LFB1 and LFB3 homeoproteins are sequentially expressed during kidney development

LFB1 is a transcriptional activator (Frain et al., 1989) also named HNF-1 (Courtois et al., 1987), APF (Cereghini et al., 1988), HP (Schorpp et al., 1988), HNF-1 α (Mendel et al., 1991), A box factor (Maire et al., 1989) required for the expression of several liverspecific genes in vitro, in cultured cell lines and in transgenic mice (De Simone et al., 1988, 1991; Monaci et al., 1988; Courtois et al., 1987, 1988; Tronche et al., 1989; Lichtsteiner et al., 1987; Schorpp et al., 1988; Feuerman et al., 1989). In a recent study (Nicosia et al., 1990), we described the isolation of cDNA clones coding for LFB1 (Frain et al., 1989; Chouard et al., 1990; Baumhueter et al., 1990), demonstrating that this protein contains a new type of homeodomain, characterized by 21 extra amino acids. We have also identified and cloned LFB3 (De Simone et al., 1991) another member of the LFB1 gene family, also referred to as vHNF-1 (Rey-Campos et al., 1991; Baumhueter et al., 1988; Cereghini et al., 1988) or HNF-lβ (Mendel et al., 1991) encoding a transcriptional activator, which also contains an extra-large homeodomain and forms heterodimers with LFB1 both in vitro and in vivo. LFB3 is expressed earlier than LFB1 during mouse and rat development, and is strongly induced by retinoic acid in F9 embryonic carcinoma cells (De Simone et al., 1991). LFB3 and LFB1 are expressed in several tissues which are not related by an embryonic lineage or by topology, including liver, kidney and the digestive tract (Baumhueter et al., 1990; De Simone et al., 1991; Rey-Campos et al., 1991). A common feature of these tissues is that they develop a specialized epithelium, and in situ hybridization experiments demonstrated that LFB3 and LFB1 are expressed in polarized epithelia (De Simone et al., 1991; Lazzaro unpublished observations).

The majority of epithelial cells originate from two of the three germ layers, the ectoderm and the endoderm. These cell populations are committed to becoming specialized epithelia, and develop as an elongation of the primary epithelial-like sheets of these germ layers. They migrate to their final location, and there establish morphologically defined units. In a few cases, such as kidney development, epithelial cells may originate from mesoderm-derived mesenchymal cells. Since kidney is one of the major sites of expression of both LFB3 and LFB1, we focused our attention on the conversion of non-polar mesenchymal cells to epithelium during rat metanephric development.

The development of the kidney anlage (metanephros) in the rat embryo begins on day 12.5 post coitum (p.c.), when the ureter bud bulges out from the Wolfian duct and invades the surrounding metanephrogenic mesenchyme. This bud then branches into the undifferentiated mesenchyme and induces the nephric mesenchyme to convert into the epithelial cells of the functioning nephron (Grobstein, 1955). During this conversion, the metanephrogenic mesenchyme expresses a specific set of proteins, including cytokeratins, polysialic acid, N-CAM, uvomorulin, cingulin and laminin A chain (Lackie, 1990; Rodriguez-Boulan and Nelson, 1989; Avner et al., 1983).

Studies on the early appearance of epithelial markers during metanephric tubulogenesis, using the transfilter organ culture technique (Grobstein, 1955), revealed that two classes of molecules are required either to establish (e.g. laminin A chain) or maintain (e.g. uvomorulin) cell polarity. The first class of molecules is expressed only during the development of the metanephros (laminin A chain, syndecan), and disappears during maturation of the nephron (Vainio et al., 1989; Klein et al., 1988; Ekblom et al., 1990). The second class of molecules, e.g. uvomorulin (Vestweber and Kemler, 1985; Vestweber et al., 1985) is detectable throughout development and in the adult kidney.

We describe here the temporal and spatial expression patterns of LFB1 and LFB3 during kidney development, which demonstrate that the two genes are expressed sequentially and that the onset of LFB3 transcription coincides with the “induction” of the metanephrogenic mesenchyme into a polarized epithelium.

Embryos and dissected organs were obtained from natural matings between Wistar rats. Midday of the day of vaginal plug appearance was considered day 0.5 p.c. Each sample was staged according to the external criteria of Witshi (1962).

The samples were fixed immediately after dissection in 4% paraformaldehyde in lx PBS, pH 7.0 for 12 hours at 4°C. Subsequently, samples were cryoprotected by immersion in 20% sucrose/1 × PBS overnight at 4°C and embedded in tissue-tek OCT compound (Miles Scientific, Naperville, EL.). Sections of 10 μm were cut using a cryostat and collected on poly-L-lysine-coated slides, according to Toth et al., 1987.

