Vertebrate kidney development is characterized by the progressive formation of several kidney types. For example, mammals form three kidney structures of increasing complexity: the pronephros, the mesonephros, and finally the metanephros, which serves as the adult organ (Dressler,2006). Other vertebrates, like fish and amphibians, initially use a pronephros and later form a mesonephros that functions during larval and adult stages (Dressler,2006). Despite the serial changes in organ complexity, the nephron is the universal structural and functional unit of all vertebrate kidneys. In most cases, nephrons consist of three parts that have discrete activities: (1) a sieve (the glomerulus) composed of an intricate arrangement of fenestrated capillaries surrounded by a capsule of podocytes that filters the blood, (2) an epithelial tubule that receives the filtrate and modifies it by reabsorbing and secreting metabolites, and (3) a duct that performs limited solute exchange while transporting the waste for excretion (Dressler,2006). The tubule epithelium is patterned along its length into functionally specialized segments that express unique gene expression profiles of solute transporters (Kopan et al.,2007; Schedl,2007; El-Dahr et al.,2008). Proximal tubule segments perform the bulk of metabolite recovery for amino acids, salts, and sugars, while distal tubule segments are specialized for fine-tuning salt levels (Hebert et al.,2001). Of interest, vertebrates that reside in diverse habitats ranging from the terrestrial to the aquatic possess a largely similar proximodistal complement of nephron tubule segments. Recent molecular analyses have demonstrated that zebrafish, frog, mouse, and human nephrons contain functionally comparable proximal and distal tubule segments and express many of the same solute transporter and transcription factor genes (Wingert and Davidson,2008). Differences in segment gene expression likely reflect physiological contrasts between species.
Nephron segmentation is the process through which segment identities are produced during nephrogenesis. Several key signals have been identified that are essential for the formation of certain nephron segments, but a comprehensive understanding of segment patterning in any one model organism has not yet been achieved (Schedl,2007; Wingert and Davidson,2008). For example, metanephric nephron segmentation in mammals requires Notch signaling for proximal tubule formation and Pou3f3/Brn1, a POU-domain containing transcription factor, for distal tubule development (Nakai et al.,2003; Cheng et al.,2007). Notch ligands and receptors are regionally expressed in the nephron anlagen (Cheng and Al-Awqati,2005), and conditional inactivation of Notch2 results in a loss of proximal tubule segments (Cheng et al.,2007). Conversely, genetic inactivation of Pou3f3/Brn1 leads to metanephric nephrons that display a dramatic reduction in distal segments (Nakai et al.,2003). Recent studies of pronephric nephron formation in lower vertebrates have identified roles for retinoid signaling and the Iroquois homeobox transcription factor Irx3. During zebrafish pronephros formation, retinoic acid (RA) is necessary for specifying proximal tubule fates, and the abrogation of RA synthesis leads to an expansion of the distal tubule and duct segments (Wingert et al.,2007). In Xenopus, knockdown of Irx3 eliminates the formation of at least one distal segment, and Irx3 has been proposed to act as a master regulator in the specification of this segment (Alarcon et al.,2008; Reggiani et al.,2007). While it remains unclear if the signals that pattern vertebrate nephrons can be integrated into one universal model of nephrogenesis, trends in the regional expression of Notch genes, Pou3f3/Brn1, and Irx3 support the enticing suggestion that nephron patterning is conserved (Wingert and Davidson,2008).
The zebrafish pronephros is an excellent model for nephron segment patterning studies. Zebrafish embryos form an anatomically simple pronephros: it is composed of two nephrons that derive from the intermediate mesoderm that can be visualized throughout their ontogeny (Drummond et al.,1998; Drummond,2003). Nephron segment lengths and boundaries can be quantified at a high resolution relative to the adjacent somites, which provide “landmarks” along the body axis (Wingert et al.,2007). Zebrafish pronephric nephrons possess at least eight discrete cell populations in common with mammals: the podocytes (P) which contribute to the glomerulus, a neck segment (N) that connects the glomerulus and tubule, two proximal segments (the proximal convoluted tubule [PCT] and proximal straight tubule [PST]), two distal segments (the distal early [DE] and distal late [DL]), and a pronephric duct (PD) that empties into the cloaca (C; Wingert et al.,2007). The parallels between fish and mammalian nephron components suggest that the zebrafish studies can generate insights into the conserved mechanisms of nephrogenesis.
In this study, we have examined the domains of gene expression in zebrafish embryo renal progenitors before the appearance of proximal and distal tubule segments. We discovered that an intricate, nested pattern of transcription factor domains precedes the appearance of mature tubule segments in the nephron, and that this pattern undergoes a series of complex spatiotemporal alterations. Using lib, a newly isolated genetic model of RA deficiency, we found that retinoid signals influence the earliest spatial patterns of transcription factor domains. We discovered that a loss of the paraxial mesoderm disrupts the expression of the RA-biosynthesis gene aldh1a2 and induces nephron segment defects similar to lib, thus demonstrating that the paraxial mesoderm is a key source of RA for nephron patterning. In addition, we found that irx3b is required at relatively late stages of nephron patterning for the differentiation of the first distal segment. These findings suggest a stepwise model whereby the sequential actions of RA and irx3b orchestrate segmentation of the zebrafish pronephros.
