Regenerating vascular mural cells in zebrafish fin blood vessels are not derived from pre-existing mural cells and differentially require Pdgfrb signalling for their development

During tissue regeneration, lost cell types need to be replaced through de novo differentiation of tissue resident stem cells or from pre-existing differentiated cells (Poss, 2010). Several vertebrate species, such as zebrafish or newts, can regenerate a wide variety of tissues, such as appendages (Sehring and Weidinger, 2020; Tanaka, 2016), heart muscle (de Wit et al., 2020; González-Rosa et al., 2017; Xiang and Kikuchi, 2016) or retina (Alunni and Bally-Cuif, 2016; Wan and Goldman, 2016). By contrast, mammals have lost most of their regenerative potential and repair injured tissues through imperfect wound healing, which can lead to scar formation and fibrosis (Eming et al., 2014; Erickson and Echeverri, 2018). During both processes, blood vessel regrowth occurs, highlighting their high regenerative potential (Chavez et al., 2016; DiPietro, 2013, 2016). However, it is not clear whether the newly formed blood vessels are equally functional in both settings, and how they might influence tissue regeneration. Blood vessels consist of endothelial cells lining the inner vessel surface and ensheathing mural cells. Studies in the zebrafish fin (Kametani et al., 2015; Tu and Johnson, 2011; Xu et al., 2014) and in mouse arteries (McDonald et al., 2018) have shown that endothelial cells in regenerating blood vessels are derived from pre-existing endothelial cells. Similarly, during wound healing, endothelial cells are being recruited from neighbouring blood vessels (Carmeliet, 2005; Tonnesen et al., 2000). These mechanisms closely resemble those observed during embryonic angiogenesis. By contrast, the ontogeny of newly forming murals cells and their contribution to blood vessel and tissue regeneration have remained unclear.

Mural cells can be subdivided into smooth muscle cells and pericytes, with intermediate cell types being present (Armulik et al., 2011; Grant et al., 2019; Holm et al., 2018). Generally, pericytes are important for guiding newly growing blood vessel sprouts during angiogenic phases (Eilken et al., 2017; Teichert et al., 2017) and later in blood vessel stabilization (Hellstrom et al., 2001). However, organ-dependent functional differences exist. In the brain, they ensure blood-brain barrier integrity (Langen et al., 2019) and regulate blood flow (Pfeiffer et al., 2021). Pericytes dying during an ischaemic attack have been implicated in preventing tissue reperfusion (Hall et al., 2014). Renal pericytes contribute to ultrafiltration and blood pressure control, while hepatic stellate cells regulate sinusoidal blood flow and have immunoregulatory properties (Holm et al., 2018). Several of these functions have been attributed to distinct mural cell shapes with smooth muscle cells showing a perpendicular wrapping morphology with respect to the blood vessel axis, while pericytes align longitudinally with this axis (Armulik et al., 2011; Holm et al., 2018). Mural cells furthermore display heterogeneity in their origins, with those in the head region being derived from neural crest (Sinha and Santoro, 2018). The mesothelium, which is the epithelium lining the coelomic cavity, generates most of the other mural cell populations with contributions from somitic tissue and the secondary heart field (Armulik et al., 2011). The degree to which these differences in origin contribute to functional differences between mural cells remains to be determined. One master regulator of mural cell development is platelet-derived growth factor β (Pdgfb) signalling (Kazlauskas, 2017). Mural cells and their progenitors express platelet-derived growth factor receptor β (pdgfrb), while endothelial cells express pdgfb ligands (Gaengel et al., 2009). Loss of pdgfb or pdgfrb results in a reduction of pericyte numbers in a tissue-specific manner without affecting their initial specification (Crosby et al., 1998; Hellstrom et al., 1999; Lindahl et al., 1997). Analysis of chimaeric animals suggests that smooth muscle cells similarly require pdgfrb signalling (Crosby et al., 1998). In zebrafish, mural cells can be detected along the trunk vasculature on the second day of development (Fortuna et al., 2015; Santoro et al., 2009), around 12 h after their appearance on the cranial vasculature (Ando et al., 2016; Wang et al., 2014). Studies addressing the origins of mural cells showed that trunk and hindbrain pericytes are mesoderm derived whereas those in more anterior regions have a neural crest origin (Ando et al., 2016; Stratman et al., 2017; Whitesell et al., 2014). Smooth muscle cells in zebrafish express similar marker genes when compared with mouse (Santoro et al., 2009), such as tagln (SM22alpha) (Li et al., 1996), acta2 (Gabbiani et al., 1981) and myh11a (Miano et al., 1994). A recent study also identified a conserved set of zebrafish pericyte genes (Shih et al., 2021). Zebrafish mural cells also require pdgfrb signalling during their development (Ando et al., 2021; Stratman et al., 2017). Together, these findings suggest that mural cell biology during developmental stages is well conserved in zebrafish.

