequently, they populate the aorticopulmonary septum (APS), myocardium (MYO) and interventricular septum (IVS). Separate channels are displayed in Figure 1—figure supplement 1. Light blue indicates known neural crest derivatives. Dark blue reflects newly discovered neural crest derivatives. Scale bars: B, C, E 400 μm; D, F, G, H 100 μm.

https://doi.org/10.7554/eLife.47929.003

Virally infected embryos were then allowed to develop for 1–9 days post injection, cryo-sectioned, and analyzed using confocal microscopy. One day after injection, whole mount imaging revealed RIA-labeled cells migrating in a stream along pharyngeal arch 3 (Figure 1B,B’), that subsequently accumulated in pharyngeal arches 3, 4 and 6 two days after infection (Figure 1C). Next, we confirmed that all labeled cells in the periphery co-localized with the migratory neural crest marker, HNK-1, demonstrating that the neural crest is the only population labeled with H2B-YFP outside the neural tube, thus verifying specificity of infection (Figure 1D, Figure 1—figure supplement 1A). With time, labeled cardiac crest cells were observed in numerous and diverse derivatives, populating the cardiovascular system in a proximal to distal progression (Figure 1E–I, Supplementary file 1a). Consistent with quail-chick chimera, we observed RIA-labeled cells adjacent to and within the walls of pharyngeal arch arteries, in the aorticopulmonary septum, outflow tract, and cardiac cushion. Moreover, we definitively observed YFP-labeled cells in the superior interventricular septum, a site for which the neural crest contribution has been controversial, although ventricular septal defects are common after cardiac neural crest ablation (Kirby et al., 1985). The cells of the outflow tract septum and pharyngeal arch arteries differentiated into smooth muscle actin (SMA) positive cells on embryonic day (E) 5 (Figure 2A,B).

Figure 2 with 1 supplement see all
Cardiac crest-derived cells differentiate into smooth muscle and cardiomyocytes in avian embryos.

(A, B) Retrovirally labeled cardiac crest cells (H2B-YFP, green) that migrate into the outflow tract (A, OFT) and pharyngeal arch arteries (B) express smooth muscle actin (SMA, magenta) marker. (C, D) Labeled cardiac crest cells that enter the ventricle express myocardial marker Troponin T (magenta) (C), and myocardial terminal differentiation marker Myosin Heavy Chain (MHC, magenta) (D) enclosed in dashed line. (E, F) Neural crest-derived cardiomyocytes are not actively dividing or undergoing apoptosis, as demonstrated by phosphohistone H3 staining (PH3, magenta) (E) and Caspase 3 staining (magenta) (F). Transverse view of E6 embryos. Separate channels are displayed in supplement 1. Scale bars: 100 μm.

https://doi.org/10.7554/eLife.47929.005

Importantly, by E3 and onward, virally labeled neural crest cells were observed in the myocardium of both the outflow tract and the ventricles, where they expressed the myocardial markers, Troponin T and Myosin Heavy Chain (Figure 1H, Figure 1—figure supplement 1B, outflow tract; Figure 2C,D, Figure 2—figure supplement 1A,B, ventricles). These neural crest-derived cardiomyocytes were not actively undergoing cell division or programmed cell death (Figure 2E,F), consistent with the stable presence of cells observed over time (Figure 1I, Supplementary file 1a,1b). Supplementary file 1a and 1b present quantification of contributions of virally labeled cells in the chick ventricular myocardium. While previous lineage tracing experiments in zebrafish showed that a stream of neural crest cells integrate into the myocardium of the primitive heart tube to give rise to cardiomyocytes, our results present the first evidence of a homologous neural crest contribution to cardiomyocytes in chick embryos (Sato and Yost, 2003; Li et al., 2003; Cavanaugh et al., 2015).

Lineage analysis in the mouse embryo

To test whether the contribution of cardiac neural crest cells to the myocardium was conserved in mammals, we examined Wnt1-Cre;ZsGreenfl/fl transgenic mice in which neural crest cells were labeled with cytoplasmic GFP (Chai et al., 2000). Embryos were fixed at E15.5 (similar to E7 in chick). Analogous to the results in chick embryos, we observed a large number of ZsGreen-positive myocardial cells in the outflow tract and ventricles, as confirmed by Troponin T expression (Figure 3A–C). To avoid ectopic expression that has been associated with the Wnt1-Cre;ZsGreenfl/fl transgenic line due to endogenous Wnt1 activation caused by in-frame ATG located upstream of Wnt1 start codon, we tested an improved Wnt1 line (Wnt1-Cre2+; R26mTmG mouse line) without ectopic activation of canonical Wnt/β-catenin pathway (Lewis et al., 2013). The results were similar to those observed with the Wnt1-Cre;ZsGreenfl/fl transgenic mice (Figure 3D,E). As in the chick embryos, murine neural crest derived cells were present in the outflow tract, interventricular septum, and myocardium of both ventricles.

Figure 3 with 1 supplement see all
Wnt1-Cre fate mapping in mice confirms the presence of cardiac crest-derived myocardium.

(A) Low magnification image to show the relative anatomical positions of a mouse heart at E15.5 (sagittal view, DAPI-blue). (B, C) In Wnt1-Cre; ZsGreenfl/fl mice, neural crest-derived cells (green, Wnt1-Cre driven ZsGreen expression is abbreviated as Wnt1-ZsGreen, enclosed in dashed line) were observed in myocardium (Troponin T, gray) of the outflow tract (B), and ventricle (VENT) (C, C’’: separate channels of inset C’). (D, E) Similar results were obtained from Wnt1-Cre2+; R26mTmG mice (Wnt1-Cre2+ driven replacement of membrane localized tdTomato (mT) by EGFP (mG) (abbreviated as Wnt1-mtmg), where cardiac crest-derived cells (green, enclosed in dashed line) were present in myocardium of the outflow tract (D) and ventricle (Troponin T, gray) (E, E’’: separate channels of inset E’). Transverse view. Spatial-temporal information and antibody staining are displayed in supplement 1. Scale bars: A 400 μm; B-E 100 μm; C’, E’ 10 μm.

https://doi.org/10.7554/eLife.47929.007

The numbers of neural crest-derived cells appear to decrease with distance along the proximal-to-distal axis (Figure 3—figure supplement 1A), such that no neural crest-derived cardiomyocytes were observed in the apex of the heart (Figure 3—figure supplement 1D,E). As in the chick, the numbers of Wnt1+ cells remain stable with time, and the cells do not appear to undergo active cell division or apoptosis (Figure 3—figure supplement 1B,B’,C,C’). This contribution persists postnatally, as Wnt1+ cells are present at postnatal day 2 (Figure 3—figure supplement 1F–H). These results are consistent with previous studies using less specific P0-cre lines and demonstrate that comparable cardiac crest contributions occur in birds and mammals (Tomita et al., 2005; Tamura et al., 2011). Quantification of numbers of neural crest lineage labeled cells in the trabeculated myocardium of mice reveals that they represent approximately 17% of the population in the proximal half of the ventricle (Supplementary file 1a).