Zebrafish nephron segments were abbreviated as follows: C cloaca CS corpuscle of Stannius DE distal early DL distal late G glomerulus N neck P podocytes PCT proximal convoluted tubule PST proximal straight tubule
Expression Domains Are Dynamic Within the Nephron Progenitor Territory
To study the origins of nephron segments, we analyzed the expression of transcription factors and signaling molecules between the time when the intermediate mesoderm is first detected around the 3 somite stage to the emergence of mature nephron segments around 24 hours postfertilization (hpf; equivalent to the 28 somite stage; Wingert et al.,2007). For each kidney gene, we determined the precise gene expression domain relative to the somites by performing double whole-mount in situ hybridization with an age-appropriate somite marker (myogenic differentiation 1 [myoD] for embryos <15 somites and myosin heavy chain [mhc] for embryos >15 somites). Consistent with our previously published observations, we found that nephron progenitors displayed uniform expression of several transcription factors including pax2a and pax8 until approximately the 5 somite stage (data not shown; Wingert et al.,2007). Between the 6 and 8 somite stages, the nephron territory was subdivided into two molecularly distinct adjacent regions that showed subtle dynamic alterations. At the 6 somite stage, we observed a rostral domain, located adjacent to somites 2–5 and marked by expression of the Notch ligand genes delta-c (dlc) and jagged-2 (jag2), as well as a caudal domain marked by expression of the zinc finger transcription factor mecom (also known as evi1) that began adjacent to somite 6 (data not shown; Wingert et al.,2007). At the 8 somite stage, the dlc and mecom domains remained mutually exclusive; however, the jag2 domain expanded such that it overlapped with mecom at somites 6–7 (Figs. 1A,C, 3). Thus, a set of overlapping rostral and caudal identities is established among nephron progenitors during early somitogenesis, and likely represents the influences of early proximodistal patterning signals occurring at this time.
By the 15 somite stage, we found a complex, nested arrangement of transcription factor domains within the intermediate mesoderm. Several gene transcripts were expressed by all nephron progenitors, such as hnf1ba, pax2a, and pax8 (Fig. 1B; data not shown). Genes with restricted expression domains broadly marked either the rostral, central, and/or caudal regions of the nephron field, although the precise extent of each gene varied within these partially overlapping areas. Expression of hnf4a was found rostrally, whereas hnf1bb transcripts were found throughout both the rostral and central areas (Fig. 1B,D). We found that irx3b transcripts marked the central area, and that several genes marked portions of the central as well as the entire caudal area, namely pou3f3a, pou3f3b, mecom, emx1, and gata3 (Fig. 1B,D; data not shown). Interestingly, we noted that some genes at this time-point have an expression boundary that is coincident with the mature segment boundaries that appear at the 28 somite stage. For example, the anterior limit of the mecom expression domain (somite 9) and the posterior limit of hnf4a (somite 11) coincide with the eventual PST segment boundaries (somites 9–11; Fig. 1D, 3). In addition, the posterior limits of hnf1bb and irx3b coincide with the posterior boundary of the DE segment (at somite 13; Fig. 3). Other gene expression domains though, like the anterior limits of irx3b, pou3f3a, and pou3f3b, were not coincident with subsequent segment boundaries and instead spanned the eventual boundary of the adjacent PCT and PST segments (Fig. 3). These data suggest that the boundaries of some segments can be predicted as early as the 15 somite stage, while others are defined at later stages.
The nested arrangement of transcription factor domains in nephron progenitors persisted at subsequent stages, and was still evident at the 28 somite stage, the time-point when segment-specific solute transporters mark the differentiated PCT, PST, DE, and DL tubule segments (Fig. 3). However, we found that spatiotemporal changes in gene expression occurred between the 20 and 28 somite stages. Interestingly, there was a common trend whereby genes that mark the rostral region showed no change after 15 somites, while genes that mark the central and caudal regions underwent dynamic alterations at subsequent time-points. For example, the domain of the rostral region gene hnf4a remained constant from the 15 somite stage onward (Figs. 1B, 2, 3). In contrast, the domains of the central and caudal region genes etv5, irx3b, pou3f3a, pou3f3b, mecom, emx1, gata3, and grhl2a showed one or more alterations at different times between 15 and 28 somites (Figs. 1B, 2, 3). In general, these changes can be summarized as posterior shifts in the anterior expression boundary of each gene, although some genes also exhibited contractions at their posterior boundary. For example, the genes irx3b, pou3f3a, and pou3f3b initially shared a common anterior boundary adjacent to somite 7 until the 20 somite stage. This boundary retracted to somite 9 for pou3f3a and pou3f3b at the 22 somite stage, followed by a similar retraction for irx3b slightly later, at the 24 somite stage. The expression domain of pou3f3a/3b underwent further refinement by the 28 somite stage, with a down-regulation of transcripts posteriorly. The net result of these changes is that irx3b, pou3f3a and pou3f3b become restricted to the PST and DE (somites 9–13) at 28 somites (Figs. 2, 3).
We found a striking correlation between the pattern of transcription factor expression domains and the location of nephron segment boundaries at the 28 somite stage (Figs. 2, 3). Surprisingly, we found only one transcription factor, etv5, restricted to a single tubule segment (the PST), while the others marked two or more nephron segments. The transcription factors formed an overlapping, nested pattern such that each tubule segment possessed a unique combination of transcripts. For example, the PST contained hnf1ba, hnf4a, hnf1bb, etv5, irx3b, pou3f3a, and pou3f3b transcripts, while the DE expressed a subset of this medley that excluded hnf4a and etv5 (Figs. 2, 3).
Cell Proliferation Is Not Essential for Nephron Segmentation During Somitogenesis
One possible explanation for the dynamic spatiotemporal alterations in gene expression domains is that regional changes in cell proliferation generate shifts in the physical arrangement of nephron tubule progenitors. To explore this possibility, we treated embryos from the 15, 20, 22, or 24 somites stages with camptothecin or nocodazole to prevent cell division, and then examined the final location of each tubule segment at 24 hpf (Supp. Fig. S1, which is available online). We found no evidence that an inhibition of cell proliferation altered the normal positioning of the proximal and distal segments. Treatment with proliferation-blocking compounds at younger stages was prohibitive to embryo development and precluded the ability for segmentation markers to be assessed (data not shown). Based on these experiments, we conclude that the emergence of nephron tubule segment identities occurs independently of growth after 15 somites.
