Pre-pattern in the pronephric kidney field of zebrafish

The first kidney to form in embryos of all vertebrate species is the pronephros. It has three components: glomerulus, tubules and nephric duct. This embryonic kidney is replaced later in development by the mesonephros, and in birds and mammals, by a metanephros (Saxén, 1987). The mesonephric and metanephric kidneys arise along the course of the nephric duct and, in fact, the metanephros is generated by mutual inductions between duct and mesenchyme (Saxén, 1987). The pronephric kidney, however, appears to arise de novo from cells in the intermediate mesoderm (IM) (Goodrich, 1930). The origin of the elements of the pronephric kidney have not been defined by lineage analysis. Fashioning of this organ requires the generation and apportionment of cells with at least three distinctive cell fates that represent the three components of the kidney, and their appropriate migration and positioning with respect to each other during morphogenesis.

Organ fields in vertebrate embryos are regions defined classically by explantation and transplantation as regions that contain organ precursors (Huxley and de Beer, 1934). Based on the assumption that tubule formation is indicative of kidney formation, explantation and transplantation in Ambystoma points to the intermediate mesoderm as a site of renal origin, at least for the tubule progenitors (Fales, 1935). The positions of molecular markers characteristic of particular cell fates do not necessarily correspond to precursor location defined by lineage analysis (e.g. Serbedzija et al., 1998). Hence, lineage analysis, by marking of cells with dye or virus (Price et al., 1987) is the only definitive way to determine precursor location, migration and cell fate assignment.

There are three modular components to the pronephros. The glomerulus is the site of blood filtration in the kidney. It consists of a capillary tuft in apposition to podocytes of the kidney. Fenestrations in the capillary endothelium and slit diaphragms in the podocytes allow the passage of the filtrate into Bowman’s space. The cells of the tubule are the site of selective reabsorption and secretion. The nephric duct carries the modified urine to the exterior (Saxén, 1987).

The pronephros of the zebrafish consists of paired glomeruli at the midline ventral to the dorsal aorta, and a set of pronephric tubules that project laterally from the glomeruli to the nephric ducts that run along the side of the embryo (Drummond et al., 1998). Histologically, the first evidence of the pronephros is a cell aggregate derived from the coelomic epithelium located ventral to the anterior somites (Goodrich, 1930). This aggregate will form the paired pronephric primordia which remain visible as flattened epithelial sacs between the 30 to 36 hours post-fertilization (hpf) stages in the zebrafish (Drummond et al., 1998). Unlike the pronephros of amphibia, the connection to the coelomic cavity in teleosts is transient and the nephrocoel of the primordia is closed to the body cavity (Drummond et al., 1998; Goodrich, 1930). The primordia then fuse at the midline and will form the sole paired nephrons of the pronephros. The glomerular blood supply is derived from the dorsal aorta and filtration begins at the 40 hpf stage (Drummond et al., 1998; Majumdar and Drummond, 1999).

Ablation experiments in Xenopus demonstrate that the tubular and duct progenitors in the organ primordia are spatially distinct (Vize et al., 1995). However, explant studies indicate that the kidney cell fate is specified much earlier in the intermediate mesoderm (Brennan et al., 1998; Brennan et al., 1999). The origins of these structures from undifferentiated intermediate mesoderm and the lineage relationship among these cells has not been determined in the normal pronephric kidney. In explanted metanephric mesenchyme, induced to initiate nephron morphogenesis but lacking vessels, retroviral tagging indicates that there is a shared lineage between glomerular and tubular cells (Herzlinger et al., 1992). In vivo labeling of the pronephric duct indicates that in some species the duct grows by accretion of cells from surrounding mesoderm, and in others by caudal migration of a duct rudiment (Cornish and Etkin, 1993; Fox and Hamilton, 1964; Obara-Ishihara et al., 1999; Poole and Steinberg, 1981).