The lineage contributions of neural crest-derived cells in chick and mouse are remarkably similar to those previously shown in zebrafish (Sato and Yost, 2003; Li et al., 2003; Cavanaugh et al., 2015). In all three species, neural crest cells contribute to cardiomyocytes of the trabecular myocardium. This homologous lineage contribution in both amniotes and anamniotes raised the intriguing possibility that neural crest cells may represent a cell population that could contribute to heart repair in adults.

Reactivation of neural crest gene regulatory genes during adult zebrafish heart regeneration

In adult birds and mammals, cardiac injury leads to scarring with little regeneration, whereas heart regeneration is common in amphibians and fish (González-Rosa et al., 2017). For example, adult zebrafish have the capacity to regenerate their hearts after removal of up to 20% of the ventricle. This has been shown to occur by dedifferentiation and proliferation of pre-existing cardiomyocytes (Poss et al., 2002; Jopling et al., 2010). Given that cardiac neural crest cells give rise to a portion of zebrafish cardiomyocytes during development (Sato and Yost, 2003; Li et al., 2003; Cavanaugh et al., 2015) similar to those we report here in chick and mouse, we next asked whether the progeny of these cells might have the ability to contribute to heart regeneration in adult zebrafish.

To address this possibility, we first turned to a transgenic line expressing GFP under the control of a sox10 promoter, Tg(−4.9sox10:eGFP), that labels all embryonic migratory neural crest lineages to address whether neural crest-derived cardiomyocytes reactivated their developmental program upon injury (Carney et al., 2006).

While sox10 is expressed in migrating zebrafish cardiac neural crest cells, it is down-regulated in the embryo shortly after these cells reach the heart (Cavanaugh et al., 2015). We confirmed this in adult hearts, finding that very few cells within the apex of the adult myocardium of control adult fish expressed sox10 one month post-sham injury, in which the body cavity was opened but no resection was made (Figure 4A, Supplementary file 1c, n = 3). However, after surgical removal of ~20% of the ventricular apex, cells in the heart reactivated the sox10 promotor sequence and began to re-express GFP in cardiomyocytes of the trabeculated myocardium near the injured site by 7 days post resection (dpa) (Figure 4A; n = 6). GFP expression was not limited to the regenerating tissue but was also observed in the uninjured part of the ventricle. By 21dpa, the hearts had undergone vast regeneration and morphologically were nearly indistinguishable from controls (Figure 4A; n = 6). Interestingly, consistent with our prediction, the regenerating apex was comprised of more sox10+ positive cells (Figure 4A,B, Supplementary file 1c), suggesting that these cells had proliferated and redeployed a neural crest gene regulatory program during the heart regeneration process.

Figure 4 with 2 supplements see all
Cardiac neural crest contributes to heart regeneration in zebrafish.

(A) In sham-operated adult zebrafish hearts from a transgenic line expressing GFP under the control of a sox10 promoter, very few cells expressed Tg(−4.9sox10:eGFP) (green) (n = 3). 7 days-post amputation (dpa), the sox10 promoter was reactivated as shown by GFP+ cells in the trabeculated myocardium near the site of injury (Tg(−4.9sox10:eGFP), green) (n = 6). 21dpa, when the resected apex regenerated, more GFP+ cells were observed in sagittal sections within and surrounding the site of injury (n = 6). Sections in A are counterstained with DAPI in blue and Myosin Heavy Chain in red. (B) Endogenous sox10 mRNA expression was observed by paraffin section in situ hybridization in uninjured, 7dpa, and 21dpa hearts. Arrows denote cells with sox10 expression. From these results, we conclude that sox10 is reactivated after injury. (C) Along with sox10, expression of neural crest marker, tfap2a, was also enriched after injury. Arrows label areas of expression in the myocardium. (D) Differential gene expression analysis of FACS-sorted Tg(sox10:mRFP)+ and FACS-sorted Tg(sox10:mRFP)- transcriptomes show n = 1093 genes are enriched at 21dpa in the sox10+ cells compared to the rest of the ventricular tissue (n = 12 ventricles per replicate). Zebrafish neural crest genes as determined by GO analysis are highlighted on the volcano plot. (E) Upregulation of neural crest gene regulatory network genes was also observed from our differential expression analysis (**p<0.05, ***p<0.001). Co-localization of sox10 mRNA expression with Tg(sox10:GAL4-UAS-Cre;ubi:Switch)+ neural crest-derived cardiomyocytes is presented in supplement 1; schematic diagram of experimental design for obtaining the regenerating neural crest transcriptome and further analysis of gene enrichments is presented in supplement 2. Scale bars: 100 μm.

https://doi.org/10.7554/eLife.47929.009

To test if sox10 and other bona fide neural crest markers such as tfap2a, were upregulated endogenously, we performed in situ hybridization on paraffin sections of regenerating and uninjured ventricles. The results reveal upregulation of expression of sox10 and tfap2a transcripts after injury, whereas they were mostly absent from uninjured ventricles (Figure 4B,C). Furthermore, we observed co-localization of sox10 transcripts with a Tg(sox10:GAL4-UAS-Cre;ubi-Switch), which permanently labels all sox10-derived lineages with mCherry (Figure 4—figure supplement 1, n = 2). The Tg(sox10:GAL4-UAS-Cre;ubi-Switch) is a double transgenic line for the sox10:GAL4-UAS-Cre transgene and the ubi:Switch reporter in which the sox10 promoter drives expression of Cre recombinase. Upon activation of sox10 expression in neural crest cells, eGFP is excised and so cells of the sox10 lineage are permanently labeled with mCherry (Cavanaugh et al., 2015). All cells expressing sox10+ transcripts also had mCherry, though not all mCherry positive cells were sox10+ at the 7 day time point (Figure 4—figure supplement 1, insets 1 and 2). Our results are consistent with recent findings from Abdul-Wajid and colleagues, who observed that ablation of the embryonic neural crest yields few or no sox10+ cells in the adult heart and results in severe heart defects (Abdul-Wajid et al., 2018). This suggests there are no subsequent post-embryonic neural crest additions to the heart and that the population we observe re-expressing neural crest genes are embryonic-derived neural crest progeny.

These results raise the intriguing possibility that the neural crest developmental gene regulatory network was being redeployed in neural crest-derived cells of the heart during regeneration. To test this, we performed transcriptional profiling of sox10:mRFP+ cells in the regenerating zebrafish hearts at 21dpa. To this end, we dissected and dissociated injured ventricles (n = 12 per replicate) into single cell suspensions and performed FAC-sorting of sox10:mRFP+ cells (Figure 4—figure supplement 2A). The results were compared with mRFP negative cardiac cells from the same injured, isolated ventricles. This led to the identification of 1093 genes that are significantly enriched (p-adj <0.05) in regenerating sox10+ cells compared to sox10- cells of the same injured ventricles (Figure 4D, Figure 4—figure supplement 2). We then compared the differentially expressed genes of isolated 21dpa sox10+ cells to: 1) our recently published chick developmental cardiac neural crest gene regulatory program, 2) known zebrafish neural crest genes, and 3) core neural crest gene regulatory network genes expressed at all axial levels (Tani-Matsuhana et al., 2018; Martik and Bronner, 2017; Lukoseviciute et al., 2018). The results revealed upregulation of many genes of the embryonic neural crest gene regulatory network at the time of regeneration (Figure 4D and E).

