Fibronectin mediates correct positioning of the interrenal organ in zebrafish


Correspondence to: Yi-Wen Liu, Department of Life Science, Tunghai University, No. 181, Sec. 3, Taichung-Port Road, Taichung, 40704, Taiwan R.O.C. E-mail:


Background: Fibronectin (Fn) forms a centripetal gradient during the fetal adrenal gland organogenesis, and modulates hormone responsiveness of adrenocortical cells in the primary culture. However, how Fn is involved in organ formation of the adrenal gland remains unclear. Results: In this study, we found that Fn accumulates around migrating ff1b-expressing interrenal cells, which were marked by the ff1b promoter-driven transgenic fluorescence, during the course of interrenal organ assembly. The interrenal cells displaying the migratory phenotype were absent in the fn1 mutant, while specification and kidney association of the interrenal tissue remained normal. The Fn deposition in the interrenal microenvironment was severely reduced in the vessel-deficient ets1b morphant, implying its origin of synthesis from the peri-interrenal vasculature. In the fn1 mutant, early-migrating chromaffin cells were capable of interacting with steroidogenic interrenal cells, yet continuous migration and midline convergence of chromaffin cells were disrupted. Migration defects of both interrenal and chromaffin lineages, in the absence of Fn, thus led to incomplete interrenal organ assembly in aberrant positions. Conclusions: Our results indicate that Fn is essential for patterning interrenal organ formation, by modulating the migratory behavior of both steroidogenic interrenal and chromaffin cells. Developmental Dynamics 242:432–443, 2013. © 2013 Wiley Periodicals, Inc.


The teleostean interrenal gland is a functional counterpart of the mammalian adrenal gland, a crucial component of the hypothalamic-pituitary-adrenal axis. The adrenal gland is composed of the outer cortex, which produces steroid hormones in a zone-specific manner, and the inner medulla, which synthesizes catecholamines. Although lacking a highly organized zonal structure, the interrenal gland demonstrates conserved functional features as the major steroid-producing organ. Functional assembly of the interrenal gland is marked by the integration of steroidogenic interrenal tissue with chromaffin cells, the functional equivalent of adrenal medulla. In the zebrafish embryo, the integration between interrenal and chromaffin cells is evident by 3 days post-fertilization (dpf), when the fish embryo is able to physiologically respond to stress and secrete steroids (To et al., 2007; Alsop and Vijayan, 2009). It means that teleostean steroid function in the embryo is initiated concurrently with highly dynamic tissue–tissue interactions. Meanwhile, a hypothalamic-pituitary-interrenal axis develops and functions as early as assembly of the interrenal organ, which is essential for maintaining the growth and differentiation of interrenal steroidogenic cells. Hence, these temporally compressed events for zebrafish interrenal organogenesis might enable us to explore those molecular and cellular mechanisms that are conserved between mammals and teleosts.

In the zebrafish, the interrenal tissue is derived from the embryonic kidney (the pronephros) during the segmentation stage, and is segregated from the pronephric field by 30 hr post-fertilization (hpf). Both the kidney and the interrenal tissue arise as bilateral clusters across the midline, and converge at the midline in a coordinated fashion (Hsu et al., 2003; Liu, 2007). While central assembly of the pronephros requires the hemodynamic force (Serluca et al., 2002), medial fusion of bilateral primordial interrenal tissues is regulated by the interrenal-associated endothelium, possibly prior to vessel formation (Liu and Guo, 2006). Bilateral interrenal tissue clusters merged at the midline, and then relocalized to the right of the midline, with the relocalization event regulated by the adjacent axial artery and vein, which are under active assembly. As a result, the interrenal tissue from 28 hpf onwards is located to the right of the midline, and juxtaposed between the dorsal aorta (DA) and the right branch of the posterior cardinal vein. From about 2 dpf onwards, the interrenal cell cluster further extends medially across the midline, and becomes integrated with the chromaffin cells, which are derived from the trunk neural crest. The assembly of the interrenal organ is evident at 3 dpf by the intimate intermingling between interrenal and chromaffin cells. Intrinsic and extrinsic molecules that determine the interrenal cell migration during the process of interrenal organ assembly have remained unclear.