Previous studies investigating mural cell function during tissue injury indicated that pericytes retained properties of mesenchymal stem cells with cultured pericytes being able to contribute an array of different cell types, such as chondrocytes (Crisan et al., 2008), adipocytes (Tang et al., 2008), skeletal muscle (Dellavalle et al., 2011, 2007) and neurons (Dore-Duffy et al., 2006). More recent results, however, showed that lineage-labelled pericytes did not contribute to these cell types in various injury settings in vivo, calling their stem cell properties into question (Guimaraes-Camboa et al., 2017). Despite these findings, work in zebrafish demonstrated that cells derived from pdgfrb-expressing mural cells contribute a specific extracellular matrix promoting axonal growth after spinal cord injury (Tsata et al., 2021). Studies examining mural cells during wound healing suggest that differentiated pericytes are activated and proliferate, subsequently supporting angiogenesis, while at other times promoting fibrosis (Ansell and Izeta, 2015; Rodrigues et al., 2019). Thus, the functional significance of pericytes during tissue regeneration remains an area of debate. In addition, studies unequivocally providing evidence concerning the origin of newly forming pericytes either during wound healing or tissue regeneration have been lacking (Rodrigues et al., 2019). We also lack an understanding of the signalling pathways that guide mural cell development during regeneration and how impaired mural cell function might impact the function of newly formed blood vessels and tissues.

Here, we have investigated mural cell morphology, ontogeny and the signalling pathways important for their development and regeneration in the zebrafish fin using newly developed transgenic lines. We show that zebrafish mural cells show two distinct morphologies depending on their proximo-distal location within the fin. They express the mammalian smooth muscle cell marker myh11a and can be lineage traced to tissue-resident progenitor cells. These express pdgfrb, but not myh11a. Furthermore, mural cell differentiation during both fin development and regeneration relies on Pdgfrb signalling. Lineage-tracing experiments, although not providing definite results, suggest that new perivascular cells in regenerating fins are not derived from pre-existing mural cells. We also do not detect significant contribution of differentiated mural cells to other cell types. Together, our studies illustrate the morphological diversity of mural cells in the zebrafish fin, while arguing against their stem cell properties. We further propose that the signalling mechanisms patterning mural cells during juvenile stages are being redeployed during tissue regeneration.

During juvenile stages, the fin vasculature expands ventrally from an artery-vein loop in the posterior part of the embryo and continues to grow during adulthood (Fig. 1A-E). To investigate mural cell development during these stages, we imaged double transgenic fish: TgBAC(pdgfrb:citrine)^s1010; Tg(-0.8flt1:RFP)^hu5333. In these animals, arterial ECs express red fluorescent protein (RFP), while pdgfrb-positive cells express citrine (Bussmann et al., 2010; Vanhollebeke et al., 2015). We observed the occurrence of citrine-expressing cells along the artery as early as 1 week post-fertilization (wpf; Fig. 1B,B′ white arrowheads). These cells were present throughout juvenile stages at 2 wpf (Fig. 1C-C″, white arrowheads) and 3 wpf (Fig. 1D,D′, white arrowheads), as well as in 4 wpf fish (Fig. 1E-H′). In addition to mural cells, we detected distinct cell types with varying levels of citrine expression. One population consisted of ovoid cells either located at the fin ray base (Fig. 1C′, yellow arrowheads) or at the bone ray joints at later developmental time points (Fig. 1G″, yellow arrowheads). Moreover, we found cuboidal-shaped cells surrounding arterial blood vessels (Fig. 1F-H′, pseudo-coloured in light green in Fig. 1G′,G″,H′). To characterize and quantify these distinct cell populations, we divided the fin into four segments (S1-S4), starting with S1 closest to the fish body towards S4 at the distal end of the fin. We then measured cellular dimensions and distribution throughout these segments. This analysis showed that, on average, citrine-expressing mural cells were longer and narrower than cuboidal-shaped cells (Fig. 1I). We also found that mural cell numbers increased from distal to more proximal locations (Fig. 1J). To examine mural cells more closely, we generated triple transgenic zebrafish: TgBAC(pdgfrb:gal4)^ncv24; Tg(UAS:GFP)^nkuasgfp1a; Tg(-0.8flt:RFP)^hu5333. Analysis of these animals revealed additional morphological differences within the mural cell population. We found that GFP-expressing cells either wrapped around arteries (Fig. S1A-C′, Movies 1 and 2) or sent out long protrusions along the vessel axis (Fig. S1D-E′, red arrowheads, see also Fig. 1G-H′, Movies 3 and 4). These cells were differentially distributed along the proximo-distal fin axis. Although S4 was mostly devoid of mural cells, S3 contained mainly mural cells with long protrusions. Mural cells in segments closer to the body (S1 and S2) were characterized by their wrapping morphology (Fig. 1K). To further characterize these morphologies, we made use of a double transgenic approach using TgBAC(pdgfrb:gal4)^ncv24; Tg(UAS:KAEDE)^s1999t fish. These fish express the photoconvertible protein Kaede in mural cell populations. Kaede photoconversion revealed complex cell shapes consisting of numerous finger-like extensions surrounding arteries (Fig. 1L,M, Movies 5 and 6). Wrapping mural cells either showed distinct borders between each other or overlapped (Fig. 1L,M). Thus, using citrine expression, as well as cell shapes as defining criteria, we were able to distinguish at least four different populations of pdgfrb-expressing cells in developing zebrafish fins.