Interestingly, numerous genes known to be responsible for cardiomyocyte proliferation also are expressed in sox10+ cells upon heart injury (Figure 4—figure supplement 2E) (González-Rosa et al., 2017). The co-expression of these genes as well as an upregulation of a cell proliferation gene signature suggests a role for sox10-derived cells in cardiomyocyte proliferation during regeneration (Figure 4—figure supplement 2C). Furthermore, these results suggest that the population of proliferating cardiomyocytes in the regenerating heart is heterogeneous and comprised of both neural crest- and mesoderm-derived cardiomyocytes (González-Rosa et al., 2017; Sánchez-Iranzo et al., 2018; Schindler et al., 2014; Kikuchi et al., 2010).

Discussion

While much attention has been paid to the molecular signals that promote myocardial dedifferentiation and proliferation during regeneration, far less is known about the cell lineages that contribute to the regeneration process. Based on our observation on the lineage relationship between cardiac neural crest cells and cardiomyocytes during development, we propose that neural crest-derived cells (progenitors and/or pre-existing cardiomyocytes) may represent a key population that proliferates and differentiates into new cardiomyocytes after injury.

Our cell lineage labeling results provide direct evidence for a neural crest contribution to the undamaged myocardium of the amniote heart. Furthermore, consistent with previous lineage tracing experiments in zebrafish (Cavanaugh et al., 2015), where a proportion of cardiac crest derived-cells were located in the trabeculated myocardium in adult fish, we show that after injury, there is activation of numerous neural crest gene regulatory transcription factors and other neural crest genes during regeneration (Figure 4). While the underlying gene regulatory network of neural crest cells is responsible for formation of cardiomyocytes during normal development, we speculate that it also does so in a similar manner upon injury by redeploying sox10 and other neural crest gene regulatory network genes. The finding that sox10-derived cells are primarily in the proximal trabecular myocardium of the zebrafish heart suggests that these cells must be migrating into to the wound site after injury. Of course, we cannot rule out the possibility that the cells that reactivate sox10 and the neural crest program may come from another adult lineage. But in the adult, their molecular signature strongly correlates with that of embryonic neural crest cells (Figure 4D and E). Whereas our data clearly show that the sox10+ cells contribute to cardiomyocytes (Figure 4A), whether they also might contribute to other lineages (e.g. hematopoietic cells) within the regenerated tissue remains to be explored.

Why was the contribution of neural crest cells to cardiomyocytes in amniotes previously missed? Interspecific quail-chick chimera are generated via transplantation of donor tissue into the host, which requires time to heal (Kirby et al., 1983). If the neural crest cells that migrate to the ventricles are the earliest migrating cells, this population may have been delayed after grafting due to wound healing and hence unable to migrate as far. Alternatively, the labeled cells may have been missed since it can be challenging to identify a small population of dispersed quail cells amongst many more numerous chick cells. Furthermore, cell behavior might be altered when transplanted quail cells are introduced into a chick environment. Our retroviral lineage labeling circumvents these issues by indelibly labeling an endogenous neural crest population without the need for grafting. Moreover, the labeled cells are easily detectable due to their fluorescent readout. For lineage labeling in mice, there were hints in the literature regarding a possible neural crest contribution to cardiomyocytes. However, the experiments were either indirect or used lineage tracing techniques that were not specific to the neural crest. For example, Tomita et al. showed that cells isolated from ‘cardiospheres’ can behave like neural crest cells when injected into chick embryos (Tomita et al., 2005). In addition, lineage analysis in mouse using a P0-cre line revealed EGFP-positive cells in the myocardium that gather at the ischemic border upon injury (Tomita et al., 2005). However, P0 is not a neural crest specific marker, making these results inconclusive at the time. In contrast, Wnt1 is the ‘gold standard’ for neural crest labeling and the improved Wnt1 line (Wnt1-Cre2+; R26mTmG) corrects possible ectopic expression problematic in the original Wnt1-Cre;ZsGreen line (Jiang et al., 2000; Chai et al., 2000; Lewis et al., 2013).

In chick and mouse, neural crest-derived cells comprise a significant portion (~17%) of the trabeculated myocardium in the proximal part of both ventricles. Interestingly, this percentage is similar to what has been reported in zebrafish (Cavanaugh et al., 2015; Abdul-Wajid et al., 2018). In amniotes, we find that the density of the cells decreases along the proximal-distal axis and appears to be stable through time (Figure 1I, Supplementary file 1a,b). The presence of neural crest-derived cardiomyocytes across vertebrates and the redeployment of a sox10+ cell population in zebrafish heart regeneration suggest that the neural crest-derived myocardium might also play a role in heart regeneration in neonatal mice, which requires further testing.

In summary, the present results show, for the first time, the common ability of cardiac neural crest cells across diverse vertebrates to contribute to heart muscle. Moreover, these cells appear to be critical for cardiac regeneration in zebrafish. If the results extrapolate to other species, the mechanisms that control the normal development of the neural crest into cardiomyocytes may be harnessed to stimulate these cells to proliferate and regenerate new cardiomyocytes, thus offering potential therapeutic approaches to repair heart damage in mammals including humans.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or referenceIdentifiersAdditional
information
Genetic reagent (Mus musculus)Wnt1-Cre; ZsGreenfl/flPMID:10725243Jackson Laboratories, Stock# 003829Drs. Xia Han and Yang Chai at University of Southern California, Center for Craniofacial Molecular Biology
Genetic reagent (Mus musculus)Wnt1-Cre2+((129S4-Tg(Wnt1-cre)1Sor/J));R26mTmGPMID: 23648512Jackson Laboratory, Stock# 22137Dr. Jeffrey Bush at
University of California, San Francisco
Genetic reagent (Danio rerio)Tg(−4.9sox10:eGFP)PMID: 17065232ZFIN ID: ZDB-TGCONSTRCT-070117–69
Genetic reagent
(Danio rerio)
Tg(sox10:GAL4-UAS-Cre;ubi-Switch)PMID: 26086691Drs. Ann M. Cavanaugh and Jau-Nian Chen at
Department of Molecular, Cell and Developmental biology, University of California,
Los Angeles
Genetic reagent (Danio rerio)Tg(sox10:mRFP)PMID: 18176560ZFIN ID: ZDB-TGCONSTRCT-080321–2
Cell line (Galllus gallus DF1)UMNSAH/DF-1 fibroblast spontaneously transformedATCC#CRL-12203, Lot number 62712171; RRID:CVCL_0570
Recombinant DNA reagentRES-H2B-YFP-DDAddgeneRRID:Addgene_96893
AntibodyMouse monoclonal anti-bovine Troponin T, IgG2a (CT3)DSHBRRID:AB_528495Dilution (1:10)
AntibodyMouse monoclonal anti-chicken Myosin Heavy Chain, IgG1 kappa light chain (ALD58)DSHBRRID:AB_528361Dilution (1:10)
AntibodyMouse monoclonal anti-
chicken Myosin Heavy Chain, IgG1 kappa light chain (F59)
DSHBRRID:AB_528373Dilution (1:10)
AntibodyMouse monoclonal anti-NH2 terminal synthetic decapeptide of alphasmooth muscle actin,
IgG2a
Sigma# A5228Dilution (1:500)
AntibodyMouse monoclonal anti-human phospho-histone H3, IgG1Abcam#ab14955Dilution (1:500)
AntibodyRabbit polyclonal anti-human Caspase 3, IgGR and D systems#AF835Dilution (1:500)
AntibodyGoat polyclonal anti -GFP, IgGAbcam#ab6673Dilution (1:500)
AntibodyGoat polyclonal anti-mouse IgG2a
Alexa-568
Molecular ProbesRRID:AB_2535773Dilution (1:1000)
AntibodyGoat polyclonal anti-mouse IgG1
Alexa-568
Molecular ProbesRRID:AB_2535766Dilution (1:1000)
AntibodyGoat polyclonal anti-rabbit IgG Alexa-568Molecular ProbesRRID:AB_2534121Dilution (1:1000)
AntibodyDonkey polyclonal anti-goat IgG Alexa-488Molecular ProbesRRID:AB_2534102Dilution (1:1000)
Software, algorithmImage processing software FIJIhttps://imagej.net/Fiji
Software, algorithmR v3.6.1https://www.r-project.org/
Software, algorithmDESeq2PMID: 25516281RRID:SCR_015687
Software, algorithmBowtie2PMID: 22388286RRID:SCR_005476
Software, algorithmfeatureCounts (Subread)PMID: 24227677RRID:SCR_009803
OtherAccumaxInnovative Cell Technologies, Inc#AM105
Commercial assay or kitSMART-seq Ultra Low Input RNA Kit V4Takara Clontech#634891