There are suggestions from studies of other organ fields that the distinctive modular elements of an organ may be set aside quite early in development and reflected in spatial domains of gene expression and cell fate in organ fields. Interestingly, single gene defects indicate that structural elements of organs may be selectively removed leaving the rest of the tissue to develop relatively normally (Fishman and Olson, 1997). Hence, both embryological and genetic lines of evidence suggest a relative independence to development of structural modules of organs. For example the no isthmus mutation removes the pronephric tubules but does not appear to affect glomerular or duct development (Majumdar et al., 2000). Thus, it is crucial to define where cell fate assignments are made.

We address the origin, patterning and regulatory attributes of the pronephric field in vivo in the zebrafish by use of an uncaging technique we used previously to define the heart field (Serbedzija et al., 1998). We find that there is a bounded region of the intermediate mesoderm that constitutes a pronephric field. Within this region, there is an antero-posterior pattern to the fate map, with pre-glomerular cells most anterior, pre-duct cells most posterior, and pre-tubular cells in between. Interestingly, this fate map corresponds to particular combinations of transcription factor gene expression in the intermediate mesoderm.

Wild-type zebrafish were maintained and raised as described (Westerfield, 1995). Dechorionated embryos were kept in 30% Danieau’s solution and staged according to somite number or hours post-fertilization (hpf) (Kimmel et al., 1995). The wild-type lines were of the TL background obtained from the laboratory of Dr Christianne Nusslein-Volhard in Tübingen, Germany.

wt1 was cloned using nested PCR on a 24 hpf embryo lambda-ZAP Express (Stratagene) cDNA library. The first reaction used the vector T3 primer at the 5′ end and WTR1 (GGTTGACCGCACAATGGCAGGAGTC), a reverse primer made against the zinc-finger region (Accession Number X85734). No clear bands were visible by ethidium bromide staining, but Southern blotting using a probe for human WT1 (a gift from Dr Jerry Pelletier, McGill University) confirmed a 1.1 kb product. That area of the gel was excised and used as the template in the second reaction with a second forward primer made against the pBK-CMV vector and WTR2 (GCCCGGTACTCTCCGCACATCATG) as the reverse primer. A 900 bp product was subcloned in pBluescript (Stratagene) and several clones were sequenced. We noted discrepancies at the 5′ end of the clones and so a primer, WT1F-Eco (CCGAATTCAACGGTAACTGTACTCTTCCAGTG), was made to reamplify the wt1 clone beginning where the sequences all agreed. The PCR product was cloned into pBluescript, resequenced and used in all in situ hybridizations of this study. EcoRI was used to linearize the template and T7 RNA polymerase to transcribe antisense riboprobes.

Zebrafish single-minded homologs were cloned using a degenerate nested PCR using adult cDNA as a template. In the first round of PCR, the primers were ATGAARGAGAARTC(A/T/C)AARAAYGCNGCNAA (forward) and TGGAA(A/T/G)GTGTCRCANCCGTG(A/ T/G)GCGTGGTGGTA (reverse), the primers for the second PCR were GAGAARGAGAACGCNGARTTCTAYGA (forward) and TCRAT(A/C/G)AGCTCCTGIGGCTCGTANGG (reverse). The resulting product was a 700 bp fragment containing the DNA-binding domain, as well as the PAS domains, this was then used to screen a 24 hpf library at medium stringency (0.2× SSC, 0.1% sodium dodecyl sulphate at room temperature) to obtain full-length clones. EcoRI and T7 RNA polymerase were used to linearize and transcribe antisense probes. The sequences for sim1 and wt1 were analyzed using the BLAST and CLUSTAL algorithms. The nucleotide sequence and predicted amino acid sequence of the clones have been deposited in GenBank (sim1, AY028626; wt1, AY028627).