The classical perception of mural cell distribution along the vascular tree posits that proximal arterioles are covered by smooth muscle cells with their distinct morphologies surrounding arteries with their entire cell bodies, while more distally located capillaries are ensheathed by pericytes (Grant et al., 2019; Holm et al., 2018). Based on this, we hypothesized that the mural cells we detected in distal segments of the fin most likely resembled pericytes, but we were unsure to what extent wrapping cells in more proximal locations would be similar to smooth muscle cells. To further investigate these identities, we developed a transgenic line expressing yellow fluorescent protein (YFP) under the control of the myosin heavy chain 11a (myh11a) promoter, a gene reported to be smooth muscle cell specific (Miano et al., 1994; Vanlandewijck et al., 2018). Accordingly, when analysing double transgenic Tg(0.8flt1:RFP)^hu5333; TgBAC(myh11a:YFP)^mu125 zebrafish embryos at 5 days post fertilization (dpf), we detected YFP expression in mural cells surrounding arteries and in circulation (Fig. S2A-C). Prominent YFP expression was also detectable in gut smooth muscle cells (Fig. S2A, arrows). Similar to Tg(pdgfrb:citrine)^s1010 fish, we detected YFP-expressing cells ensheathing fin arteries as early as 1 wpf (Fig. S2D-E′, yellow arrowheads) and throughout juvenile stages up to 4 wpf (Fig. S2F-H′, yellow arrowheads). We then quantified their morphology and distribution along the proximo-distal axis in 4 wpf larvae (Fig. 2A-F). This analysis revealed a striking similarity with respect to their distribution when compared with citrine-expressing mural cells in TgBAC(pdgfrb:citrine)^s1010 fish. We detected both, wrapping cells (Fig. 2B,B′) and cells with protrusions (Fig. 2C,C′, red arrowheads) expressing YFP. A greater number of cells displayed wrapping morphology in proximal regions, whereas more distal regions contained mostly cells exhibiting protrusions (Fig. 2D,D′, red arrowhead). The most distal segment (S4) was again mostly devoid of mural cells (Fig. 2F). Thus, in the zebrafish fin, mural cells without a characteristic smooth muscle cell morphology also express YFP driven by the myh11a promoter.

To determine the relationship between pdgfrb- and myh11a-expressing cells, we generated double transgenic animals TgBAC(pdgfrb:mCherry)^ncv23; TgBAC(myh11a:YFP)^mu125, expressing mCherry under the control of the pdgfrb promoter and YFP in myh11a-positive cells (Fig. 2G-L). All YFP-expressing mural cells examined were positive for both transgenes. By contrast, mCherry-expressing cuboidal cells did not express YFP (Fig. 2L′, magenta asterisks). We again detected a higher number of wrapping double-positive cells in proximal fin regions, whereas distal fin regions mostly contained cells with protrusions (Fig. 2M). To determine whether pdgfrb and myh11a co-expression was a specific feature of fin mural cells, we analysed mural cells in the trunk of embryonic zebrafish at 6 dpf (Fig. S3A-E). In this setting, we found pdgfrb-expressing mural cells on both arterial and venous intersegmental vessels (ISVs). A greater number of mural cells on arterial ISVs also expressed the myh11a transgene when compared with venous ISVs, while mural cells only expressing pdgfrb were more abundant on venous ISVs. Together, these findings reveal that distinct mural cell morphologies in the zebrafish fin do not correlate with differences in the expression of the contractile protein Myh11a. Our findings further show that pdgfrb is more broadly expressed when compared with myh11a (Fig. 2N).

As pdgfrb and myh11a were surprisingly expressed in all mural cell populations, we sought to identify markers that would label different subsets of fin mural cells. Acta2 is a known smooth muscle cell marker both in mice (Gabbiani et al., 1981) and zebrafish (Whitesell et al., 2014). We combined Tg(acta2:mCherry)^ca8 with TgBAC(myh11a:YFP)^mu125 fish and analysed fluorescent protein expression in their progeny. At embryonic stages, we found distinct and overlapping expression domains of these two transgenes (Fig. S3F-H). The outflow tract of the heart showed myh11a expression, while the ventricle was expressing acta2 (Fig. S3G). In intersegmental vessels, we detected both transgenes in mural cells on arteries (Fig. S3H). We then analysed the fin vasculature at 4 wpf (Fig. 3A-D) and at 6 wpf (Fig. 3E-H). This analysis revealed the existence of acta2/myh11a double-positive mural cells on proximally located arteries, while acta2 expression was absent in more distal regions of the fin (Fig. 3D,H). Thus, mural cells with elongated morphology do not express acta2, while wrapping cells can express either myh11a or myh11a together with acta2. These findings illuminate morphological and gene expression differences within mural cells of the zebrafish fin. They furthermore suggest that the previously established proximo-distal hierarchy in terms of smooth muscle cell to pericyte transition along the vascular tree is conserved in the zebrafish fin. We thus define those fin mural cells that express acta2, myh11a and pdgfrb and exhibit a wrapping morphology as smooth muscle cells. Mural cells expressing myh11a together with pdgfrb and showing an elongated morphology, we define as pericytes. Cells with wrapping morphology displaying myh11a and pdgfrb expression, but no acta2 expression might constitute an intermediate mural cell population.

