1. Developmental Biology
  2. Stem Cells and Regenerative Medicine
Download icon

Distinct origins and molecular mechanisms contribute to lymphatic formation during cardiac growth and regeneration

  1. Dana Gancz
  2. Brian C Raftrey
  3. Gal Perlmoter
  4. Rubén Marín-Juez
  5. Jonathan Semo
  6. Ryota L Matsuoka
  7. Ravi Karra
  8. Hila Raviv
  9. Noga Moshe
  10. Yoseph Addadi
  11. Ofra Golani
  12. Kenneth D Poss
  13. Kristy Red-Horse
  14. Didier YR Stainier
  15. Karina Yaniv  Is a corresponding author
  1. Weizmann Institute of Science, Israel
  2. Stanford University, United States
  3. Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, United States
  4. Max Planck Institute for Heart and Lung Research, Germany
  5. Duke University, United States
  6. Duke University School of Medicine, United States
Research Article
  • Cited 10
  • Views 2,640
  • Annotations
Cite this article as: eLife 2019;8:e44153 doi: 10.7554/eLife.44153

Abstract

In recent years, there has been increasing interest in the role of lymphatics in organ repair and regeneration, due to their importance in immune surveillance and fluid homeostasis. Experimental approaches aimed at boosting lymphangiogenesis following myocardial infarction in mice, were shown to promote healing of the heart. Yet, the mechanisms governing cardiac lymphatic growth remain unclear. Here, we identify two distinct lymphatic populations in the hearts of zebrafish and mouse, one that forms through sprouting lymphangiogenesis, and the other by coalescence of isolated lymphatic cells. By tracing the development of each subset, we reveal diverse cellular origins and differential response to signaling cues. Finally, we show that lymphatic vessels are required for cardiac regeneration in zebrafish as mutants lacking lymphatics display severely impaired regeneration capabilities. Overall, our results provide novel insight into the mechanisms underlying lymphatic formation during development and regeneration, opening new avenues for interventions targeting specific lymphatic populations.

Introduction

The embryonic origins of lymphatic vessels have been debated for over a century, with a model claiming a venous origin for the lymphatic endothelium being predominant (Semo et al., 2016). However, recent work in zebrafish and mouse has uncovered additional surprising sources for lymphatic endothelial cells (LECs) during embryonic development (Nicenboim et al., 2015; Martinez-Corral et al., 2015; Stanczuk et al., 2015; Mahadevan et al., 2014; Klotz et al., 2015; Pichol-Thievend et al., 2018; Stone and Stainier, 2019; Eng et al., 2019). These include among others, vein resident angioblasts in the zebrafish trunk (Nicenboim et al., 2015), c-kit+ cells in the mouse mesentery (Stanczuk et al., 2015), VavCre-derived yolk sac hemogenic endothelium (Klotz et al., 2015), and the local capillary plexus in the skin (Pichol-Thievend et al., 2018). Nevertheless, there is still a gap in our knowledge regarding the cellular mechanisms by which these initial lymphatic structures give rise to unique organotypic networks.

As most organs in the body, the heart contains a dense network of lymphatic vessels (Klotz et al., 2015; Johnson and Blake, 1966). Normal cardiac function relies on the cardiac lymphatic system for controlling myocardial fluid homeostasis, lipid transport, and the immune response (Brakenhielm and Alitalo, 2019). While the existence of cardiac lymphatics was reported already in the 17th century (Rudbeck, 1653), only recently have data emerged describing their development and cellular origins in mammals (Klotz et al., 2015; Stone and Stainier, 2019; Flaht-Zabost et al., 2014; Norman and Riley, 2016). In the mouse for instance, LECs were shown to enter the heart on the ventral side along the outflow tract (OFT) at approximately embryonic day 12.5 (E12.5), and later emerge on the dorsal side near the sinus venosus at E14.5. This is in response to VEGF-C expressed in the OFT mesenchyme and cardiac epicardium (Klotz et al., 2015; Chen et al., 2014a). From E14.5, LYVE1/PROX1/VEGFR3-expressing lymphatics sprout at the surface of the ventricle expanding from the base toward the apex, and reaching full maturity at postnatal day 15 (P15) (Flaht-Zabost et al., 2014; Angeli and Harvey, 2015). While lineage-tracing studies have uncovered different cell types giving rise to cardiac lymphatics (Klotz et al., 2015; Stone and Stainier, 2019; Maruyama et al., 2019), the putative link between cellular origins and functional properties, as well as the specific contribution of different LECs to neo-lymphangiogenesis under pathological conditions, remain largely unknown.

Acute myocardial infarction (MI) is one of the most common cardiac pathologies and a leading cause of death worldwide. Following MI in mammals, massive loss of cardiomyocytes (CMs), combined with limited regenerative capacity of the myocardium result in the formation of a collagen-based scar leading to reduced cardiac function. In recent years, there has been growing interest in the role of the lymphatic system in this process, as pro-lymphangiogenic treatments in mice have been shown to promote healing of the heart after MI by reducing fluid retention and improving inflammatory cell clearance (Klotz et al., 2015; Vuorio et al., 2017; Henri et al., 2016; Ishikawa et al., 2007; Vieira et al., 2018).

Zebrafish, unlike mammals, exhibit a remarkable capacity to regenerate their hearts (González-Rosa et al., 2017), making it an ideal model to study the origins and functions of the lymphatic endothelium following cardiac injury. With this in mind, we set out to characterize the cardiac lymphatic system of the zebrafish. Our results identify two distinct lymphatic populations in the zebrafish heart: one forming through sprouting lymphangiogenesis, and a second one, established through coalescence of isolated LEC clusters. Interestingly, these cardiac lymphatics differ in their response to signaling cues. We further demonstrate the presence of similar LEC populations in the mouse heart, suggesting that the mechanisms underlying cardiac lymphatic development are evolutionarily conserved. Finally, we address the behavior of LECs following cardiac injury in zebrafish, and show that not all lymphatics respond equally to injury, and that coalescence of isolated LECs represents the main mechanism of lymphatic growth following acute cardiac damage. Notably, we demonstrate that lymphatic vessels are required for cardiac regeneration in zebrafish, as mutants devoid of lymphatics display severely impaired regeneration capabilities. Taken together our findings provide novel insights into the origins, mechanisms of formation and heterogeneity of the cardiac lymphatic vasculature during development and regeneration. A better understanding of cardiac lymphatic formation holds great promise for developing new therapeutic interventions targeting specific lymphatic subsets.

Results

Morphological and molecular heterogeneity of zebrafish cardiac lymphatics

In contrast to the zebrafish systemic lymphatic vasculature, which develops within the first days of embryonic development (Nicenboim et al., 2015; Yaniv et al., 2006), we find that cardiac lymphatics start reaching the heart only after the larva-to-juvenile transition (Parichy et al., 2009). The first lymphatic vessels, labeled by the TgBAC(prox1a:KalTA4,UAS:uncTagRFP) (hereafter termed ‘prox1a’) and Tg(lyve1b:dsRed2) (hereafter termed ‘lyve1b’) transgenes, are detected in the OFT at ~21–34 days post-fertilization (dpf) (fish size 9–12 mm) (Figure 1a–d). These large collecting lymphatics expand during the following weeks (Figure 1c,d), until they cover the entire OFT by ~8 weeks post-fertilization (wpf) (fish size 14–24 mm) (Figure 1b) and reach their mature form, which remains stable throughout adulthood (Figure 1—figure supplement 1a–d). It is not until ~12–16 wpf (fish size 25–32 mm), however, that blind-ended lymphatics arising at the base of the OFT sprout towards the ventricle, in close proximity to the major coronary vessels (Figure 1e–h, white arrows). At similar stages, we also detected isolated clusters of LECs spread throughout the ventricle, which were not connected to the OFT or ventricular lymphatics (Figure 1e,f,h, blue arrows), as well as a dense lymphatic network associated with the epicardial adipose tissue (Figure 1—figure supplement 2a, yellow arrows). As with other lymphatic vessels in zebrafish (Yaniv et al., 2006; Jung et al., 2017), cardiac lymphatics lack open connections with the blood vasculature as confirmed by intravascular injection of Qdots 705 (Figure 1k, arrows).

Figure 1 with 4 supplements see all
Lymphatic vessel heterogeneity in the zebrafish heart.

(a) Diagram of ~7 wpf zebrafish heart depicting the outflow tract (OFT), ventricle (V) and lymphatics (red). (b) At 10 wpf (fish size 16–24 mm) prox1a-labeled collecting lymphatics are clearly detected in the OFT, but absent on the ventricular surface (n = 9). (c,d) Quantification of 21–34 dpf (fish size 7–14 mm) OFT lymphatic development in prox1a (c) (n21dpf=48, n28dpf=38, n34dpf=19) andlyve1b (d) (n21dpf=12, n28dpf=11, n34dpf=13) transgenic zebrafish. (e) Diagram of adult zebrafish heart depicting ventricular lymphatics (white arrows), fat-associated lymphatics (yellow arrows) and isolated lymphatic clusters (blue arrows). OFT and ventricular lymphatics as well as isolated lymphatic clusters in 22–24 wpf (fish size 25–30 mm) are labeled by the flt4 (f), lyve1b (g), prox1a (f,h) and mrc1a (g,h) (nf = 4, ng = 5, nh = 5) transgenic reporters. (i) 1–5 cells at the tip of ventricular lymphatics are labeled primarily by the flt4 transgene (arrows). Nuclei are labeled by DAPI (blue) (48wpf, fish size 25–30 mm, n = 5). (j) Ventricular lymphatics are not labelled by the blood vessel/endocardial- marker Tg(kdrl:nls-mCherry) (23wpf, fish size 25–30 mm, n = 5). (k) Angiogram of 28 wpf (fish size 28 mm) Tg(flt1_9a_cFos:GFP);Tg(lyve1b:dsRed2) heart. Cardiac lymphatics (arrows) are not labeled following intravascular injection of Qdot705 (blue) (n = 6). Scale bars are 200 µm in b, f-h, j, k; 50 µm in i. Posterior view in b, anterior view in f-k.

Examination of transgenic zebrafish with labeled LECs, revealed clear expression of lyve1b (Okuda et al., 2012), prox1a (van Impel et al., 2014), Tg(flt4BAC:mCitrine) (hereafter termed ‘flt4’) (van Impel et al., 2014), and Tg(mrc1a:EGFP) (hereafter termed ‘mrc1a’) (Jung et al., 2017), in all cardiac lymphatic subsets (Figure 1f–h, Figure 1—videos 1 and 2). Since these markers are absent from the blood vascularized ventricles prior to the appearance of lymphatic vessels (Harrison et al., 2015), we concluded that they specifically highlight LECs in the adult heart. Interestingly, we noticed that 1–5 cells at the tip of ventricular lymphatic vessels were labeled primarily by the flt4 transgene (Figure 1i, arrows). Finally, we could verify that the main cardiac lymphatics were not labeled by the blood vessel marker Tg(kdrl:nls-mCherry) (Figure 1j, Figure 1—figure supplement 1e,f) and arterial-specific marker Tg(flt1_9a_cFos:GFP) (Figure 1k), which were clearly detected in coronary arteries and arterial capillaries.

Strikingly, we also detected sporadic expression of the prox1a transgene in few arterioles, that were co-labeled by the arterial enhancer Tg(flt1_9a_cFos:GFP) and highlighted by intravascular injection of Qdot705 (Figure 1—figure supplement 2b, inset). In similar fashion, assessment of hearts extracted from prox1a;flt4 double transgenic zebrafish following Qdot705 angiography revealed two vessel populations labelled by the prox1a transgene– one, where the expression fully overlapped with that of the flt4 reporter and was devoid of Qdot705 labeling (Figure 1—figure supplement 2c, white arrows), and a second one labelled only by prox1a and Qdot705 (Figure 1—figure supplement 2c, yellow arrows). In order to investigate whether both populations indeed express Prox1a, we carried out immunostaining with anti-Prox1 antibody. As seen in Figure 1—figure supplement 2d only prox1a positive LECs, but not prox1a-positive blood ECs were labelled by the Prox1 antibody (Figure 1—figure supplement 2e, insets), suggesting that the expression in blood ECs could be a result of post-transcriptional regulation or may represent an artefact of the transgenic reporter.

Development of OFT lymphatics

Live imaging of the different LEC-transgenic reporters revealed that the first lymphatic sprouts reach the OFT at 21–28 dpf (Figure 2a,b, arrows). At this stage, no Tg(flt1_9a_cFos:GFP) or Tg(fli1:EGFP)-labeled blood vessels are detected in the heart (Figure 2a,b). Surprisingly, we found that these sprouts originated in four facial lyve1b+ vessels that run parallel to the ventral aorta (VA) (Figure 2c,d, yellow arrows) (Isogai et al., 2001) before reaching both sides of the OFT (Figure 2d; white arrows). These vessels, which we named ‘ventral facial lymphatic (VFLs)', are also labeled by the prox1a transgene (Figure 2e, arrow) and connect to the lymphatics of the branchial arches (Figure 2c (LAA); e, inset). The VFLs were devoid of blood flow, as reflected by the absence of Tg(gata1a:dsRed2) labelled erythrocytes (Figure 2f, arrows), which were readily detected in the surrounding aorta and gills.

Figure 2 with 3 supplements see all
Establishment of OFT lymphatics.

(a–b) Blood vessels are not detected in the OFT (outlined) of 21–28 dpf (fish size 5–7 mm) (a) Tg(flt1_9a_cFos:GFP); Tg(lyve1b:dsRed2) hearts (n = 6) or (b) Tg(fli1:EGFP);Tg(prox1a:KalTA4-UAS:uncTagRFP) (n = 4) fish, prior to OFT lymphatic sprouting (white arrows). (c) Diagram depicting a ventral view of a zebrafish larval head, indicating the approximate region imaged in (d-f). Facial lymphatics are colored in red (adapted from Okuda et al., 2012), yellow arrow points to the VFL. (d) OFT lymphatic sprouts (white arrows) arising from the VFL (outlined, yellow arrow) are detected at 24 dpf (Fish size 5–7 mm) in Tg(kdrl:EGFP);Tg(lyve1b:dsRed2);casper larvae (n = 5). (e) The VFL (arrow) connects to the LAA (inset). (f) No blood flow is detected in the VFL (arrows) of 22 dpf (fish size 5—7 mm) Tg(fli:EGFP);Tg(gata1a:dsRed2) larvae (n = 10). (g) Quantification of OFT lymphatics in 35 dpf (fish size 9–13 mm) prox1a transgenic zebrafish treated with 100 µM Atenolol (ncontrol = 43, nAtenolol = 43, *p<0.001). (h) mRNA levels of vegfc, cxcl12a and cxcl12b (nindependent experiments=5, *p<0.01) in the OFTs of 14–34 dpf larvae. (i–k) OFT of 19 wpf (22–25 mm) wt sibling (i) and flt4-/- (j) in the background of lyve1b demonstrating severe lymphatic defects in flt4 -/- hearts, quantified in (k) (nwt = 4, nflt4 -/-=5, *p<0.001). (l–n) OFT of 15 wpf (20–21 mm) wt sibling (l) and vegfc +/- (m) in the background of lyve1b showing malformed lymphatics in vegfc +/-, quantified in (n) (nwt = 8, nvegfc+/-=6, *p<0.01, **p<0.05). (o–q) OFT of 12 wpf (19–21 mm) age-matched wt control (o) and vegfd -/- (p) in the background of lyve1b showing normal OFT lymphatics in vegfd -/- hearts, quantified in (q) (nwt = 4, nvegfd-/-=5) (r–t) OFT lymphatics of 9.5 wpf (20–23 mm) wt sibling (r) and cxcr4a-/- (s) in the background of lyve1b showing mild defects in cxcr4a-/- OFT lymphatics, quantified in (t) (nwt = 4, ncxcr4a-/-=10, *p<0.01). (u–w) OFT lymphatics of 20 wpf (19–24 mm) wt sibling (u) and cxcl12b-/- (v) in the background of prox1a showing normal OFT lymphatics in cxcl12b-/- hearts, quantified in (w) (nwt = 4, ncxcl12b -/-=8). VFL, ventral facial lymphatics; LFL, lateral facial lymphatic; LAA, lymphatic branchial arches; MFL, medial facial lymphatic. Scale bars are 50 µm in a-f, 200 µm in i-v. Error bars, mean ± S.E.M.

In zebrafish, the first facial lymphatic sprouts (FLS) were shown to arise in the common cardinal vein (CCV) and primary head sinus (PHS) (Eng et al., 2019; Okuda et al., 2012). Subsequently, a population of lymphangioblasts of unknown origins joins the sprouts emanating from the FLS to form individual facial lymphatic vessels. These angioblast cells, which form close to the ventral aorta and are initially devoid of both venous and lymphatic markers, contribute not only to facial lymphatics but also to the hypobranchial artery, confirming their multipotent ability (Eng et al., 2019). Thus, facial lymphatics, most probably including the VFLs from which the OFT lymphatics sprout, originate from local sources within the facial domain.

The appearance of the first lymphatic sprouts in the OFT correlates with the transition from the larval to the juvenile stages (Parichy et al., 2009). We therefore wondered whether these two processes could be somehow associated. In particular, we hypothesized that increased heart rate and cardiac output during the larval to the juvenile transition, could result in elevated blood pressure and fluid extravasation, and potentially trigger the growth and or/attraction of lymphatic vessels (Boardman and Swartz, 2003). To test this hypothesis, we measured the heart rate of zebrafish between 5–40 days post-fertilization (dpf), and found that it indeed peaks during the larva to juvenile transition (Figure 2—figure supplement 1a). Moreover, heart-rate attenuation following administration of the ß-blocker Atenolol (Figure 2—figure supplement 1b) (Hein et al., 2015) resulted in delayed appearance, and impaired branching of OFT lymphatics (Figure 2g, Figure 2—figure supplement 1c,d). To identify molecular candidates mediating lymphatic recruitment, we analyzed gene expression on isolated OFTs at the relevant stages, and detected clear upregulation of the well-established pro-lymphangiogenic factor vegfc, and of the chemokines cxcl12a, and cxcl12b (Figure 2h). In order to investigate the potential role of these factors in OFT lymphatic development, we analyzed genetic mutants. We observed marked defects in OFT lymphatics following depletion of the Vegfc receptor, Vegfr3/Flt4 (Figure 2i–k). flt4 mutants bear a truncated form of the Flt4 receptor and were shown to lack a thoracic duct at 5dpf (Kok et al., 2015). While in general facial lymphatics were slightly shorter in flt4-/- hearts (Figure 2—figure supplement 2a,b, yellow arrows), no major defects were observed in the VFL, from which OFT lymphatics sprout (Figure 2—figure supplement 2a,b, white arrows). In contrast, lymphatic vessels were almost completely absent from the OFTs of flt4 -/- hearts (Figure 2i–k, white arrow). Flt4 is the receptor for the pro-lymphangiogenic growth factors Vegfc and Vegfd. We therefore examined the contribution of each ligand to OFT lymphatic development by assessing vegfc+/- (Villefranc et al., 2013) and vegfd-/- hearts. Since vegfc homozygous mutants are embryonic lethal, we examined heterozygous animals that, similar to their homozygous siblings, display reduced formation of the thoracic duct at five dpf (Villefranc et al., 2013), but survive through adulthood. In spite of bearing a wild type (wt) vegfc allele, loss of one copy of this gene resulted in maldeveloped lymphatic vessels and overall reduction of OFT lymphatic coverage in vegfc+/- fish (Figure 2l–n). vegfd-/- hearts (Figure 2—figure supplement 3d,f) in contrast, did not display any substantial defects in OFT lymphatics (Figure 2o–q) or in the VFL (Figure 2—figure supplement 2e, white arrow) at three wpf, despite of displaying minor defects during early facial lymphatic development (Figure 2—figure supplement 3a–c) (Bower et al., 2017a).