Cell culture and retrovirus preparation

Request a detailed protocol

Using a standard transfection protocol, chick DF1 cells (ATCC, Manassas, VA; #CRL-12203, Lot number 62712171, RRID:CVCL_0570, Certificate of Analysis with negative mycoplasma testing at the ATCC website) were transfected with RIA-H2B-YFP plasmid (RRID:Addgene_96893) and ENV-A plasmid in 15 cm culture dishes. Cell culture medium was collected 24 hr post-transfection, and twice per day for four days, then centrifuged at 26,000 rpm for 1.5 hr. The supernatant was dried with aspiration, and the pellet was dissolved in 20–30 μl of DMEM to a final titer of 1 × 107 ifu/mL. Viral aliquots were stored in −80°C until the time of injection.

Chick embryo processing and viral injection

Request a detailed protocol

Viral stock was diluted 1:2 with Ringer’s solution (0.9% NaCl, 0.042%KCl, 0.016%CaCl2 • 2H2O wt/vol, pH7.0) to generate the working solution, which was mixed with 0.3 μl of 2% food dye (Spectral Colors, Food Blue 002, C.A.S# 3844-45-9) as indicator. The lumen of the neural tube adjacent to the middle of the otic vesicle to the level of somite three was injected with 0.2 μl of working in HH8-10 chicken embryos. Embryos were sealed with surgical tap and incubated at 37°C for 1–9 days, harvested at HH14 (n = 5), HH18 (n = 5), HH21 (n = 4), HH25 (n = 4), HH28 (n = 12), HH32 (n = 4) and E10 (n = 4). At the time of harvesting, chick embryos were dissected, fixed in 4%PFA in PBS for 30 mins at 4°C, then embedded in gelatin and sectioned (Microm HM550 cryostat).

Wnt1-Cre mouse and tissue preparation

Request a detailed protocol

The Wnt1-Cre; ZsGreenfl/fl mice described in Chai et al. (2000) (gift from Drs. Xia Han and Yang Chai at University of Southern California, Center for Craniofacial Molecular Biology) were harvested and fixed at E15.5 (n = 8) and P2 (n = 2). The hearts were dissected, fixed in 4%PFA in PBS for 30mins at 4°C. E15.5 Wnt1-Cre2+; R26mTmG mice (Lewis et al., 2013) (129S4-Tg(Wnt1-cre)1Sor/J, gift from Dr. Jeffrey Bush at University of California, San Francisco, n = 3) were fixed with 4% PFA overnight before dissection. The hearts were embedded in gelatin, and sectioned.

Quantification of neural crest contributions to the ventricular myocardium and regeneration

View detailed protocol

To quantify RIA-labeled cells in chick embryos, three consecutive sections of the same axial level were imaged per embryo. The number of YFP-positive cells was averaged to account for variability due to sampling. n = 4–6 embryos were analyzed at each stage as biological replicates. The results are presented as presence or absence of virally labeled cardiac neural crest derivatives at different anatomical locations in Figure 1I and as numerical values in Supplementary file 1a, 1b. To quantify Wnt1-Zsgreen+ cells in E15.5 mouse heart, three consecutive sections of the same axial level were imaged per embryo (n = 4). Automated particle analysis was conducted with FIJI program to estimate the total number of Zsgreen+ cells in the image. For the percentage of neural crest-derived cells in the ventricle, the same procedure was performed with the DAPI channel which represents total cell population. % Zsgreen/DAPI was calculated, and averaged to the result presented in the text of Supplementary file 1a. Same analysis was conducted to estimate the number of sox10:eGFP+ cells in 7dpa (n = 3), 21dpa (n = 3) and sham operated (n = 3) hearts in an area of 2 × 105 μm2 at the apex. One section per heart at the middle of the apex was quantified and presented in Supplementary file 1c.

Zebrafish cardiac injury and tissue collection

Request a detailed protocol

Adult zebrafish heart resection was conducted with the Tg(−4.9sox10:eGFP) or Tg(sox10:mRFP) line, according to published protocols (Poss et al., 2002). Resected and sham operated fish hearts (n = 24) were collected at 7 days post injury (dpi) (n = 18), and 21 dpi (n = 53) at which time the fish were euthanized and the hearts were removed for further analysis. The hearts were fixed in 4%PFA in PBS for overnight at 4°C prior to processing for staining. Adult zebrafish were maintained in the Beckman Institute Zebrafish Facility at Caltech, and all animal and embryo work were completed in compliance with California Institute of Technology Institutional Animal Care and Use Committee (IACUC) protocol 1764.

Immunohistochemistry and image analysis

Request a detailed protocol

After cryosectioning, slides were incubated in 1xPBS at 42°C to remove gelatin. 0.3% vol/vol Triton-X100in 1xPBS was used to permeabilize the tissue. Sections were incubated with primary antibody underneath a parafilm layer at 4°C overnight (primary antibody dilutions: 1:10 Troponin T CT3, DSHB (RRID:AB_528495); 1:10 Myosin Heavy Chain ALD58, DSHB (RRID:AB_528361); 1:10 Myosin Heavy Chain F59, DSHB (RRID:AB_528373); 1:500 Mouse anti-smooth muscle actin, Sigma-Cat# A5228-200uG; 1:500 Mouse anti phospho-histone H3, Abcam-ab14955; 1:500 rabbit anti caspase-3,R and D Systems # AF835; 1:500 goat anti GFP, Abcam Cat#ab6673, all in blocking reagent 1xPBS with: 5% vol/vol normal donkey serum, 0.3% vol/vol Triton-X100). Subsequently, sections were washed for 3 times with 1xPBS, incubated with secondary antibody for 40 mins at room temperature and counterstained with DAPI. Secondary antibodies include: Goat anti-mouse IgG2a Alexa-568 (RRID:AB_2535773), Goat anti-mouse IgG1 Alexa-568(RRID:AB_2535766), Goat anti-rabbit IgG Alexa-568(RRID:AB_2534121), Donkey anti-goat IgG Alexa-488(RRID:AB_2534102); 1:1000, Molecular Probes. Zeiss AxioImager.M2 with Apotome.2 and Zeiss LSM 800 confocal microscope were utilized for imaging. Images were cropped, rotated, and intensity was linearly adjusted for visualization.