Notch signalling is implicated in regulating mural cell specification (Ando et al., 2016) and proliferation (Wang et al., 2014). We therefore asked whether we might observe differences in Notch signalling pathway activation in different mural cell populations. Analysis of Tg(TP1bglob:VenusPEST)^s940; Tg(acta2:mCherry)^ca8 animals revealed Notch pathway activation (indicated through venusPEST expression) predominantly in mural cells in distal fin regions, whereas acta2-expressing mural cells were mostly devoid of notch pathway activation (Fig. 3I-L). We also analysed the overlap of pdgfrb expression with Notch pathway activation using TgBAC(pdgfrb:gal4ff)^ncv24; Tg(UAS:GFP)^nkuasgfp1a; Tg(TP1bglobin:H2B-mCherry)^s939 animals (Fig. S4). This analysis revealed Notch pathway activation in pdgfrb-expressing mural cells in distal fin regions, while in more proximal fin regions, we detected pdgfrb-positive cells that did not exhibit notch pathway activity. Thus, pericytes in distal fin regions show Notch pathway activation, whereas smooth muscle cells in more proximal regions do not.

We then set out to determine the cell population giving rise to fin mural cells. To do so, we generated transgenic fish that express the photoconvertible Dendra2 protein in the nuclei of pdgfrb-positive cells: TgBAC(pdgfrb:H2B-dendra2)^mu158 (Fig. S5A-D). Proof of concept studies in embryos showed that Dendra2 was expressed in mural cell populations in different embryonic regions (Fig. S5B′-D, white arrowheads) and could be photoconverted upon blue light exposure (Fig. S5B′-D, blue arrowheads). Based on their location close to arteries, we focused on cuboidal-shaped cells as potential mural cell precursors. To examine possible changes in cell shapes, we crossed TgBAC(pdgfrb:H2B-dendra2)^mu158 fish with TgBAC(pdgfrb:gal4ff)^ncv24; Tg(UAS:EGFP)^nkuasgfp1a fish, expressing cytoplasmatic EGFP in pdgfrb-positive cells. Examination of individual photoconverted cuboidal cells 5 days post-photoconversion (dpc; Fig. 4) revealed that a proportion of photoconverted cells had undergone cell shape changes, now extending long processes along the vessel axis. As we did not photoconvert single cells, we analysed how many photoconverted cells transitioned towards the mural cell lineage. This analysis revealed that most cells maintained their cuboidal morphology after photoconversion (Fig. 4E-G). These results suggest that fin mural cells can differentiate from local precursors that express pdgfrb.

Previous studies showed that col1a2 expressing fibroblasts can give rise to pericytes during embryonic development (Rajan et al., 2020). We therefore asked whether cuboidal-shaped cells belonged to the fibroblast lineage. To do so, we used a transgenic line expressing GFP under the control of the col1a2 promoter (Rajan et al., 2020) and analysed GFP expression in the fin. Mural cells in proximal fin regions only expressed pdgfrb, while we detected col1a2 expression in cuboidal-shaped cells (Fig. 5A,B). In middle regions of the fin, double positive cells of cuboidal morphology were readily detectable (Fig. 5C). Mural cells did not show col1a2 expression in this fin region, either. In most distal regions, we found double positive cells of cuboidal (Fig. 5D,E, red arrowheads) and mural cell morphology (Fig. 5D,E, blue arrowheads). To further investigate these changes, we sorted mural cells based on their co-expression of pdgfrb and myh11a and single pdgfrb expressing cells from zebrafish fins and compared the expression patterns of fibroblast and mural cell genes (Fig. 5F,G). Double positive cells showed an upregulation of mural cell markers, while trending towards a decrease in fibroblast markers when compared to single pdgfrb expressing cells (Fig. 5G). Thus, like in embryonic zebrafish, mural cells in the fins of juvenile zebrafish might be partially derived from cells of fibroblastic origin, which downregulate fibroblast markers during their maturation.