By using a fluorescent transgenic line driven by the promoter of nuclear receptor ff1b, the functional ortholog of adrenal 4 binding protein/steroidogenic factor 1 (Ad4BP/SF-1; officially designated NR5A1) as well as the key gene for specification and differentiation of the interrenal tissue (Chai et al., 2003; Hsu et al., 2003; Quek, 2009), we found in this study that interrenal cells appeared to display an active migratory phenotype during the stage of interrenal organ assembly. Interestingly, a rich enrichment of extracellular matrix protein Fn was deposited around migrating ff1b-expressing cells during the interrenal morphogenetic movement. Fn, a major adhesive molecule of the extracellular matrix, interacts with the Integrin receptor family to mediate a wide range of cellular processes including cytoskeletal organization, cell-substratum adhesion and spreading, and cell migration (Hynes, 1992; Astrof and Hynes, 2009). In the human fetal adrenal gland, Fn forms a centripetal gradient of accumulation, and is highly enriched at the fetal zone (Chamoux et al., 2001). Experiments on primary cultures derived from whole fetal adrenal glands have revealed that Fn orchestrates with other extracellular matrix components including Laminin and Collagen IV to modulate hormonal responsiveness, steroid secretion, and cell turnover (Chamoux et al., 2002a, 2002b). On the other hand, the Fn distribution of the fetal adrenal gland appears highly parallel with the centripetal migration of fetal adrenocortical cells (Ishimoto and Jaffe, 2011). However, whether Fn participates in the fetal adrenal centripetal migration remains unknown. It is also unclear how the Fn accumulation in the fetal adrenal gland is regulated.

Studies performed on the zebrafish fn1 mutant (natter) have revealed essential roles of Fn for the development of somites and the heart. Fn accumulates at the somite boundaries in an Integrinα5-dependent manner, and is required for maintenance but not specification of somite boundaries (Koshida et al., 2005). Fn is also detected around the myocardial precursors, and at the midline region between the endoderm and endocardial precursors (Trinh and Stainier, 2004). The peri-myocaridal Fn appears to be generated by the endothelium, and regulates epithelial organization for the migrating myocardial cells. We, therefore, hypothesized that the Fn deposition associated with steroidogenic interrenal cells might be essential for the interrenal morphogenetic movement, and hence implicating a role of Fn in the interrenal–chromaffin interaction. Therefore, we utilized the genetic fn1 mutant to check whether the interrenal morphogenetic movements were perturbed in the absence of Fn. We found that the fn1 mutant displayed a normal specification of the interrenal tissue, which was successfully segregated from and remained associated with the pronephros. However, defects of interrenal morphogenetic movements were exhibited in the fn1 mutant. Furthermore, the Fn accumulation in the interrenal microenvironment was severely reduced in embryos deficient in Ets1b, which is essential for blood vessel formation, suggesting that the source of Fn in the interrenal microenvironment was at least in part contributed by the neighboring vasculature. Interestingly, the convergence of chromaffin cells at the midline was also perturbed in the fn1 mutant, with initial but not continuous recruitment of chromaffin cells to the interrenal region being successful. Defective morphogenetic movements of both interrenal and chromaffin cells in the fn1 mutant together led to incomplete assembly of the interrenal organ in aberrant locations. Our study thus supported a role for Fn to be essential for cell migratory events that ensure correct interrenal organ assembly in teleosts.


The Fn Protein Accumulates Around Migrating ff1b-Expressing Interrenal Cells During the Stage of Interrenal Organ Assembly

Following the early midline fusion and lateral relocalization, the steroidogenic component of the interrenal organ develops into a bilobed organ lateral to the notochord from 3 dpf onwards, with the right lobe generally larger than the left lobe (Chai et al., 2003; To et al., 2007). It remains unclear whether the asymmetric bilobed organ structure is due to uneven expansion, or to the migratory behavior of steroidogenic cells during a general expansion of developing interrenal tissue. To check whether steroidogenic cells display migratory activity following the early morphogenetic movements, a transgenic line Tg(ff1bEx2:GFP) was used to mark both primordial and differentiated interrenal cells (Fig. 1). Our earlier study showed that the reporter gene expression driven by the ff1b promoter could recapitulate the expression of ff1b during zebrafish embryogenesis, and has been used to delineate morphogenetic movements of the developing interrenal tissue (Chou et al., 2010).

Figure 1.

The distribution of Fn around migrating interrenal cells. Interrenal cells (GFP as driven by ff1b promoter in A–D and A″–D″) formed a protruding extension at 2 dpf (B, B″) and dispersed clusters at 2.5 (C, C″) and 3 dpf (D, D″), on transverse sections of Tg(ff1bEx2:GFP) embryos. Migrating interrenal cell clusters (denoted by yellow arrows) were closely associated with the Fn protein (red fluorescence in A'–D' and A″–D″; Fn enrichment in the interrenal microenvironment highlighted by white arrowheads) throughout the medial extension process of the interrenal tissue. All sections are oriented with the dorsal side toward the top. White dotted lines indicate the position of the midline. NC, notochord; DA, dorsal aorta. Scale bar = 50 μM.

Vibratome sections of the Tg(ff1bEx2:GFP) embryos demonstrated that while ff1b-expressing cells formed a single cluster to the right of the midline by 1.5 dpf (Fig. 1A), a protruding extension toward the midline was detected at 2 dpf (Fig. 1B), which was possibly constituted by migrating interrenal cells. At 2.5 dpf, interrenal cell clusters showing migratory behavior were detected near the midline region (Fig. 1C), and some steroidogenic cells were situated at the left of the midline, obviously representing those cells that have migrated across the midline. The migration and extension of the interrenal tissue led to a dispersed distribution of interrenal morphology at 3 dpf (Fig. 3D), which was accompanied by a continuous size expansion of the ff1b-expressing domains. It thus suggested that the asymmetric bilobed structure of the interrenal organ from 3 dpf onwards was actually due to a continuous migratory behavior of interrenal cell clusters from 2 to 3 dpf.