Whole-mount in situ hybridization was performed as previously described (Oxtoby and Jowett, 1993) with some minor modifications. The probes used in this study have been previously described: pax2.1 (Krauss et al., 1991), pax8 (Pfeffer et al., 1998) and alpha-tropomyosin (Ohara et al., 1989). For the double label experiments, we adapted an existing protocol from mouse (Cygan et al., 1997). Embryos were hybridized with both digoxigenin- and fluorescein-labeled riboprobes. Anti-DIG AP (1:2000) and BM Purple (Boehringer Mannheim) were used to detect the first probe. The phosphatase was inactivated by incubating in 10 mM EDTA while heating to 65°C for 30 minutes. The fluorescein-labeled probe was detected using anti-fluorescein AP (1:5000) and the INT/BCIP substrate (Boehringer Mannheim). After the color reaction, embryos were equilibrated in 70% glycerol and photographed on a Leica dissecting microscope or a Zeiss Axiophot microscope equipped with DIC optics. For histological analysis, the stained embryos were equilibrated in PBS with 0.1% Tween-20 (PBT), dehydrated in series into ethanol, embedded in JB-4 (Polysciences) and sectioned.

Cell tracing experiments were performed essentially as described (Serluca and Fishman, 1999) with some modifications. Embryos obtained from natural spawning were dechorionated using Pronase (Sigma) and injected with approximately 1 to 5 pl of 1% DMNB-caged fluorescein dextran (Molecular Probes) in 160 mM KCl at the one- to four-cell stage and raised until early somitogenesis. At the eight-somite stage, embryos were placed on a 2% agarose ramp and positioned to allow a dorsal view of the notochord and developing somites. The lineage tracer was activated using a pulsed nitrogen laser (LaserScience) tuned to a wavelength of 362 nm mounted on a Zeiss Axioskop microscope equipped with a 40× water-immersion lens (Photonics, Arlington Heights, IL). The activation was followed using a fluorescein filter until patches of the desired size appeared labeled. The embryos were allowed to develop at 28.5°C until the 40-48 hpf stage and then fixed in 4% formaldehyde/PBS. In vivo visualization of the label is complicated at this stage as it lies in deep tissues and is quenched by autofluorescence from the yolk. For visualization in fixed tissue, the embryos were first permeabilized in acetone for 30 minutes at –20°C, then rehydrated in series in phosphate buffered saline+0.1% Tween-20 (PBT). Endogenous peroxidase activity was eliminated by incubating in 0.5% normal sheep serum (NSS) and 0.5% hydrogen peroxide for 30 minutes in PBT. The embryos were blocked for 30 minutes using 10% NSS in PBT and then incubated with a 1:500 dilution of anti-fluorescein-POD (Boehringer Mannheim) in 10% NSS for 2 hours. After several quick washes and one overnight wash in PBT, the antibody was detected using 3,3′ diaminobenzidine (DAB) SigmaFast tablets (Sigma). To facilitate the identification of labeled cells, the embryos were bleached in 3% hydrogen peroxide and 1% potassium hydroxide for 30 minutes at room temperature. Embryos were scored by visual inspection on a dissecting microscope and either photographed or embedded in JB4 plastic for further histological analysis. Serial section were stained with Methelyne Blue-Azure II and coverslipped (Humphrey and Pittman, 1974). In each set of experiments, approximately one third of the embryos were sectioned, in order to confirm the identity of the labeled structure.

For the tissue ablation studies, embryos were mounted and positioned on agarose ramps at the six- to eight-somite stage as described above. The laser was set to 452 nm, and we focused on the intermediate mesoderm cells adjacent to the first and second somite of the left side of the embryo. We used a low pulsing frequency in order to carefully monitor changes in cell shape. The ablated patch contracted, then retained a granular appearance as described (Serbedzija et al., 1998).