To determine the signalling pathways required for fin mural cell differentiation, we examined pdgfrb^um148 mutant zebrafish (Kok et al., 2015). We again divided fin arteries into four segments and quantified the distribution of mural cells using TgBAC(pdgfrb:gal4)^ncv24; Tg(UAS:GFP)^nkuasgfp1a; Tg(-0.8flt1:RFP)^hu5333 transgenic animals. In wild-type siblings, we observed a decrease of mural cell numbers along the proximo-distal fin axis (Fig. 6A-C′, yellow arrowheads, quantified in Fig. 6G). Cuboidal cells were more homogenously distributed (Fig. 6B′,C′, segments S3 and S4, blue arrowheads, quantified in Fig. 6H). In pdgfrb^um148 mutant fish (Fig. 6D-F′), we observed only a small reduction in mural cell numbers in proximal segments (Fig. 6E,E′, yellow arrowheads, quantified in Fig. 6G, Movies 7 and 8), whereas those in distal segments were strongly decreased (Fig. 6F,F′, quantified in Fig. 6G). Of interest, cuboidal cells in distal fin regions increased in number in mutant animals (Fig. 6F,F′, blue arrowheads, quantified in Fig. 6H, Movies 9 and 10). To corroborate our findings, we also analysed the distribution of mural cells in TgBAC(pdgfrb:citrine)^s1010; Tg(-0.8flt1:RFP)^hu5333 and TgBAC(myh11a:YFP)^mu125; Tg(-0.8flt1:RFP)^hu5333 fish (Fig. S6). We detected a similar change in mural/cuboidal cell distribution in these transgenic fish and additionally observed an increase in citrine expression levels in cuboidal cells in pdgfrb^um148 mutants (Fig. S6K). This suggests a block in differentiation of cuboidal-shaped cells into the mural cell lineage in pdgfrb^um148 mutants, while at the same time leading to an upregulation of pdgfrb promoter activity in this precursor lineage. Therefore, also in the zebrafish fin, Pdgfrb signalling is instrumental for proper mural cell specification. However, proximal blood vessel segments were still ensheathed with mural cells expressing transgenes driven by both pdgfrb and myh11a promoters. This resembles early larval stages, where some smooth muscle cell populations also appear unaffected in pdgfrb^um148 mutants (Ando et al., 2021).

We then determined possible changes in vascular function in pdgfrb^um148 mutants. To do so, we measured blood vessel diameters along the proximo-distal fin axis. This analysis revealed an increase in blood vessel diameters in distal segments in mutant animals (Fig. 6I). This indicates that, similar to mice (Hellstrom et al., 2001) and zebrafish brain blood vessels (Ando et al., 2021), loss of mural cells in pdgfrb^um148 mutant animals results in fin blood vessel dilation.

To examine whether the same signals contribute to mural cell development in tissue regeneration, we performed two types of fin regeneration experiments in wild-type and pdgfrb^um148 mutant animals. We either removed a piece of tissue containing a bone ray and an artery in the centre of a fin ray (Fig. 7A), or amputated about 50% of the fin (Fig. S7). We then allowed the tissue to regenerate and determined mural cell numbers. In wild-type fish, we observed a newly forming artery as early as 3 days post-injury (dpi; Fig. 7B,C), with a fully regenerated artery ensheathed by mural cells being present at 5 dpi (Fig. 7D-F). When comparing wild-type fish (Fig. 7G-I,M) at 7 dpi to pdgfrb^um148 mutants, we observed a significant reduction in mural cell recruitment to the artery in mutants (Fig. 7J-M). We also detected a reduction in mural cell coverage around regenerated arteries in the fin amputation assay, both in TgBAC(pdgfrb:citrine)^s1010 (Fig. S7A-D′) and in TgBAC(myh11a:YFP)^mu125 transgenic animals (Fig. S7E-H′, quantified in I). Although these defects did not lead to a reduction of regenerate outgrowth (Fig. S8A-C), we observed an increase in connections between regenerated arteries and veins in pdgfrb^um148 mutant fish (Fig. S8A′,B′, red asterisks, quantified in D). Thus, also in regenerative settings, pdgfrb signalling is a prerequisite for proper mural cell differentiation. These findings suggest that developmental signals are re-used during mural cell specification in regenerating tissues.

During regeneration, either tissue-resident stem stells or dedifferentiated lineage-restricted precursor cells are the source for new tissues and cell types (Tanaka and Reddien, 2011). To analyse whether new mural cells are derived from pre-existing ones, we tracked mural cells during regeneration through photoconverting Dendra2 protein in cells around the injury site in TgBAC(pdgfrb:H2B-dendra2)^mu158 fish (Fig. 8A-C). We then determined the contribution of photoconverted mural cells to the newly formed tissue at 5 dpi (Fig. 8D,E). We exclusively detected newly made green Dendra2 protein in the regenerated region of the fin, while the surrounding tissue still contained an abundance of previously photoconverted red cells (Fig. 8E).

As transgenic lines can show mosaicism and not all Dendra2-expressing cells might be photoconverted in our experiments, we sought to verify our lineage-tracing results using an alternative approach. We decided to directly image mural cells on blood vessels next to generating arteries on consecutive days (Fig. 9). This analysis revealed only subtle changes in the location of pre-existing mural cells on uninjured blood vessels (Fig. 9A), while we readily detected the emergence of new mural cells on nascent arteries (Fig. 9A, day 5-7, asterisks). We also analysed whether newly forming cuboidal cells would give rise to regenerating mural cells, as we had observed in growing fins. Examination of TgBAC(pdgfrb:mcherry)^ncv23; TgBAC(col1a2)^ca103 fish indeed showed the existence of cells expressing both transgenes at early stages of artery regeneration (Fig. 9F,G). However, at later stages, the number of double-positive cells diminished, indicating that cells had differentiated in either the fibroblast or mural cell lineage. Together, these findings indicate that pre-existing mural cells are not a major source of newly forming mural cells during tissue regeneration in the zebrafish fin (see summary of findings in Fig. S9). They also suggest that these cells do not serve as stem cells during the regeneration of other fin tissues.