We then investigated the potential role of the Cxcr4/Cxcl12 axis in OFT lymphatic development. CXC chemokines play a well-established role in guidance of various cell types, including LECs (Cha et al., 2012). In addition, cxcl12a and cxcl12b were found to be expressed on the surface of the OFT at ~4 wpf, prior to the formation of OFT blood vessels (Harrison et al., 2015). Therefore, Cxcr44/Cxcl12 signaling can potentially mediate LEC sprouting over the OFT. Analysis of adult cxcr4a -/- hearts (Siekmann et al., 2009) carrying the lyve1b transgene revealed only minor defects in OFT lymphatics as compared to wt siblings (Figure 2r–t). Moreover, homozygous mutants for the cxcl12b ligand (cxcl12b-/-) did not display any noticeable phenotypes (Figure 2u–w). Of note, the VFL was not affected in either cxcr4a -/- or cxcl12b-/- fish (Figure 2—figure supplement 2f–h).

Altogether, our results suggest that changes in heart rate during larva to juvenile transition modulate OFT lymphatic development in a Vegfc/Flt4-dependent manner, whereas Cxcr4a/Cxcl12b involvement is restricted to remodeling of the OFT lymphatic plexus.

Establishment of ventricular lymphatics

Despite the fact that OFT lymphatics are established at 3–4 wpf, it is not until ~2 months later that they begin sprouting toward the ventricle (Figure 1e–h; 3a, inset). Preceding lymphatic sprouting, the zebrafish myocardium undergoes significant expansion, which was proposed to trigger the formation of the coronary vasculature (Harrison et al., 2015; Gupta et al., 2013). Interestingly, we found that ventricular lymphatics sprout in close proximity to the major coronary vessels and continue to grow and branch over the following months (Figure 3b). Similar association between developing lymphatics and the blood vasculature has been described in the mouse heart (Klotz et al., 2015). To test whether coronary vessels play an active role in ventricular lymphatic development, we treated juvenile zebrafish with Phenylhydrazine hydrochloride (PHZ), which induces CM hypertrophy and enhanced vascularization (Sun et al., 2009). PHZ treatment starting at ~8 wpf- coinciding with the initial development of the coronary plexus-, resulted in enlarged hearts accompanied by enhanced growth and remodeling of coronary vessels (Figure 3c,d). Interestingly, we also detected significantly longer lymphatic vessels in the ventricles of these animals, which closely followed major coronary vessels (Figure 3c,d), suggesting a possible role for the coronary vasculature in lymphatic vessel growth. To further confirm these results, we induced hypervascularization by conditionally over-expressing Vegfaa in CMs using Tg(cmlc2:CreER);(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa) fish (hereafter termed Vegfaa-OE) (Karra et al., 2018) (Figure 3e–g, insets; Figure 3—figure supplement 1a). Vegfaa-OE was induced at ~7 wpf by administration of 4-hydroxytamoxifen and hearts were analyzed 17 days later. As previously reported (Karra et al., 2018), we detected a significant increase in blood vessel coverage, especially those of small caliber (Figure 3e–g, insets; Figure 3—figure supplement 1b), which was accompanied by a massive expansion of the lymphatic plexus (Figure 3e–g), supporting a tight association between blood and lymphatic vessel growth.

Figure 3 with 1 supplement see all
Coronary arteries serve as a scaffold for ventricular lymphatic sprouting.

Insets are magnifications of dashed boxes. (a) 13 wpf (fish size 17 mm) Tg(fli1:EGFP);Tg(prox1a:KalTA4-UAS:uncTagRFP) hearts showing ventricular lymphatic sprouting at the base of the OFT (inset) (n = 11). (b) Ventricular lymphatics grow in close proximity to coronary arteries (inset) in 116 wpf (fish size 25–32 mm) Tg(flt1_9a_cFos:GFP);Tg(lyve1b:dsRed2) hearts (n = 15). (c) Heart of 11wpf (fish size 16–22 mm) Tg(fli1:EGFP);Tg(prox1a:KalTA4-UAS:uncTagRFP) fish treated with 100 µg/ml Phenylhydrazine hydrochloride (PHZ), showing increased ventricle size and total length of blood and lymphatic vessels following PHZ treatment, quantified in (d) (ncontrol = 21, nPHZ = 15, *p<0.005). (e,f) 12 wpf (fish size 19–22 mm) Tg(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa)pd262; Tg(cmlc2:CreER) fish in the background of Tg(flt1_9a_cFos:GFP);Tg(prox1a:KalTA4-UAS:uncTagRFP). Induction of Vegfaa-OE results in increased blood vessel (insets) and lymphatic vessel coverage, as compared to vehicle (Vegfaa-VEH) treated sibling control, quantified in (g) (nVegfaa-VEH = 9, nVegfaa-OE = 10, *p<0.001). (h–j) Immature coronary plexus, lacking the stereotypical tree-patterning results in nearly absent ventricular lymphatics in cxcr4a-/- hearts of 22 wpf (fish size 25–28 mm) (i), as compared to wt siblings (h), quantified in (j) (nwt = 5, ncxcr4a-/-=6, *p<0.05). (k–m) Ventricular lymphatics are absent in Tg(fli1:EGFP); Tg(lyve1b:dsRed2); flt4-/- hearts (l) at 19–23 wpf (fish size 25–30 mm) as compared to wt siblings (k). (m) Quantification of blood and lymphatic vessel phenotype in flt4 -/- hearts (nwt = 4, nflt4 -/-=5 *p<0.01). (n–p) Tg(fli1:EGFP); Tg(lyve1b:dsRed2); vegfc +/- hearts at 26 wpf (fish size 25–30 mm) display severely defective ventricular lymphatics. (p) Quantification of blood vessel coverage and lymphatic sprout length and number, in vegfc +/- hearts (nwt = 3, nvegfc+/-=5 *p<0.005). (q–t) 12 wpf (fish size 19–22 mm) Tg(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa)pd262; Tg(cmlc2:CreER) fish in the background of Tg(fli1:EGFP);Tg(lyve1b:dsRed2). The increase in lymphatic, but not blood vessel coverage induced by Vegfaa-OE (r), is reversed in vegfc+/- heterozygous animals (s). (t) Quantification of blood and lymphatic vessel coverage in (q–s) (nVegfaa-VEH = 3, nVegfaa-OE = 3, nVegfaa-VEH-blood vessel coverage=3, nVegfaa-OE-lymphatic vessel coverage=10 nVegfaa-OE;vegfc +/- -lymphatic vessel coverage=4, *p<0.01, relative to vehicle treated sibling control, **p<0.001 relative to Vegfaa-OE). Scale bars are 200 µm. Error bars, mean ± S.E.M. All panels show anterior views.

In order to understand whether coronary vessels are required for proper ventricular lymphatic sprouting, we first examined hearts of cxcr4a-/- fish, which have previously been reported to fail to develop coronary vasculature (Harrison et al., 2015). Surprisingly, we found that neither blood vessel coverage (Figure 3h–j), nor ventricle size (Figure 3—figure supplement 1c) was affected in cxcr4a-/- hearts. Instead, the coronary plexus appeared immature, lacked its stereotypic hierarchical pattern and the main arteries that run over the OFT and connect to the gills were severely malformed or absent (Figure 3—figure supplement 1d, yellow arrow; e, asterisk). These defects impede the establishment of proper connections with the circulatory network and rendering the cardiac muscle only partially perfused (Figure 3—figure supplement 1f, yellow arrows). Similar phenotypes have been described in Cxcl12 and Cxcr4 mutant mice, where ECs of the peritruncal plexus fail to properly connect to the aortic endothelium, resulting in impaired coronary artery stem formation and establishment of an immature coronary plexus (Ivins et al., 2015). Interestingly, the underdeveloped coronary plexus observed in cxcr4a -/- hearts was associated with severe ventricular lymphatic abnormalities, characterized by markedly shorter and less branched lymphatic sprouts (Figure 3h–j, white arrow, Figure 3—figure supplement 1g), suggesting that the presence of blood ECs per se, is not sufficient to support lymphatic growth, but rather a mature coronary tree is required. To further investigate the nature of the interaction between coronary vessels and ventricular lymphatics, we assessed hearts of flt4 -/- and vegfc+/- animals (Figure 3k–p). In contrast to cxcr4a-/- hearts, we detected no significant differences in blood vessel coverage and patterning (Figure 3k–p, Figure 3—figure supplement 1h,i) as well as in ventricle size (Figure 3—figure supplement 1j,k) in these animals. Yet, ventricular lymphatics were severely affected in vegfc+/- and nearly absent in flt4-/- hearts (Figure 3k–p), indicating that Vegfc/Flt4 signaling is absolutely required for ventricular lymphatic formation. Taken together, the mutant analyses support the idea that a mature coronary plexus could serve as a source of Vegfc, a hypothesis substantiated also by the observation that mouse coronary ECs begin expressing Vegfc as they differentiate into arteries (Su et al., 2018). In order to test this hypothesis, we over expressed Vegfaa in vegfc+/ - fish. While the reduced levels of Vegfc did not impede Vegfaa-OE induced cardiomegaly and hypervascularization (Figure 3q–t Figure 3—figure supplement 1l,m), the increase of lymphatic coverage (Figure 3e–g) was abrogated in these animals (Figure 3q–t), supporting the idea that a mature coronary plexus serves as a scaffold for ventricular lymphatic development, in a Vegfc-dependent manner. Moreover, these results demonstrate that the increased lymphatic growth observed in Vegfaa-OE hearts results from the expanded blood vasculature and, most probably, from enhanced Vegfc production, and not from a direct effect of Vegfaa on LECs. In the future, it will be interesting to investigate whether the Cxcl12/Cxcr4 signaling pathway somehow regulates Vegfc secretion from the coronary ECs.

A novel population of isolated LECs is detected in the hearts of zebrafish and mice

Besides the ‘regular’ lymphatic vessels, the zebrafish ventricle holds an additional population of isolated LECs (Figure 1e,f,h), labeled by the prox1a, lyve1b, mrc1a and flt4 transgenic reporters (Figure 1f,h Figure 4a-d, Figure 4—video 1), and positively stained by Prox1 antibody (Figure 4e, arrows). Light sheet imaging of explanted hearts revealed that these lymphatic structures initially appear as single cells (~12–13 wpf, fish size 20–22 mm), that later on expand to form isolated capillaries (~16 wpf, fish size 23–28 mm) (Figure 4b–d, Figure 4—video 1), through a process that resembles lymph-vasculogenesis (Semo et al., 2016). While we could not determine the exact source of these isolated LECs, we could confirm the lack of connections to ventricular lymphatics (Figure 1f,h; Figure 4c,f) and/or to the arterial network (Figure 4f, inset) suggesting that they may have a separate origin.

Figure 4 with 3 supplements see all
A novel population of isolated LECs is detected in the adult zebrafish heart.

Insets are magnification of dashed boxes. (a) Tg(fli1:EGFP);Tg(prox1a:KalTA4-UAS:uncTagRFP) 16 wpf (fish size 23–28 mm) heart showing isolated lymphatic clusters (n = 8). (b,c) Double labeled prox1a+;flt4+ isolated LECs are first detected at ~13 wpf (20–22 mm) (b, arrows) and coalesce to generate isolated capillaries by 16wpf (fish size 25–28 mm) (c, arrows). (d) Quantification of double-labeled prox1a+;flt4+ isolated LECs in the ventricles of 14–28 mm fish (n14-19mm=4, n20-22mm=5, n23-28mm=5). (e) prox1a+ isolated LECs are also labeled by Prox1 antibody (inset, arrow). (f) 20 wpf (fish size 28 mm) double-transgenic prox1a;flt4 hearts demonstrate that isolated LECs are not labeled following intravascular injection of Qdot705 (inset, arrow) (image in f) is an additional view of Figure 2—figure supplement 1c). (g) Isolated LEC clusters develop normally in 22 wpf (fish size 25–30 mm) Tg(flt1_9a_cFos:GFP);Tg(lyve1b:dsRed2); cxcr4a-/- hearts, quantified in (h) (nwt20-22mm=8, ncxcr4a-/- 20-22mm = 6, nwt23-28mm=7, ncxcr4a-/-23-28mm = 8). (i,j) lyve1b+ isolated LEC clusters are not precociously detected in Vegfaa-OE hearts (12.5 wpf, fish size 19–22 mm), in (i) (nveh = 7, nTam = 8) or PHZ treatment (j) (ncontrol = 9, nPHZ = 9) (k). No isolated LECs are detected in 19-23wpf (fish size 23–28 mm) Tg(fli1:EGFP);Tg(lyve1b:dsRed2);flt4 -/- hearts, quantified in (l) (nwt = 5, nflt4-/-=6). (m) Significantly reduced numbers of isolated LECs are detected in Tg(fli1:EGFP);Tg(lyve1b:dsRed2);vegfc+/- animals at 26 wpf (fish size 25–28 mm), quantified in (n) (nwt = 3, nvegfc+/-=5). Scale bars are 200 µm. Error bars, mean ± s.e.m. Anterior view in a-c, e,f,k,m. Posterior view in g.

Unlike ventricular lymphatics, the isolated LEC clusters appeared normal in cxcr4a-/- hearts (Figure 4g,h, Figure 3—figure supplement 1e, blue arrows, Figure 4—videos 2 and 3). Moreover, while Vegfaa-OE induced precocious sprouting of ventricular lymphatics (Figure 3q,r,t) at ~12.5, when they were still absent from control hearts (Figure 3q,t), no effects were detected in the numbers and/or distribution of the isolated LEC following Vegfaa-OE (Figure 4i) or PHZ treatment (Figure 4j). Thus, coronaries appear to be dispensable for the formation of this lymphatic subpopulation. On the other hand, isolated LECs were completely absent from flt4-/- hearts (Figure 4k,l) and markedly reduced in vegfc+/ - hearts (Figure 4m,n). Taken together, these results suggest that distinct lymphatic populations respond differently to the same molecular cues, as is the case of Cxcl12/Cxcr4 signaling which prevented ventricular lymphatic growth but did not impair the appearance of the isolated LEC clusters. In contrast, Vegfc/Flt4 signaling was absolutely required for formation of both populations, indicating its crucial role as key regulator of lymphatic development.

In order to investigate whether similar isolated LECs are also present in mammals, we examined mouse hearts stained with PROX1 and VE-Cadherin antibodies at different embryonic stages (Figure 5a–f). At embryonic day E13.5, coronary vessels were already present in the dorsal side of the heart (Figure 5a, red), but no PROX1+ LECs were detected. CMs were the only PROX1+ cells detected over the ventricle at this stage (Figure 5a, green). One day later at E14.5, lymphatic sprouts emanating from the sinus venosus region were clearly visualized on the dorsal aspect of the heart, consistent with previous reports (Klotz et al., 2015) (Figure 5b, white arrows). In addition to these lymphatics with clear vessel morphology, we also detected isolated PROX1+ LECs throughout the surface of the ventricle (Figure 5b, inset, blue arrows), similar to the isolated LECs found in the zebrafish heart (Figure 4d). Both the main lymphatic vessels and the isolated LECs also expressed VE-Cadherin. Staining with the lymphatic membrane resident receptor LYVE1 at E15.5, confirmed that these LECs clusters, positive for both PROX1 and LYVE1, were not connected to the main lymphatic vessels (Figure 5c, inset, arrows, Figure 5—figure supplement 1a). Clusters could still be found at E17.5 (Figure 5e, inset, arrows, Figure 5—figure supplement 1b–c, Figure 5—video 1), but were no longer detected at day P23 (Figure 5f). Interestingly, similar to flt4 mutant fish, Ccbe1 mutant mice lacked both ventricular sprouts and isolated LECs (Figure 5g,h). Ccbe1 is required for lymphangiogenesis in mice (Bos et al., 2011) and fish (Hogan et al., 2009), acting as a regulator of VEGF-C processing (Jha et al., 2017). Thus, the requirement for VEGF-C/Vegfr3 signaling in both cardiac lymphatic populations is evolutionarily conserved.

Figure 5 with 2 supplements see all
Both lymphangiogenesis and lymph-vasculogenesis contribute to cardiac lymphatic development in mammals.

(a,b) Whole mount confocal images of mouse hearts immunonstained for VE-Cadherin and PROX1. (a) At E13.5 coronary (red) but not lymphatic vessels (green) are present on the ventricle surface (n = 6) (Low levels of PROX1 are detected in cardiomyocytes). (b) In addition to regular lymphatic vessels (white arrows), isolated PROX1+ LECs are observed in close proximity to the coronaries at E14.5 (inset, blue arrows) (n = 6). (c) Whole mount confocal images of E15.5 mouse hearts immunonstained for LYVE1 and PROX1. Isolated PROX1+/LYVE1+ LECs (blue arrows) are not connected to the main lymphatic sprouts (white arrow). (d,e) Whole mount confocal images of mouse hearts immunonstained for VE-Cadherin and PROX1. Isolated LECs gradually expand to form multicellular lumenized structures (inset, arrows) (n = 4). (f) PROX1+ isolated LECs are no longer detected at P23 (n = 4). (g,h) Whole mount confocal images of E15.5 mouse hearts immunonstained for LYVE1 and PROX1. PROX1+ LECs are absent from Ccbe1 mutant hearts (h) (nwt = 3, nCcbe1 = 3). (i) Lineage-tracing strategies for identification of cardiac LEC origins. (j) Dorsal view of AplnCreERT2,RosamTmG heart from embryo dosed with tamoxifen at E12.5 and analyzed at E14.5, showing no AplnCreERT2;PROX1+ LECs. Cre recombination is labeled in red, ECs in blue (Erg) and lymphatics in green (PROX1). Insets are magnification of dashed boxes. (n = 7). (k) Dorsal view of ApjCreERT2,RosamTmG heart, showing that PROX1+ LECs are not labeled by ApjCreERT2 in embryos dosed with tamoxifen at E12.5 and analyzed at E14.5. Cre recombination is marked in blue, ECs in red (ERG) and lymphatics in green (PROX1). Insets are magnification of dashed boxes (n = 4). (l) Ventral view of heart from BmxCreERT2, RosaTdTom embryos dosed with tamoxifen at E8.5 and analyzed at E15.5, showing that PROX1+ LECs are not labeled by BmxCreER. Cre recombination is marked in green, ECs in red (ERG) and lymphatics in blue (PROX1). Inset is magnification of dashed boxes. (n = 5). Dorsal (m) and ventral (n) views of ApjCreERT2,RosamTmG embryos dosed with tamoxifen at E9.5 and 10.5 and analyzed at E15.5, showing PROX1+ isolated LECs and lymphatic vessels, labeled by ApjCreERT2. Cre recombination is marked in blue, ECs in red (ERG) and lymphatics in green (PROX1). (o) Quantification of mTmG labeling shows reduced ApjCreERT2 lineage traced PROX1+ cells as compared to ERG ECs (recombination efficiency). D;Dorsal, V;Ventral (nD = 5, nV = 6) Insets are magnification of dashed boxes. (nDorsal = 6, nVentral = 5). Scale bars are 200 µm.