Interestingly, the protruding extension of the interrenal tissue cluster at 2 dpf was directed toward an enriched deposition of Fn protein accumulated near the midline region and ventral to the notochord and the DA (Fig. 1B'', white arrowheads), while the Fn protein was also deposited around the notochord as well as the DA. From 2.5 to 3 dpf, interrenal clusters migrating across the midline continued to associate with the Fn deposited along the course of interrenal medial extension (Fig. 1C'',D''). It thus indicated that the Fn accumulation in the interrenal microenvironment displayed a temporal correlation with the interrenal cell migration from 2 to 3 dpf. Therefore, it is of interest to explore whether Fn could modulate morphogenesis of the interrenal tissue during the process of medial extension and functional assembly.

Although the fluorescent intensity of a 24-hpf Tg(ff1bEx2:GFP) embryo in whole-mount could be captured by confocal microscopy (Chou et al., 2010), it was too weak on vibratome sections for locating bilateral interrenal clusters. Nevertheless, indirect evidence has pointed out the possibility of Fn deposition in the microenvironment of bilateral interrenal tissues. At around 24 hpf, bilateral interrenal tissues are associated with those flk1-expressing angioblasts that could be involved in the assembly of axial vasculature (Liu and Guo, 2006). Fibronectin has been found to deposit around the flk1-expressing anigoblasts as early as 16 hpf (Jin et al., 2005), and hence would likely accumulate in the vicinity of the endothelium-associated interrenal clusters at 24 hpf.

Morphogenetic Movements of Steroidogenic Interrenal Cells Are Defective, Despite a Normal Interrenal Specification, in the fn1 Mutant

To examine whether Fn plays a role in the morphogenesis of developing interrenal gland, we checked the morphology of steroidogenic interrenal tissue in the fn1 mutant as well as its wild-type siblings (Fig. 2A, B, C, E). The ventral flat-mount view (Fig. 2A) of deyolked wild-type embryos demonstrated an onset of central migration for steroidogenic interrenal cells by 33 hpf. The migrating steroidogenic cells formed a coherent protruding extension from the aggregated interrenal tissue cluster, which became more evident by 56 hpf (Fig. 2A, white arrows). Along with an expansion of interrenal size, those interrenal cells displaced away from the main cluster appeared loosely-associated at 74 hpf, which was temporally correlated with an intermingling of steroidogenic interrenal and differentiated chromaffin cells at this stage (Chai et al., 2003; To et al., 2007). For all the stages assayed, most of the fn1 mutants displayed an interrenal phenotype where no migrating cells were detected (fn1 class I: 70% at 33 hpf, n = 22; 70% at 56 hpf, n = 53; 71% at 74 hpf, n = 45), whereas the rest showed a more severe morphological defect by the presence of bilateral interrenal tissue clusters (fn1 class II). In contrast, the majority of wild type siblings displayed a migratory interrenal tissue phenotype (78% at 33 hpf, n = 67; 99% at 56 hpf, n = 145; 96% at 74 hpf, n = 125). Since interrenal progenitors appeared as bilateral clusters in either the fn1 mutant (100%, n=25) or its wild-type siblings (100% in 31 randomly-selected samples) at 24 hpf (Fig. 2B), the phenotype of fn1 class II was apparently due to an unsuccessful fusion of bilateral interrenal progenitors after their specification. Cross-sections of the fn1 mutant and its wild type siblings at 3 dpf revealed that the Fn protein expression was severely reduced in both classes of fn1 mutants (Fig. 2C). While interrenal cells migrated along the ventral side of the DA in the wild-type embryo, those in the fn1 mutant did not migrate and remained as single (class I) or bilateral (class II) clusters associated with the lateral side(s) of the DA. Although Integrin-mediated signaling is involved in the interaction between mesoderm and endoderm during gastrula (Nair and Schilling, 2008), the migration phenotypes of the interrenal tissue in the fn1 mutant were not due to an early effect, as the embryos treated with Fn antagonist RGD peptide from 16 hpf (14-somite stage) onwards also led to migration defects of the interrenal tissue, albeit in a slightly more severe manner (see Supp. Fig. S1, which is available online). The migration of interrenal cells in the RGD-treated embryo (n=38) was arrested either at the stage of midline fusion (class II; 29%) or during the subsequent lateral relocalization event (class I; 71%).

Figure 2.