We generated transgenic zebrafish lines to investigate blood vessel mural cell populations in the zebrafish fin during adult stages and in tissue regeneration. We discovered the existence of several distinct mural cell populations that differed with respect to their morphology and location: in fin regions proximal to the fish body, we observed mural cells ensheathing arteries, while those in distal regions extended long processes along the proximo-distal vessel axis. These morphologies closely resemble those recently observed for subsets of mural cells on mammalian brain blood vessels (Grant et al., 2019). Furthermore, we detected differences in gene expression patterns within fin mural cells of different morphologies. All mural cells expressed pdgfrb and myh11a. This was a surprising finding, as myh11a was previously reported to be exclusively expressed in differentiated smooth muscle cells (Miano et al., 1994). We detected expression of acta2 in myh11a-positive mural cells in proximal fin regions, suggesting the existence of smooth muscle cells that express acta2 and myh11a, while mural cells in distal regions share pericyte characteristics.

Single-cell sequencing data of mouse brain tissue revealed that some pericytes also expressed myh11, while no acta2 expression was detected in this cell population (Vanlandewijck et al., 2018). In line with this assessment, single-cell sequencing of mural cell populations of different mouse organs has revealed a surprising diversification of pericytes, also in respect to the expression of components of the SMC contractile machinery, such as myh11, tagln and acta2 (Muhl et al., 2020; Vanlandewijck et al., 2018). In organs such as the bladder or colon, pericytes expressed tagln and acta2, while in the brain and lung they did not. Therefore, the repertoire of contractile protein expression varies between pericytes and smooth muscle cell populations, as we now also show for mural cells in the fin of zebrafish. It will be of interest to investigate the expression of recently identified pericyte-specific zebrafish transgenes (Shih et al., 2021) in fin mural cells in order to better characterize the distribution of mural cells displaying smooth muscle or pericyte features.

We show that cuboidal-shaped cells that express fibroblast genes, such as col1a2, can give rise to mural cells in distal regions of the fin. These findings are in line with results in embryonic zebrafish, where col1a2-expressing cells contribute to mural cells on intersegmental blood vessels (Rajan et al., 2020). We find a similar change in gene expression from col1a2 high to pdgfrb high during fin regeneration, suggesting that the lineage transformation from cells of fibroblastic origin towards the mural cell lineage is well conserved. At present it is unclear whether other cell populations can similarly contribute to the mural cell lineage, as suggested by the differential effects of mutations in pdgfrb on mural cells in proximal and distal fin regions. More elaborate lineage-tracing studies will be needed to address these questions.

Fin blood vessels consist of arteries and veins that run through the entirety of the fin and are connected by inter-vessel commissures, but an elaborate capillary bed is missing, in particular in distal fin regions (Huang et al., 2003; Kametani et al., 2015; Xu et al., 2014). Our blood vessel diameter measurements revealed that fin arteries become larger towards distal regions of the fin, which contrasts with other vascular beds. Of note, in these distal regions, we predominantly detect blood vessel diameter increases in pdgfrb mutants. Further studies will be needed to decipher the functions of fin mural cells on vascular patterning and function, and how they compare with other organs and tissues.

The finding that newly forming mural cells in the fin were not derived from pre-existing ones was surprising, as previous studies suggested that pericytes can function as mesenchymal stem cells, even though this notion was later called into question (Birbrair et al., 2017). During wound healing, several sources of pericytes, such as pre-existing pericytes or mesenchymal stem cells, are being discussed (Morikawa et al., 2019). Although a previous study showed that bone marrow-derived mesenchymal stem cells could contribute to the pericyte lineage during wound repair (Sasaki et al., 2008), definitive lineage tracing data concerning the contribution of pre-existing pericytes is still lacking. Our studies did not find significant contribution of pre-existing mural cells to other cell types within regenerating fin tissue. This supports previous lineage-tracing results in mice showing limited stem cell properties of mural cells (Guimaraes-Camboa et al., 2017). Other studies, however, showed that pericytes can give rise to scar tissue during spinal cord injury (Göritz et al., 2011). This tissue prevented axonal regrowth in mice (Dias et al., 2018). In embryonic zebrafish, progeny of pdgfrb-expressing mural and myoseptal cells prevented scarring during spinal cord injury and thereby promoted axon regeneration (Tsata et al., 2021). Therefore, differences in the potential of perivascular cells during tissue regeneration and or healing might exist.

In keeping with the requirement for Pdgfrb signalling during pericyte development in embryonic stages (Armulik et al., 2011), inhibition of such signalling after wounding led to impaired pericyte recruitment (Rajkumar et al., 2006). These findings, together with our work showing deficient mural cell recruitment in pdgfrb mutants during tissue regeneration, underline the role of Pdgfrb signalling as a master regulator of pericyte biology in different biological settings.