We then asked what is the origin of these isolated LECs. The majority of mouse cardiac lymphatics (78%) has been shown to originate from TIE2+ ECs, which emerge from the common cardinal vein (CCV) and migrate toward the heart (Klotz et al., 2015). Additionally, a contribution from the hemogenic endothelium has also been proposed, albeit the exact identity of these cells has remained controversial (Klotz et al., 2015; Ulvmar and Mäkinen, 2016). Individual vessels contained cells from both origins, making cardiac lymphatics a mosaic of different sources. In order to identify the cellular origins of the new population of isolated LECs, we used different lineage-tracing strategies (Figure 5i). Since these LEC clusters are often found in close proximity to the developing coronary vessels (Figure 5b,d), a feature that was also evident in the zebrafish heart (Figure 4g), we decided to investigate whether they originate from the local vasculature (i.e. coronary vessels). To this end, we used AplnCreERT2 mice, which specifically labels sprouting ECs, but not endocardial or lymphatic cells (Liu et al., 2015). AplnCreERT2 mice were crossed to the Cre-dependent fluorescent reporter ROSA26mTmG/+ (Muzumdar et al., 2007) and Cre-mediated recombination was induced by Tamoxifen administration at day E12.5, after the coronary vasculature has formed, but prior to the appearance of lymphatics over the ventricle. Staining with PROX1 and the EC-specific ERG antibodies at E14.5 rendered no PROX1+ cells that were labeled also by the mTmG reporter (Figure 5j). This result was also obtained with another Cre that labeled coronary vessels when induced at later stages, ApjCreERT2 dosed at E12.5 (Figure 5k, inset), further confirming that LECs do not originate from the local coronary vasculature, as previously reported (Angeli and Harvey, 2015). Yet, we cannot exclude the possibility that ECs acquire a lymphatic fate before exiting the blood vasculature, and as a consequence, were not labeled by AplnCreERT2 or ApjCreERT2.

An additional potential source is the endocardial/endothelial cells that line the lumen of the heart, which also give rise to part of the coronary blood vasculature in mouse (Wu et al., 2012; Red-Horse et al., 2010). We have recently observed that that the BmxCreERT2 line is highly specific for the endocardium in the heart, when induced with Tamoxifen prior to coronary vessel development (data not shown). BmxCreERT2 was originally reported as inducing recombination in embryonic arteries specifically, but it is also highly expressed in the endocardium and almost completely excluded from the sinus venosus (Ehling et al., 2013). BmxCreERT2 mice were crossed to ROSA26TdTom/+ and Cre-mediated recombination was induced at E8.5. Hearts were harvested at day E15.5 and stained with antibodies against PROX1 and ERG. Despite almost complete endocardial recombination (data not shown), no PROX1+;TdTom+ cells were detected on the surface of the ventricle (Figure 5l, insets), suggesting no significant contribution of the endocardium or embryonic arteries to cardiac lymphatics. Finally, we turned back to APJCreERT2 mice (Chen et al., 2014b), but this time induced labeling at E9.5 and E10.5 to specifically mark the major embryonic veins (SV and CCV) (Chen et al., 2014b) (Figure 5—figure supplement 1d,e). Hearts were harvested at day E15.5 and stained as described above. This manipulation resulted in lineage labeling of a portion of the lymphatic vessels and isolated clusters on both the dorsal and ventral aspects of the heart (Figure 5m–o), suggesting that these LECs arose from an early venous source.

Strikingly, we noted that lineage labeling was lower than expected in APJCreERT2 mice dosed at E9.5/10.5 if most vessels were to derive from a venous source. We have previously shown that multiple doses of Tamoxifen result in near complete labeling of the sinus venosus (Chen et al., 2014b). We show here that dosing at E9.5/10.5 labels ~ 90% of sinus venosus-derived ECs on the dorsal side of the heart (Figure 5m,o) and a majority of the cardinal vein (Figure 5—figure supplement 1e). However, the relative incidence of PROX1+/mTmG+ LECs on the dorsal side of the heart was 25% within lymphatic vessels and 21% in the isolated clusters (Figure 5m). Similar labeling was detected on the ventral side (Figure 5n), with 35% PROX1+/mTmG+ cells within lymphatic vessels and 17% in the isolated clusters (Figure 5o). In addition, we detected a large population of lymphatics proximal to the OFT region that was not traced by APJCreERT2 (Figure 5—figure supplement 1f), which was also reported to be negative for Tie2Cre labeling (Klotz et al., 2015). These data suggest that both ventricular lymphatics and isolated LECs are only partially derived from the APJ+ lineage. Since the isolated LECs first appear as single cells associated with the blood vasculature, it is tempting to speculate that they may originate from cells carried by the blood circulation. Alternatively, they could arise from others, yet unknown non-endothelial sources.

Taken together, our results show that mammalian cardiac lymphatics form by both sprouting lymphangiogenesis and coalescence of isolated LECs (lymph-vasculogenesis). Using multiple lineage-tracing strategies, we show that neither the local coronary vasculature nor the endocardium, contribute to the different cardiac lymphatic subsets. Yet, it is important to keep in mind that despite the high recombination efficiency of the BmxCreERT2, we cannot rule out the existence of additional progenitor populations nested within the endocardium that are not labeled by this tracer.

Differential response of cardiac lymphatics to injury

Given the importance of lymphatic vessels for immune surveillance and fluid homeostasis, it seems reasonable to hypothesize that they play similar roles during organ regeneration. Previous studies have highlighted a clear association between increased lymphatic vessel density following MI and improved cardiac function (Klotz et al., 2015; Henri et al., 2016; Ishikawa et al., 2007). In addition, enhancing the lymphangiogenic response after MI has been shown to augment immune cell trafficking (Vieira et al., 2018). Nevertheless, the cellular origins and mechanisms of response of different lymphatic subtypes to cardiac injury have not been addressed. Unlike mammals, zebrafish have a remarkable ability to regenerate their heart, making it an ideal model to study the origins and functions of the lymphatic system during organ regeneration. We employed the well-established model of cardiac cryoinjury, which closely models MI and allows visualization of the vascularization process (González-Rosa et al., 2011; Marín-Juez et al., 2016) (Figure 6a), to investigate the response of different cardiac lymphatic subsets to injury. Injuries were performed on 6–18 mpf (fish size 25–32 mm) transgenic zebrafish, hearts were harvested at 40 hr post cryoinjury (hpci) or at 7, 14, 21, and 73 days post cryoinjury (dpci). Notably, we found that prox1a+ sprouts (Figure 6b,c, white arrows), as well as new prox1a+ isolated LECs (Figure 6b,c, yellow arrows) were the first to be detected in the injured area, as early as at ~40 hpci. While part of these sprouts were also labeled by the blood EC marker flt1_9a (Figure 6c insets, white arrows) and may represent blood arterioles (Marín-Juez et al., 2016), the isolated LECs in the injured area were not marked by the arterial-specific transgene (Figure 6c, insets, yellow arrows), supporting their lymphatic identity. As a whole, the large majority of lymphatics in the injured area between 40 hpci-14 dpci, were not connected to large collecting lymphatics of the OFT, or to ventricular lymphatics (Figure 6d–i, Figure 6—figure supplement 1a,b,d,f). While we could not determine the origins of the injury-specific LEC clusters, which did not appear in sham operated hearts (Figure 6—figure supplement 1c,e,g), they were highly heterogeneous both in morphology and gene expression. We could detect cells expressing combinations of LEC markers, such as prox1a and flt4 (Figure 6f, inset, white arrow, Figure 6—video 1), as well as cells expressing prox1a (Figure 6f) or flt4 only (Figure 6f, inset, blue arrow). Interestingly, isolated LECs in the injured area developed into long sprouts (Figure 6h–j, inset, Figure 6—video 3), which continued growing throughout the regenerative phase and could be detected even after 2 months pci (Figure 6l,m, Figure 6—figure supplement 1g,h, Figure 6—video 2).

Figure 6 with 5 supplements see all
Differential response of cardiac lymphatics to injury.

(a) Diagram depicting the cryoinjury procedure. Injured area is outlined in all images, insets show high-magnification of dashed boxes. (b) flt4;prox1a transgenic hearts at 40 hpci showing prox1a+ sprouts (yellow arrow) and isolated LECs (blue arrow) in the injured area (n = 5). (c) Tg(flt1_9a_cFos:GFP);Tg(prox1a:KalTA4-UAS:uncTagRFP) double labeled coronary sprouts (inset, yellow arrow), as well as prox1a+ isolated LECs (inset, blue arrow) are detected in the injured area at 40 hpci. (d–f) seven dpci injured hearts of (d) Tg(kdrl:nls-mCherry);Tg(flt4BAC:mCitrine) (n = 5), (e) Tg(kdrl:nls-mCherry);Tg(mrc1a:EGFP) (n = 5), and (f) Tg(prox1a:KalTA4-UAS:uncTagRFP);Tg(flt4BAC:mCitrine) (n = 5) fish, with white arrows pointing to OFT-connected ventricular lymphatics, and blue arrows pointing to isolated LECs in the injured area. (g–i) 14 dpci ventricles of (g) Tg(flt1_9a_cFos:GFP;Tg(lyve1b:dsRed2) (n = 8), (h) Tg(prox1a:KalTA4-UAS:uncTagRFP) (n = 5) and (i) Tg(prox1a:KalTA4-UAS:uncTagRFP);Tg(flt4BAC:mCitrine) (n = 5) showing isolated lymphatic sprouts in the injured area (blue arrows), which are not connected to ventricular lymphatics (white arrow) (j) Double labeled prox1a;flt4 ventricular lymphatic sprouts invade the injured area at 14 dpci (inset, white arrows) (n = 3). (k) 73 dpci Tg(kdrl:nls-mCherry);Tg(flt4BAC:mCitrine) heart showing increased lymphatic coverage in the injured vs. uninjured areas of the ventricle, quantified in (l). Error bars, mean ± s.e.m. *p<0.001. Scale bars are 200 µm. Fish size 25–30 mm.

Contrary to the rapid lymph-vasculogenic response, OFT-derived ventricular lymphatics reached the regenerating area only at ~1–3 weeks pci (Figure 6j, arrows), mostly following their own-damage, as lymphatic vessels that were remote from the injury site, did not respond at all (data not shown).

We then asked whether the same cues controlling the development of cardiac lymphatics are reactivated during injury. In line with the reduction in ventricular lymphatics (Figure 3m–p) and isolated lymphatic clusters (Figure 4j–m) observed in flt4 -/- and vegfc+/- hearts, these were nearly absent in the vicinity, as well as within the injured area of ~1 month post-cryoinjury (mpci) mutant hearts (Figure 7a–f, arrows). Even at ~2 mpci, when an extensive lymphatic network covered the injury in wt hearts, vegfc+/- hearts remained largely devoid of lymphatic vessels (Figure 7g–i, arrows). In line with the impaired post-injury neo-lymphangiogenesis (Figure 7a–i), the scar area was increased in injured flt4 -/- and vegfc+/- hearts (Figure 7j–o, arrow). Acid Fuchsin Orange-G (AFOG) staining of heart sections showed that while in control hearts the fibrotic scar was mostly resolved by ~1 mpci and replaced by new muscle tissue, flt4 -/- hearts displayed a prominent scar composed of a collagenous core and a thin layer of fibrin at the border of the wound (Figure 7j–l, arrow). Similarly, the scar was not resolved in vegfc+/- hearts even at ~2 mpci (Figure 7m–o, arrow).

Figure 7 with 1 supplement see all
cardiac regeneration is impaired in flt4/vegfc mutant fish.

Injured area is outlined in all images (a,b) lyve1b+ lymphatic capillaries (blue arrows) are detected in the injured area of 36 dpci Tg(fli1:EGFP); Tg(lyve1b:dsRed2) wt sibling (a) but not in flt4-/- hearts (b). (c) Quantification of lyve1b+ lymphatic vessel coverage in the injured area of 35–38 dpi flt4-/- hearts (nwt = 3, nflt4-/-=8, *p<0.001). (d–i) lyve1b+ lymphatic capillaries (blue arrows) are detected in the injured area of 28 dpci (d) and 65 dpci (g) Tg(fli1:EGFP); Tg(lyve1b:dsRed2) wt siblings, but not in vegfc+/- heterozygous fish (e,h). (f,i) Quantification of lyve1b+ lymphatic vessel coverage in the injured area of 21–28 dpi vegfc+/- hearts (nwt = 3 hearts, nvegfc +/-=4, *p<0.005) and 65–66 dpi vegfc +/- hearts (nwt = 10 hearts, nvegfc+/-=10, *p<0.005). (j,k) AFOG-stained sections at 36 dpci showing lack of regeneration in flt4 -/- hearts (k) as compared to wt siblings (j). Collagenous scar is stained in blue, fibrin in red, and cardiac muscle in orange. Black arrow points to scar. (l) Increased scar area (calculated as percent of ventricle) in flt4-/- hearts (nwt = 8 hearts, nflt4-/-=11, *p<0.05). (m,n) AFOG-stained sections at 65 dpci showing lack of regeneration in vegfc+/- hearts (n) as compared to wt siblings (m). Black arrow points to scar. (o) Increased scar area (calculated as percent of ventricle) in vegfc+/- hearts (nwt = 6 hearts, nvegfc+/-=3, *p<0.001). (p,q) lyve1b+ isolated LEC clusters (blue arrows) are normally detected in the injured area of 29 dpci Tg(flt1_9a_cFos:GFP);Tg(lyve1b:dsRed);cxcr4a-/- (q) and wt sibling (p) hearts. (r) Quantification of lyve1b+ lymphatic vessel coverage (relative to wt sibling) in 29 dpi cxcr4a -/- hearts (nwt = 4 hearts, ncxcr4a-/-=5). (s,t) AFOG-stained sections at 77 dpci showing impaired regeneration in cxcr4a-/- mutant (t) as compared to wt siblings (s). Black arrow points to scar. (u) Increased scar area (calculated as percent of ventricle) in cxcr4a-/- mutant hearts (nwt = 5 hearts, ncxcr4a-/-=5, *p<0.005). Scale bars are 200 µm. All fish size are 25–30 mm.

In contrast to these results, isolated lymphatic clusters, which develop normally in cxcr4-/- hearts (Figure 4h–i), could be readily detected following injury as well (Figure 7p–r), suggesting that similar programs control lymphatic growth during development and repair. Nevertheless, scar area was increased in cxcr4-/- mutant hearts at 77 dpci (Figure 7s–u, arrow), supporting previous studies suggesting a cell-autonomous requirement of Cxcl12/Cxcr4 signaling for CM and/or coronary migration (Harrison et al., 2015; Itou et al., 2012).

Discussion

In this study, we investigated the mechanisms underlying the development of the cardiac lymphatic system in zebrafish and mouse, and analyzed the response of the lymphatic system to cardiac injury. We demonstrate that the cardiac lymphatic vasculature is composed of distinct subsets, which respond differently to lymphangiogenic cues and display unique behaviors during cardiac injury and regeneration. Our results suggest that cardiac injury induces the de novo formation of lymphatics, through a process reminiscent of lymph-vasculogenesis rather than sprouting lymphangiogenesis. The heterogeneous molecular nature of the newly formed lymphatic sprouts, combined with the fact that they are initially not connected to the pre-existing cardiac lymphatic vasculature, may reflect on alternative LEC origins, specifically activated in response to injury. Finally, our results indicate that neo-lymphangiogenesis following cardiac injury is necessary for proper scar resolution and cardiac regeneration.

Recent studies have reported the presence of isolated LECs in different organs during embryogenesis. In zebrafish for instance, a novel population of perivascular LECs was found in the meninx. These cells differentiate into dispersed, non-lumenized structures that act as scavenger cells and persist throughout life (Bower et al., 2017b; van Lessen et al., 2017). In the mouse, isolated LECs have been shown to contribute to lymphatic vessels in the skin (Martinez-Corral et al., 2015; Pichol-Thievend et al., 2018), the mesentery (Stanczuk et al., 2015) and most recently also the intracranial and spinal meninges (Antila et al., 2017). We find that in the mouse heart, isolated LECs represent transient structures that are no longer detected by P23, once the lymphatic system is fully formed. In zebrafish on the other hand, the presence of isolated LECs is detected at all stages, most probably correlating with the late onset and continuous growth of cardiac lymphatics throughout adult life. Moreover, these clusters may serve as an available source of LECs to support lymphatic growth in response to cardiac injury or other tissue needs.

Our analyses of the cellular origins of cardiac lymphatics in zebrafish and mice provide novel insights into the ontogeny of this system. We find that in zebrafish, the OFT lymphatics along with their derived ventricular lymphatics, originate in the VFL (Figure 2), a late forming facial lymphatic vessel. Interestingly, the zebrafish facial lymphatic network itself has recently been shown to form from three progenitor populations. Two of them are of venous origins (CCV and PHS) while the third one is angioblast-derived, and contributes not only to facial lymphatics but also to the hypobranchial artery (Eng et al., 2019). Our results therefore, indicating that OFT lymphatics sprout from facial lymphatics and not from the trunk vasculature, rise the appealing hypothesis that certain components of the heart, such as the OFT, along with their accompanying lymphatic vasculature could share common cardiopharyngeal mesoderm (CPM) origins (reviewed in Diogo et al., 2015). This idea, receives further support from recent lineage-tracing experiments in mice revealing that a population of Isl1-expressing CPM progenitors contribute LECs to the ventral side of the ventricles and the OFT (Maruyama et al., 2019).

The multiple spatiotemporally controlled lineage-tracing strategies revealed that in mice, both lymphatic vessels and isolated clusters have, at least in part, a venous origin. These two LEC populations may originate from different veins (such as the cardinal vein and sinus venosus). Alternatively, they may arise from the same vein, employing different mechanisms of migration and sprouting to reach the heart, thus ensuring proper and perhaps faster lymphatic coverage. Finally, it could also be possible that some cells detach from the parent lymphatic vessel and migrate through as an isolated cluster, as it has been shown for LECs in the lung, which bud from extra-pulmonary lymphatics and migrate as single cells or small clusters into the developing lung (Kulkarni et al., 2011).