We cloned wt1 as a potential marker of glomerular podocytes, because it is known to mark these cells in both mice and humans (Armstrong et al., 1993; Pritchard-Jones et al., 1990). wt1 is a zinc-finger transcription factor involved in the development of multiple kidney forms in many species (Carroll and Vize, 1996; Pelletier et al., 1991; Pritchard-Jones et al., 1990; Sainio et al., 1997). We examined the developmental pattern of wt1 and compared it with that of pax2.1/no isthmus (noi), a zebrafish ortholog of the Pax2 gene, known to be expressed in the pronephric kidney, and with sim1, which we cloned because it is a marker of the developing duct in other species (Brand et al., 1996; Fan et al., 1996; Krauss et al., 1991; Pfeffer et al., 1998; Pourquie et al., 1996). sim1 is a zebrafish homologue of the Drosophila single-minded gene and encodes a basic helix-loop-helix transcription factor containing a PAS (period, ARNT, single-minded) motif (Crews and Fan, 1999).

wt1 expression first appears at the two- to three-somite stage as bilateral stripes in the intermediate mesoderm, extending from the first to fourth somite, as shown in Fig. 1A,B. We have used the somitic pattern of alpha-tropomyosin as an axial reference (Fig. 1B). Double staining with wt1 and pax2.1 indicates that the anterior limit of wt1 expression extends further than that of pax2.1 (Fig. 1C).

At the 11 somite stage, wt1 expression additionally appears in cells medial to the original stripes (Fig. 1D). It is in this cell population, ventral to the somites (arrow in Fig. 1E,F) that wt1 expression persists as, by midsomitogenesis, the wt1-expressing lateral stripes begin to fade (arrowhead in Fig. 1E). Their location suggests that this aggregate (Fig. 1F) may include those cells that will give rise to the paired pronephric primordia and which later express wt1 (Figs 1G,H), as shown previously (Drummond et al., 1998). Between 36 and 48 hpf, the pattern of wt1 expression is refined to include only the more medial cells until, from 40 hpf onwards, wt1 expression becomes limited to the cells of the glomerulus, presumably the podocytes (Fig. 1I-K). This expression is consistent with that observed during mouse and human kidney development, where WT1 is ultimately restricted to the podocytes (Armstrong et al., 1993; Pritchard-Jones et al., 1990). At the 48 hpf stage, pax2.1 labels the tubules and the anterior portion of the duct, but not the glomerulus (Fig. 1I,J). By 72 hpf, expression of pax2.1 recedes from the duct, and becomes confined to the pronephric tubule, while wt1 still marks the glomerulus (Fig. 1L-N).

sim1 is first expressed at the two somite stage (Fig. 2A). Double labeling experiments using wt1 and sim1 probes indicate that the posterior end of wt1 expression coincides with the rostral end of sim1 expression (Fig. 2B). Co-staining alpha-tropomyosin, indicates that pax2.1 expression extends to the second somite, while sim1 expression extends to somite four or five (Figs 2C,D). We have also looked at the expression of a second paired-box gene expressed in the kidney, pax8. The anterior border of pax8 expression is identical to that of pax2.1 (Fig. 2E). Both pax2.1 and pax8 expression in the kidney begins prior to somitogenesis and thus precedes the onset of both wt1 and sim1 expression (Krauss et al., 1991; Pfeffer et al., 1998). The intermediate mesoderm continues to express sim1 throughout the segmentation period. At this stage it is expressed also in the lateral portion of the somite (Fig. 2F). sim1 expression is absent from the pronephros after 30 hpf.

We were curious as to whether this region of presumptive renal markers corresponds to the site of origin of pronephric precursors, and whether the anterior-posterior patterning reflected elements of a cell fate map. In other words, we were interested in knowing if this region constituted a pronephric field.

We used laser activation of a caged lineage tracer to follow the fate of cells and to construct a fate map for the zebrafish embryonic kidney. We injected caged fluorescein in early cleavage stage embryos, which then develop until early (8-10S) somitogenesis. Using a pulsed laser, we then uncaged the fluorescein-dextran in a patch of three to five cells. We found that, as anticipated by the expression studies, cells destined to become the pronephros lay in the intermediate mesoderm adjacent to the somites. We found no cells outside the field that contribute to the pronephros. Lateral to the pronephric area we found that mesodermal cells were fated to become blood, while more anterior cells adopt mesenchymal fates in the pharyngial arches (S. J. Childs and F. C. S. and M. C. F., unpublished).