The analysis of interrenal morphogenesis in the fn1 mutant. A: Morphogenetic movements of the steroidogenic interrenal tissue in the fn1 mutant (fn1) and its wild type siblings (WT) at 33, 56, and 74 hpf were detected by whole-mount 3β-Hsd staining. Ventral flat-mount views of deyolked samples showed that the steroidogenic interrenal tissue appeared as a single aggregated cell cluster in the wild-type embryo, with the medially-migrating cells forming a protruding extension (white arrows) evident as early as 33 hpf, at the level between the third (S3) and fourth (S4) somites, which was extended across the midline (yellow dotted lines) by 56 hpf and became more loosely associated at 74 hpf. The medial extension and dispersion of single interrenal cell cluster was defective in the class I fn1 mutant, while the class II mutant displayed a more severe phenotype where bilateral interrenal primordia did not converge. B: Bilateral interrenal clusters expressing ff1b transcripts were detected in both the fn1 mutant and its wild-type sibling at the stage of 24 hpf, as assayed by an ISH analysis. C: The interrenal migration phenotype of the fn1 mutant at 72 hpf, which displayed disrupted Fn expression (red), was further demonstrated by vibratome sections with the dorsal side oriented toward the top. The lumen of the DA is marked by yellow dots. D: Relative density (top chart) and area (bottom chart) of steroidogenic interrenal tissues as assessed by 3βHsd activity staining from the ventral surface, in the fn1 mutant compared with its wild-type sibling control. The number of embryos for each population is indicated in parentheses. Both density and area of 3βHsd activity-positive interrenal cells were mildly yet significantly reduced in the fn1 mutant. *P < 0.05; **P < 0.005; ***P < 10−5. E: Double ISH assays showing the colocalization of ff1b (red) with wt1a and wt1b (dark brown), respectively, at 33 hpf in the fn1 mutant as well as its wild-type sibling. F: The Fast Red fluorescent intensity of ff1b ISH signals, and the number of ff1b-expressing interrenal cells as measured from the ventral surface, are quantified in left and right charts, respectively. Expression levels of ff1b transcripts, as well as the number of ff1b-expressing interrenal cells, are essentially invariable between the fn1 mutant (n = 25) and its wildtype siblings (n = 18) at the early stage of interrenal differentiation. G: The morphology of the developing gut tube in the fn1 mutant. Vibratome sections at the level of anterior gut region in the fn1 mutant (fn1−/−) as well as its wild-type sibling (WT) were stained with the antibody against Acta2 for demarcating smooth muscle cells located at the outer layers of gut tube (Gut) and swim bladder (SB), while the antibody also detected myofibrils at the somites (S). Morphology but not laterality of the gut tube is affected in the absence of Fn. NT, neural tube. NC, notochord. Scale bar = 50 μM.

Figure 3.

The effect of ets1bMO on the Fn distribution in the interrenal region. Top: Transverse sections of Tg(kdrl:GFP)s843 embryos either uninjected (A, A') or injected with ets1bMO (B, B') were harvested at 34 hpf and assayed for 3β-Hsd activity (black), GFP (green), and Fn expression (red in A, B; white in A', B'). The result of ets1b morphant in B, B' is a representative of 9 sectioned samples, which all showed disrupted vasculature at the level of the midtrunk. Bottom: Transverse sections of Tg(ff1bEx2:GFP) embryos either uninjected (C, C') or injected with ets1bMO (D, D') were harvested at 60 hpf and assayed for GFP (green), Fn expression (red), and pFAK (blue). C', D' are magnified views of the outlined areas in C, D. The result of ets1b morphant in D, D' is a representative of 20 sectioned samples. The accumulation of Fn in the interrenal microenvironment is severely reduced in the ets1b morphant. All sections are oriented with the dorsal side toward the top. DA, dorsal aorta; IR, interrenal tissue; NC, notochord; NT, neural tube; S, somite; SB, swim bladder.

Apart from the evident defect of interrenal morphogenetic movements, functional differentiation of the interrenal tissue demonstrated a mild yet significant reduction in the fn1 mutant. Both density and area of 3βHsd activity, as estimated by densitometry analysis from the ventral surface of the stained interrenal tissue, showed a trend of constant increase from 33 to 74 hpf in either fn1 mutant or its wild-type siblings (Fig. 2D). However, for all the stages assayed, the fn1 mutant persistently demonstrated a slightly yet significantly lower level of 3βHsd activity than its wild-type sibling.

Despite a downregulation of interrenal steroidogenic activity in the fn1 mutant as shown in Figure 2D, specification of the interrenal tissue from the pronephric field is neither perturbed nor delayed (Fig. 2E). Quantification of the ff1b-expressing domains in randomly selected fn1 mutants (n = 25; 19 class I and 6 class II) and their wild-type siblings (n = 18) showed that there was no significant difference in the expression levels of ff1b mRNA. Moreover, the number of ff1b-expressing cells showed no apparent difference between the fn1 mutant and its wild-type sibling. Since the medial extension of interrenal cells was initiated by 33 hpf (Fig. 2A), of which in the fn1 mutant could not be entitled to an insufficient cell number at this stage. In the situation where proliferative activity of interrenal cells is reduced due to a repressed hypothalamic-piuitary-interrenal axis (To et al., 2007), medial extension and organ assembly of the interrenal gland remain unperturbed, further supporting that the factors that determine migratory activity of interrenal cells might be decoupled from those promoting interrenal cell proliferation.