Notably, the relative low fraction of LECs labeled by ApjCreERT2 (20–30%), compared to the high recombination efficiency of this Cre driver (~90%), hints at putative additional sources with major contribution to both isolated and ‘traditional’ cardiac lymphatics. One possibility is that the isolated LEC clusters originate from non-venous migratory progenitor cells as previously shown for other organs (Martinez-Corral et al., 2015; Stanczuk et al., 2015). Most recently, yolk sac born erythro-myeloid progenitors (EMPs) were shown to contribute to blood ECs in multiple organs, including the heart (Plein et al., 2018), raising the possibility that they could also contribute to lymphatic ECs in the heart. Finally, whether the isolated LEC clusters in mouse and zebrafish derive from the CPM remains an open question.

While debate continues over the venous vs. non-venous origins of cardiac lymphatics: a large proportion of cardiac LECs has recently been traced by Pax3Cre, a well-established marker of the dermomyotome (a subset of the somitic paraxial mesoderm). In this case, PROX1-expressing LEC precursors could be traced by Pax3Cre while still located in the dorsolateral wall of the CV (Engleka et al., 2005), suggesting that even the venous-derived cardiac lymphatics could be more heterogeneous than previously appreciated. In the future, it would be interesting to ascertain whether specific molecular characteristics or functions are ascribed to LECs of distinct origins.

Our results uncover significant differences in the series of events leading to the establishment of distinct lymphatic populations within the heart, and exemplify how each lymphatic subset responds differently to the same molecular cues. We find that the upregulation of pro-lymphangiogenic cues in the OFT, highly correlates with changes in heart rate associated with larva to juvenile transition. Hence, the increased demand for fluid drainage appears to elicit a lymphangiogenic response leading to the formation of OFT lymphatics. Interestingly, while lymphatics colonize the OFT prior to the appearance of the blood vasculature, ventricular lymphatics closely follow coronary vessels in both fish and mice, and fail to sprout in zebrafish cxcr4a mutants, which bear an immature coronary plexus. These results, along with the correlated increase in blood and lymphatic vessel growth induced by PHZ and Vegfaa-OE, strongly support a role for the coronary vasculature in ventricular lymphatic growth. We further show that lymphatics fail to form in zebrafish vegfc mutants, even when hearts are hyper-vascularized as a result of Vegfaa-OE, suggesting that a mature coronary plexus may act as a scaffold for ventricular lymphatic development, in a Vegfc-dependent manner. In contrast to ventricular lymphatics, the isolated LEC clusters are spread throughout the ventricle, and do not follow a specific path. Therefore, their formation was not affected by changes in the coronary vasculature, that is they were normally detected in cxcr4a mutant hearts and were not precociously induced by Vegfaa-OE or PHZ treatment (Figure 4). Yet, this population was fully absent from flt4 -/- and vegfc+/ - hearts, pointing to an additional source, other than coronary ECs, for Vegfc. While Vegfc has been previously shown to be expressed in maturing coronary artery ECs (Su et al., 2018), supporting a role of blood vessels in guiding lymphatic growth, Vegfc and Ccbe1 (Bonet et al., 2018) expression has also been detected in the epicardium in mice (Chen et al., 2014b), as well as in fish CMs under certain conditions (Marín-Juez et al., 2016). As a whole, the formation of the different lymphatic subsets in zebrafish, clearly follows the tissue requirements. OFT lymphatics develop during larva to juvenile transition in response to hemodynamic changes eliciting the demand for fluid drainage. Later on, ventricular lymphatics sprouts in response to cardiac muscle expansion and coronary formation. Finally, the LEC clusters emerge and incorporate into a growing capillary network, thereby increasing myocardial lymphatic coverage.

While previous reports have described an increase in lymphatic vessel density following MI (Klotz et al., 2015; Henri et al., 2016), little is known about the cellular and molecular mechanisms by which cardiac lymphatics form in response to injury. Interestingly, we find that the majority of lymphatics in the zebrafish regenerating area form de novo, whereas only a few arise from pre-existing ventricular lymphatics. In similar fashion, recruitment of isolated LECs was shown to take place during wound healing in the adult mouse skin, which later interconnect through a process reminiscent of lymph-vasculogenesis (Boardman and Swartz, 2003; Rutkowski et al., 2006). In contrast, at the periphery of the wound, lymphangiogenesis occurs by sprouting of pre-existing lymphatic vessels (Paavonen et al., 2000). As wound healing and regeneration share common features, including an immediate inflammatory response, revascularization, innervation and formation of a fibrotic scar (Richardson, 2018) it seems reasonable to speculate that similar mechanisms may control the formation of lymphatics in both of these processes. Nonetheless, our results indicate that all lymphatics in the injured area derive from intra-cardiac sources. These could include the isolated LECs, or LECs detaching from ventricular lymphatics, as recently shown for collateral artery assembly following MI in mice (Das et al., 2019). Another potential source is the endocardium. A subset of hemogenic angioblasts, expressing cardiac markers has been previously shown to contribute to the endocardium and to serve as a source for transient definitive hematopoietic progenitors (Nakano et al., 2013; Zamir et al., 2017). Moreover, the endocardium itself was found to possess angiogenic capabilities (Wu et al., 2012; Chen et al., 2014b). While we could not detect definitive endocardial contribution to developing cardiac lymphatics (Figure 5), the endocardium may hold the ability to give rise to LECs under pathological conditions.

Regardless of their origin, studies emerging in recent years have ascribed a beneficial role for lymphangiogenesis promoting therapies in different aspects of MI pathology, including cardiac edema, inflammation and scarring (Klotz et al., 2015; Henri et al., 2016; Vieira et al., 2018). Moreover, exogenous supply of VEGF-C has been shown to improve cardiac function after MI in mice (Klotz et al., 2015). Here, we show that an intact lymphatic system is instrumental for cardiac regeneration in zebrafish, as flt4 and vegfc+/- animals fail to regenerate their hearts, despite the presence of a normal coronary vasculature (Figure 7). While we cannot exclude the possibility that Vegfc/Flt4 signaling is autonomously required in other cell populations (e.g. blood ECs and/or macrophages), our data point to the absence of lymphatic vessels as the main reason precluding heart regeneration.

Most recently, an additional study has also examined cardiac lymphatics in zebrafish carrying a hypomorphic allele of vegfc (Le Guen et al., 2014) and a null mutation in vegfd (Bower et al., 2017a). While double heterozygote animals showed a marked reduction in ventricular lymphatics similar to the vegfc+/- mutants used in this study, double mutants were almost completely devoid of cardiac lymphatics, as were the flt4-/- fish described herein. Strikingly, cardiac regeneration was normal in most of the vegfc+/ -;vegfd -/- double mutant hearts, with only some displaying impaired regeneration capacities (Vivien et al., 2019). While the penetrance of the phenotypes described by Vivien et al. was lower than the observed in this study, perhaps due to the use of different mutant alleles, both studies describe a robust lymphangiogenic response to cardiac cryoinjury and point to a role for cardiac lymphatics in cardiac regeneration. In the future, further analyses will be required in order to ascertain what is the precise function that the different lymphatic subsets play in cardiac regeneration.

Materials and methods

Key resources table
Reagent type
(species)
or resource
DesignationSource or referenceIdentifiersAdditional
information
Strain
(Danio rerio)
Tg(fli1:EGFP)yl(Nicenboim et al., 2015)ZDB-ALT-011017–8
Strain
(Danio rerio)
Tg(lyve1b:dsRed2)nz101(Nicenboim et al., 2015)ZDB-ALT-120723–3
Strain
(Danio rerio)
Tg(gata1a:dsRed)sd2(Nicenboim et al., 2015)ZDB-ALT-051223–6
Strain
(Danio rerio)
TgBAC(prox1a:KalTA4-4xUAS-E1b:uncTagRFP)nim5(Nicenboim et al., 2015)ZDB-ALT-140521–3
Strain
(Danio rerio)
Tg(flt1_9a_cFos:GFP)wz2(Nicenboim et al., 2015)ZDB-ALT-150723–14
Strain
(Danio rerio)
Tg(flt4BAC:mCitrine)hu7135(van Impel et al., 2014)ZDB-ALT-140521–1
Strain
(Danio rerio)
Tg(mrc1a:EGFP)y251(Jung et al., 2017)ZDB-ALT-170717–2
Strain
(Danio rerio)
Tg(kdrl:nls-mCherry)y173(Fujita et al., 2011)ZDB-ALT-110429–4
Strain
(Danio rerio)
Tg(kdrl:EGFP)s843(Jin et al., 2005)ZDB-ALT-050916–14
Strain
(Danio rerio)
cxcr4aum20(Siekmann et al., 2009)ZDB-ALT-091124–1
Strain
(Danio rerio)
cxcl12bmu100(Bussmann et al., 2011)ZDB-ALT-110513–2
Strain
(Danio rerio)
vegfcum18(Villefranc et al., 2013)ZDB-ALT-130718–3
Strain
(Danio rerio)
Tg(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa)pd262; Tg(cmlc2:CreER)(Karra et al., 2018)ZDB-ALT-181129–18
Strain
(Danio rerio)
flt4um203(Kok et al., 2015)ZDB-ALT-160721–30
Strain
(Danio rerio)
vegfdbns257This paperN/ACRISPR/Cas9 generated. Prof. Didier YR Stainier (Max Planck Institute for Heart and Lung Research, Germany)
Strain
(Mus musculus)
CD1 (wild type)Charles River LaboratoriesStrain# 022
Strain
(Mus musculus)
FVB (wild type)Charles River LaboratoriesStrain# 207
Strain
(Mus musculus)
ApjCreERT2(Chen et al., 2014b)MGI:5689869
Strain
(Mus musculus)
BmxCreERT2(Ehling et al., 2013)MGI:5513853
Strain
(Mus musculus)
AplnCreERT2(Liu et al., 2015)MGI:5637737
Strain
(Mus musculus)
Ccbe1(Bos et al., 2011)N/A
Strain
(Mus musculus)
RosamTmG(Muzumdar et al., 2007)Stock# 007676
Strain
(Mus musculus)
RosaTdTomato(Muzumdar et al., 2007)Stock #007909
Sequence-based reagentcxcl12a_FThis paperPCR primersCGTAGTAGTCGCT CTGATGG
Sequence-based reagentcxcl12a_RThis paperPCR primersTGGGACTGTGTTG ACTGTGGAA
Sequence-based reagentcxcl12b_FThis paperPCR primersGGAGCATCCGAGA GATCAAG
Sequence-based reagentcxcl12b_RThis paperPCR primersTGTTCTTCAGCTT GGCAATG
Sequence-based reagentVegfc_F(Astin et al., 2014)PCR primersAAGGGCCCTAACA GAATGTC
Sequence-based reagentVegfc_R(Astin et al., 2014)PCR primersTTTGAATGAAGGG TGTCAGG
Antibodyanti-PROX1 (Rabbit polyclonal)AbcamCat# 11941IF(1:700)
Antibodyanti-VE-Cadherin (Rat polyclonal)BD PharmingenCat# 550548IF(1:100)
Antibodyanti-PROX1 (Goat polyclonal)R and D SystemsCat#: AF2727IF(1:300)
Antibodyanti- ERG (Rabbit monoclonal)AbcamCat#: ab92513IF(1:1000)
Antibodyanti- LYVE-1 (Rat monoclonal)eBiosciencesCat#: 14-0443-80IF(1:100)
AntibodyAlexa Fluor Conjugated Secondary Antibodies (488,594,633,635,647)Life TechnologiesN/AIF(1:250)
Chemical compound, drugAtenololSigma AldrichA7655
Chemical compound, drugPhenylhydrazine hydrochloride (PHZ)Sigma Aldrich78690
Chemical compound, drug4-hydroxytamoxifenSigma AldrichH7904
Commercial assay or kitQtracker705InvitrogenQ21061MP
Commercial assay or kitAcid Fuchsin Orange-G (AFOG)DIAPATH010307
Software, algorithmAngiotool(Zudaire et al., 2011)N/A
Software, algorithmImage JNIH (https://www.nih.gov/ij/)N/A

Zebrafish husbandry and transgenic lines

Request a detailed protocol

Zebrafish were raised by standard methods (Gibbs-Bar et al., 2016) and handled according to the guidelines of the Weizmann Institute Animal Care and Use Committee. Zebrafish lines used in this study were: Tg(fli1:EGFP)yl, Tg(lyve1b:dsRed2)nz101, Tg(gata1a:dsRed)sd2, TgBAC(prox1a:KalTA4-4xUAS-E1b:uncTagRFP)nim5 and Tg(flt1_9a_cFos:GFP)wz2 (Nicenboim et al., 2015); Tg(flt4BAC:mCitrine)hu7135 (van Impel et al., 2014), Tg(mrc1a:EGFP)y251 (Jung et al., 2017), Tg(kdrl:nls-mCherry)y173 (Fujita et al., 2011), Tg(kdrl:EGFP)s843 (Jin et al., 2005), Tg(myl7:GFP) (González-Rosa et al., 2011), cxcr4aum20 (Siekmann et al., 2009), cxcl12bmu100 (Bussmann et al., 2011), vegfcum18 (Villefranc et al., 2013) flt4um203 (Kok et al., 2015), Tg(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa)pd262; Tg(cmlc2:CreER) (Karra et al., 2018). vegfdbns257 mutants were generated by targeted genome editing using the CRISPR/Cas9 system as previously described (Matsuoka et al., 2016). The second exon of vegfd, which encodes part of the Vegfd protein coding sequences prior to the Vegf homology domain was targeted. The vegfdbns257 mutant allele harbors a 59 base-pair insertion in the exon 2 of vegfd predicted to lead to a premature stop codon at tyrosine residue 56 thus yielding a truncated polypeptide containing a stretch of 55 amino acids of Vegfd. The following guide RNA (gRNA) sequence was used to target the exon 2 of vegfd: 5’- GATGTTGACCGAATACC −3’. 1 nl of a solution containing 250 ng/µl of Cas9 mRNA and 100 ng/µl of gRNA was injected at the one-cell stage. Wt and vegfdbns257 animals were identified by PCR using the following primers at expected PCR product sizes (127 bp for wt allele and 186 bp for vegfdbns257 allele): vegfdbns257 forward primer: 5’- GACACAAATCAGGAAAAGTGG −3’ vegfdbns257 reverse primer: 5’- CATCGAAGTGCTTCAGCTTG −3’

Experiments were conducted on fish from the same clutch, which were of same age (weeks post-fertilization), and size -based on standard body length (the distance from the snout to the caudal peduncle) (Parichy et al., 2009). Initially fish were selected based on their age (~22 dpf VFL development, ~3–8 wpf OFT lymphatics development,~10–20 wpf OFT lymphatics in mutants, ~22–24 wpf ventricular lymphatics). Subsequently, they were anesthetized by immersion into 0.04% tricaine and placed on a ruler for measuring standard body length. Fish above or below a certain size range, as stated in the text, were excluded from the experiment.

For imaging of up to four wpf larvae, embryos were either treated for 7 days with 0.003% N-phenylthiourea (PTU) (Sigma, St Louis, MO) to inhibit pigment formation or casper (roy-/-;nacre-/-) mutant fish were used (White et al., 2008).

Mice

Mouse use followed Stanford IACUC guidelines. Strains used were wild type (CD1 and FVB, Charles River Laboratories), ApjCreERT2 (51), BmxCreERT2 (Ehling et al., 2013), AplnCreERT2 (Liu et al., 2015) and Ccbe1 (Bos et al., 2011).

Cre lines were crossed to RosamTmG or to RosaTdTomato (Muzumdar et al., 2007). Pregnant females were dosed intraperitoneally (4 mg of tamoxifen in corn oil at E8.5 for BmxCreERT2, E9.5+E10.5 ApjCreERT2 or E12.5 for AplnCreERT2 and ApjCreERT2). Hearts were analyzed at E13.5-P23.

Cryoinjury

Request a detailed protocol

Cryoinjury was performed as described (González-Rosa et al., 2011). Briefly, 6–12 mpf (month post fertilization) fish were anesthetized by immersion into 0.04% tricaine (Sigma, St Louis, MO). A small incision was made through the body wall and the pericardium using microdissection scissors, and a cryoprobe cooled in liquid nitrogen, was placed on the ventricular surface until thawing was observed. Fish were then returned to fresh water for recovery.

Angiography

Request a detailed protocol

Angiography was performed on anesthetized fish by retro-orbital injection (Pugach et al., 2009) of Qtracker705 (Invitrogen Q21061MP). Fish were euthanized 2–5 min following injection.

qRT–PCR

Request a detailed protocol

qRT–PCR was carried out as previously described (Avraham-Davidi et al., 2012) using the following primers: vegfc (Astin et al., 2014)

cxcl12a_F 5’- CGTAGTAGTCGCTCTGATGG-3’

cxcl12a_R 5’- TGGGACTGTGTTGACTGTGGAA −3’

cxcl12b_F 5’- GGAGCATCCGAGAGATCAAG-3’

cxcl12b_R 5’- TGTTCTTCAGCTTGGCAATG-3’

Pharmacological treatments

Request a detailed protocol

To reduce heart rate, zebrafish larvae were treated daily with 100 µM of the beta-adrenergic antagonist Atenolol (Sigma-Aldrich, A7655) added to the fish water for 2.5 weeks starting at 17 dpf (Hein et al., 2015).

To induce cardiomegaly, 8 wpf fish were treated for 3 weeks with 2.5 µg/ml phenylhydrazine hydrochloride (PHZ, Sigma, 78690). To acclimate the fish to PHZ, the first treatment was of 30 min in 1.25 µg/ml PHZ solution. Every other day thereafter, fish were incubated for 1 hr in 2.5 µg/ml PHZ solution followed by 30 min wash in fish water (Sun et al., 2009).

To induce Vegfaa-OE, 67 dpf Tg(βactin2:loxP-mTagBFP-STOP-loxP-vegfaa)pd262 fish were treated for 24 hr with 5 µM 4-hydroxytamoxifen (Sigma, H7904) or 0.05% ethanol (vehicle control).

Immunohistochemistry and imaging

Request a detailed protocol

Zebrafish hearts were isolated from terminally anesthetized zebrafish and fixed briefly (30 s) in 4%PFA/PBS. For immunostaining, hearts were fixed in 4%PFA/PBS overnight, incubated in blocking solution (1% goat serum, 2% Triton X-100, 1%BSA in PBS) for 2–3 days and then with anti-PROX1 antibody (Abcam, 11941, 1:700) in staining solution (1% goat or donkey serum, 0.25% Triton X-100 in PBS) for 4–5 days at 4°C. Following PBST (PBS with 0.25% Triton X-100) washes for 6 hr, hearts were incubated with Alexa Fluor 488 conjugates secondary antibodies (Jackson ImmunoResearch, 111-485-045) diluted in blocking solution for 3–4 days at 4°C and washed again. Hearts were then mounted into glass capillaries (Brand) in 1.5% low-melting point agarose/PBS solution (ROTH) and imaged using a light sheet Z.1 microscope (Zeiss Ltd.) equipped with 2 sCMOS cameras PCO- Edge, 10X/0.2 excitation objectives and W-Plan Apochromat 20x/0.1 detection (water immersion). In vivo confocal imaging of larvae up to 4 wpf was performed using a Zeiss LSM 700 upright confocal microscope (Carl Zeiss) with a W-Plan Apochromat 20 × objective, NA 1.0.