In situ hybridization of adult zebrafish hearts

Request a detailed protocol

After fixation, hearts were embedded in paraffin and sections were prepared at 10 µm thickness on a Zeiss microtome. After paraffin removal with histosol, sections were washed and then hybridized with 1 ng/µl anti-sense digoxygenin-labeled probes overnight at 70°C in a humidifying chamber. After hybridization, sections were washed with 50% formamide/50% 1X SSCT buffer followed by washes with MABT and a blocking step in 1% Roche blocking reagent. Sections were then incubated overnight at room temperature with a 1:2000 dilution of anti-DIG-Alkaline Phosphatase antibody (Roche). After several washes with MABT, chromogenic color was developed using NBT/BCIP precipitation (Roche).

Transcriptome analysis of regenerating neural crest cells

Request a detailed protocol

For each replicate (n = 2), regenerating ventricles (n = 12) were isolated at 21 days post injury and dissociated into a single cell suspension using a pestle-A tissue homogenizer followed by incubation in Accumax (Innovative Cell Technologies, Inc) at 30°C. sox10-mRFP-positive and sox10-mRFP-negative cells were collected by FAC-sorting on a BD Biosciences FACSAriaFusion Cell Sorter. cDNA from mRFP-positive and negative cells was prepared using SMART-seq Ultra Low Input RNA Kit V4 (Takara) according to the manufacturer’s protocol. Sequencing libraries were built according to Illumina Standard Protocols and sequenced using an Illumina HiSeq2500 sequencer at the Millard and Muriel Jacobs Genetics and Genomics Laboratory (California Institute of Technology, Pasadena, CA). 50 million, 50 bp, single-ended reads from two biological replicates were mapped to the zebrafish genome (GRCz10) using Bowtie2 (Langmead and Salzberg, 2012). Transcript counts were calculated using featureCounts (Subread) and differential gene expression analysis was performed using DESeq2 (Liao et al., 2014Love et al., 2014). Protein classification analysis was performed using PANTHER (Mi et al., 2019). Heatmaps of normalized counts were generated using Heatmap2. Databases have been deposited to NCBI (BioProject # PRJNA526570).

Data availability

All data is available in the main text, the supplementary materials. Databases have been deposited to NCBI (BioProject # PRJNA526570).

The following data sets were generated
    1. Weiyi Tang
    2. Megan L Martik
    3. Yuwei Li
    4. Marianne E Bronner
    (2019) BioProject/SRA
    ID PRJNA526570. Cardiac neural crest contributes to cardiomyocytes in amniotes and heart regeneration in zebrafish.

References

    1. Chai Y
    2. Jiang X
    3. Ito Y
    4. Bringas P
    5. Han J
    6. Rowitch DH
    7. Soriano P
    8. McMahon AP
    9. Sucov HM
    (2000)
    Fate of the mammalian cranial neural crest during tooth and mandibular morphogenesis
    Development 127:1671–1679.
    1. Jiang X
    2. Rowitch DH
    3. Soriano P
    4. McMahon AP
    5. Sucov HM
    (2000)
    Fate of the mammalian cardiac neural crest
    Development 127:1607–1616.

Decision letter

  1. Didier Y Stainier
    Senior and Reviewing Editor; Max Planck Institute for Heart and Lung Research, Germany

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Cardiac neural crest contributes to cardiomyocytes in amniotes and heart regeneration in zebrafish" for consideration by eLife. Your article has been reviewed by Didier Stainier as the Senior Editor and Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As noted by all three reviewers, the manuscript addresses an important issue, that of neural crest contribution to the myocardium, and provides interesting and significant new findings. While some experiments were suggested that would strengthen the paper, the consensus from the discussions was that they were not essential.

However, as detailed in reviewer #2's comments, the interpretation of some of the zebrafish data needs to be adjusted.

I have pasted below the full reviews as all the comments should be taken into account when revising the manuscript.

Reviewer #1:

The current manuscript reports interesting findings on the role of the cardiac neural crest cells to the development of the heart in amniotes. Using a series of elegant lineage tracing analysis in chicken and mouse embryos Bronner and colleagues describe a previously unknown contribution of the cranial neural crest cells to cardiomyocytes of the outflow tract and ventricle in vertebrates. The analyses are well performed and provide interesting data. Remarkably, the authors also report in a model with heart regenerative capacities, the adult zebrafish, a reactivation of embryonic programs related to the neural crest development upon a cardiac injury (both by in situ hybridization and transcriptome analysis). Then they speculate that the cranial neural crest-derived cardiomyocytes of the ventricles could contribute to the regenerative process in the injured adult heart. While these speculations are certainly relevant, the manuscript would benefit from further analysis to support this proposal.

Essential revisions:

- Since both proliferation and trans-differentiation are prominent processes during heart regeneration, it is relevant to assess whether there is also reactivation of a proliferation gene signature (beyond gata4) and early cardiac transcriptional regulators such as Nkx2.5, Mef2 and Tbx5 in the neural crest-derived cardiomyocytes in response to injury.

- Can the authors comment on why have they decided to do the transcriptomic analysis at 21 dpi instead of 7 dpi, which is the early onset of reactivation of the developmental neural crest program?

- In order to rule out an epicardial origin (this cell population is also reactivated upon cardiac injury and re-expresses gene developmental programs (Cahill et al., 2017)), it is also relevant to determine whether or not the neural crest-derived cardiomyocytes express bona fide epicardial markers such as Tbx18 or Tcf21.

- Given that the neonatal mammalian heart is also capable of regeneration after various injury models (apical resection and coronary ligation injuries, see Porrello et al., 2011), assessing the reactivation of neural crest gene regulatory programs in mice would further support the most exciting conclusions of this study.

Reviewer #2:

In this manuscript, Tang, Martik, and colleagues examine the contributions of the cardiac neural crest to the myocardium. This is an important and controversial topic with substantial relevance to the causes of congenital heart disease and to the evolution of neural crest derivatives, as well as potential relevance to mechanisms underlying tissue regeneration. It is therefore valuable that the authors provide new lineage analysis data sets that document the contribution of some cardiac neural crest cells to the myocardium in both chick and mouse. This work is carefully performed and interpreted and provides new information of interest to the field. In addition, the authors demonstrate that sox10, as well as other genes associated with the migratory neural crest, are upregulated in the myocardium of the regenerating zebrafish heart, following injury. These new gene expression data add to our understanding of how the adult zebrafish heart responds to injury, and the authors are interested in drawing a connection between these observations and the contributions of cardiac neural crest-derived cells to regeneration. These data are also valuable to those seeking to understand the regulation of cardiac regeneration, but would benefit from some interpretative adjustments. Specifically:

1) The induction of expression of sox10 and other neural crest-associated genes is very interesting, but it is inappropriate to interpret this as evidence that cardiac neural crest cells contribute to the regenerative capacity of the zebrafish heart without evidence that the cells activating expression of these genes are indeed neural crest-derived. The injury stimulus may induce expression of sox10 et al. in cardiomyocytes that were not originally neural crest-derived, and a different type of lineage tracing experiment would be required to determine whether neural crest-derived cardiomyocytes indeed respond to injury differently than lateral mesoderm-derived cardiomyocytes do. Alternatively, the authors could amend their interpretation – these changes in gene expression are certainly interesting whether or not the responding cells are necessarily neural crest-derived. (Saying that the sox10:GAL4-UAS-Cre line permanently labels all neural crest lineages is an overstatement – it permanently marks sox10-expressing cells, but there may be instances of sox10 expression outside of the neural crest.)