Mutation of the lightbulb Gene Disrupts Proximodistal Patterning of Nephron Segments
We next sought to gain insight into the genetic pathways that establish nephron segment fates. We isolated the lightbulb (lib) mutant in a forward genetic screen of ENU-mutagenized zebrafish for recessive embryonic kidney defects. lib embryos have a truncated cervical region at 24 hpf based on the abnormally close proximity of the otic vesicle to the first somite compared with wild-types (Fig. 4A). At 48 hpf, lib embryos exhibited a kink in the cervical region, pericardial edema, lacked pectoral fins, and had a pronounced trunk curvature along with a distorted lightbulb-shaped yolk (Fig. 4A). These morphological characteristics of lib are very similar to the hallmarks of RA deficiency in zebrafish embryos, leading us to hypothesize that lib had a deficit in RA bioavailability (Begemann et al.,2001,2004; Grandel et al.,2002; Alexa et al.,2009).
In the developing embryo, RA bioavailability is controlled by enzymatic production and degradation (Duester,2008). RA is produced by the sequential action of retinol dehydrogenases (RDHs) and aldehyde dehydrogenase (Aldh or Raldh) synthesizing enzymes, and degraded by the action of cytochrome P450 enzymes belonging to the Cyp26 family (Duester,2008). We compared the phenotype of lib mutant embryos with neckless (nls), which have a point mutation in the catalytic domain of aldh1a2, and wild-type embryos treated with diaminobenzaldehyde (DEAB), a reversible chemical inhibitor of Aldh enzymes. nls embryos are thought to be Aldh1a2-deficient but have less severe tissue defects compared with DEAB treatment, perhaps due to maternal Aldh1a2 activity present in the yolk (Begemann et al.,2004; Alexa et al.,2009). At 48 hpf, nls embryos had a less severe cervical kink, pericardial edema, and body curvature compared with lib, while DEAB-treated embryos and lib were more similar (Fig. 4A). Next, nephron segmentation in lib, nls, and DEAB-treated embryos was evaluated using double whole-mount in situ hybridization with segment-specific solute transporters and mhc to mark the somites. Compared with wild-types, 24 hpf lib embryos had reduced expression of the podocyte markers, wt1b, wt1a, and mafb, and the podocytes were located more rostrally, suggesting that podocyte development was partially affected (Fig. 5A; data not shown). At 48 hpf, lib embryos had reduced expression of both wt1b and nephrin, suggesting that podocyte numbers were not restored as development progressed (data not shown). lib embryos also formed shorter proximal segments, as shown by the PCT and PST markers slc20a1a and trpm7 (Fig. 5A). In addition, lib formed expanded distal segments, as shown by the DE, DL, and PD markers slc12a1, slc12a3, and gata3, respectively (Fig. 5B). nls embryos had a milder nephron segment phenotype than lib mutants, with reduced podocytes located at the correct location along the trunk, a slightly reduced PCT segment, and a slightly expanded DE segment (Fig. 5). DEAB-treated embryos had nephrons composed entirely of the distal segments (DE, DL, and PD) that were each expanded in length (Fig. 5). These data indicate that lib mutants have a nephron patterning defect more severe than nls, but less severe than DEAB treatment.
lib Is a New Model of Aldh1a2 Deficiency
We next undertook a positional cloning strategy to determine the genetic defect in lib. Using bulk segregant analysis, the lib gene was mapped to zebrafish chromosome 7 (Fig. 4B). Linkage analysis of 2698 meioses placed the lib locus between the SSLP markers z8693 and z11894, a region that includes the aldh1a2 gene. This was surprising as the kidney phenotype of lib mutants is more severe than nls mutants, and because nls and its phenotypically similar alleles no-fin and aldh1a2um22 are all considered to represent null alleles of aldh1a2 in the zebrafish (Begemann et al.,2001,2004; Grandel et al.,2002; Alexa et al.,2009). Nevertheless, based on the mapping and phenotype, we hypothesized that lib was a novel, more severe mutant allele of aldh1a2.
To investigate if lib was an aldh1a2 mutation, we mated lib and nls adult heterozygotes and found that the mutations failed to complement one another (Fig. 4A). Compound nls/lib heterozygote embryos had morphological characteristics most similar to nls mutant embryos, with a kink in the cervical region, pericardial edema, and absent pectoral fins (Fig. 4A). In addition, compound nls/lib heterozygote embryos had a nephron segment pattern of intermediate severity between the nls and lib phenotypes. Compared with nls, compound nls/lib heterozygotes had a slightly shorter PST and slightly larger DE, but had no change in DL size or podocyte location as seen in lib (Fig. 5). Next, reverse transcriptase-polymerase chain reaction (RT-PCR) was used to isolate aldh1a2 cDNA transcripts from lib mutants. Sequence analysis identified a C to A transversion at nucleotide 174, predicting a Tyr → Stop nonsense mutation at amino acid 58 (Fig. 4C). The wild-type zebrafish aldh1a2 peptide is 518 amino acids in length, and includes a series of nucleotide binding domains, tetramerization domains, and catalytic domain (Alexa et al.,2009). The lib mutation is predicted to truncate the first nucleotide binding domain and eliminate all subsequent domains. Of interest, examination of aldh1a2 expression in lib revealed that transcripts were markedly reduced both in lib homozygous and heterozygote embryos, consistent with nonsense-mediated decay (Fig. 4D).