We found that mural cells in proximal fin regions still formed in pdgfrb mutants, while at the same time displaying a morphology distinct from perivascular cells in distal regions. This might suggest that distinct signalling pathways controlling mural cell recruitment and differentiation exist in different areas of the fin. We now find activation of the Notch signalling pathway only in mural cells in distal fin regions, supporting this hypothesis. These observations might further indicate separate embryonic origins for mural cells along the proximo-distal fin axis. Alternatively, distally located perivascular cells might give rise to proximal mural cell populations in wild-type settings, as suggested by a recent study investigating smooth muscle cell development in coronary arteries (Volz et al., 2015). Here, pericytes located on distal vessel branches are a precursor population for smooth muscle cells. In pdgfrb mutants, this route of distal to proximal differentiation would be compromised, leading to the activation of alternative sources for mural cells not normally used in wild-type fish. Detailed linage-tracing studies in combination with single-cell sequencing approaches will be needed to answer these questions and to determine how they relate to mural cells in other tissues and organisms.

Zebrafish embryos were maintained in 1×E3 under standard husbandry conditions for 5 days in an incubator at 28.5°C. After 5 days, the embryos were transferred to embryo tanks in the zebrafish facility, where they were raised until adulthood. Previously described zebrafish lines were Tg(0.8flt1:RFP)^hu5333 (Bussmann et al., 2010), TgBAC(pdgfrb:citrine)^s1010 (Vanhollebeke et al., 2015), TgBAC(pdgfrb:mCherry)^ncv23 (Ando et al., 2016), TgBAC(pdgfrb:gal4ff)^ncv24 (Ando et al., 2016), Tg(kdrl:TagBFP)^mu293 (Matsuoka et al., 2016), pdgfrb^um148 (Kok et al., 2015), TgBAC(col1a2:GFP)^ca103 (Ma et al., 2018), Tg(acta2:mCherry)^ca8 (Whitesell et al., 2014), Tg(TP1bglob:VenusPEST)^s940 (Ninov et al., 2012), Tg(TP1bglob:H2BmCherry)^s939 (Ninov et al., 2012) and Tg3(fli1:LIFEACT-EGFP)^mu240 (Hamm et al., 2016). All animal experiments were performed in compliance with the relevant laws and institutional guidelines, and were approved by local animal ethics committees of the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen.

Transgenic lines were generated using Bacterial artificial chromosome (BAC) recombineering. To generate the TgBAC(myh11a:YFP)^mu125 and TgBAC(pdgfrb:H2B-dendra2)^mu158, the start codons of myh11a or pdgfrb in BAC clone CH73-223E22 or CH1073-606I16 were replaced with YFP or H2B-Dendra2 cassette using Red/ET recombineering (GeneBridges). The YFP cassette was amplified by PCR from pCS2+Citrine_kanR with the primers myh11a _HA1_citrine_fw (5′-ttatcaatcagtttttctccttaaatttgcagttgtcttaacccagcaccACCATGGTGAGCAAGGGCGAGGAG-3′) and myh11a_HA2_kanR_rev (5′-tctttgtccgtgaaaaggaatttctcatcatcgctcaagcctttcttcgtTCAGAAGAACTCGTCAAGAAGGCG-3′); homology to the BAC vector is depicted in lowercase throughout. iTol2_amp cassette was inserted into the vector back bone using a construct amplified with primers pTarBAC_iTol2_fw (5′-gcgtaagcggggcacatttcattacctctttctccgcacccgacatagatCCCTGCTCGAGCCGGGCCCAAGTG-3′) and pTarBAC_iTol2_rev (5′-gcggggcatgactattggcgcgccggatcgatccttaattaagtctactaATTATGATCCTCTAGATCAGATC-3′). The H2B-Dendra2 cassette was amplified by PCR from pCS2+H2B-Dendra2_kanR with primers Pdgfrb_HA1_dendra2_fw (5′-tgttgttttctctccgtctgcagtgttgaatgtgtcctgctctagaagaaCCACCATGCCAGAGCCAGCGAA) and Pdgfrb_HA1_dendra2_rev (5′-ttgtgatagcagtgaataggaagtggatgcggctgatggtcgaactcttTCAGAAGAACTCGTCAAGAAGGCG-3′). The iTol2_amp cassette was inserted into the vector back bone using a construct amplified with primers pCC1FOS_ iTol2 _fw (5′-tctctgtttttgtccgtggaatgaacaatggaagtccgagctcatcgctaCCCTGCTCGAGCCGGGCCCAAGTG-3′) and pCC1FOS_ iTol2 _rev (5′-cgacacccgccaacacccgctgacgcgaaccccttgcggccgcatcgaatATTATGATCCTCTAGATCAGATC). BAC DNA was purified using Midiprep (Invitrogen) on the day of injection. A cocktail of 100 pg BAC DNA and 50 pg of tol2 transposase mRNA per embryo was injected into wild-type embryos.