2) Along the same lines, the data shown here associate the induction of sox10 expression (and the expression of other neural crest-associated genes) with regeneration but do not directly demonstrate the significance of these genes to the regenerative response. Demonstrating their significance would surely require substantial additional experimentation; here, the authors should simply take caution to avoid implying that these particular genes are associated with regulation of regeneration.

3) Could the authors comment on the relationship of the differential gene expression that they document in Figure 4H with other published datasets that examine the transcriptomic changes triggered by injury? Is it the case that the same neural crest-associated genes are differentially expressed in those datasets (but haven't been highlighted in the analysis done thus far)?

Reviewer #3:

This is an interesting paper showing additional contributions of the neural crest to parts of the heart across vertebrates. Moreover, the authors present evidence that during regeneration of the zebrafish heart, neural crest-derived heart cells contribute to the regenerate and redeploy a neural crest gene expression programme. This suggests previously unappreciated heterogeneity in the cells contributing to regeneration.

Overall, the findings are important, but the manuscript could be substantially improved by more explanations and quantifications of data.

1) The manuscript has two parts and it is not entirely clear how these are connected. In the first part, the authors use lineage tracing to find new and known populations of crest cells that contribute to heart development. These cells seem to be rare in the apex "…"no neural crest-derived cardiomyocytes were observed in the apex of the heart"…"). However, in the second part of the manuscript they remove the apex for regeneration purposes. How are the two parts of the manuscript connected?

2) The authors say that they have found novel crest-derived cell populations in the heart. At the same time, they state that previous research was controversial, and tools were inadequate due to high background. It is unclear whether previous papers already described these populations and this is here confirmed with more robust tools, or whether these populations are entirely new. Moreover, it would be good to at least discuss in more detail how they know that their tools are superior in their hands (e.g. "the improved Wnt1 line").

3) Many results rely on scoring for the presence or absence of cells. How were tissues scored? More quantifications of the relative contribution to the tissue in question should be shown for lineage tracing and zf heart regeneration, such that the relative importance of the crest contribution can be appreciated.

4) Figure 4—figure supplement 1: Were there also cells that expressed sox10 mRNA de novo? The prediction would be that only lineage-traced cells re-express sox10 – otherwise non-neural crest derived cells contribute to the population of sox10+ cells in regeneration. Can this be clearly observed in the combination of colorimetric ISH and transgenic fluorescence?

5) More information is needed on how the "the cardiac neural crest migratory gene regulatory module" was identified in the expression profile of regenerated heart cells in zf. e.g. did they use GO analysis? They highlight 20 out of 656 genes – what are the other genes? Why is 21 days post-injury the best time point?

https://doi.org/10.7554/eLife.47929.017

Author response

Reviewer #1:

[…] Remarkably, the authors also report in a model with heart regenerative capacities, the adult zebrafish, a reactivation of embryonic programs related to the neural crest development upon a cardiac injury (both by in situ hybridization and transcriptome analysis). Then they speculate that the cranial neural crest-derived cardiomyocytes of the ventricles could contribute to the regenerative process in the injured adult heart. While these speculations are certainly relevant, the manuscript would benefit from further analysis to support this proposal.

Essential revisions:

- Since both proliferation and trans-differentiation are prominent processes during heart regeneration, it is relevant to assess whether there is also reactivation of a proliferation gene signature (beyond gata4) and early cardiac transcriptional regulators such as Nkx2.5, Mef2 and Tbx5 in the neural crest-derived cardiomyocytes in response to injury.

Thank you for this suggestion. We now include information about genes associated with cell proliferation as well as early cardiac transcription factors, including Nkx2.5, Gata4 and Tbx5 in our dataset. From our RNA-seq data, we do not see enrichment of these factors compared to injured, sox10-negative cells. While not enriched, these genes are expressed at low levels, perhaps because of the time point examined.

To address the reviewer’s comment, we have included a heatmap of normalized counts in our supplemental data (Figure 4—figure supplement 2) to show enrichment of known regeneration genes in our 21dpa dataset. As suggested by the reviewer, the RNAseq data reveal expression of cell cycle and proliferation genes, consistent with an important role for proliferation of sox10+ cells. We now include data on enrichment of cell cycle genes in Figure 4—figure supplement 2.

- Can the authors comment on why have they decided to do the transcriptomic analysis at 21 dpi instead of 7 dpi, which is the early onset of reactivation of the developmental neural crest program?

We selected the 21 day time point for practical reasons, since we noted a significant number of Sox10-expressing cells in the heart at this time after regeneration as opposed to fewer cells at 7 days. We validated gene expression by in situ hybridization at both time points but focused the RNA-seq to the later time point. Moreover, at 21 days, we see the full progression of neural crest gene regulatory states from induction to migration that may have been missed at an earlier time point. However, the reviewer raises an excellent point and we will try to profile more time points for future experiments.

- In order to rule out an epicardial origin (this cell population is also reactivated upon cardiac injury and re-expresses gene developmental programs (Cahill et al., 2017)), it is also relevant to determine whether or not the neural crest-derived cardiomyocytes express bona fide epicardial markers such as Tbx18 or Tcf21.

Thank you for this suggestion. To address this possibility, we have looked through our datasets for early cardiac transcription factors. The results show epicardial markers such as Tbx18 or Tcf21 are not enriched in our Sox10+ cells compared to Sox10- cells in the regenerating heart.

- Given that the neonatal mammalian heart is also capable of regeneration after various injury models (apical resection and coronary ligation injuries, see Porrello et al., 2011), assessing the reactivation of neural crest gene regulatory programs in mice would further support the most exciting conclusions of this study.

We agree with the reviewer that it would be logical to extend this to neonatal mice. As this will require learning a new injury model, we would prefer to pursue this in a future publication, but have added this as a discussion point to be addressed in the future.

Reviewer #2:

[…] These new gene expression data add to our understanding of how the adult zebrafish heart responds to injury, and the authors are interested in drawing a connection between these observations and the contributions of cardiac neural crest-derived cells to regeneration. These data are also valuable to those seeking to understand the regulation of cardiac regeneration, but would benefit from some interpretative adjustments. Specifically:

1) The induction of expression of sox10 and other neural crest-associated genes is very interesting, but it is inappropriate to interpret this as evidence that cardiac neural crest cells contribute to the regenerative capacity of the zebrafish heart without evidence that the cells activating expression of these genes are indeed neural crest-derived. The injury stimulus may induce expression of sox10 et al. in cardiomyocytes that were not originally neural crest-derived, and a different type of lineage tracing experiment would be required to determine whether neural crest-derived cardiomyocytes indeed respond to injury differently than lateral mesoderm-derived cardiomyocytes do. Alternatively, the authors could amend their interpretation – these changes in gene expression are certainly interesting whether or not the responding cells are necessarily neural crest-derived. (Saying that the sox10:GAL4-UAS-Cre line permanently labels all neural crest lineages is an overstatement – it permanently marks sox10-expressing cells, but there may be instances of sox10 expression outside of the neural crest.)