Patches of cells were activated lateral to the somite at the eight- to ten-somite stage (Fig. 3A,B). At the 12-somite stage, a portion of the labeled patch is found to lie below the somite, indicating a medial movement of the cells (Fig. 3C). However, by the 18 somite stage, the cells are difficult to detect by fluorescence (Fig. 3D) as they probably lie below the somitic mesoderm and above the developing gut. In order to visualize the labeled cells in these later stages, we used an antibody directed against fluorescein. The timing of the early cell movement correlated well with the medial appearance of wt1-expressing cells.

Using smaller regions of activation within the area we are defined as the pronephric field, we discovered that cell fate is predicted by axial position. These data are summarized in Table 1. Intermediate mesoderm adjacent to the first somite contains cells contributing exclusively to the glomerular segment (Figs 3E, 4A,B). At the level of somite two, we found both tubular progenitor cells (Fig. 4C,D) and glomerular progenitors. In some embryos, we observed label in both the glomerulus and tubules. We cannot determine whether these two fates derive from individual progenitors because we cannot be completely confident that we label only single cells. Cells adjacent to somite three contributed to the pronephric tubules (Fig. 3F) and, to a lesser extent, the anterior nephric duct. Caudally, we found only duct progenitors. Intermediate mesoderm at the level of the first three somites also contains progenitors for fin and gut mesenchyme. The rare labeling of periderm directly along the tract of the laser seems clearly to result from inexact focusing of the laser beam.

The mechanism of nephric duct elongation in other species involves a combination of directed cell migration and in situ generation of the duct from intermediate mesoderm cells (Cornish and Etkin, 1993; Poole and Steinberg, 1981). We examined this process in the zebrafish embryo by labeling intermediate mesoderm cells at somite five (S5), to see if they later migrated to form part of the more caudal duct. In separate experiments, we labeled more posterior undifferentiated mesodermal cells, at the level of the somite ten (S10), in order to assess the role of in situ differentiation (Fig. 5A,B). In some experiments, we labeled adjacent myocytes as a positional reference.

In both types of experiments, either labeling duct or posterior undifferentiated mesoderm, the labeled patch gave rise to a discrete continuous segment of the duct, typically less than one somite width in length. The labeled cells remained contiguous, without appearance of gaps (Fig. 5C,D) and the labeled cells in the duct remained in the same position as the site of IM labeling. Histological analysis of the embryos confirmed the identity of the labeled structure as nephric duct (Fig. 5E). Thus, in the zebrafish embryo the pronephic duct appears to elongate entirely by the recruitment of a subset of intermediate mesoderm cells and their in situ epithelialization without involvement of cell migration from the duct.

Embryonic fields have the capacity to regulate (Huxley and de Beer, 1934). That is, local injury or destruction of cells within the field is compensated and there is restoration of the field and eventual formation of a normal organ. To investigate the regulatory capacity of the zebrafish kidney field, we performed unilateral tissue ablations between the eight- and ten-somite stages. We focused on cells within the intermediate mesoderm that form the wt1-expressing region. We confirmed the extent of the injury to both glomerular and tubular precursors by in situ hybridization with the wt1 probe (Fig. 6B). Following the ablation procedure, the patch first contracted and became granular in appearance (Fig. 6A). Two hours following the operation, the ablated side has recovered its normal wt1 expression (Fig. 6C). We allowed the embryos to develop until the 48 hpf stage. The surviving embryos (28/31) were not edematous and did not exhibit any of the abnormal morphology associated with kidney defects (Brand et al., 1996; Drummond et al., 1998). Histological analysis of the post-ablated embryos revealed the presence of a complete pronephric kidney, including the glomerulus and tubules (Fig. 6D-F), indicating that this pronephric progenitor region of the intermediate mesoderm has regulatory ability characteristic of other embryonic fields.