To provide further evidence that the lib phenotype is caused by a RA synthesis defect, we investigated whether elevating RA levels could rescue lib. Embryos from lib heterozygous incrosses were injected at the 1-cell stage with wild-type aldh1a2 cRNA or treated with exogenous RA from 60% epiboly to the 28 somite stage, then scored for morphology and pectoral fin development at 48 hpf (Supp. Fig. S2A). Overexpression of aldh1a2 was sufficient to rescue lib morphology (Supp. Fig. S2A,B). Of interest, exposure to a high dose of RA (10−8M) completely rescued lib mutant embryos, but a lower RA dose (10−9M) was insufficient to rescue lib (Supp. Fig. S2B). In contrast, the lower RA dose was able to fully rescue nls and nls/lib compound heterozygote embryos (Supp. Fig. S2B). These results support the notion that lib quantitatively possesses the most severe RA deficiency among these genotypes and that RA dosage is responsible for the graded phenotypes in lib, nls/lib, and nls mutants.
Next, we determined if exogenous RA application from 60% epiboly to the 28 somite stage was sufficient to rescue proximal nephron segments in lib. Nephron segmentation was evaluated at 24 hpf using double whole-mount in situ hybridization with segment-specific solute transporters and the somite marker mhc. RA-treated lib mutants developed expanded proximal segments (PCT, PST) and reduced distal segments (DE, DL, PD; Supp. Fig. S2C,D). In addition, RA treatment rescued podocyte number and positioning in lib embryos (Supp. Fig. S2C,D). These results demonstrate that nephron progenitors in lib mutants are competent to respond to an RA signal(s), and suggest that a paucity of retinoid ligands underlies the lib phenotype. Taken together, our analysis of the lib and nls mutants highlights that RA production by Aldh1a2 is essential for proximodistal patterning of the pronephros.
lib Mutants Have Altered Renal Progenitor Domains During Early Somitogenesis
To investigate how RA deficiency in lib mutants influenced nephron segment patterning, we examined the spatiotemporal expression of renal genes between the 8–28 somite changes. At the 8 somite stage, lib mutant embryos had slightly shortened domains of dlc and jag2 expression that were shifted anteriorly compared with wild-types (Fig. 1A,C). In addition, lib embryos had a mecom expression domain that was slightly expanded anteriorly compared with wild-types (Fig. 1A). Overall, the combination of the dlc, jag2, and mecom domains in lib mutants still formed an overlapping rostral-caudal pattern similar to wild-types, but shifted anteriorly (Fig. 1A,C). Because nephron progenitor patterning is altered in lib mutants at the earliest stages that regional differences can be detected, it suggests that RA regulates the initial proximodistal patterning of nephron progenitors, consistent with our previous observations (Wingert et al.,2007).
More significant alterations in nephron progenitor patterning were detected in lib mutants at the 15 somite stage. Notable changes included anterior shifts in the rostral (hnf4a), central (irx3b), and caudal (mecom) expression domains (Fig. 1B,D). These shifts were not equivalent for each domain, with hnf4a shifting anteriorly by 1 somite, irx3b by 3 somites, and mecom by 2 somites. The posterior boundaries of the hnf4a and irx3b domains were similarly shifted anteriorly, and also by unequal extents (3 somites for hnf4a and 1 somite for irx3b; Fig. 1B, D). The net effect of these changes was a reduction in the rostral domain and expansion in both the central and caudal domains (Fig. 1D). Despite these patterning changes, the correlations between the rostral-central-caudal domains at 15 somites and the positions of the final nephron segments that we noted in wild-type embryos were likewise the case in lib mutants. For instance, in both wild-types and lib mutants, the boundaries of the PST segment correlate to the region of overlap between the rostral (hnf4a) and caudal (mecom) domains, despite this region being smaller and anteriorly shifted in lib mutants. These results provide further support for the notion that renal transcription factor domains during early pronephros development are predictive of nephron segment fates at later stages.
Between the 20 and 28 somite stages, lib embryos exhibited similar dynamic spatiotemporal shifts in central and caudal domain markers that we observed in wild-types (Supp. Figs. S4, S5). For example, the anterior expression boundary of the central domain gene irx3b and caudal domain marker mecom retracted in lib and wild-types (Supp. Figs. S4, S5). However, there was one exception: the caudal markers pou3f3a and pou3f3b failed to retract their anterior expression boundaries and remained ectopically expressed in the PCT and PST segments. Taken together, these results provide genetic evidence that RA has an early role in establishing the rostral and caudal domains within the intermediate mesoderm, and that RA is also required at later time-points to refine the expression patterns of genes such as pou3f3a and pou3f3b.
Paraxial Mesoderm Is Necessary for Proximodistal Nephron Patterning
Based on our previous work showing that aldh1a2 was highly expressed in developing somites, we hypothesized that the paraxial mesoderm was a major source of RA responsible for establishing the proximodistal pattern of the pronephros (Wingert et al.,2007). To test this, we knocked down tbx16/spadetail, which encodes a T-box transcription factor required for paraxial mesoderm formation (Amacher et al.,2002; Morley et al.,2009), and then examined the segmentation pattern of the pronephros at 24 hpf. A deficiency in tbx16 was associated with reduced podocytes and proximal tubule segments (Fig. 6A). Conversely, tbx16 morphants formed an expanded DE segment, as shown by the marker slc12a1, while the DL (slc12a3+) and PD (gata3+) segments were not significantly altered (Fig. 6A). These patterning changes were associated with corresponding alterations in the sizes and/or positioning of the rostral, central, and caudal domains. For instance, the rostral domain marker hnf4a was reduced, while the central domain marker irx3b and the caudal domain marker mecom were expanded anteriorly (Fig. 6B). Precise mapping of the segment size differences relative to somite number was not possible in tbx16 morphants due to defective somitogenesis (Fig. 6C). Analysis of aldh1a2 expression in tbx16 morphants at the 5 and 28 somite stages showed a significant reduction in transcript levels, consistent with the loss of paraxial mesoderm that characterizes the tbx16 mutant phenotype (Amacher et al.,2002). The persistent presence of aldh1a2 transcripts in tbx16 morphants likely represents expression in the lateral plate mesoderm along with a small number of muscle precursors that form in tbx16 mutants (Amacher et al.,2002). These data reveal that tbx16 deficiency induces similar proximodistal nephron patterning defects to lib/aldh1a2 mutants, thus supporting our hypothesis that the paraxial mesoderm provides a major source of the RA that patterns the pronephros.