Mouse embryos from timed pregnancies (morning of plug designated E0.5) were fixed in 4% paraformaldehyde (PFA) for 1 hr. Fixed tissues were left intact or sectioned. Immunofluorescence staining was performed in either 1.5 ml tubes with constant rotation (whole mount) or on microscope slides (tissue sections). Primary antibodies in blocking solution (5% goat or donkey serum, 0.5% Triton X-100 in PBS) were incubated overnight at 4°C followed by PBT (PBS with 0.5% Triton X-100) washes for 6 hr. Secondary antibodies diluted in blocking solution were incubated overnight at 4°C and washed again.

Antibodies: VE-Cadherin (BD Pharmingen, 550548; 1:100); PROX1 (R and D Systems, AF2727; 1:300); ERG (Abcam, ab92513, 1:1000); LYVE-1 (eBiosciences, 14-0443-80, 1:100). Secondary antibodies were Alexa Fluor conjugates (488, 555, 594, 633, 635, 647, Life Technologies; 1:250).

Samples were imaged in Vectashield (Vector Labs) using either a Zeiss LSM-700 or Axioimager A2 epifluorescence microscope.

Acid Fuchsin Orange-G (AFOG) staining was performed on paraffin-embedded tissue sections using an AFOG staining kit (DIAPATH, 010307) following manufacturer’s instructions. Samples were imaged using Panoramic SCAN II (3DHISTECH) slide scanner.

Statistical analyses

Request a detailed protocol

Coronary blood and lymphatic vessel coverage area and total length, as well as ventricle and OFT size, were calculated from max. projected confocal images of the hearts, using Angiotool (Zudaire et al., 2011). OFT lymphatic sprouts and loops were manually counted using ImageJ/Fiji. Color coded local thickness maps of cxcr4 mutant hearts were generated using ImageJ/Fiji. Scar area following cryoinjury was measured in single section images using ImageJ, and the percent scar area was calculated with respect to the ventricle total area.

Data was analyzed using the unpaired two-tailed Student’s t-test assuming unequal variance from at least three independent experiments, unless stated otherwise. Numerical data represent mean ± s.e.m., unless stated otherwise. For the Atenolol experiment ordered logistic regression test was performed.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
    Nova Exervitatio Anatomia Exhibens Ductus Hepatico Aquosos Et Vasa Glandularum Serosa, Nune Primum Inventa, Aenesisque figuris Delineata
    1. O Rudbeck
    (1653)
    Västerås: Eucharius Lauringer.
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80

Decision letter

  1. Marianne E Bronner
    Senior and Reviewing Editor; California Institute of Technology, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

The study by Gancz et al. investigates the cardiac lymphatic vasculature in the zebrafish. The authors describe in staged images of transgenic lines that the lymphatics develop from the out flow tract (or bulbous arteriosis) progressively in juvenile stages. They show that the OFT lymphatics originate earlier from the ventral facial lymphatics. Several mutants for known regulators of lymphangiogenesis are presented that form abnormal OFT lymphatics in juvenile stages. They then suggest a proposed second origin for cardiac lymphatics in a population of Prox1 (transgene) positive cells that are isolated over the heart blood vasculature and are proposed to later integrate into the growing coronary lymphatics. In mouse, isolated Prox1+ cells of venous origin are identified and lineage traced to a venous origin. Finally, they describe a possible role for lymphatics following injury to the heart. The data provide interesting evidence for different origins of cardiac lymphatics in development and injury. This is a particularly topical question and certainly of interest to the cardiac and vascular fields.

Decision letter after peer review:

[Editors’ note: the authors were asked to provide a plan for revisions before the editors issued a binding decision. What follows is the editors’ letter requesting such plan.]

Thank you for sending your article entitled "Distinct origins and molecular mechanisms regulating lymphatic formation during cardiac growth and repair" for peer review at eLife. Your article is being evaluated by three peer reviewers, and the evaluation is being overseen by Marianne Bronner as the Senior and Reviewing Editor.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond within the next two weeks with an action plan and timetable for the completion of the additional work. We plan to share your responses with the reviewers and then issue a binding recommendation.

In particular, the reviewers found the mouse work very preliminary and suggest omitting it. It would be important to refine experimentation and convincingly demonstrate the nature and contribution of the isolated clusters. They suggest that you demonstrate the origins of lymphatics in the cryo-injury model, further characterize markers, and provide insight into their potential function. There is also the need for accurate and consistent staging of animals throughout as well as careful controls for the four mutants to ensure phenotypic specificity.

We hope you find the full reviews, attached below, helpful.

Reviewer #1:

In this study Gancz and co-workers make use of different zebrafish reporter and mutant lines to document the existence of lymphatics in the zebrafish heart and to investigate mechanisms underlying cardiac lymphatic growth during heart development and following adult (cryo-) injury. Specifically, the authors used confocal and light-sheet microscopy imaging to characterise lymphatic vessel formation and expansion in the outflow tract (OFT) and ventricular wall of hearts derived from juvenile to adult compound fli1, flt1_9a, kdrl flt4, lyve1b, prox1a, mrc1, myl7, kfl2abns11, kfl2bbns12, cxcr4um20, cxcl12bmu100, vegfcum18 or vegfdbns257 double reporter/mutant transgenic fish. This approach led to the identification of two distinct cardiac lymphatic populations in the zebrafish ventricle, which responded differently to pro-lymphangiogenic cues (e.g. VegfC) and which arise separately by sprouting lymphangiogenesis (from the OFT lymphatics) or coalescence of isolated lymphatic clusters located towards the apex of the ventricle. The latter cardiac lymphatic sub-population in particular was found to represent the main source for lymphatic growth following cardiac (cryo-)injury in zebrafish. Lastly, the authors used immunostaining against standard endothelial markers (VE-Cadherin, Prox1), different lineage-tracing models and confocal microscopy imaging to determine the existence of equivalent cardiac lymphatic populations in the developing mouse heart. Isolated Prox1-positive lymphatic clusters were present in the apical region of the developing ventricular wall (both dorsal and ventral sides) until post-natal (P) day 23, a time-point when these clusters became fully integrated in the cardiac lymphatic network. In addition, the authors traced back a modest contribution from the cardinal vein to the isolated lymphatic clusters, excluded contributions from existing coronary vessels/endocardium and suggested a requirement for an additional, yet unidentified, cell source.

In recent years the interest in cardiac lymphatics has increased significantly, particularly after findings demonstrating heterogeneity within the cardiac lymphatic vasculature in mouse and roles for lymphangiogenesis in cardiac repair following myocardial injury, which has important therapeutic implications. There is no report published thus far on the cardiac lymphatics in zebrafish, either during development of following heart injury, and there remains an important need to fully understand the cellular and molecular mechanism(s) governing cardiac lymphatic growth as such this study timely, novel and of widespread interest. Having said that, there are a number of substantial issues with the manuscript in its current form which need addressing:

1) Overall, the findings reported here come across as being somewhat preliminary. The authors aimed to characterise cardiac lymphatic growth in zebrafish and mouse using an array of transgenic models (including reporter, fate-mapping, mutants) at different developmental timepoints, but this is reported in an inconsistent manner with different reporters being used at different developmental timepoints and then compared together, or only one lymphatic marker, rather than binary combinations (at least), being used at each timepoint. This is particularly evident in the characterisation of cardiac lymphatics in zebrafish (Figure 1), where the authors concluded that the isolated ventricular lymphatic clusters only expressed prox1a (Figures 1F and 4A), but do not show equivalent images (same magnification and heart orientation) of hearts from prox1a/flt4, prox1a/mrc1, lyve1b/prox1a, lyve1b/mrc1 or prox1a/lyve1b (e.g. pending availability of a lyve1b:GFP or prox1a:GFP line) double transgenic fish at a single developmental stage (e.g. 16wpf as in Figure 1F). Also, for some of the mutant lines the authors characterised OFT lymphatics (Figure 2) at 10wpf (kfl2a and kfl2b), others at 20wpf (vegfc and vegfd) and others no specific detail of developmental stage (cxcr4a and cxcl12; also, no images of control OFTs are provided in Figure 2R, S) and no information on controls are provided in the corresponding text or figure legend. The analyses of ventricular lymphatics at 16wpf (Figure 3) appears problematic: given that OFT lymphatic formation (3-4wpf) takes places before lymphatics sprout into the ventricle (12-16wpf), why do the authors investigate OFT lymphatics at 20wpf and ventricular lymphatics at 16wpf in vegfc and vegfd mutants? To exclude potential developmental delays in the lymphatic growth/expansion the authors need to carry out their time-course studies more consistently across the different mutants.

2) The evidence for isolated lymphatic clusters in the ventricular wall (Figure 4) is not compelling. These are first described as being located towards the apex, but images in Figures 4B (12wpf) and 4E, G (20wpf) highlight prox1a-expressing LECs in the vicinity of the OFT region; how do the authors distinguish between sprouts from OFT lymphatics and putative isolated clusters if only the prox1a reporter line is used here? This is relevant for the cxcr4a (Figure 4E) and vegfc (Figure 4G) mutant analysis: in Figure 3 the authors showed reduced OFT lymphatics, but how can the authors exclude the possibility that the LECs observed at 20wpf are in fact delayed sprouts from OFT lymphatics? Also, in Figure 4D the authors included a graph detailing numbers of isolated LEC clusters from 8wpf to 16wpf but did not discuss this in the manuscript, nor did they include representative images for these developmental stages. The inclusion of such images is important to support the author's conclusions and make the findings more robust.

3) An important and outstanding question is what is the origin of the isolated LEC clusters in zebrafish? This may be somewhat beyond scope, but is an important omission which would significantly strengthen the overall findings.

4) There is an imbalance between the characterisation of zebrafish and mouse cardiac lymphatics, with the mouse data being descriptive (without functional targeting) and at a more preliminary stage. Arguably the mouse data could be removed without significantly detracting from the manuscript; or if it is to be retained the analyses need to be more in-depth and rigorous. As above, the authors should include (at least) binary combinations of lymphatic markers in their characterisation of cardiac lymphatics to establish the presence of isolated LEC clusters, in particular combine nuclear (Prox1) with membrane (VEGFR3, LYVE1, PDPN, NRP2) staining patterns. Also, images such as shown in Figure 5 would be more informative if channels were shown separately to properly assess whether LECs are isolated from the sprouting plexus and their localisation. An important outstanding question is when do the clusters first appear in the developing mouse heart? This requires a more detailed time- course of assessment.

5) In the mouse the main result arising from the different lineage-tracing approaches employed appears to be that isolated LECs are from a venous source, derived from the cardinal vein or sinus venosus. How do the authors explain the existence of isolated lymphatic clusters that share the same origin as the main lymphatics sprouts? This is a key question that relates to point 3) above for the zebrafish studies, in that the origin of these clusters across species still remains questionable.

6) With regards to the differences between the ApjCreERT2 and Tie2-Cre drivers in labelling murine cardiac lymphatics, the authors need to test the former at earlier time-points to ensure that the venous endothelium which will give rise to LECs is being targeted in an efficient manner. For instance, tamoxifen should be administered at E8.5 plus E9.5 to determine labelling of the jugular lymphatic sac followed by subsequent tracing into the cardiac lymphatics.

7) The isolated LEC clusters apparently become fully integrated within the lymphatic network by P23; what is their functional role in the postnatal heart (intact and following injury)? The authors should at least speculate on this in the Discussion.

8) In the cryo-injury zebrafish studies (Figure 6) the assessment of a role for these isolated lymphatic clusters in repair it is difficult to interpret in the absence of a fate-mapping model, since it is not possible to determine whether the same cell clusters observed during development are being analysed. Also, the authors need to include sham-operated controls and, as above, use the different binary combinations of reporter lines in a consistent manner across the different time-points analysed. In Figure 6G, the authors reported flt4-expressing vascular structures, which are negative for prox1a, sprouting into the injured area; could these be blood vessels? Flt4 expression is shared between blood and lymphatic endothelium during development, becoming restricted to the latter at later stages of development, but flt4 is also activated in neoangiogenic sprouts. This is supported by the observation of vascular structures positive for kdrl and flt4 in Figure 6J.

9) The regions highlighted in Figure 6G-K appear to be in close proximity to the main lymphatic sprouts, rather than isolated in apical areas as is suggested by the authors- this requires further clarification: a combination of membrane and nuclear LEC staining would enable a clear assessment of the clusters and would serve to identify if they are indeed isolated LECs or just leading cells from the sprouting lymphatic plexus.

Reviewer #2:

The study by Gancz et al. investigates the cardiac lymphatic vasculature in the zebrafish. The authors describe in staged images of transgenic lines that the lymphatics develop from the out flow tract (or bulbous arteriosis) progressively in juvenile stages. They show that the OFT lymphatics originate earlier from the ventral facial lymphatics. Several mutants for known regulators of lymphangiogenesis are presented that form abnormal OFT lymphatics in juvenile stages. They then suggest a proposed second origin for cardiac lymphatics in a population of Prox1 (transgene) positive cells that are isolated over the heart blood vasculature and are proposed to later integrate into the growing coronary lymphatics. In mouse, isolated Prox1+ cells of venous origin are identified and lineage traced to a venous origin. Finally, evidence is provided that suggests a possible role for lymphatics following injury to the heart.

The main interest is the proposal for different origins of cardiac lymphatics in development and injury. This is a particularly topical question and certainly of interest to the cardiac and vascular fields. The data provided are suggestive and interesting. However, there are a number of gaps and major concerns throughout the study. Specifically, from many key data points there are important controls missing. Also there are points where interpretation is not fully supported by the current data. These points need to be addressed before further considering publication of this study.

1) Zebrafish develop in a highly variable manner during the larval to juvenile transition. This variation occurs between and within clutches and is very well documented, thus staging with days post fertilisation once animals are beyond larval stages creates inaccuracies (see Parichy et al., 2009). As such, the standard approach to stage juvenile zebrafish is using body length – the authors should see the definitive study and guidelines in Parichy et al., 2009 (especially findings in Figures 3 and 4).

The use of days post fertilisation in this study rather than body length creates concerns because for many comparisons animals may not be of the same developmental age. For example: are the mutants in Figure 2 (vegfc, vegfd, cxcr4a, cxcl12) all the same length or are some mutants developmentally delayed? Likewise for the PHZ treated animals. If they are not truly stage matched, the OFT lymphatic phenotypes may just be secondary to developmental delay.

Can the authors please provide accurate staging using length measurements? Minimally this needs to be provided for key observations that could be significantly skewed by natural variation in fish populations at juvenile stages (such as for the mutant and drug treatments).

2) In Figure 2, the authors show that the OFT lymphatics derive from the ventral facial lymphatics. For all mutants analysed at the level of the OFT, the phenotypes could be due to (secondary to) reduced ventral facial lymphangiogenesis rather than suggesting a specific role for these genes in OFT lymphangiogenesis. Especially suggestive is the fact that vegfc heterozygotes already have systemically reduced lymphangiogenesis at 5 dpf (Villefranc et al., 2013). Please provide careful quantification and images of the ventral facial lymphatics in these mutants preceding the formation of the OFT lymphatics. Alternatively, temporally inducible transgenic models could be used to show that there are ongoing roles for these genes in OFT lymphangiogenesis.

3) In the mutant comparisons, mutants should be compared with siblings from the same clutch because there can be variation between clutches. This does not appear to be the case for vegfc and vegfd mutants, which are compared to just one population termed "wt". Please ensure comparisons and statistics are comparing mutants with clutch matched siblings.

4) The prox1a:RFP transgene uses a Gal4, UAS cassette to enhance signal. This system is known to lead to non-specific expression in tissues in some contexts. It has been previously noted that this transgene does not always accurately reflect endogenous Prox1 protein levels (see Koltowska et al., 2015 Figure 2 and Supplementary Figure 1).

In Figure 4, the proposed "clusters" do not express any lymphatic markers other than this transgene. The images appear (to this reviewers eye) to show blood vasculature that is co-expressing RFP rather than genuine distinct clusters of positively identified lymphatic endothelial cells. One interpretation of the current data from zebrafish could be that this expression is non-specific transgene expression in blood vessels over the heart.

Please show more comprehensive analysis of expression of multiple blood vascular and lymphatic markers in these clusters. Do they contain blood? Please also validate the cluster expression of Prox1 with antibody staining.

5) The use of exclusively nuclear markers in Figure 5 in defining the isolated Prox1 positive LEC clusters in mouse is problematic. Some of these "isolated" nuclei are not that far apart (e.g., in Figure 5B and C). Please provide a much more thorough marker analysis of these clusters in mice with more lymphatic and blood vascular markers and including reliable LEC markers that are membranous such as Nrp2, Lyve1 or Podoplanin. The concern is that cells could be connected by long, cytoplasmic/membranous connections not seen with nuclear stains. Further marker and anatomical analyses would significantly clarify the nature of these clusters.

6) In Figure 6, the authors see lymphatics arising in a wound site following cryoinjury using fixed stage analysis. The origins of these lymphatics have not been lineage traced or time-lapse imaged to allow any claims as to their cellular origins. The wound site in Figure 6B for example is adjacent to the major ventricular lymphatic and only one side of the heart is imaged so there could be major lymphatics in the non-imaged half.

These cells could have arisen by detaching from and migrating from the major OFT derived ventricular lymphatics or could have come from other sources based on the analysis provided.

Further data are required to support the authors claims that "… cardiac injury induces the de-novo formation of lymphatics, through a process reminiscent of lymph-vasculogenesis rather than sprouting lymphangiogenesis."

Reviewer #3:

This study by Gancz and colleagues assesses the mechanisms regulating cardiac lymphatic development in zebrafish and mouse. Using several zebrafish transgenic lines, the authors identified two distinct lymphatic populations expressing different transgenic markers and with different origins. One population of lymphatic vessels appears in the OFT at 3-4wpf, expand to cover the entire OFT and finally sprouts towards the ventricle. The authors found that OFT lymphatic vessels originate from a pre-existing "ventral facial lymphatic" vessel that invades the OFT during metamorphosis. The other population of lymphatic vessels derives from coalescence of isolated LECs clusters of unknown origin in the ventricle.