Adult zebrafish were anaesthetized with 0.02% tricaine and transferred to a Petri plate. Bone resection was performed by removing a piece of bone ray enclosing an artery in the third or fourth fin ray of the dorsal or ventral lobe, using a scalpel. For fin amputation, ∼50% of the fin was amputated and the third fin ray from the dorsal or ventral lobe was used for analysis. The injured fish were kept in individual tanks until experiments were carried out.

Larvae or amputated fins were embedded in 1% low melting point agarose. Fluorescent confocal images were acquired using a Zeiss LSM 780/880 (objective lens: 20× Plan Apo NA 0.80 or 40× LD C Apochromat NA 1.10) or Leica SP8 (objective lens: HC PL Fluotar 20×/0.50 or HCX APO L 63).

For lineage tracing of cuboidal cells, we performed photoconversion in the distal regions of the juvenile fin. The anesthetized larvae were embedded in 0.75% agarose containing 0.02% tricaine. Cuboidal-shaped cells were identified based on their morphology. We then chose regions in the fin that contained only cuboidal-shaped cells and were devoid of cell with elongated mural cell morphologies. We marked these regions containing between 5 and 20 cells in the Zeiss Zen software. The selected cells were exposed to 8% of 405 nm laser excitation wavelength with 40 iterations for 25 cycles. Immediately after photoconversion, we imaged fins to determine the number of photoconverted cells. This image also served as 0 dpc. The larvae were removed from the glass plates and placed in fresh E3 medium with no light exposure and reimaged after 5 days. For lineage tracing of photoconverted Dendra2 cells during adult fin regeneration, bone resection was performed in adult fins; at 12 h post-injury, anesthetized fish were transferred to glass plates and immobilized with 1.5% agarose. To keep fish sedated, E3 media containing 0.02% tricaine was added to the glass plate and the agarose around the gills, heart and mouth of the fish was carefully removed. Photoconversion was performed by selecting regions in Zeiss Zen, around the resected bone with 10% of 405 nm laser with 40 iterations for 25 cycles. After photoconversion, fish were placed in individual tanks containing fresh E3. To speed up recovery, we used a pipette to transfer E3 to the mouth and gills of the fish. Tanks containing photoconverted fish were kept in dark with minimal exposure to light provided only during feeding. For photoconversion of Kaede-expressing cells, amputated fin segments were embedded in agarose and photoconverted using a 405 nm laser with laser power of 8% and 30 iterations. Images for native or photoconverted Dendra2 and Kaede were acquired using laser settings for EGFP and RFP. For repetitive imaging, adult fish were immobilized on glass plates as described before and recovered after every session.

Acquired images were stitched using Zeiss ZEN software or Leica LAS-X. Imaris software (Bitplane) was used to generate maximum intensity projections. The dimensions of cuboidal and mural cells with protrusions were determined by measuring their length and width. The distribution of mural cells that wrap or extend protrusions was quantified using Imaris software. Artery diameter along the proximal distal axis was calculated by measuring vessel diameter in at least ten regions for each of the four segments. Data were analysed using Prism 9 (Graphpad) and graphs were plotted as mean±s.d. P<0.05 was considered statistically significant. ImageJ (NIH) was used to generate monochrome images of different fluorophores and to calculate fluorescent intensity. Adobe Illustrator software was used to compile images.

Caudal fins of 4 wpf zebrafish were amputated and washed twice with 1× HBSS (Gibco, 14185052) in a 1.5 ml microcentrifuge tube. Amputated fins were treated with TrypLE Express Enzyme (1×) (Gibco, 12604013) at 37°C for 30 min with gentle agitation. TrypLE-treated samples were centrifuged at 500 g for 5 min at 4°C, and the supernatant was carefully removed and discarded. The samples were further treated with 1 ml of 0.50 mg/ml of Liberase DL (Roche, 5401160001) in 1× HBSS at 37°C for 30 min with agitation to obtain a single-cell suspension. Dissociated cells were pelleted by centrifugation at 350 g for 5 min at 4°C. The cell pellet was resuspended in 1× HBSS buffer and centrifuged twice. The cells were resuspended in 500 µl of 1× HBSS and passed through a 40 μm nylon filter. Fluorescence-activated cell sorting (FACS) was performed for pdgfrb- (RFP), myh11a- (YFP) and RFP/YFP-expressing mural cells (RFP/YFP). Sorted cells were collected in RLT buffer followed by RNA isolation using the RNeasy Micro kit (Qiagen, 74004), and cDNA synthesis was performed using the iScript cDNA Synthesis Kit (BioRad, 1708891). qPCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems, 4367659). The ddCt method was used to calculate relative expression using RPL13a as endogenous control, dCt data were used to perform statistical analysis. Primer sequences are listed in Table S1.

We thank Naoki Mochizuki, Sarah Childs and Peng Huang for sharing transgenic lines. We also thank Roman Tsaryk for performing FACS. We are grateful to Reinhild Bussmann, Mona Finch-Stephen, Bill Vought and Nadine Greer for excellent fish care.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199640.