On the other hand, the simultaneous detection of either wt1a or wt1b transcripts with those of ff1b revealed that the unperturbed interrenal specification in the fn1 mutant was followed by proper segregation from, and association with, the wt1-expressing pronephric glomerulus (Fig. 2E). It further supports that the function of Fn for the interrenal morphogenesis is mainly on the cell migration and independent of either organ specification or even functional differentiation.

The interrenal phenotype in the fn1 mutant is reminiscent of the previous finding regarding how Fn regulates the cardiomyocyte migration during the process of heart formation (Trinh and Stainier, 2004). It thus raises the question as to whether Fn is commonly involved in migratory events during the process of organ formation. While Fn is also expressed by and deposited at the early endoderm of the zebrafish embryo (Jin et al., 2005), the Fn deficiency did not seem to affect either endodermal migration or gut tube formation, although differentiation and gross morphology of the gut tube were obviously defective (Fig. 2G). On the other hand, the swim bladder as another endoderm-derived organ is surrounded by abundant Fn deposition and appeared severely disrupted in the absence of Fn. We thus speculate that while Fn participates in the microenvironment of many organ or cell types, it might play differential roles for each specific type during development.

The Fn Distribution in the Interrenal Microenvironment Is at Least in Part Contributed by the Neighboring Endothelial Structure

The interrenal tissue is tightly associated with angioblasts and the vascular endothelium since the onset of organ formation and throughout its development (Liu and Guo, 2006). It is thus of interest to explore whether the Fn deposition in the interrenal microenvironment was synthesized by the adjacent vasculature. Consistent with the previous study by Jin et al. (2005), Fn was found to deposit around the DA segment associated with the steroidogenic interrenal tissue, as well as at somites and the swim bladder, at 34 hpf (Fig. 3A, A'). Later at 60 hpf, Fn appeared more highly enriched around the DA and the swim bladder, in the vicinity of the interrenal tissue, which was marked by ff1b promoter-driven fluorescence, and also accumulated at the ventral side of the gut tube (Fig. 3C, C'). We then examined the Fn distribution in the embryo injected with the antisense morpholino nucleotides (MO) against the ets1b gene (Fig. 3B, B'; D, D'). The ets1b gene is required for vasculogenesis in the zebrafish embryo (Sumanas and Lin, 2006), a knockdown of which leads to defects in either midline fusion or lateral relocalization of the developing interrenal tissue (Chou et al., 2010). In this work, all the ets1b morphants injected at a dosage of 1.2 pmol per embryo displayed severely disrupted vasculogenesis at the midtrunk at 34 hpf (n=28). The vibratome sections at the level of the midtrunk showed that Fn was severely reduced in the ets1b morphant, where the peri-interrenal vasculature was largely deficient (Fig. 3 B, B'). In the ets1b morphant, a significant reduction of Fn deposition in the interrenal microenvironment was evident at 60 hpf (Fig. 3D, D'). In contrast, Fn could still be detected around the swim bladder and ventral to the gut tube. Consistently, the phosphorylated Focal Adhesion Kinase (pFAK), an activated downstream component of Integrin signaling (Crawford et al., 2003), also appeared greatly reduced in the interrenal area but not in the swim bladder or the gut tube structures. Nevertheless, it is to be noted that the swim bladder and the gut tube were hypomorphic in the ets1b morphant, implying that the deletion of vasculature from the embryo might affect organogenesis at multiple levels, however not necessarily through the down-regulation of Fn. Our result is in agreement with the previous report that the RNA expression of fn1 is absent at the anterior trunk region of the endothelium-free cloche (clo) mutant (Trinh and Stainier, 2004). It also suggests that Fn might participate in the vessel-derived extracellular matrix microenvironment, which shapes the developing interrenal gland in the fish embryo.

Defective Migration Defects of Both Interrenal and Chromaffin Cells in the fn1 Mutant Lead to Incomplete Interrenal Gland Assembly in Aberrant Locations

The central migration of steroidogenic interrenal cells from 1.5 dpf onwards is temporally correlated with the functional integration between interrenal and chromaffin cells. While the steroidogenic interrenal and chromaffin cells are originated from the intermediate mesoderm and the neural crest, respectively, co-development of these two interrenal organ components starts before 2 dpf, and the resulting interrenal gland assembly is evident at around 3 dpf (Chai et al., 2003; To et al., 2007). To examine whether the interrenal gland assembly was disrupted in the fn1 mutant where the interrenal morphogenesis was defective, the morphology of differentiated chromaffin cells was delineated by the RNA expression of dopamine β-hydroxylase (dbh) gene, which functions to convert dopamine to noradrenaline.