By use of in vivo lineage markers, we define here the mesodermal site of origin and fate of cells that fashion the pronephros of the zebrafish. These cells constitute a classical organ field: a region of progenitors with internal pattern and regulative ability (Huxley and de Beer, 1934). The pronephric field is in the intermediate mesoderm. Axial position within the field predicts ultimate pronephric fate. Pre-glomerular cells, pre-tubular and pre-nephric duct cells occupy progressively more posterior domains within the field. There is little evidence for intermixing between the populations. This fate map corresponds to compartments of gene expression, such that pre-glomerular cells are wt1+/pax2.1−/sim1−; pre-tubular cells are wt1+/pax2.1+/sim1−; and pre-nephric duct cells are wt1−/pax2.1+/sim1+, indicating that molecular distinctions between the subpopulations are in place prior to nephrogenesis.

The origin of the pronephros has not been previously defined by lineage analysis. Fales demonstrated that explants from the anterior intermediate mesoderm of Ambystoma can give rise to tubules when grafted to an ectopic location, but this corresponds to only one component of the pronephros (Fales, 1935). In particular we were interested in the origin of the glomerulus. There was reason to believe that it might arise from a location distinctive from tubules, despite the proximity and association in the final organ. In evolutionary terms, the glomerular podocytes appear to be derived from coelomic epithelium, a specialization that arose presumably so adjacent vessels could excrete ultrafiltrate into the coelom (Smith, 1943). This direct coelomic excretion is, in fact, the case in lower chordates such as the lamprey (de Beer, 1928). It is also the situation for the pronephros of certain amphibia in which tubules are separated from a glomerulus (or glomus) by the coelom (Vize et al., 1995).

This work indicates that there is a contiguity of pre-podocyte, pre-tubular and pre-nephric duct cells in the IM, but that they are spatially separated. Ablation experiments have shown that the organ primordia is subdivided into discrete domains giving rise to tubules and duct (Vize et al., 1995). However, there are no prior data for direct comparison, because analysis of lineage in the kidney has been largely confined to the duct segment, most particularly the mechanism by which it elongates (Cornish and Etkin, 1993; Obara-Ishihara et al., 1999; Poole and Steinberg, 1981). Indeed observations here on the pronephric field may not be directly comparable with the development of mesonephric or metanephric kidneys, because in species where such kidneys occur they form later in development and by processes of mesenchymal induction from a pre-existing duct. It has been noted, using explants of metanephric mesenchyme organ cultures labeled by retrovirus, that a single cell can contribute to both the glomerulus and tubule if marked prior to the induction event (Herzlinger et al., 1992). In the zebrafish we did observe a few cases (6/131) with mixed fates from one labeling, but because we label three to five cells at a time we suspect that cells are destined for one fate and the rare instances of multiple fates are due to labeling at the borders of these domains.

Thus, it appears that pronephric cell fates are assigned in the intermediate mesoderm prior to or during early somitogenesis. This is similar to the situation for the heart, where progenitors for atrium or ventricle are spatially segregated while still in the lateral plate mesoderm heart field (Fishman and Chien, 1997) and for the limb where proximal-distal identities are specified within the limb field (Murray, 1926).

Regulative ability has been described for the limb, ear and cardiac fields (Copenhaver, 1926; Harrison, 1918; Serbedzija et al., 1998; Tokura, 1925) and for migratory neural crest (Raible and Eisen, 1996). We did not determine whether the embryo can replace an entire pronephric field, because embryos died immediately after ablation of very large areas of the zebrafish intermediate mesoderm. We believe this to be due to the inadvertent ablation of the yolk cell membrane. Ablations of smaller areas, however, were well tolerated. We focused our attention upon the anterior-most portion of the field: those cells that express wt1 and that normally give rise to the glomerular and tubular segments. The destruction of these cells did not prevent the formation of a glomerular structure, as confirmed by histological criteria.