In the revised manuscript, we raise the caveat that we cannot rule out the possibility that these cells are reactivating Sox10 and the neural crest program but may come from another lineage.

In support of a neural crest origin, Abdul-Wajid, 2018 have also shown that ablation of the embryonic neural crest yields few or no sox10+ cells in the adult heart. This suggests there are no post-embryonic neural crest additions to the heart. This is now discussed.

On a personal note, I find it very unlikely that the Sox10+ cells would come from another lineage. To my knowledge, there are no non-neural crest-derived Sox10-expressing cells in the periphery. There are a few Sox10+ non-neural crest-derived cells in the CNS, but they are confined to the brain. Therefore, I strongly suspect that these are indeed neural crest-derived. That said, I agree that this would need to be more rigorously validated and therefore we are more circumspect in the conclusions drawn in the manuscript.

2) Along the same lines, the data shown here associate the induction of sox10 expression (and the expression of other neural crest-associated genes) with regeneration but do not directly demonstrate the significance of these genes to the regenerative response. Demonstrating their significance would surely require substantial additional experimentation; here, the authors should simply take caution to avoid implying that these particular genes are associated with regulation of regeneration.

We agree that a much more detailed analysis would be necessary to directly demonstrate mechanism. We have amended the manuscript to make this point and be more circumspect.

3) Could the authors comment on the relationship of the differential gene expression that they document in Figure 4H with other published datasets that examine the transcriptomic changes triggered by injury? Is it the case that the same neural crest-associated genes are differentially expressed in those datasets (but haven't been highlighted in the analysis done thus far)?

By investigating the expression of our sox10+ enriched genes on RegenDbase.org (an online database of both published and unpublished regeneration transcriptome analyses), we find many of our enriched genes are present at 21dpa, but not statistically significant when compared to the 0dpa vs 21dpa dataset.

Reviewer #3:

This is an interesting paper showing additional contributions of the neural crest to parts of the heart across vertebrates. Moreover, the authors present evidence that during regeneration of the zebrafish heart, neural crest-derived heart cells contribute to the regenerate and redeploy a neural crest gene expression programme. This suggests previously unappreciated heterogeneity in the cells contributing to regeneration.

Overall, the findings are important, but the manuscript could be substantially improved by more explanations and quantifications of data.

1) The manuscript has two parts and it is not entirely clear how these are connected. In the first part, the authors use lineage tracing to find new and known populations of crest cells that contribute to heart development. These cells seem to be rare in the apex "…"no neural crest-derived cardiomyocytes were observed in the apex of the heart"…"). However, in the second part of the manuscript they remove the apex for regeneration purposes. How are the two parts of the manuscript connected?

Thank you for pointing out this issue. We have tried to integrate the two parts of the manuscript better. Basically, given that neural crest cells are a well-known stem cell population, the finding that the they can contribute to cardiomyocytes in amniotes similar to what had been previously shown in anamniotes made us wonder if they might contribute to heart regeneration. Zebrafish seemed like the most appropriate model for this due to their extensive capacity for heart regeneration.

To improve the connection between the two parts of the paper, we now more thoroughly reference and discuss several previously published papers that use the sox10 promoter for labeling cardiac neural crest cells (Carney et al., 2006; Cavanaugh et al., 2015; Abdul-Wajid et al., 2018). These studies show that in zebrafish, the neural crest contribution to the ventricles is primarily to the trabecular myocardium, similar to what we have found in chick and mouse. The study by Cavanaugh and colleagues also shows that sox10 is downregulated in the heart as development proceeds (Cavanaugh et al., 2015). We confirm this by showing minimal sox10 expression in control adult hearts.

The finding that sox10-derived cells are primarily in the proximal trabecular myocardium of the zebrafish heart suggests that these cells must be migrating into to the wound site after injury. We now discuss this interesting possibility more clearly.

2) The authors say that they have found novel crest-derived cell populations in the heart. At the same time, they state that previous research was controversial, and tools were inadequate due to high background. It is unclear whether previous papers already described these populations and this is here confirmed with more robust tools, or whether these populations are entirely new. Moreover, it would be good to at least discuss in more detail how they know that their tools are superior in their hands (e.g. "the improved Wnt1 line").

Thank you for pointing out these issues. We have added further discussion of why the contribution to cardiomyocytes may have been missed in birds and previous hints from the literature of a neural crest contribution to the mammalian heart. In birds, this population was totally missed. There is no mention of a contribution of neural crest cells to cardiomyocytes in any of the quail/chick chimera papers. In mice, the lineage tracing was done with P0 mice which is not a neural crest specific marker, but rather labels Schwann cells and notochord cells, making these results inconclusive at the time. In fact, Fukada and colleagues recently published a paper (Fukada et al., 2017) describing the differences between P0 and Wnt1 lineage labeling in mice and described Wnt1 as the “gold standard” for neural crest labeling.

In our study, we use two different Wnt1 lines to examine the contribution of neural crest cells to the murine heart. The reason for using the “improved” Wnt1 line (Wnt1-Cre2+: R26mTmG) is because the original line (Wnt1-Cre:ZsGreen), used in most publications, is known to have ectopic expression in some locations due to Wnt1 activation caused by an in-frame ATG located upstream of Wnt1 start codon, thus complicating the results. The Wnt1-Cre2+ corrects this issue and is therefore a more reliable reporter. We think that previous authors missed the neural crest contribution due to the myocardium in Wnt1 mice because the early experiments done with Wnt1-cre lineage tracing used colorometric lacZ as a readout. As there is high background from endogenous enzyme activity in the heart, it is likely that any labeling observed in this location was ignored as high background. Thus, with our tools, including retroviral lineage tracing in chick and fluorescent Wnt1-cre2+ mouse lines, we both confirmed all other previously described contributions of cardiac neural crest to the heart (Kirby et al., 1983, Jiang et al., 2000) as well as discovering a novel contribution to the myocardium. We have clarified this point in the paper and thank the reviewer for point out this confusion.

3) Many results rely on scoring for the presence or absence of cells. How were tissues scored? More quantifications of the relative contribution to the tissue in question should be shown for lineage tracing and zf heart regeneration, such that the relative importance of the crest contribution can be appreciated.

We thank the reviewer for the suggestion. To address this, we quantitated the numbers of labeled neural crest cells we observe in various cardiac locations. As we do not label the entire neural crest population in chick, it’s most accurate to compare the% of labeled cells in each location (explained Supplementary file 1 legend). For the mouse, we label the entire neural crest population and we find that 16.8% of the labeled neural crest cells are in the proximal ventricles (quantified by automated particle analysis in FIJI program). Supplementary file 1A presents the quantified virally labelled cells in the chick embryo and E15.5 mouse and include the average percentage of neural crest contribution in the ventricular myocardium in mouse. We clarify that tissue were scored for numbers of GFP+ cells and clarify how counting was done in sections through chick, mouse and fish tissue in the methods. We have added approximate numbers of GFP+ cells to Supplementary file 1C in the zebrafish regeneration experiments as well.