Co-localization of dbh and ff1b transcripts by double in situ hybridization (ISH) at 34 and 54 hpf showed that the initial integration between interrenal and chromaffin cells was not disrupted in the fn1 mutant (Fig. 4). We found that interrenal and chromaffin cells were closely associated as early as 34 hpf (Fig. 4A–A'', B–B''), a stage shortly after the onset of steroidogenesis; and the association was not affected in the fn1 class I (Fig. 4C–C'', D–D'') and II (Fig. 4E–E'', F–F'') mutants where morphogenetic movements of the interrenal tissues was defective. More dβh-expressing chromaffin cells were detected at the dorsal side of the ff1b-expressing interrenal tissue cluster in the wild-type embryo at 54 hpf, apparently representing those chromaffin cells that are migrating to colonize the interrenal organ (Fig. 4H–H''). In contrast, this continuous recruitment of chromaffin cells from the dorsal side into the interrenal area could barely be detected in either class I (Fig. 4J–J'') or class II (Fig. 4L–L'') of the fn1 mutant.

Figure 4.

The interaction between interrenal and chromaffin cells in the fn1 mutant at 34 and 54 hpf. Double ISH assays showing the colocalization of ff1b (red) with dbh (black) in the class I and II fn1 mutants, as well as their wild-type siblings. Ventral flat mount (A–A'', C–C'', E–E'', G–G'', I–I'', K–K'') and lateral (B–B'', D–D'', F–F'', H–H'', J–J'', L–L″) views are shown for the representative embryo of each phenotypic type (n =17, 10, and 24 for the 34-hpf fn1 mutant class I, class II, and their wild-type siblings, respectively; n = 6, 3, and 18 for the 54-hpf fn1 mutant class I, class II, and their wild-type siblings, respectively). The anterior (A) versus posterior (P) orientation of each sample is indicated, and the midline is marked by yellow dotted lines. The embryos in the lateral view panels are oriented with the dorsal side to the top, and migrating chromaffin cell clusters are highlighted by white arrowheads. The initial interaction between interrenal and chromaffin cell populations appears normal; while the continuous recruitment of chromaffin cells to the interrenal region is defective in the fn1 mutant. Scale bar = 50 μM.

Co-localization of dbh and ff1b transcripts at 77 hpf further displayed that the continuous integration between interrenal and chromaffin cells was severely disrupted in the absence of Fn (Fig. 5). At this stage, continuously-migrating chromaffin cells have reached and assembled in the interrenal region, forming a bilobed structure of chromaffin cell cluster immediately dorsal to the ff1b-expressing interrenal tissue (Fig. 5A″, B″). In contrast, the dbh-expressing cells in the fn1 mutant failed to converge at the midline, and distributed as discrete clusters in two rows bilaterally along the midline (Fig. 5C″–G″). It implied that the continuous migration of chromaffin progenitor cells was defective in the absence of Fn. Nevertheless, those dbh-expressing cell clusters situated in close proximity to interrenal tissues in the fn1 mutant demonstrated intimate interactions with the ff1b-expressing cells, implicating that the Fn deficiency did not disrupt the interaction between interrenal cells and those chromaffin cells that have colonized the interrenal area at the early stage. Our results thus suggested that Fn is required for correct positioning of interrenal organ assembly, by ensuring normal morphogenetic movements for both steroidogenic interrenal and chromaffin cells.

Figure 5.

The interaction between interrenal and chromaffin cells in the fn1 mutant at 77 hpf. Double ISH assays showing the colocalization of ff1b (red) with dbh (black) at 77 hpf in the class I and II fn1 mutants, as well as their wild-type sibling. Ventral flat mount (A–A'', C–C'', E–E'') and lateral (B–B'', D–D'', F–F'', G–G″) views are shown for the representative embryo of each phenotypic type (n = 14, 7, and 58 for the fn1 mutant class I, class II, and their wild-type siblings respectively). The anterior (A) versus posterior (P) orientation of each sample is indicated. The embryos in the lateral view panels are oriented with the dorsal side to the top. Defective migration and convergence of chromaffin cells in the fn1 mutant result in incomplete interrenal organ assembly. Scale bar = 50 μM.

It remains to be explored why the convergence of differentiated chromaffin cells was defective in the fn1 mutant. As trunk neural crest cells in the zebrafish migrate along the middle of the medial surface of each somite (Honjo and Eisen, 2005), rather than along the somite boundary where Fn accumulates (Koshida et al., 2005), the migration defect of chromaffin cells in the fn1 mutant might not be due to a loss of Fn along the migration course of trunk neural crest cells. However, the early defect of somite boundary formation in the fn1 mutant leads to a disorganized myofibril pattern, where fast- and slow-twitch muscle lengths are uncoupled (Snow et al., 2008). It is, therefore, possible that the disorganized muscle fiber organization in the fn1 mutant might result in defective migration of trunk neural crest cells, and so their derived chromaffin cells.