Through qPCR, the authors observed the upregulation of the pro-lymphangiogenic factors klf2a, vegfc, cxcl12a and cxcl12b in the OFT during metamorphosis, coinciding with the establishment of the cardiac lymphatic system. Genetic mutants were used to investigate the potential role of these factors for this process. Because of embryonic lethality, the authors had to perform their experiments with heterozygous mutants. They found that the distinct lymphatic subsets differentially respond to these pro-lymphangiogenic signals during growth.

The ventricular lymphatics deriving from OFT vessels sprouts in close association with the coronary vasculature. The requirement of the coronary vasculature for lymphatic sprouting was assessed through analysis of cxcr4a mutants and PHZ-treated animals.

Similar LEC population was identified in the mouse heart. The origin of these cardiac lymphatics system was assessed using cell lineage tracing experiment, showing that the different subsets do not derives from the local coronary vasculature, nor from the endocardium.

Finally, the regeneration of lymphatic system was assessed in zebrafish using the cryoinjury model. The authors observed differences in the response of the various lymphatic populations to the injury. Prox1a+ vessels rapidly invade the injury area. However, the majority of lymphatic vessels in the regenerating area form de-novo as isolated clusters, suggesting lymph-vasculogenesis rather than sprouting lymphaniogenesis as the main mechanism for lymphatics regeneration.

Overall, this is an interesting study of high standards. The findings contribute to our knowledge about development of cardiac lymphatics in zebrafish. However, the conclusions based on analysis of mutants are sometimes not convincingly supported by the results. Furthermore, the part with the cryoinjury model contains descriptive data, without testing the role of lymphatics in cardiac regeneration.

- "we detected a significant reduction in OFT lymphatic coverage (Figure 2J-M), with under-developed vessels displaying fewer sprouts that those of their wt siblings". In the Figure 2M, the graph indicates that the percentage of lymphatic coverage was higher in the mutants than in wt, which is in contradiction with the statement of the authors.

- ".., we could not observe any defects in OFT lymphatics of vegfd mutant fish (Figure 2N, P, Q)…" The vegfd mutant was generated for this study using the Crispr/Cas9 system. Do these mutants display any phenotype, indicative of a gene loss of function?

- "Our results suggest that while VegfC is a key determinant of OFT lymphatic formation, the Cxcr4/Cxcl12 axis is mostly involved in patterning and remodelling of the OFT lymphatic plexus". I don't see any differences between the phenotype of vegfc and cxcr4a mutants (Figure 2O and R). I found that this conclusion is too strong for these data. How can we conclude that VegfC is a key determinant of OFT lymphatic formation, given that lymphatic vessels were still formed? The authors should describe more precisely the phenotype of the mutants and prove that the lymphatic function is also altered.

It would be also interesting to test if cardiac lymphatic vessel development in the ventricle is also impaired in these mutants.

- ".… suggesting a possible role for the major coronary vessels in guiding lymphatic sprouts". The authors cannot exclude that the treatment itself influences directly lymphatic development.

[Editors' note: Formal revisions were requested, following approval of the authors’ plan of action. Further revisions were then requested prior to acceptance, as described below.]

Thank you for submitting your article "Distinct origins and molecular mechanisms regulate lymphatic formation during cardiac growth and repair" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Marianne Bronner as the Senior and Reviewing Editor. The reviewers have opted to remain anonymous.

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

Summary:

The major findings, describing the development and response to injury of the zebrafish cardiac lymphatics are of sufficient interest for publication in eLife and it is noted this is to be accompanied by the "sister" paper by Harrison and colleagues, also under revision. While the paper is much improved, the reviewers feel that a few changes are still needed as summarized below. The full reviews also are included below for further clarification.

Essential revisions:

i) Additional work is required to definitively prove the existence of LEC clusters in zebrafish hearts as distinct from the main venous-derived lymphatic plexus; specific experiments using multiple stainings and z-stack confocal projections, etc are suggested in the report.

ii) The mouse studies remain quite preliminary and still require additional work to add any insight over and above what has already been published. The clusters are not clearly demonstrated and there is no insight into origin. At the moment it detracts from the fish studies, so the mouse data should be dropped and instead the authors should focus on definitively demonstrating that LEC clusters exist in the fish heart.

Reviewer #1:

In this revised manuscript, Gancz and co-workers have improved the original submission with new data and convincingly document the development of lymphatic vessels in the zebrafish heart after the larva-to-juvenile transition, as well as their response/requirement following cryoinjury in the adult heart. The new data include the binary combination of markers to characterise cardiac lymphatics (including antibody staining to complement their transgenic lines), the inclusion of images from age/size-matched controls, removal of data deemed too preliminary (e.g. kfl2a/b data) and inclusion of new models, such as flt4-/- and vegfAa-OE. Having said that, some of the issues noted in the original submission still remain to be addressed. In particular, the authors still need to more convincingly demonstrate the following:

1) The existence of isolated lymphatic clusters in the developing zebrafish heart and in the adult zebrafish heart following cryo-injury.

The observation that the prox1a:RFP transgene is also active in blood endothelium (subsection “Morphological and molecular heterogeneity of zebrafish cardiac lymphatics”, last paragraph) makes it challenging to determine whether the putative clusters reported here are truly lymphatic in nature, or represent an artefact of the transgenic reporter (Figure 4 and Figure 1—figure supplement 2). The authors employed immunostaining using an antibody against prox1 to validate their transgenic approach, and whilst this revealed that the OFT and connecting ventricular lymphatic branches were clearly positive for nuclear expression of prox1 the majority of RFP-labelled ventricular vessels were negative leaving a question mark as to the source of the prox1a:RFP+ cells (Figure 1—figure supplement 2D versus E). Moreover, a similar concern rests with the other two reporters employed (flt4BAC:mCitrine, lyve1b:dsRed2), in that both flt4 and lyve1 have been shown to be expressed in blood endothelium. Consequently, how can the authors be sure that the putative ventricular lymphatic clusters reported are not subsets of blood endothelial cells activating the mCitrine and dsRed reporters, respectively? Equally, how do these transgenic lines "behave" in the cryoinjury setting, i.e. does injury (hypoxia, cell death?) induce ectopic expression of the transgenes? These are critical issues confounding the interpretation of the data generated through the use of the different transgenic reporters and need to be fully addressed to support the notion of isolated LEC clusters. In this regard, the authors should carefully map out the overlap in blood and lymphatic endothelium, both in development and post-injury, for each of the reporters in combination with specific antibodies against lymphatic markers, akin to prox1 in Figure 4 and Figure 1—figure supplement 2.

2) A requirement for coronary vessels in cardiac lymphatic expansion in the ventricular wall.

The new data derived from the vegfAa-OE model in a vegfC+/- genetic background does not support the authors' conclusion for a requirement for coronary vessels in ventricular lymphatic vessel expansion. In this model vegfC haploinsufficiency is ubiquitous, rather than specific to coronary endothelium, and other cell types have been reported to express vegfC in the developing heart, such as epicardial cells (Chen et al., 2014). Therefore, the reduction of lymphatic growth is quite likely to be downstream of a general reduction in levels of vegfC arising from multiple cell types. Likewise, the observation of increase lymphatic expansion following vegfAa-OE may reflect an overall increase in ventricle size and the need for an adequate lymphatic coverage, akin to PHZ treatment. The authors need to discuss these possibilities and tone down their conclusions.

3) The existence of isolated lymphatic clusters in the developing mouse heart.

Unfortunately the mouse data still remain preliminary and inconclusive regarding the presence of discrete clusters of LECs of non-venous origin. The authors have included a new panel c in Figure 5, but this is suboptimal as the region shown is a high-power view and, therefore, the relationship to the main lymphatic network developing in the E15.5 murine heart is unclear. The authors should provide a low-power view containing the entire ventricular surface. Also, in order to confirm that the "multicellular structures" marked with the blue arrowheads are detached from the unmarked vascular structures, the authors need to include a representative z-stack. Specifically, re Figure 5E the z-stack depth in the lower power image is different from the high-power inset, and the region of interest appears to be connected to the main lymphatic network as suggested by the continuous VE-Cadherin labelling in the inset. It is evident that VE-Cadherin labels lymphatic endothelium at later stages of heart development (compare Figure 5A-D with Figure 5E) and as such the authors need to highlight this in the manuscript. Statements such as "the Prox1+ cells gradually expanded between days E15.5-E17.5 to form multicellular structures (Figure 5D, E)" need to be revised and toned down, as the authors are not lineage-tracing these clusters (nor their progeny). Finally, the inclusion of the Ccbe mutant data here does not add to the overall novelty of the findings and is limited to the mouse studies. Taken together, the main issues raised for this part of the study still remain and if the authors are intent on including these mouse studies (not required in the opinion of this reviewer) then they need to revise further to go beyond simply confirming previous reports in mouse, describing non-venous sources contributing to cardiac lymphatics (lineage-tracing data in Figure 5).

Specific points:

- In Figures 1, 3, 4, 6 and Figure 1—figure supplement 2, the inset boxes representing high-power views of regions of interest are covering the low-power/whole heart views and need to be separated into distinct panels.

- To confirm lack of connections of putative clusters to proximal ventricular lymphatics, the authors need to provide a representative confocal z-stack.

- Why is the coverage of prox1a:RFP labelled vessels in the ventricle so extensive, when compared to lyve1b:dsRed or flt4:mCitrine reporters? This links to point 1) above.

- For the cryoinjury model, why is it that all the sham images exhibited reduced or almost no ventricular lymphatic vessels, compared to injured heart at similar age/stage (Figure 6—figure supplement 1)? Are these images truly representative of the stages at which cryoinjury took place?

- Whilst in revision, a new study from the Porrello lab has been published reporting that Vegfc/d-dependent regulation of the lymphatic vasculature during cardiac regeneration is influenced by injury context (Vivien et al., 2019); the authors should discuss this work in the revised manuscript.

Reviewer #2:

The authors have provided a much-improved manuscript in response to reviewer questions. The additional transgenic markers, antibody stains and angiograms have clarified cell identity and improved the description of cardiac lymphangiogenesis. The mouse lineage tracing experiments are improved with further markers. The genetic studies using VegfA-Oe and Vegfc mutants in zebrafish, and observations that isolated clusters and major lymphatic tracts respond differently in different mutants, improve the value of this study.

https://doi.org/10.7554/eLife.44153.sa1

Author response

Reviewer #1:

[…] There are a number of substantial issues with the manuscript in its current form which need addressing:

1) Overall, the findings reported here come across as being somewhat preliminary. The authors aimed to characterise cardiac lymphatic growth in zebrafish and mouse using an array of transgenic models (including reporter, fate-mapping, mutants) at different developmental timepoints, but this is reported in an inconsistent manner with different reporters being used at different developmental timepoints and then compared together, or only one lymphatic marker, rather than binary combinations (at least), being used at each timepoint. This is particularly evident in the characterisation of cardiac lymphatics in zebrafish (Figure 1), where the authors concluded that the isolated ventricular lymphatic clusters only expressed prox1a (Figures 1F and 4A), but do not show equivalent images (same magnification and heart orientation) of hearts from prox1a/flt4, prox1a/mrc1, lyve1b/prox1a, lyve1b/mrc1 or prox1a/lyve1b (e.g. pending availability of a lyve1b:GFP or prox1a:GFP line) double transgenic fish at a single developmental stage (e.g. 16wpf as in Figure 1F).

We fully agree with the reviewer on that the presentation of our data was not “clean” enough in the original version of our manuscript. We have added images of combined lymphatic transgenes at comparable stages (based on age and body length), using similar magnifications and heart orientations for all lymphatic subsets, and have made sure that the data are presented this way throughout the entire manuscript.

Also, for some of the mutant lines the authors characterised OFT lymphatics (Figure 2) at 10wpf (kfl2a and kfl2b), others at 20wpf (vegfc and vegfd) and others no specific detail of developmental stage (cxcr4a and cxcl12; also, no images of control OFTs are provided in Figure 2R, S) and no information on controls are provided in the corresponding text or figure legend.

We apologize for the inconsistencies in the presentation of our data. We now provide additional images and information for all mutants: All OFT lymphatic analyses were performed on 16-20 wpf/16-24mm fish. WT siblings or age-matched control images were added as well. These data are included in new Figure 2.

The analyses of ventricular lymphatics at 16wpf (Figure 3) appears problematic: given that OFT lymphatic formation (3-4wpf) takes places before lymphatics sprout into the ventricle (12-16wpf), why do the authors investigate OFT lymphatics at 20wpf and ventricular lymphatics at 16wpf in vegfc and vegfd mutants? To exclude potential developmental delays in the lymphatic growth/expansion the authors need to carry out their time-course studies more consistently across the different mutants.

The variability among the analysed stages derives primarily from the fact that some of these fish were compared based on body length, in order to exclude potential developmental delays resulting from size vs. age differences. To avoid potential misunderstanding, we restricted all the OFT lymphatic analyses to 16-24mm fish (WTs and mutants) and ventricular lymphatic analyses to 25-30mm fish (WTs and Mutants). All figures and quantitations have been updated accordingly and explanations about measurement procedures have been added to the Materials and methods section.

2) The evidence for isolated lymphatic clusters in the ventricular wall (Figure 4) is not compelling. These are first described as being located towards the apex, but images in Figures 4B (12wpf) and 4E, G (20wpf) highlight prox1a-expressing LECs in the vicinity of the OFT region; how do the authors distinguish between sprouts from OFT lymphatics and putative isolated clusters if only the prox1a reporter line is used here? This is relevant for the cxcr4a (Figure 4E) and vegfc (Figure 4G) mutant analysis: in Figure 3 the authors showed reduced OFT lymphatics, but how can the authors exclude the possibility that the LECs observed at 20wpf are in fact delayed sprouts from OFT lymphatics?

We thank the reviewer for this comment. The initial distinction between OFT/ventricular lymphatic sprouts and isolated LECs, in both wt and mutant animals, was based on whether they are connected or not to a pre-existing lymphatic vessel. While ventricular lymphatics remain connected to their OFT lymphatic source, isolated clusters are not connected to any vessel. To further confirm these results, we now provide several images of the isolated clusters, labeled by different marker combinations, and throughout different stages, which demonstrate that these clusters are not connected to other lymphatic vessels. The data can be found in Figures 1F, H and Figure 4.

We have also carried out a thorough characterization of the numbers of isolated LECs in cxcr4, vegfc and flt4 mutants, as well as following VegfAa over-expression and PHZ treatment. We found that unlike ventricular lymphatics, isolated lymphatic clusters are normally present in cxcr4 mutants, and are not affected by vegfaa over-expression and PHZ treatment. In contrast, they are absent from flt4 mutant hearts and are markedly reduced in vegfc mutants, emphasizing the requirement of Flt4/VegfC signaling for development of all cardiac lymphatic subsets. These data are included in new Figure 4.

Also, in Figure 4D the authors included a graph detailing numbers of isolated LEC clusters from 8wpf to 16wpf but did not discuss this in the manuscript, nor did they include representative images for these developmental stages. The inclusion of such images is important to support the author's conclusions and make the findings more robust.

We have added to Figure 4 representative images for the analysed stages as well as quantification of isolated clusters in double LEC transgenic lines.

3) An important and outstanding question is what is the origin of the isolated LEC clusters in zebrafish? This may be somewhat beyond scope, but is an important omission which would significantly strengthen the overall findings.

We agree with the reviewer about the importance of identifying the origins of the isolated clusters, but concur also with the appreciation that these experiments fall beyond the scope of this manuscript, especially given the range of the revision period in eLife. While we have attempted to analyse the putative contribution of additional cell types, we find the data too preliminary and not conclusive to be added to this manuscript.

4) There is an imbalance between the characterisation of zebrafish and mouse cardiac lymphatics, with the mouse data being descriptive (without functional targeting) and at a more preliminary stage. Arguably the mouse data could be removed without significantly detracting from the manuscript; or if it is to be retained the analyses need to be more in-depth and rigorous. As above, the authors should include (at least) binary combinations of lymphatic markers in their characterisation of cardiac lymphatics to establish the presence of isolated LEC clusters, in particular combine nuclear (Prox1) with membrane (VEGFR3, LYVE1, PDPN, NRP2) staining patterns. Also, images such as shown in Figure 5 would be more informative if channels were shown separately to properly assess whether LECs are isolated from the sprouting plexus and their localisation. An important outstanding question is when do the clusters first appear in the developing mouse heart? This requires a more detailed time- course of assessment.

We believe that the identification of isolated lymphatic clusters in mammals in addition to fish, is an important finding which contributes to the understanding of cardiac lymphatic formation across species. As shown in Figure 5, the clusters first appear at E14.5, and continue to expand between days E15.5-E17.5, forming multicellular lumenized structures. By day P23 the isolated lymphatic clusters are no longer detected. Binary combination of nuclear (Prox1) and membrane (LYVE1) were added to Figure 5, which include also channel separation to better demonstrate the lack of connections between isolated lymphatic clusters and the main lymphatic vessels. In addition, we now demonstrate that the formation of both “regular” cardiac lymphatics and isolated lymphatic clusters is VEGFR3/VegfC-dependent, and therefore abolished in Ccbe KO mice, thus supporting the data acquired in fish. These data are included in new Figure 5.

5) In the mouse the main result arising from the different lineage-tracing approaches employed appears to be that isolated LECs are from a venous source, derived from the cardinal vein or sinus venosus. How do the authors explain the existence of isolated lymphatic clusters that share the same origin as the main lymphatics sprouts? This is a key question that relates to point 3) above for the zebrafish studies, in that the origin of these clusters across species still remains questionable.

Our lineage tracing analyses in mouse indeed revealed that both isolated clusters and main lymphatic sprouts are partially derived from a venous origin. One explanation for these results could be that these two lymphatic populations originate from different veins (such as the cardinal vein and sinus venosus). Alternatively, they could arise in the same vein but employ different mechanisms of migration and sprouting to reach the heart, thus ensuring proper and perhaps faster lymphatic coverage of the heart. Finally, it could also be possible that some cells detach from the parent lymphatic vessel and migrate through, as has been shown for LECs in the lung, which bud from extra-pulmonary lymphatics and migrate as single cells or small clusters into the developing lung (Kulkarni et al., 2011). We will add these points to the Discussion.

Notably, the relative low fraction of LECs labeled by ApjCreERT2 (20-30%), compared to the high recombination efficiency of this Cre driver (~90%), hints at putative additional non-venous sources with major contribution to both isolated and “traditional” cardiac lymphatics. Yet, at this stage, we have not been able to definitely prove the alternative origins of these cells

6) With regards to the differences between the ApjCreERT2 and Tie2-Cre drivers in labelling murine cardiac lymphatics, the authors need to test the former at earlier time-points to ensure that the venous endothelium which will give rise to LECs is being targeted in an efficient manner. For instance, tamoxifen should be administered at E8.5 plus E9.5 to determine labelling of the jugular lymphatic sac followed by subsequent tracing into the cardiac lymphatics.