RA Receptor Activity Is Also Essential for Nephron Segment Patterning
Retinoid ligands transduce their effects by means of heterodimeric complexes of nuclear receptors that consist of a retinoic acid receptor (RAR) and retinoid X receptor (RXR; Duester,2008). To explore whether modulations in retinoid receptor activity can alter nephron segmentation, we tested the effects of retinoid receptor antagonists and agonists in wild-type zebrafish embryos. Wilt-type embryos treated with an RAR-α antagonist from 60% epiboly to the 28 somite stage formed shortened proximal segments (PCT, PST) and expanded distal segments (DE, DL, PD; Supp. Fig. S3). Conversely, treatment with an RAR-α agonist from 60% epiboly to the 28 somite stage had the opposite effect, and led to expanded proximal segments along with absent or dramatically reduced distal segments (Supp. Fig. S3). These findings provide independent evidence that RA signaling is essential for nephron segmentation, and suggest that RAR-α may be a key functional component in nephron patterning.
irx3b Acts Downstream of RA to Specify the DE Segment in the Zebrafish Pronephros
A striking feature of lib mutants is the significantly enlarged DE segment. Previous studies in Xenopus have demonstrated that Irx3 is needed to specify the distal segments during frog pronephros nephrogenesis (Alarcon et al.,2008; Reggiani et al.,2007). While the function of the zebrafish orthologue irx3b has not been assessed, our analysis of early transcription factor expression domains traced irx3b expression in medial nephron progenitors and showed a shift of this expression in lib mutants. To explore the functional importance of irx3b for specifying DE segment fate downstream of RA, we used morpholinos to knockdown this factor and examined nephron segmentation compared with mismatch control-injected animals. We observed pericardial edema in irx3b morphants at 72 hpf, suggestive of renal failure or salt imbalance (data not shown), and analyzed nephron segmentation in irx3b morphants by whole-mount in situ hybridization. irx3b-deficient embryos lacked a mature DE segment, as revealed by the absence of slc12a1 and romk expression at 24 hpf (Fig. 7; data not shown). In place of the DE segment, the proximal segments were slightly expanded, shown by lengthened domains of the PCT marker slc20a1a and PST marker trpm7, such that the boundary between these segments was shifted caudally along the trunk (Fig. 7A,C). In addition, the corpuscle of Stannius (CS), a gland derived from the tubular epithelium between the DE and DL segments and marked by gata3, was dramatically expanded (Fig. 7A,C). These results indicate that irx3b is essential for the expression of DE-specific markers and suggest that irx3b regulates the PCT/PST boundary, as well as restrict CS fate.
Consistent with the onset of irx3b expression at 15 somites, irx3b-deficient embryos showed no changes in renal gene expression until late somitogenesis. At 28 somites, we found changes in transcription factor domains that correlated with the alterations in the mature nephron segmentation pattern at 28 somites (Supp. Fig. S6). For example, irx3b morphants possessed an expanded domain of hnf4a that coincided with the expansion of PCT and PST segments (Supp. Fig. S6B). In addition, the domains of pou3f3a and pou3f3b were shifted posteriorly, such that they were located next to somite 10 compared with somite 9 in wild-type embryos, also corresponding with the shifted position of the PST (Supp. Fig. S6). Of interest, the posterior boundary of pou3f3a was located at somite 14, coincident with the location of the expanded CS population observed in irx3b morphants, and possibly representing ectopic expression in these cells. Lastly, the domains of mecom, emx1, and gata3 were unchanged in irx3b morphants, consistent with the normal location and size of the DL and PD segments (Supp. Fig. S6). The finding that transcription factor domains were not altered until approximately the 28 somite stage suggests that irx3b function is required late, likely between the 24 and 28 somite stages, to establish a DE-specific gene program and restrict the expression of proximal tubule markers.
The observation that the PCT/PST boundary was posteriorly shifted in irx3b morphants led us to test whether irx3b knockdown could rescue the shortened shortened proximal segment(s) at the expense of the expanded DE in lib mutants. lib heterozygous incrosses were injected with either an irx3b morpholino (MO) or mismatch control, and then nephron segmentation was examined at the 28 somite stage. Knockdown of irx3b in lib abrogated the DE segment, and significantly expanded the length of both the PCT and PST segments (Fig. 7A,C). In particular, the PST segment was doubled in length in lib/irx3b doubly deficient animals, occupying the portion of the nephron that would otherwise have formed the DE in lib mutants. These findings indicated that, while RA signaling is needed to establish the initial PCT and PST progenitor domains, the ultimate size of these segments is not irrevocably set during early nephron patterning and can be altered by modulating the activity of the later-acting gene irx3b.