Our results support that the Fn-containing microenvironment in the interrenal region contributes to promote morphogenetic movements of both steroidogneic interrenal and chromaffin lineages, yet might not play an essential role for reciprocal interactions between these two cell types. It is interesting to note that the location of interrenal–chromaffin integrations seemed to be defined by the aberrantly positioned interrenal tissues, arguing for a role of interrenal-derived factors for recruiting chromaffin cells.

Do Interrenal-Derived Factors Play a Role in the Interrenal–Chromaffin Interaction?

Previous studies in mice have implicated a role of developing adrenocortical cells for the migration of chromaffin cells. Ectopic adrenocortical cells that resulted from transgenic expression of Ad4BP/SF-1 are capable of recruiting sympathoadrenal progenitor cells, albeit in a relatively inefficient manner (Zubair et al., 2009). In the adult mice, the transgenic Ad4BP/SF-1 expression leads to eccentric chromaffin cell location and thus a disorganized adrenal organ architecture. Similarly, reduced size of the adrenal cortical anlagen in Ad4BP/SF-1 heterozygous mice leads to unsuccessful immigration of chromaffin progenitors into the cortical anlagen (Lohr et al., 2006). On the other hand, differentiation of the chromaffin cell lineage was not affected upon a lack of the adrenal cortex in Ad4BP/SF-1 homozygous mice (Gut et al., 2005). Our results in the zebrafish model are consistent with those in mice in that the presence of steroidogenic cells is essential for recruiting chromaffin cells during the organ assembly, yet not required for the generation and differentiation of chromaffin cells. This implicates that steroidogenic tissue-derived factor(s) might play a role in the guidance of chromaffin cell migration. A disruption of pituitary–interrenal interactions in zebrafish, whether by genetic mutation, dexamethasone treatment, or morpholino knockdown, does not affect the migration of, as well as integration between, interrenal and chromaffin cells (To et al., 2007). It thus ruled out the possibility of interrenal-derived steroids playing any role in regulating the interrenal–chromaffin interaction.

The endothelium-free zebrafish clo mutant has been found defective in both the migration of steroidogenic cells, and the interrenal–chromaffin interactions (Liu and Guo, 2006). While the interrenal specification and differentiation are not perturbed in the clo mutant, bilateral interrenal primordia fail to fuse at the midline in the absence of peri-interrenal vasculature, and the differentiated chromaffin cells converge at an extra-interrenal location, which is well separated from the interrenal region. Although the interrenal tissue phenotype of class II fn1 mutants highly resembles that of the clo mutant, steroidogenic interrenal cells in the class II fn1 mutant, however, are able to recruit chromaffin cells (Figs. 4E″, F″, K″, L″, 5E″–G″; for a comparison between the clo and fn1 mutants see Fig. 6). Since the vessel–interrenal association was not disrupted in the fn1 mutant (Chiu et al., 2012) as well as in embryos treated with the Fn antagonist RGD peptide (Supp. Fig. S1), it indicates that the adrenal/interrenal-derived guidance cue for chemo-attracting chromaffin cells might be generated during the process of interrenal–vessel interaction. Much similar to the adrenal cortex, the adrenal medulla is a highly vascularized endocrine organ that provides a well-developed network of capillaries, venules, and large veins for chromaffin cells to release their product (Kikuta and Murakami, 1984). For the teleostean interrenal organ where steroidogenic and chromaffin cells intermingle instead of forming discrete compartments, both cell types might be exposed to the same set of endothelium-derived signals, and a dissection of such would help us to understand the mechanism that governs assembly of the adrenal/interrenal organ.

Figure 6.

The clo and fn1 mutants display various types of phenotypic defects in the interrenal organ assembly. The panels represent ventral views of wild-type or mutant embryos at 3 dpf, oriented with anterior to the top. The phenotype of the clo mutant is depicted based on the result of Liu and Guo (2006). NC, notochord. 3S and 4S, the third and fourth somite respectively.


Ethics Statement

All of the zebrafish-use protocols in this research were reviewed and approved by the Institutional Animal Care and Use Committee of Tunghai University (IRB Approval NO. 96–05).

Zebrafish Husbandry

Zebrafish (Danio rerio) were raised according to standard protocols (Westerfield, 2000). Embryos were obtained from natural crosses of wild-type, transgenic, or heterozygous mutant fish, and staged as previously described (Kimmel et al., 1995). The following lines were used in this study: fn1kt259 (Koshida et al., 2005) (gift of Prof. Shinji Takada, National Institutes of Natural Sciences, Okazaki, Japan); Tg(ff1bEx2:GFP) (Quek, 2009) (gift of Dr. Woon-Khiong Chan, NUS, Singapore); Tg(kdrl:EGFP)s843 (gift of Didier Stainier, University of California, San Francisco, CA); and Tg(fli1:EGFP)y1 (Zebrafish International Resource Center, Eugene, OR).