We agree with the reviewer that ensuring efficient targeting of the venous endothelium giving rise to LECs, namely the CCV, by early activation of ApjCreERT2 (E8.5+e9.5) is an important control. We lineage traced Apj CreER expressing cells using the mTmG reporter. Tamoxifen was dosed to the mother at e9.5 and e10.5, and embryos were collected at e11.5. Using this dosing strategy, we saw efficient labeling of the cardinal vein, and labeling in the jugular lymph sac. Some of the mTmG+, Prox1+ cells appear to be budding off the cardinal vein. These results are consistent with the conclusion that the ApjCreER lineage traced cardiac lymphatics could be derived from the cardinal vein and lymph sacs. These results were added to Figure 5—figure supplement 1.

7) The isolated LEC clusters apparently become fully integrated within the lymphatic network by P23; what is their functional role in the postnatal heart (intact and following injury)? The authors should at least speculate on this in the Discussion.

This is indeed an interesting question. Unlike zebrafish, in which the cardiac lymphatic system continues growing through adulthood, mouse cardiac lymphatics are fully formed by P23 and the isolated LEC clusters are fully integrated by then. Yet, the fact that at least part of them appear to arise from different sources, may hint at different specializations during pathological conditions. A paragraph discussing these possibilities has been added to the text.

8) In the cryo-injury zebrafish studies (Figure 6) the assessment of a role for these isolated lymphatic clusters in repair it is difficult to interpret in the absence of a fate-mapping model, since it is not possible to determine whether the same cell clusters observed during development are being analysed. Also, the authors need to include sham-operated controls and, as above, use the different binary combinations of reporter lines in a consistent manner across the different time-points analysed.

We agree with the reviewer about the importance of identifying the origins of the isolated clusters following cardiac injury. We have performed additional experiments and could not find LECs at the injury site co-expressing the myeloid marker Lys or the retinoic acid synthesizing enzyme raldh2, known to be present in the epicardium and endocardium following injury. The images are provided in Author response image 1 for the reviewers’ convenience, but we have opted to omit them from the manuscript. We included a paragraph in the Discussion listing potential origins for LECs following injury including recruitment of isolated LECs. In addition, sham-operated animals and binary combinations of reporter lines across the different time points were added to Figure 6—figure supplement 1 as requested.

Author response image 1
LECs in the injured area do not co-express endocardial, epicardial or myeloid markers.

(a) Section of a Tg(flt4:mCitrine) heart at 7 dpi. Cardiomyocytes are immunostained with anti-MHC antibody (red). Epicardium and injury activated endocardium are immunostained with anti-retinoic acid (RA)-synthesizing enzyme Raldh2 (white). White dotted lines delineate the injured area. Insets show high-magnification of yellow dashed box. Flt4 positive cells in the injured area are not labeled with Raldh2 (b) Tg(flt4:mCitrine);Tg(lyz:dsRed) heart at 6dpi. White dotted lines delineate the injured area. Insets show high-magnification of yellow dashed box. Flt4 positive cells in the injured area do not express the myeloid marker lys. Scale bars are 200µm.

In Figure 6G, the authors reported flt4-expressing vascular structures, which are negative for prox1a, sprouting into the injured area; could these be blood vessels? Flt4 expression is shared between blood and lymphatic endothelium during development, becoming restricted to the latter at later stages of development, but flt4 is also activated in neoangiogenic sprouts. This is supported by the observation of vascular structures positive for kdrl and flt4 in Figure 6J.

We agree with the reviewer that some of the flt4 positive sprouts may represent blood endothelial cells. while, previous publications (He et al., 2017; Marín-Juez et al., 2016) demonstrated that neovascularization following cardiac injury involves only sprouting from pre-existing blood vessels, a new study now demonstrates collateral assembly via migration and coalescence of isolated arterial cells following MI in mice (Das et al., 2019). While both prox1a and flt4 transgenes may individually label blood as well as lymphatic ECs, we found that cells expressing both transgenes are of lymphatic identity and are not labeled by intravascular injection of Qdot705 (Figure 1—figure supplement 2C). We therefore used prox1a/flt4 (Figure 6F, I, J) as well as prox1a/mrc1a (Figure 6H) transgenic combinations to confirm the lymphatic identity of isolated LECs in the injured area.

9) The regions highlighted in Figure 6G-K appear to be in close proximity to the main lymphatic sprouts, rather than isolated in apical areas as is suggested by the authors- this requires further clarification: a combination of membrane and nuclear LEC staining would enable a clear assessment of the clusters and would serve to identify if they are indeed isolated LECs or just leading cells from the sprouting lymphatic plexus.

We thank the reviewer for this suggestion. We complemented our data with additional views of the injured hearts, clearly demonstrating the lack of connections with other vessels. These were added to the new Figure 6—figure supplement 1A, B, D, F.

Reviewer #2:

[…] There are a number of gaps and major concerns throughout the study. Specifically, from many key data points there are important controls missing. Also there are points where interpretation is not fully supported by the current data. These points need to be addressed before further considering publication of this study.

1) Zebrafish develop in a highly variable manner during the larval to juvenile transition. This variation occurs between and within clutches and is very well documented, thus staging with days post fertilisation once animals are beyond larval stages creates inaccuracies (see Parichy et al., 2009). As such, the standard approach to stage juvenile zebrafish is using body length – the authors should see the definitive study and guidelines in Parichy et al., 2009 (especially findings in Figures 3 and 4).

The use of days post fertilisation in this study rather than body length creates concerns because for many comparisons animals may not be of the same developmental age. For example: are the mutants in Figure 2 (vegfc, vegfd, cxcr4a, cxcl12) all the same length or are some mutants developmentally delayed? Likewise for the PHZ treated animals. If they are not truly stage matched, the OFT lymphatic phenotypes may just be secondary to developmental delay.

Can the authors please provide accurate staging using length measurements? Minimally this needs to be provided for key observations that could be significantly skewed by natural variation in fish populations at juvenile stages (such as for the mutant and drug treatments).

We fully agree with the reviewer’s comment and apologize for not having added this information in our original submission. In this revised version in addition to age, we provide standard body length measurements (the distance from the snout to the caudal peduncle) for all the experiments presented, including those comparing wt vs. mutants and control vs. treatments. Importantly, throughout all our experiments, special care was taken to include only fish of same age and size to avoid bias resulting from potential developmental differences.

2) In Figure 2, the authors show that the OFT lymphatics derive from the ventral facial lymphatics. For all mutants analysed at the level of the OFT, the phenotypes could be due to (secondary to) reduced ventral facial lymphangiogenesis rather than suggesting a specific role for these genes in OFT lymphangiogenesis. Especially suggestive is the fact that vegfc heterozygotes already have systemically reduced lymphangiogenesis at 5 dpf (Villefranc et al., 2013). Please provide careful quantification and images of the ventral facial lymphatics in these mutants preceding the formation of the OFT lymphatics. Alternatively, temporally inducible transgenic models could be used to show that there are ongoing roles for these genes in OFT lymphangiogenesis.

We appreciate the reviewer’s comment. We have documented the development of the ventral facial lymphatic vessel in all mutants (vegfc, vegfd, cxcr4, cxcl12b and flt4) and have detected no defects in its formation. These data are included in Figure 2—figure supplement 2.

3) In the mutant comparisons, mutants should be compared with siblings from the same clutch because there can be variation between clutches. This does not appear to be the case for vegfc and vegfd mutants, which are compared to just one population termed "wt". Please ensure comparisons and statistics are comparing mutants with clutch matched siblings.

Throughout all experiments included in our revised version, mutants are compared with clutch matched siblings or aged/size-matched WTs in the case of certain homozygous adult mutants (e.g. vegfd and cxcl12b).

4) The prox1a:RFP transgene uses a Gal4, UAS cassette to enhance signal. This system is known to lead to non-specific expression in tissues in some contexts. It has been previously noted that this transgene does not always accurately reflect endogenous Prox1 protein levels (see Koltowska et al., 2015 Figure 2 and Supplementary Figure 1).

In Figure 4, the proposed "clusters" do not express any lymphatic markers other than this transgene. The images appear (to this reviewers eye) to show blood vasculature that is co-expressing RFP rather than genuine distinct clusters of positively identified lymphatic endothelial cells. One interpretation of the current data from zebrafish could be that this expression is non-specific transgene expression in blood vessels over the heart.

Please show more comprehensive analysis of expression of multiple blood vascular and lymphatic markers in these clusters. Do they contain blood? Please also validate the cluster expression of Prox1 with antibody staining.

We appreciate the reviewer’s comment. In this revised version of our manuscript we provide a thorough characterization of the prox1a expressing vessels and assess the specificity of the transgene. As pointed out by the reviewer, we detected sporadic expression of the prox1a transgene in few arterial capillaries, that were co-labelled by the arterial enhancer Tg(flt1_9a_cFos:GFP)wz2 and highlighted by intravascular injection of Qdot705 (Figure 1—figure supplement 2B). In contrast, prox1a/flt4 double labeled vessels (in prox1a;flt4 double transgenic zebrafish) were devoid of intravascular injected Qdots, confirming their lymphatic identity. Hence, we identify two vessel populations labelled by the prox1a transgene- one, where its expression fully overlapped with that of the flt4 reporter and was devoid of Qdot705 labelling (Figure 1—figure supplement 2C),and a second one labelled only by prox1a and Qdot705 (Figure 1—figure supplement 2C).We further analysed the isolated lymphatic clusters during development (Figure 4) and following cardiac injury (Figure 6) using this transgenic combination. We injected Qdots intravascularly and verified that neither the ventricular lymphatics nor the isolated clusters co-labeled by prox1a and additional transgene are connected to the blood circulation (see new Figure 1—figure supplement 2D, E and Figure 4F).

In order to investigate whether both populations indeed express Prox1, we carried out immunostaining with anti-Prox1 antibody. As seen in Figure 1—figure supplement 2onlyprox1a positive LECs, but not prox1a positive blood ECs were labelled by the Prox1 antibody (Figure 1—figure supplement 2E, insets), suggesting that the expression in blood ECs represents an artefact of the transgenic reporter.

5) The use of exclusively nuclear markers in Figure 5 in defining the isolated Prox1 positive LEC clusters in mouse is problematic. Some of these "isolated" nuclei are not that far apart (e.g., in Figure 5B and C). Please provide a much more thorough marker analysis of these clusters in mice with more lymphatic and blood vascular markers and including reliable LEC markers that are membranous such as Nrp2, Lyve1 or Podoplanin. The concern is that cells could be connected by long, cytoplasmic/membranous connections not seen with nuclear stains. Further marker and anatomical analyses would significantly clarify the nature of these clusters.

Binary combination of nuclear (Prox1) and membrane (LYVE1) staining were added to Figure 5. These clearly demonstrate the lack of connections between isolated lymphatic clusters and the main lymphatic sprouts.

6) In Figure 6, the authors see lymphatics arising in a wound site following cryoinjury using fixed stage analysis. The origins of these lymphatics have not been lineage traced or time-lapse imaged to allow any claims as to their cellular origins. The wound site in Figure 6B for example is adjacent to the major ventricular lymphatic and only one side of the heart is imaged so there could be major lymphatics in the non-imaged half.

We agree with the reviewer about our inability to define the cellular origins of the isolated clusters, and this is clearly mentioned throughout the manuscript. However, there are certain facts that we have carefully documented and quantified, for which we can make clear statements. For instance, while the wound site may sometimes be adjacent to major ventricular lymphatics, we carefully verified that the isolated lymphatic clusters at the injury site are not directly connected to any pre-existing lymphatic vessel. Our conclusions are based on imaging of the heart from multiple angles using lightsheet microscopy, which confirm that the isolated LECs are not connected to pre-existing ventricular lymphatics. The images showing different views of injured hearts have been added to Figure 6D, F, H, I and Figure 6—figure supplement 1A, B, D, F.

These cells could have arisen by detaching from and migrating from the major OFT derived ventricular lymphatics or could have come from other sources based on the analysis provided.

Further data are required to support the authors claims that "… cardiac injury induces the de-novo formation of lymphatics, through a process reminiscent of lymph-vasculogenesis rather than sprouting lymphangiogenesis."

We believe that the data presented in Figure 6D, F, H, I and Figure 6—figure supplement 1A, B, D, F demonstrate that LECs in the injury site, regardless of their origin, are not connected to other lymphatic vessels, and initially appear as isolated clusters. Therefore, we concluded that their appearance is “reminiscent of lymph-vasculogenesis”. As stated above, we have included additional views of the injured hearts in new Figure 6—figure supplement 1A, B, D, to further support this claim.

We agree with the reviewer that without live imaging or defined lineage-tracing we cannot claim anything about their origins (this caveat appears in the Discussion), yet we believe our data strongly supports the fact that these vascular structures do not form through lymphangiogenesis.

Reviewer #3:

[…] Overall, this is an interesting study of high standards. The findings contribute to our knowledge about development of cardiac lymphatics in zebrafish. However, the conclusions based on analysis of mutants are sometimes not convincingly supported by the results. Furthermore, the part with the cryoinjury model contains descriptive data, without testing the role of lymphatics in cardiac regeneration.

- "we detected a significant reduction in OFT lymphatic coverage (Figure 2J-M), with under-developed vessels displaying fewer sprouts that those of their wt siblings". In the Figure 2M, the graph indicates that the percentage of lymphatic coverage was higher in the mutants than in wt, which is in contradiction with the statement of the authors.

We thank the reviewer for pointing out this mistake. The sentence should state, "we detected a significant increase in OFT lymphatic coverage”. However, since our new data revealed major roles for the Vegfc/Flt4 and Cxcl12/Cxcr4 pathways in OFT lymphatic development, we decided to omit the klf2 data that was inconclusive and perhaps less relevant to this process. If however, the reviewers think otherwise, we will add them back to the manuscript.

- ".., we could not observe any defects in OFT lymphatics of vegfd mutant fish (Figure 2N, P, Q)…" The vegfd mutant was generated for this study using the Crispr/Cas9 system. Do these mutants display any phenotype, indicative of a gene loss of function?

We appreciate the reviewer’s comment. vegfdbns257 homozygous mutants indeed exhibit compromised facial lymphatic development, consistent with previous reports, however no major defects were observed in the VFL, from which OFT lymphatics sprout. The position and morphology of these vessels, which run in parallel to the ventral aorta was similar in wt and mutant fish. We now provide additional data describing the generation and early phenotypic characterization of these mutants in Figure 2—figure supplement 3.

- "Our results suggest that while VegfC is a key determinant of OFT lymphatic formation, the Cxcr4/Cxcl12 axis is mostly involved in patterning and remodelling of the OFT lymphatic plexus". I don't see any differences between the phenotype of vegfc and cxcr4a mutants (Figure 2O and R). I found that this conclusion is too strong for these data. How can we conclude that VegfC is a key determinant of OFT lymphatic formation, given that lymphatic vessels were still formed? The authors should describe more precisely the phenotype of the mutants and prove that the lymphatic function is also altered.

In the revised version of our manuscript we show that while vegfc+/- animals display significantly reduced OFT lymphatic coverage, cxcr4 mutants show defective patterning of OFT lymphatics, with no overall changes in lymphatic coverage. Since vegfc+/- animals still carry a wt copy of the vegfc gene (homozygous mutants do not survive through adulthood), which could lead to a partial phenotype, we analysed also OFT lymphatic development in homozygous adult flt4 mutants. As seen in new Figure 2I-K, OFT lymphatics are nearly absent in flt4 mutants, strongly supporting and expanding our original findings.

It would be also interesting to test if cardiac lymphatic vessel development in the ventricle is also impaired in these mutants.

Impaired ventricular lymphatic development was also observed in vegfc+/- and cxcr4 mutants. In addition, our new data reveal that ventricular lymphatics are completely absent in flt4 mutants.

These results were added to new Figure 3.

- ".… suggesting a possible role for the major coronary vessels in guiding lymphatic sprouts". The authors cannot exclude that the treatment itself influences directly lymphatic development.

Following the reviewer comment, we carried out additional experiments to further verify the involvement of coronary vessels in guiding lymphatic vessels:

1) In addition to PHZ treatment, we induced myocardial expansion and increased blood vessel coverage by over-expressing the angiogenic factor VegfAa (Karra et al., 2018). Our new results demonstrate that lymphatic coverage, as well as the diameter of ventricular lymphatics, is significantly increased in these hearts (new Figure 3E-G).

2) In order to ascertain whether VegfAa overexpression enhances lymphangiogenesis indirectly (through increased angiogenesis) or by direct action of VegfAa on LECs, we over-expressed VegfAa in vegfc+/- heterozygous animals and assessed whether despite reduced VegfC production, there is still lymphatic over-growth, which could have pointed to a direct effect of VegfAa on LECs. As seen in new Figure 3Q-T and Figure 3—figure supplement 1J, while depletion of VegfC had no effect on VegfAa induced cardiomegaly and hyper-vascularization, ventricular lymphatic growth was abolished, suggesting that enhanced coronary formation induces ventricular lymphatic growth in a VegfC-dependent manner.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Essential revisions:

i) Additional work is required to definitively prove the existence of LEC clusters in zebrafish hearts as distinct from the main venous-derived lymphatic plexus; specific experiments using multiple stainings and z-stack confocal projections, etc. are suggested in the report.

Our manuscript provides solid evidence for the existence of isolated LEC clusters in zebrafish. We base this statement in the following lines of evidence:

1) We provide images of different transgene combinations, at different stages, which clearly demonstrate that the isolated LEC clusters are not connected to neither blood nor lymphatic vessels (Figure 1F, H and Figure 4). This is in contrast to the main ventricular lymphatics that remain connected to their OFT lymphatic source.

2) In the previous assessment of our manuscript, certain concerns were raised regarding the specificity of the prox1a transgene in labeling lymphatic vessels. In order to address these comments, we carried out Prox1 immunostaining experiments in combination with intravascular injection of Qdot705. Comparison between the prox1 transgenic reporter and Prox1 immunostaining, revealed the presence of two different vessel populations labelled by the prox1a transgene: (i) Lymphatic ECs, where the expression of the prox1a transgene fully overlaps with that of the flt4 reporter and is devoid of Qdot705 labelling (Figure 1—figure supplement 2C), and (ii) a population of arterial capillaries, co-labelled by the arterial enhancer Tg(flt1_9a_cFos:GFP)wz2 (Figure 1—figure supplement 2B), which is connected to the blood circulation as demonstrated by Qdot705 labelling (Figure 1—figure supplement 2C). Based on these results, we decided to use this combinatorial matrix to assess the nature of the isolated LECs: intravascular injection of Qdot705 into double transgenic lines (flt4BAC:mCitrine;prox1a:RFP) confirmed that neither the ventricular lymphatics nor the isolated clusters are connected to the blood circulation (Figure 1—figure supplement 2D, E and Figure 4F). It is important to note that all analyses of the isolated LEC clusters during development (Figure 4) and following cardiac injury (Figure 6), were carried out using double transgenic combinations.