Recent work in Xenopus has shown that overexpression of Irx3 induces ectopic expression of mature distal tubule markers (Alarcon et al.,2008; Reggiani et al.,2007). However, it is not clear whether Irx3 plays an early patterning role during nephron development to specify the DE domain or whether it acts later within this segment to up-regulate or restrict gene expression. This issue is challenging to address in the zebrafish as early markers specific to DE progenitors have not been identified and irx3b transcripts are found in both PST and DE segments. To resolve this, we made use of DEAB to create embryos with nephrons composed of DE and DL tubule segments only (Wingert et al.,2007; Fig. 7A). Because the PST segment is absent in DEAB-treated embryos (Wingert et al.,2007), expression of irx3b in these animals will specifically mark the DE progenitors. Consistent with this, we found irx3b transcripts in the same region of the nephron as the DE marker slc12a1 at the 28 somite stage (Fig. 7B). Next, we investigated whether irx3b+ DE progenitors were still present in DEAB-treated embryos that were also injected with the irx3b morpholino. Expression of irx3b was found in DEAB-treated/irx3b-deficient embryos in a region equivalent to where the DE segment would normally form, despite this region now expressing the DL marker slc12a3 (Fig. 7A,B). These results suggest that two distal domains are patterned in DEAB-treated/irx3b-deficient embryos, but that in the absence of irx3b activity the DE region fails to express DE-specific solute transporters and instead expresses DL markers. Thus, we conclude that irx3b in zebrafish is unlikely to be involved in the early patterning of the DE segment but rather plays a role in its terminal differentiation by regulating the expression DE-specific genes and/or repressing the gene programs of neighboring segments.
There are many gaps in our knowledge about the mechanisms that accomplish nephron patterning during kidney development. Here, we show that the nephron segments in the zebrafish pronephros arise following an elaborate series of spatiotemporal gene expression changes in the intermediate mesoderm. We have constructed a detailed transcription factor map that describes this process and is predictive of the mature segmentation pattern of the pronephros. Spatiotemporal expression domains in nephron progenitors during mid- to late-somitogenesis were unaffected by blocking cell proliferation, suggesting that cellular turnover does not significantly account for the dynamic alterations that we detected. Furthermore, our genetic evidence demonstrates that RA is essential for delineating the earliest rostrocaudal patterning events in the intermediate mesoderm, and suggest that RA dosage influences the size and positioning of subsequent transcription factor domains and thereby the mature nephron segmentation pattern. We also show that irx3b, acting at relatively late stages of nephron development, is involved in determining the position of the PST/DE boundary and modulating gene expression in the DE segment (Fig. 8). Together, our study provides a new molecular map that can be used to guide future interrogations into the workings of nephrogenesis, and adds additional insights to previous reports about the ways that RA and irx3b influences nephron formation.
RA Signals Demarcate an Early Subdivision of Pronephros Progenitors
Several lines of evidence now indicate that retinoid signaling is crucial for pronephros patterning, and that the location of RA, time of production, and dosage are all crucial. We recently demonstrated that the caudal homeobox genes regulate the locale of RA by establishing the expression domains of the aldh1a2 and cyp26a1 genes (Wingert et al.,2007). Axial shifts in the expression of these enzymes were associated with shifts in pronephros position along the embryo trunk, suggesting that the position of RA synthesis had important inductive consequences for nephron patterning (Wingert et al.,2007). aldh1a2 expression in the paraxial mesoderm has been suggested to serve as the source of retinoids that pattern the pronephros along with other nearby tissues, like the hindbrain, and our analysis of tbx16 morphants provides further support for this hypothesis. It should be noted, however, that the disruption of the formation of the paraxial mesoderm by means of tbx16 knockdown alters both aldh1a2 expression and leads to other broad defects, thus we cannot eliminate the possibility that the change in nephron patterning in tbx16 morphants is affected by disruptions in global patterning. Further studies are needed to delineate between RA-specific and other changes. Given the outcomes of genetic or chemical alterations in RA signaling, we favor the hypothesis that RA produced by paraxial mesoderm acts to pattern nephron progenitors (Fig. 8). Proximal tubule segment identities rely on RA for their specification, and dramatic alterations in segment fates ensue when RA levels are greatly elevated or repressed: zebrafish embryos treated with exogenous RA develop nephrons made entirely of proximal segments, while the abrogation of RA production with DEAB leads to nephrons composed entirely of distal segments (Wingert et al.,2007). Interestingly, segmentation changes only occur if RA bioavailability was altered from the onset of gastrulation through to early somitogenesis stages, establishing an early role for retinoid signaling in pronephros development.
In this study, our analysis of the lib (presumptive null allele) and nls (hypomorph allele) mutants provides genetic evidence that modulating RA levels produces graded effects on the nephron segmentation pattern. By morphological criteria, lib mutants exhibit a more severe RA-deficiency phenotype than nls, and form a slightly shorter PST segment concomitant with an expanded DE segment. The rescue of the lib mutant phenotype accordingly requires treatment with higher levels of exogenous RA than the nls mutant. Taken together, these lines of evidence suggest that RA dosage determines the size of the proximal tubule domain. Intriguingly, partial effects of RA were also observed when embryos were treated with an inhibitor that targets RAR-α activity. Interference with RAR-α led to reduced proximal segments and expanded distal tubule segments, but these changes were not nearly as dramatic as DEAB exposure. This finding could reflect the involvement of other RA receptors during nephron progenitor patterning, although such an interpretation may be complicated by partial drug efficacy in zebrafish cells as compared to other organisms for this particular compound. Future studies, such as RA receptor knockdown experiments, are needed to tease out the specific roles of RA receptors during nephrogenesis, and to address fully whether RA receptor activity is necessary in nephron progenitors themselves.