3β-Hydroxysteroid Dehydrogenase (3β-Hsd) Staining, ISH, Densitometry and Immunohistochemistry (IHC)

Embryos to be subject to histological analysis were treated with 0.03% phenylthiourea (Sigma, St. Louis, MO) from 12 hpf onwards to inhibit pigment formation. 3β-Hsd activity staining, ISH, and IHC were performed essentially according to Grassi Milano et al. (1997) Liu and Guo (2006), and Trinh and Stainier (2004), respectively, with modifications.

To delineate the morphology of steroidogenic interrenal tissue, histochemical staining for 3β-Hsd enzymatic activity was performed on whole embryos, and Nomaski images were captured using an Olympus BX51 microscope system.

For whole-mount ISH assays, digoxigenin-labeled antisense riboprobes were synthesized from linearized plasmids of wt1a, wt1b, and dβh respectively. Fluorescein-labeled antisense riboprobes were synthesized from linearized ff1b plasmids. DIG-labeled riboprobes were detected with alkaline phosphatase conjugated anti-DIG antibody (Roche, Indianapolis, IN) while fluorescein-labeled probes were detected by alkaline phosphatase conjugated anti-fluorescein antibody (Roche). Visualization was performed either with BCIP/TNBT (Millipore, Billerica, MA), or with Fast Red (Roche). Stained embryos were post-fixed in 4% paraformaldehyde in PBS and washed in PBS supplemented with 0.1% of Tween 20. Stained embryos were subject to yolk-sac removal and flat-mount analysis. The specimens were cleared in 50% glycerol in PBS, mounted on glass slides, and photographed under Nomaski optics on an Olympus (Center Valley, PA) BX51 microscope system or a Zeiss (Thornwood, NY) Axioplan II microscope equipped with LSM510.

For the quantification of 3β-Hsd activity, photos of embryos from each respective group were taken with identical illumination and magnification, by using Axioskop 2 plus microscope equipped with AxioVision 3.0 (Carl Zeiss) software. Areas and density of the respective signal were measured by the Image Gauge Program, Version 4.0 (Fuji Photo Film).

For the quantification of ff1b expression levels, the Fast Red fluorescence signals of the ff1b transcripts were assessed from image stacks acquired at a 8-μM z-steps, by using a Zeiss Axioplan II microscope equipped with LSM510. Intensity values from the consecutive stacks encompassing the interrenal region were summated to obtain the relative ff1b expression level for each sample.

For IHC experiments performed on Tg(ff1bEx2:GFP) and Tg(kdrl:EGFP)s843 embryos, fixed embryos were embedded in 4% NuSieve GTG low-melting agarose, cut into 100-μM sections with a Leica VT1000M vibratome, and permeabilized with PBS containing 1% Triton X-100, before antibody staining in PBDT (1% BSA, 1% DMSO, 0.1% Triton X-100 in PBS). Rabbit polyclonal anti-human Fibronectin (Sigma), mouse monoclonal anti-human FAK [pY397] (BD Transduction Laboratories, San Diego, CA), and rabbit polyclonal anti-zebrafish Acta2 (GeneTex, San Antonio, TX) were used at 1:200, 1:100, and 1:100 dilutions; respectively. Dylight™594-conjugated or Dylight™650-conjugated goat anti-rabbit IgG was used as the secondary antibody at a 1:200 dilution. Images of vibratome-sectioned embryos were captured with the confocal microscopy.

Microinjection of Antisense Morpholino Oligonucleotides (MOs)

MOs targeting the ets1b gene, ets1bMO1, and ets1bMO2, were synthesized at Genetools LLC (Philomath, OR), diluted in 1× Danieau solution, and co-injected in equimolar amounts for a total of 1.2 pmol per embryo, into one- to two-cell stage embryos by using a Nanoject (Drummond, Broomall, PA). The nucleotide sequences of the ets1bMO1 and ets1bMO2 are 5′ – TTG GTA CAT TTC CAT ATC TTA AAG T - 3′ (Sumanas and Lin, 2006) and 5′ – CAC TGA GTC CTT ATT TCA CTA TAT C - 3′ (Sumanas and Lin, 2006), respectively.

Statistical Analysis

All quantitative data are expressed as the mean±SE of the mean (SEM). Statistical analysis of the data was performed using analysis of variance, followed by Student's t-test. A probability of P < 0.05 was considered statistically significant.


We thank Dr. Woon-Khiong Chan, Prof. Shinji Takada, and Prof. Didier Steinier for the kind gifts of Tg(ff1bEx2:GFP), fn1kt259, and Tg(kdrl:EGFP)s843 strains, respectively; You-Lin Zhuo and Zhe-Yu Jiang for excellent technical assistance; and the Taiwan Zebrafish Core Facility (TZCF), the Zebrafish Core in Academia Sinica (TZCAS), and the Zebrafish Core Facility at NTHU and NHRI (TZeTH) for assistances on fish culture.