Duct elongation of the zebrafish pronephros occurs by recruitment of cells from the intermediate mesoderm. This process appears to have a species-dependent mechanism. In axolotl, the duct elongates due to migration of cells from the pronephric duct rudiment (Poole and Steinberg, 1981). In Xenopus, the elongation involves both cell migration and a contribution from IM cells located posteriorly, cells that are recruited into the elongating duct structure (Cornish and Etkin, 1993; Fox and Hamilton, 1964).

The clustering of cell fates into spatially distinct zones correspond to different patterns of expression of three transcription factors: wt1, pax2.1 and sim1 (Fig. 7).

In mice, Wt1 is expressed in the intermediate mesoderm that will form the mesonephros and in the uninduced mesenchyme which gives rise to the metanephros (Armstrong et al., 1993). Targeted inactivation of Wt1 leads to an absence of the metanephric kidney and gonads as well as a defect in the caudal mesonephric segments (Kreidberg et al., 1993; Sainio et al., 1997). Pax2 in the mouse is found in the nephric duct and later in induced mesenchyme. Mouse embryos homozygous for a null allele manifest a failure of normal duct growth and epithelialization of the mesenchyme (Torres et al., 1995). Overexpression of Pax2 causes an abnormal and dysfunctional renal epithelium, similar to that observed in congenital nephrotic syndrome (Dressler et al., 1993). Thus, Pax2 in the mouse appears to be involved in the generation and maintenance of the kidney tubular epithelium. sim1 is expressed in the developing duct of the mouse and chick. Inactivation of the sim1 locus does not lead to any overt kidney defects (Michaud et al., 1998). However this analysis may be complicated by the presence of a closely related homolog, sim2, in that same tissue (Fan et al., 1996). The roles of these genes in the pronephric kidney are not known.

In zebrafish, wt1 begins as a broad stripe in intermediate mesoderm, from which a cluster appears to break off and migrate medially. The position of this cluster corresponds roughly to that of glomerular progenitors. After formation of the glomerulus, wt1 expression in the kidney is limited to the podocytes. Whereas it appears that expression of wt1 alone is indicative of the glomerular fate, co-expression of pax2.1 with wt1 seems to preclude the glomerular fate, and rather is associated with the tubular cell fate. Naturally, we examined the effect of interfering with wt1 function by morpholino antisense oligonucleotides, but found no effect (F. C. S. and M. C. F., unpublished). There is genetic evidence to indicate a direct role for the pax2.1 gene in control of a cell fate switch. no isthmus zebrafish embryos have a mutation in the pax2.1 gene and lack pronephric tubules. Tissue found in its place has characteristics of podocytes, including expression of vascular endothelial growth factor (VEGF) (Majumdar et al., 2000). Although pax2.1 also is expressed in the duct-forming cell, it is not required for duct formation as noi embryos do form a nephric duct.

The signals that partition the IM into different zones remain to be determined. The fact that there appears to be an antero-posterior organization suggests that axial information might be used. There is precedent for linkage between axial information and organ field patterning. The anterior notochord regulates the posterior extent of the heart field in zebrafish (Goldstein and Fishman, 1998) and Hox gene expression is reflected in the pattern of the limb field in the chick (Tabin, 1995).

We are greatly indebted to Sarah Childs for assistance, discussions and encouragement especially in the early stages of the project and for comments on the manuscript. We also thank John Mably and Iain Drummond for insightful discussions and criticisms. The pax2.1 and pax8 probes were provided by Iain Drummond. Peter Moffett and Jerry Pelletier kindly provided a mouse sim2 probe prior to publication. This work was supported in part by NIH grants RO1HL49579, RO1DK55383 and RO1HL63206 to M. C. F. This work was carried out in partial fulfillment of the requirements for the PhD degree (F. C. S.).