Our studies raise several intriguing questions about how RA actually affects the nephron progenitors. Cell-autonomy has yet to be assessed regarding RA activity in the nephron progenitors. Another pressing question is whether RA and its receptors cooperate to directly or indirectly modulate the expression of renal genes. Interestingly, a recent study provided evidence that RA participates actively to trigger gene expression of the kidney podocyte lineage. Podocyte precursors express the Wilm's tumor 1a (wt1a) gene, and wt1a promoter analysis discovered the presence of functional RARE elements that were shown to mediate responsiveness of the promoter to RA in vivo (Bollig et al.,2009). Abrogation of RA synthesis in the lib mutant (this study) and with chemical inhibitors to aldh1 enzymes (Wingert et al.,2007) blocks or reduces podocyte formation, respectively. These findings illuminate the possibility that RA may act directly, by activating genes required for proximal tubule fate and also by repressing distal (caudal domain) genes. However, these data do not rule out the notion that indirect mechanisms are also operative.
Of interest, changes in RA synthesis are associated with alterations in the earliest regionalization of the intermediate mesoderm. Nephron progenitors display at least two regional identities between the 6 and 8 somite stages, which we have termed rostral and caudal. We found a fascinating, partial overlap of these domains at the 8 somite stage, perhaps suggestive of a more complex arrangement of molecular identities that will be better understood as more early renal markers are discovered. Nevertheless, in the lib setting of decreased RA production, the rostral nephron progenitor population is reduced while the caudal population is expanded, and we noted previously that DEAB treatment abrogates the rostral domain entirely (Wingert et al.,2007). These data support the notion that RA ultimately works to induce a rostral progenitor identity among nephron progenitors. While the genes involved in nephron patterning downstream of RA are still unclear, the notch ligand genes dlc and jag2 are excellent candidates, given the role of the Notch pathway in promoting proximal tubule fates during mammalian kidney development (Cheng and Al-Awqati,2005; Cheng et al.,2007).
Conserved and Unique Elements of Irx3 Function During Zebrafish and Frog Nephrogenesis
Irx genes encode homeodomain transcription factors, and members of this gene family have functions in the development of many tissues. In diverse settings ranging from imaginal discs in flies to the vertebrate neural plate, Irx genes act to pattern regional differences amongst cell territories and can both activate and repress target gene expression (Cavodeassi et al.,2001). Irx3 has recently been identified as a requisite player in two aspects of Xenopus pronephros development: first, it maintains the pronephric territory at early stages, and second, it is required to form slc12a1-expressing distal tubule cells (Reggiani et al.,2007; Alarcon et al.,2008).
In this work, we show a conserved requirement for irx3b in distal tubule differentiation in the zebrafish pronephros. Knockdown of irx3b results in an absence of DE marker expression and concomitant expansions in the lengths of the PCT and PST segments. This phenotype does not appear to be caused by a failure to correctly pattern the nephron into a discrete DE domain as irx3b transcripts, which mark presumptive DE progenitors, are still localized to a DE-sized subdomain in DEAB-treated/irx3b-deficient embryos. Instead, we propose that in the absence of Irx3b, the DE segment adopts either a PST expression profile, in the case of irx3b knockdown in wild-type embryos, or a DL expression profile, in the case of DEAB-treated/irx3b-deficient embryos. Consistent with a late role for irx3b during nephron development, alterations in gene expression domains were only detectable in irx3b morphants at time-points that immediately preceded the onset of the segment-specific solute transporter genes. Thus, the transcriptional targets of Irx3b may include these transporters (acting as an activator of DE-specific transporters and/or repressor of PST and DL transporters) as well as equivalent transcription factors acting in neighboring segments. In support of this, there is precedence in other tissues for such diverse targets of Irx factors; regional patterning in the developing vertebrate brain uses a system of transcription factor antagonism between Irx3 and Six3, while Irx4 and Irx5 in myocardial cells alters potassium ion channel gene expression (Kobayashi et al.,2002; Constantini et al., 2005; He et al.,2009). We have identified putative Iroquois binding sites upstream of zebrafish slc12a1, consistent with the possibility that irx3b regulates the expression of this DE-specific transporter (Wingert and Davidson, unpublished observations).
In contrast, the role of Irx3 in the maintenance of the pronephros territory does not appear to be conserved in zebrafish, highlighting one functional difference between the frog and zebrafish Irx3 genes. In zebrafish, at least four irx orthologues are expressed in the zebrafish pronephros, irx2a, irx3b, irx4a, and irx5a, which may explain the difference between these species (Lecaudey et al.,2005). Furthermore, Irx3 expression in Xenopus embryos appears to be positively regulated by retinoids, as fewer transcripts are detected in the absence of retinoids while exogenous RA is associated with elevated transcript numbers (Alarcon et al.,2008). These alterations appear linked to early changes in the mesoderm destined to form the pronephros rather than later proximodistal patterning events (Alarcon et al.,2008). While RA similarly appears to function upstream of irx3b in the zebrafish pronephros, changes in the irx3b domain correlate with altered nephron patterning.
Insights Into Conserved Nephron Segmentation Mechanisms Among Vertebrates
This study has identified critical roles for RA and irx3b during zebrafish pronephros segmentation, but other vital signals have yet to be determined. Our characterization of transcription factor gene expression in the intermediate mesoderm has revealed a complex molecular profile that is spatially and temporally dynamic, and future functional analysis of these genes will reveal key determinants of segment identity and cross-regulatory interactions. A similar level of transcriptional complexity is beginning to be realized during mammalian nephrogenesis, largely in part due to the efforts of the GenitoUrinary Development Molecular Anatomy Project (GUDMAP), which has mapped the expression of mouse genes to different compartments of the developing metanephric kidney (Brunskill, et al,2008; McMahon et al.,2008; Georgas et al.,2009). Findings from the GUDMAP, as well as previous studies, have documented a set of genes with regionalized domains within murine nephrons. For example, Pou3f3 is restricted to prospective distal regions of the nephron, similar to our observations of the zebrafish orthologues pou3f3a and pou3f3b (Nakai et al.,2003). It is likely that many of the same transcriptional pathways underlie proximodistal nephron patterning in all vertebrates. Therefore our work, establishing a detailed molecular description of gene expression during nephrogenesis, should provide a rich source of data for generating hypothesis-driven experiments in other model organisms such as the mouse. A greater knowledge of the molecular basis of vertebrate nephron development will help elucidate the causes of congenital kidney defects, which are often poorly understood, as well as further our understanding of the regulatory networks that maintain nephron function in the adult.