4) Figure 4—figure supplement 1: Were there also cells that expressed sox10 mRNA de novo? The prediction would be that only lineage-traced cells re-express sox10 – otherwise non-neural crest derived cells contribute to the population of sox10+ cells in regeneration. Can this be clearly observed in the combination of colorimetric ISH and transgenic fluorescence?

Thanks for raising this. We now clarify that only cells that lineage trace with the sox10 marker reexpress sox10 during regeneration. However, not all mCherry+ cells express sox10, presumably because they lose expression as they differentiate into cardiomyocytes. We now include an updated figure that is clearer.

5) More information is needed on how the "the cardiac neural crest migratory gene regulatory module" was identified in the expression profile of regenerated heart cells in zf. e.g. did they use GO analysis? They highlight 20 out of 656 genes – what are the other genes? Why is 21 days post-injury the best time point?

Thank you for raising this issue as we now realize that we were unclear. For comparison, we use data from a recently published paper in which we profiled migratory cardiac neural crest genes in chick (Tani-Matsuhana et al., 2018). We then looked for similar gene signatures in our zebrafish sox10+ dataset, focusing on transcription factors associated with modules of the neural crest gene regulatory network (Martik and Bronner, 2017). We have also highlighted known zebrafish neural crest genes on our volcano plot (Figure 4D). This gene list was acquired by a literature search as well as ZFIN. We now clarify this in the text.

https://doi.org/10.7554/eLife.47929.018

Article and author information

Author details

  1. Weiyi Tang

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Conceived the project, performed virus preparation, lineage analysis in chick and mouse, immunohistochemistry, quantification, and wrote the manuscript
    Contributed equally with
    Megan L Martik
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1279-1001
  2. Megan L Martik

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Conceived the project, performed the heart regeneration experiments and RNAseq, and wrote the manuscript
    Contributed equally with
    Weiyi Tang
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1186-4085
  3. Yuwei Li

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Formal analysis, Validation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Performed molecular cloning for virus preparation, provided consultation for the manuscript
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7753-4869
  4. Marianne E Bronner

    Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, United States
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Writing—original draft, Project administration, Writing—review and editing, Conceived the project, assisted with lineage analysis in chick, and wrote the manuscript
    For correspondence
    mbronner@caltech.edu
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4274-1862

Funding

National Institutes of Health (NIHR01DE027568)

  • Marianne E Bronner

National Institutes of Health (NIHRO1HL14058)

  • Marianne E Bronner

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We would like to thank Drs. Xia Han and Yang Chai at University of Southern California, Center for Craniofacial Molecular Biology for being extremely supportive and kindly providing Wnt1-Cre; ZsGreenfl/fl cardiac tissue. Many thanks to Dr. Jeffrey Bush at University of California, San Francisco who generously sent us Wnt1-Cre2+; R26mTmG mouse embryos. We appreciate the help from Drs. Ann M Cavanaugh and Jau-Nian Chen at Department of Molecular, Cell and Developmental biology, University of California, Los Angeles in sharing Tg (NC: mCherry) transgenic fish line for Sox10:GAL4-UAS-Cre;ubi:Switch. We would also like to acknowledge the Caltech Millard and Muriel Jacobs Genetics and Genomics Laboratory, in particular, Igor Antoshechkin for sequencing of our RNAseq libraries. We thank Rochelle Diamond and Diana Perez of the the Caltech Flow Cytometry Cell Sorting Facility for cell sorting assistance. We thank David Mayorga and Ryan Fraser of the Beckman Institute Zebrafish Facility for help with zebrafish husbandry and Joanne Tan-Cabugo and Constanza Gonzalez for technical assistance, and Beckman Institute Biological Imaging Facility for equipment. This work is supported by NIHR01DE027568 and NIHRO1HL14058 to MEB and a Helen Hay Whitney Post-doctoral Fellowship to MLM.

Ethics

Animal experimentation: Adult zebrafish were maintained in the Beckman Institute Zebrafish Facility at Caltech, and all animal and embryo work were completed in compliance with California Institute of Technology Institutional Animal Care and Use Committee (IACUC) protocol 1764.

Senior and Reviewing Editor

  1. Didier Y Stainier, Max Planck Institute for Heart and Lung Research, Germany

Version history

  1. Received: April 25, 2019
  2. Accepted: August 8, 2019
  3. Accepted Manuscript published: August 8, 2019 (version 1)
  4. Version of Record published: September 3, 2019 (version 2)

Copyright

© 2019, Tang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 6,248
    Page views
  • 861
    Downloads
  • 40
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Weiyi Tang
  2. Megan L Martik
  3. Yuwei Li
  4. Marianne E Bronner
(2019)
Cardiac neural crest contributes to cardiomyocytes in amniotes and heart regeneration in zebrafish
eLife 8:e47929.
https://doi.org/10.7554/eLife.47929

Further reading

    1. Developmental Biology
    2. Evolutionary Biology
    Nico Posnien, Vera S Hunnekuhl, Gregor Bucher
    Review Article

    Gene expression has been employed for homologizing body regions across bilateria. The molecular comparison of vertebrate and fly brains has led to a number of disputed homology hypotheses. Data from the fly Drosophila melanogaster have recently been complemented by extensive data from the red flour beetle Tribolium castaneum with its more insect-typical development. In this review, we revisit the molecular mapping of the neuroectoderm of insects and vertebrates to reconsider homology hypotheses. We claim that the protocerebrum is non-segmental and homologous to the vertebrate fore- and midbrain. The boundary between antennal and ocular regions correspond to the vertebrate mid-hindbrain boundary while the deutocerebrum represents the anterior-most ganglion with serial homology to the trunk. The insect head placode is shares common embryonic origin with the vertebrate adenohypophyseal placode. Intriguingly, vertebrate eyes develop from a different region compared to the insect compound eyes calling organ homology into question. Finally, we suggest a molecular re-definition of the classic concepts of archi- and prosocerebrum.

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Tao Zhang, Liyang Wan ... Hongbin Lu
    Research Article Updated

    The attachment site of the rotator cuff (RC) is a classic fibrocartilaginous enthesis, which is the junction between bone and tendon with typical characteristics of a fibrocartilage transition zone. Enthesis development has historically been studied with lineage tracing of individual genes selected a priori, which does not allow for the determination of single-cell landscapes yielding mature cell types and tissues. Here, in together with open-source GSE182997 datasets (three samples) provided by Fang et al., we applied Single-cell RNA sequencing (scRNA-seq) to delineate the comprehensive postnatal RC enthesis growth and the temporal atlas from as early as postnatal day 1 up to postnatal week 8. And, we furtherly performed single-cell spatial transcriptomic sequencing on postnatal day 1 mouse enthesis, in order to deconvolute bone-tendon junction (BTJ) chondrocytes onto spatial spots. In summary, we deciphered the cellular heterogeneity and the molecular dynamics during fibrocartilage differentiation. Combined with current spatial transcriptomic data, our results provide a transcriptional resource that will support future investigations of enthesis development at the mechanistic level and may shed light on the strategies for enhanced RC healing outcomes.