3) Following assessment of our revised version, reviewer #1 raises new concerns about two of the other transgenic reporters used in this study (“a similar concern rests with the other two reporters employed (flt4BAC:mCitrine, lyve1b:dsRed2), in that both flt4 and lyve1 have been shown to be expressed in blood endothelium”. In this case, we disagree with the reviewer’s comment:

a) Both the flt4BAC:mCitrine and lyve1b:dsRed2 reporters have been shown to be expressed by both venous and lymphatic endothelial cells in the developing embryo (Van Impel et al., 2014; Nicenboim et al., 2015; Okuda et al., 2012; Marin-Juez et al., 2016). The expression of these transgenes in the adult zebrafish heart however, has not been thoroughly characterized. Our results, following intravascular injection of Qdot705 indicate that the expression of these reporters is restricted to lymphatic vessels in the adult zebrafish heart (Figure 1I, K and Figure 4F). Specifically, we detect the presence of perfused blood vessels (Figure 1K, blue only) that are not labelled by the flt1_9a_cFos:GFP arterial enhancer (Figure 1K, green +blue) or by the lyve1b:dsRed2 reporter(Figure 1K, red), which we conclude are venous vessels.Hence, we are confident that the cells labeled by the flt4BAC:mCitrine and lyve1b:dsRed2 reporters in the adult heart are for the most part LECs.

b) To further confirm these results, we analysed cardiac lymphatics using an additional transgenic reporter- mrc1a:GFP, whichhas been shown to label LECs in juvenile zebrafish (Jung et al., 2017), and find that its expression fully overlaps with that of the other reporters.

c) Finally, it is important to note that the expression of the lymphatic reporters is first detected over the ventricle at around ~12wpf, when ventricular lymphatics begin sprouting, despite of the fact that a fully functional blood vasculature (veins and arteries) is present much earlier (~8 wpf) (Figure 3 and Harrison et al., 2015).

4) Regarding the reviewer’s request that “the authors should carefully map out the overlap in blood and lymphatic endothelium, both in development and post-injury, for each of the reporters in combination with specific antibodies against lymphatic markers….”

As explained above, we provide extensive evidence for the overlapping expression of the 4 different lymphatic reporters utilized in this study. Moreover, since we noticed that the prox1 reporter displayed additional blood vessel expression, we carried out Prox1 immunostaining to further confirm our findings. I must emphasize that this was not an easy task- Most available antibodies do not work properly in zebrafish, especially in adult animals, being this one of the main reasons for the wide use of transgenic reporters instead. To the best of our knowledge there is only one additional study describing immunostaining of lymphatic vessels in adult zebrafish (Shimoda and Isogai, Acta Histochem Cytochem. 2012), whose results fully concur with our own. Unfortunately, no antibodies against the zebrafish proteins are available for the other markers. Hence, it will not be feasible to carry out these experiments in the timeframe of this revision.

5)“Equally, how do these transgenic lines "behave" in the cryoinjury setting, i.e. does injury (hypoxia, cell death?) induce ectopic expression of the transgenes?”

We believe that the use of multiple transgenic lines, all labeling the same set of LEC clusters after injury, significantly minimizes the chances that they represent ectopic expression of the reporters. It seems unlikely that 4 different transgenic reporters will render ectopic expression in similar sets of cells, all resembling lymphatic vessels, but we cannot completely rule this out. While the behavior of these lines in settings of hypoxia and/or cell-death might by itself be interesting, we believe this characterization falls beyond the scope of this manuscript.

6) “To confirm lack of connections of putative clusters to proximal ventricular lymphatics, the authors need to provide a representative confocal z-stack’)”.

We provide full confocal z-stacks for the following images: (Figure 1F, H; Figure 4B, G; Figure 6G, F, H; Figure 6—figure supplement 1H; Figure 7Q) showing lack of connections between the isolated LEC clusters and the ventricular lymphatics. Yet, given the curved architecture of the heart, we believe that single plane images are more informative than the multiple z-stack confocal planes. Such image of a single z-stack is provided in Figure 4E inset.

7) Finally, we have also carried out a thorough characterization of the isolated LECs in cxcr4, vegfc and flt4 mutants, as well as following VegfAa over-expression and PHZ treatment. We found that unlike ventricular lymphatics, isolated lymphatic clusters are normally present in cxcr4 mutants, and are not affected by VegfAa over-expression and PHZ treatment. In contrast, they are absent from flt4 mutant hearts and are markedly reduced in vegfc+/- animals (Figure 4). These results confirm that the isolated LECs represent a separate lymphatic population, which responds differently than ventricular lymphatics/blood vessels to the same signaling cues.

Overall, we believe that the thorough characterization of the OFT, ventricular and isolated LECs presented in our manuscript, which includes analysis of different combinations of all available lymphatic transgenic reporters (lyve1, flt4, mrc1a and prox1), antibody immunostaining, and functional assays (microangiography) to distinguish between blood and lymphatic vessels, provides strong foundation for the existence of a separate population of isolated LECs.

ii) The mouse studies remain quite preliminary and still require additional work to add any insight over and above what has already been published. The clusters are not clearly demonstrated and there is no insight into origin. At the moment it detracts from the fish studies, so the mouse data should be dropped and instead the authors should focus on definitively demonstrating that LEC clusters exist in the fish heart.

We strongly disagree with the reviewer’s comment. The findings describing the existence of a separate population of isolated lymphatic cells in the murine heart have not been previously reported, and as such they are novel.

We agree with the reviewer that we were unable to provide insight into the origins of the lymphatic clusters that appear during cardiac development. However, we have thoroughly analyzed the morphology of the clusters at many different time points during embryonic and postnatal development. Numerous confocal images were taken of over 100 hearts during preliminary investigations and in the data reported here in which we find that developing hearts contain lymphatic clusters. In the spirit of eLife, we believe it is important that these findings are published as they will surely open new avenues for future research.

In response to the below comments, please see that we have included additional images that better depict what we have observed many times during our experiments (I. E., Orthogonal sections of our confocal images as well as low and high-power Z stack examples). We will note that although a non-venous origin was shown for cells within the lymphatic sprouts (Klotz et al., 2015), the authors did not describe the presence of lymphatic clusters and no other publication since then has either. We therefore believe that although we could not show origins of the clusters, reporting their presence is an important part of this study as it shows an analogous structure to that thoroughly investigated in zebrafish. We hope that our request to keep these data will be favorably considered.

Reviewer #1:

[…] Some of the issues noted in the original submission still remain to be addressed. In particular, the authors still need to more convincingly demonstrate the following:

[…] 3) The existence of isolated lymphatic clusters in the developing mouse heart.

Unfortunately the mouse data still remain preliminary and inconclusive regarding the presence of discrete clusters of LECs of non-venous origin.

We have added additional panels to Figure 5—figure supplement 1A-C showing that the clusters are indeed discrete. Using Lyve1 to mark lymphatic cell membranes, we found that some lymphatic cells had no connections with any other neighboring lymphatic cell (Figure 5—figure supplement 1A). Furthermore, using VE-cadherin and Prox1 immunofluorescence, we provide a whole ventricle, 20X, and 40X images of a typical cluster with associated orthogonal views and a video through all the Z-stacks, which confirms that it does not communicate with other lymphatic sprouts or coronary vessels (Figure 5—figure supplement 1B and C, Figure 5—video 1). It is important to note that we do not make any claims regarding venous vs. non-venous origins of these cells, but just describe the results from our different lineage-tracing studies, which show mostly negative (but accurate) results.

The authors have included a new panel C in Figure 5, but this is suboptimal as the region shown is a high-power view and, therefore, the relationship to the main lymphatic network developing in the E15.5 murine heart is unclear.

We have added a low magnification image of the heart that shows the main lymphatic vessel so it is now possible to understand the location of the isolated clusters relative to the other vessels (Figure 5—figure supplement 1A).

The authors should provide a low-power view containing the entire ventricular surface.

A low power image of the entire ventricle is now included for both Lyve1 immunofluorescence and VE-cadherin/Prox1 (Figure 5—figure supplement 1A-C, Figure 5—video 1).

Also, in order to confirm that the "multicellular structures" marked with the blue arrowheads are detached from the unmarked vascular structures, the authors need to include a representative z-stack. Specifically, re Figure 5E the z-stack depth in the lower power image is different from the high-power inset, and the region of interest appears to be connected to the main lymphatic network as suggested by the continuous VE-Cadherin labelling in the inset.

We reimaged this heart and have now included images that contain Z stacks through the same depth of tissue (Figure 5—figure supplement 1B and C). We have also highlighted a typical isolated LEC cluster (with orthogonal views and a video through the Z stack, Figure 5—figure supplement 1B and C, Figure 5—video 1) which is not connected to VE-Cadherin+ blood or lymphatic vessels. The clusters are frequently close to coronary vessels, in this case two cell lengths away (Figure 5—figure supplement 1C), but this is likely because coronary vessels are dense within the ventricle.

It is evident that VE-Cadherin labels lymphatic endothelium at later stages of heart development (compare Figure 5A-D with Figure 5E) and as such the authors need to highlight this in the manuscript.

Thank you for this suggestion. We have included this point in the text.

Statements such as "the Prox1+ cells gradually expanded between days E15.5-E17.5 to form multicellular structures (Figure 5D, E)" need to be revised and toned down, as the authors are not lineage-tracing these clusters (nor their progeny).

This statement was removed and instead we describe the clusters as simply being present during development and not observable in three-week-old hearts.

Finally, the inclusion of the Ccbe mutant data here does not add to the overall novelty of the findings and is limited to the mouse studies. Taken together, the main issues raised for this part of the study still remain and if the authors are intent on including these mouse studies (not required in the opinion of this reviewer) then they need to revise further to go beyond simply confirming previous reports in mouse, describing non-venous sources contributing to cardiac lymphatics (lineage-tracing data in Figure 5).

We feel that the manuscript benefits from the Ccbe1 data because it is evidence that, like in Zebrafish, the VEGFc pathway is required for the presence of isolated LEC clusters. Other studies reported that isolated lymphatic clusters of non-venous origins in the mesentery are independent of the VEGFC receptor VEGFR3 (Stanczuk, 2015). Therefore, it is a key piece of information regarding the developmental mechanisms of cardiac LEC clusters.

Specific points:

- In Figures 1, 3, 4, 6 and Figure 1—figure supplement 2, the inset boxes representing high-power views of regions of interest are covering the low-power/whole heart views and need to be separated into distinct panels.

We separated all high-power views into distinct panels as requested

- To confirm lack of connections of putative clusters to proximal ventricular lymphatics, the authors need to provide a representative confocal z-stack.

We provide full confocal z-stacks for the following images: (Figure 1F, H; Figure 4B, G; Figure 6G, F, H; Figure 6—figure supplement 1H; Figure 7P, Q) showing lack of connections between the isolated LEC clusters and the ventricular lymphatics. Yet, given the curved architecture of the heart, we believe that single plane images are more informative than the multiple z-stack confocal planes. Such image of a single z-stack is provided in Figure 4E inset.

- Why is the coverage of prox1a:RFP labelled vessels in the ventricle so extensive, when compared to lyve1b:dsRed or flt4:mCitrine reporters? This links to point 1) above.

As explained above, the prox1a:RFP transgene labels also a subset of arterial capillaries, highlighted also by the arterial enhancer flt1-9a:GFP and therefore its coverage is more abundant than other lymphatic markers. This observation further supports the fact that the other reporters (flt4, lyve1, mrc1a) specifically label LECs and not blood ECs.

- For the cryoinjury model, why is it that all the sham images exhibited reduced or almost no ventricular lymphatic vessels, compared to injured heart at similar age/stage (Figure 6—figure supplement 1)? Are these images truly representative of the stages at which cryoinjury took place?

The sham operations were carried out together with the cryo-injury experiments and are indeed representative. We have replaced the image in Figure 6—figure supplement 1C. In addition, a recent paper also reports no differences in cardiac lymphatics following sham operation (Vivien et al., 2019).

- Whilst in revision, a new study from the Porrello lab has been published reporting that Vegfc/d-dependent regulation of the lymphatic vasculature during cardiac regeneration is influenced by injury context (Vivien et al., 2019); the authors should discuss this work in the revised manuscript.

We have added a paragraph to the Discussion referring this study.

https://doi.org/10.7554/eLife.44153.sa2

Article and author information

Author details

  1. Dana Gancz

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing—original draft, Writing—review and editing, Conducted zebrafish experiments
    Competing interests
    No competing interests declared
  2. Brian C Raftrey

    1. Department of Biology, Stanford University, Stanford, United States
    2. Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Stanford, United States
    Contribution
    Formal analysis, Investigation, Visualization, Writing—review and editing, Conducted experiments with mouse hearts
    Contributed equally with
    Gal Perlmoter
    Competing interests
    No competing interests declared
  3. Gal Perlmoter

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Formal analysis, Investigation, Visualization, Writing—review and editing, Conducted zebrafish experiments
    Contributed equally with
    Brian C Raftrey
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2017-3008
  4. Rubén Marín-Juez

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Investigation, Visualization, Writing—review and editing, Conducted zebrafish experiments
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5903-7463
  5. Jonathan Semo

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Investigation, Conducted zebrafish experiments
    Competing interests
    No competing interests declared
  6. Ryota L Matsuoka

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Present address
    Department of Cardiovascular and Metabolic Sciences, Lerner Research Institute, Cleveland Clinic, Cleveland, United States
    Contribution
    Generated and validated the vegfdbns257 mutants
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6214-2889
  7. Ravi Karra

    1. Regeneration Next, Duke University, Durham, United States
    2. Department of Medicine, Duke University School of Medicine, Durham, United States
    Contribution
    Generated and provided transgenic lines
    Competing interests
    No competing interests declared
  8. Hila Raviv

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Conducted zebrafish experiments
    Competing interests
    No competing interests declared
  9. Noga Moshe

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Project administration, Managed fish work
    Competing interests
    No competing interests declared
  10. Yoseph Addadi

    Department of Life Sciences Core Facilities, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  11. Ofra Golani

    Department of Life Sciences Core Facilities, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Software
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9793-236X
  12. Kenneth D Poss

    Regeneration Next, Duke University, Durham, United States
    Contribution
    Supervision
    Competing interests
    No competing interests declared
  13. Kristy Red-Horse

    1. Department of Biology, Stanford University, Stanford, United States
    2. Institute for Stem Cell Biology and Regenerative Medicine, Stanford University School of Medicine, Stanford, United States
    Contribution
    Visualization, Writing—review and editing, Supervised mouse experiments
    Competing interests
    No competing interests declared
  14. Didier YR Stainier

    Department of Developmental Genetics, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany
    Contribution
    Writing—review and editing, Supervised part of the zebrafish experiments
    Competing interests
    Senior editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0382-0026
  15. Karina Yaniv

    Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel
    Contribution
    Conceptualization, Supervision, Funding acquisition, Visualization, Writing—review and editing
    For correspondence
    Karina.Yaniv@weizmann.ac.il
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5638-7150

Funding

H2020 European Research Council (335605)

  • Karina Yaniv

H2020 European Research Council (818858)

  • Karina Yaniv

United States-Israel Binational Science Foundation (2015289)

  • Karina Yaniv

Minerva Foundation (712610)

  • Karina Yaniv

H and M Kimmel Institute for Stem Cell Research, the Estate of Emile Mimran (SABRA program)

  • Karina Yaniv

National Institutes of Health (R01 HL081674)

  • Kenneth D Poss

National Institutes of Health (R01 HL131319)

  • Kenneth D Poss

National Institutes of Health (R01 136182)

  • Kenneth D Poss

American Heart Association

  • Kenneth D Poss

Fondation Leducq

  • Kenneth D Poss
  • Didier YR Stainier

National Institutes of Health (RO1-HL128503)

  • Kristy Red-Horse

New York Stem Cell Foundation (Robertson Investigator)

  • Kristy Red-Horse

Max-Planck-Gesellschaft

  • Didier YR Stainier

Deutsche Forschungsgemeinschaft (394046768 – SFB 1366)

  • Rubén Marín-Juez
  • Didier YR Stainier

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

Acknowledgements

The authors thank R Hofi, G Almog, N Stettner and A Harmelin (WIS) for excellent animal care. We thank B Weinstein for the Tg(mrc1a:EGFP)y251, N Mercader for the Tg(myl7:GFP), A Siekmann for the cxcr4um20 and cxcl12bum100, and E Tzahor for critical reading of the manuscript. Light sheet Imaging was carried out at the ‘de Picciotto-Lesser Cell Observatory in memory of Wolf and Ruth Lesser’.

The authors are grateful to all members of the Yaniv laboratory for discussion, technical assistance and continuous support. This works was supported by European Research Council (335605) and (818858) to KY, Binational Science Foundation (2015289) to KY, Minerva Foundation (712610) to KY, the H and M Kimmel Inst. for Stem Cell Research, the Estate of Emile Mimran (SABRA program). KDP is supported by grants from NIH (R01 HL081674, R01 HL131319, and R01 136182), American Heart Association, and the Leducq Foundation. KR is supported by the NIH (RO1-HL128503) and is a New York Stem Cell Foundation - Robertson Investigator. DYRS is supported by funds from the Max Planck Society, the Leducq Foundation and the DFG (project number 394046768) SFB1366/project A4.

Ethics

Animal experimentation: This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All of the animals were handled according to approved institutional animal care and use committee (IACUC) protocols (#01470218-2) of the Weizmann Institute of Science. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Weizmann Institute of Science. All surgery in fish was performed under tricaine anesthesia, and every effort was made to minimize suffering.

Senior and Reviewing Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Publication history

  1. Received: December 10, 2018
  2. Accepted: November 5, 2019
  3. Accepted Manuscript published: November 8, 2019 (version 1)
  4. Version of Record published: November 27, 2019 (version 2)
  5. Version of Record updated: December 4, 2019 (version 3)

Copyright

© 2019, Gancz 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

  • 2,640
    Page views
  • 494
    Downloads
  • 10
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, 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)

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

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

Further reading

    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Catherine Pfefferli, Anna Jaźwińska
    Insight

    Experiments on zebrafish show that the regeneration of the heart after an injury is supported by lymphatic vessels.

    1. Cell Biology
    2. Developmental Biology
    Cuie Chen, Yukiko M Yamashita
    Research Article

    Asymmetrically dividing stem cells often show asymmetric behavior of the mother versus daughter centrosomes, whereby the self-renewing stem cell selectively inherits the mother or daughter centrosome. Although the asymmetric centrosome behavior is widely conserved, its biological significance remains largely unclear. Here we show that Alms1a, a Drosophila homolog of the human ciliopathy gene Alstrom syndrome, is enriched on the mother centrosome in Drosophila male germline stem cells (GSCs). Depletion of alms1a in GSCs, but not in differentiating germ cells, results in rapid loss of centrosomes due to a failure in daughter centriole duplication, suggesting that Alms1a has a stem cell-specific function in centrosome duplication. Alms1a interacts with Sak/Plk4, a critical regulator of centriole duplication, more strongly at the GSC mother centrosome, further supporting Alms1a's unique role in GSCs. Our results begin to reveal the unique regulation of stem cell centrosomes that may contribute to asymmetric stem cell divisions.