Zebrafish Husbandry, Genetic Strains, and lib Mutant Mapping
Zebrafish were maintained and staged as described (Kimmel et al.,1995). Tübingen strain embryos were used for all experiments, and collected from natural matings. lib was recovered in a diploid ENU (N-ethyl-N-nitrosurea) screen for mutations affecting kidney organogenesis, performed as described (Solnica-Krezel et al.,1994). Families were screened to isolate mutants with kidney development defects, as manifested by edema between 24 and 72 hpf. lib was cloned using meiotic mapping. lib mapping strains were generated by mating lib mutant carriers, generated and maintained on the Tübingen strain, were to the polymorphic WIK strain. Embryos were collected from pairwise matings of lib Tü/WIK heterozygotes, and scored at 72 hpf for edema. A panel of 239 CA simple sequence length polymorphisms (SSLP) markers were used to scan chromosomes and establish linkage as described (Wingert et al.,2005). aldh1a2 transcripts were isolated from wild-types and lib by reverse transcription PCR, using the following primers: forward ATGACCTCC AGTG-AAGTTGAACTG and reverse, TTAAGACGTCTTGCCTGACATCTT, and subcloned into pGEMT-easy (Promega) for sequence analysis. For lib rescues (and aldh1a2 expression studies, see below), full-length wild-type aldh1a2 was subcloned into the pCS2 expression vector using the EcoRI and XhoI sites. Whole-mount in situ hybridization aldh1a2 probe was generated by linearizing the plasmid with EcoRI and performing antisense RNA synthesis with T7 RNA polymerase (Roche).
Embryo Chemical Treatments, Morpholinos, and In Situ Hybridization Analysis
Reported expression patterns show representative results as gathered from analysis of between 10 and 15 animals per each time point and genotype (mutant studies and knockdown). Gene expression domains as reported by somite boundaries were based on counts of at least 5 separate samples to ensure accurate documentation.
For chemical treatments, wild-type embryos were incubated with camptothecin and nocodazole, as described (Murphey et al.,2006). Briefly, cell cycle inhibitors (Tocris) were dissolved to make stock concentrations of 10 mg/ml dissolved in 100% dimethyl sulfoxide (DMSO), and embryos treated at the established effective dose (Murphey et al.,2006). For lib rescue studies, embryos from lib or nls heterozygous incrosses were incubated in RA treatment, as described (Wingert et al.,2007). Briefly, all-trans retinoic acid (Sigma) was dissolved in 100% DMSO to make a 1 M stock, aliquots stored at −80°C, and embryos treated with 1 × 10−8M or 1 × 10−9M RA/DMSO, which were made by performing serial dilution of the 1M stock in E3 embryo water. For RA agonist and antagonist studies, RO 41-5232 and AM-580 (Tocris) were similarly dissolved in 100% DMSO to make a 1 M stock, aliquots stored at −80°C, and wild-type embryos were incubated in 1 × 10−8M RO 41-5232/DMSO or 1 × 10−7M AM-580/DMSO diluted in E3 embryo water. All chemical treatments were fully penetrant and produced consistent results at the doses and treatment windows that were examined.
For Aldh1a2 rescue studies in lib, the full-length open reading frame of wild-type and lib aldh1a2 alleles were subcloned into the EcoRI/XhoI sites of the pCS2 expression vector. Plasmids were linearized with Not1, and synthetic capped RNA was synthesized using SP6 mMessage machine (Ambion). Embryos were injected with 100 pg of cRNA at the 1-cell stage.
For all morpholino experiments, morpholinos were solubulized as recommended by Gene-Tools at and stored at a 4 mM concentration. Dilution of morpholino stocks were injected into 1-cell stage wild-type embryos. The tbx16 morpholino (AAGACAAGTACTCACCTCTGATAGC) targets the exon 1 donor splicing site and produces consistent results at the injected dose (Burns et al.,2009). irx3b morpholinos were designed to target the 5′ untranslated region and start site morpholinos, and gave similar results upon injection of ∼500 pl of a 0.15 mM working dose and 0.2 mM dose, respectively: (ACCGGGAGGAC TGCGGGGAAACTCG) and (ATAGCC TAGCTGCGGGAGAGACATG), respectively. A 5 base-pair mismatch start site morpholino (ATTGCCTACCTGG GGGACAGAGATG) was injected for a control comparison. Whole-mount in situ hybridization for cdh17, clck, dlc, gata3, jag2, mecom, mhc, myoD, nbc1, pax2a, pax8, raldh2, wt1a, and wt1b were performed as described (Wingert et al.,2007). Expression of emx1, etv5, grhl2a, hnf1ba, hnf1bb, hnf4a, irx3b, pou3f3a, pou3f3b were previously reported (Hauptmann and Gerster,2000; Thisse et al.,2001; Kawahara and Dawid,2002; Lecaudey et al.,2005).
R.A.W. thanks the staff of the Center for Zebrafish Research at the University of Notre Dame for providing ongoing husbandry of the lib zebrafish strain. R.A.W. was supported by the Polycystic Kidney Foundation, a Harvard Stem Cell Institute Seed Grant, and an NIH-NIDDK award. A.J.D. was supported by a NIH-NIDDK award and grants from the Harvard Stem Cell Institute.