Single-stranded RNA probes were prepared and labelled with ³⁵S-UTP >1,000 Ci/mmol, Amersham) according to Toth et al., 1987 to a specific activity of 8.5 × 10⁸ disints/min//xg. All probes were subcloned into Bluescript vector (Stratagene) and transcribed using T3 or T7 RNA polymerase (Stratagene), as described in De Simone et al., 1991.

These probes were previously tested by RNAse mapping (De Simone et al., 1991) and the specificity of the in situ hybridization signals was assessed by the use of LFB1 and LFB3 sense and antisense probes on serial sections of each sample.

In situ hybridisation was carried out as described in Toth et al. (1987). After washing and RNAse treatment according to Toth et al. (1987), the sections were dehydrated with graded ethanols containing 300 mM ammonium acetate and processed for autoradiography using Kodak N1B2 emulsion (Eastmann Kodak, Rochester, NY) diluted 1:1 with water. Slides were developed after 2–4 weeks with D19 developer (Kodak), washed in distilled water and fixed in AL-4 fixer (Kodak). The slides were stained with toluidine blue and coverslips were mounted with Eukitt (O. Kindler GmbH, Freibourg, Germany).

For transfilter culture, the kidney anlagen were dissected from 11-day-old embryos. After microsurgical ablation of the ureter bud, the metanephric mesenchyme was cocultured with the spinal cord as an inducer (Saxén and Lehtonen, 1978) in I-MEM (Iscove-Minimal Essential Medium) (Gibco Laboratories, Eggenstein, FRG) supplemented with transferrin and 10% FCS.

Total RNA was extracted from cell cultures, tissues and transfilter organ cell cultures by using the guanidinethiocyanate-acid (GTC) phenol method (Chomczynski and Sacchi, 1987). Fresh tissues were homogenized directly in GTC buffer by a motor-operated homogenizer. Cells and transfilter cultures were resuspended in GTC buffer and disrupted manually by 3-to 4-fold aspiration through an extrathin needle.

In RNAse mapping experiments, 10 μg of total RNA were annealed with 1.5 ×10⁵ cts/minute of the specific riboprobe at 45°C O/N, then digested with a mixture of RNAse A + T1 and separated on a 6% sequencing gel. The amount of RNA analyzed per lane was quantitated by OD measurement and ethidium bromide staining agarose gel. The rat and mouse LFB3 riboprobes were both 614 bp long, yielding a 510 bp protected band (aal46–316); the rat LFB1 riboprobe was 308 bp, yielding a 224 bp protected band (aa207–282), while the mouse LFB1 riboprobe was 191 bp, yielding multiple protected bands clustered around the expected position for the LFB1 transcript (156 bp) (aal40–192).

The first contact between the ureter bud and the undifferentiated kidney blastema in rat embryo occurs around day 12–13 p.c. The ureteric bud originates from the lower end of the Wolfian duct and rapidly grows into the metanephric blastema to form the renal pelvis. From the renal pelvis, several ureteric ducts penetrate further into the metanephric mesenchyme. A signal from the kidney mesenchyme stimulates the ureter epithelium to branch and subdivide until several generations of ureteric ducts have been formed. At the same time, the ureter epithelium “induces” the meta-nephrogenic mesenchyme to differentiate into a new epithelium, which will eventually become the kidney tubules (Lehtonen, 1976; Lehtonen et al., 1985; Saxén, 1987; Ekblom, 1989). After induction, the mesenchymal cells clustered around the tip of the ureter undergo conversion to a typical polarized epithelium.

Several kidneys were collected from rat embryos at day 12.5 p.c. and serial sections were hybridized in situ with LFB1- and LFB3-specific cRNA probes. In Fig. 1A, D, we show two adjacent sections of the same ureter branching at the periphery of the metanephric mesenchyme. Two condensed epithelial structures can be observed in close proximity to the ureter tips (arrows). The corresponding dark-field micrographs of sections hybridized with LFB3 cRNA probes (Fig. IE, F) show that both the ureter and the condensed epithelial structures contain LFB3 transcripts. At the same stage, LFB1 transcripts could not be detected in the corresponding structures (Fig. 1B, C).

At day 15.5 p.c., after a few days of ureter induction, aggregates of nephric mesenchyme become progressively more differentiated, and several successive stages can be distinguished during the differentiation of the nephron. The initially condensed mass (condensation stage) becomes vesicular (comma-shape stage), and then elongates to form an S-shaped tubular structure. The distal extremity of this S-shaped structure, Bowman’s capsule, will differentiate into the renal corpuscle (glomerulus) by fusing with the vascular capillary tufts. The proximal and distal convoluted tubules and Henle’s loop originate from the remaining segments. After the distal tubules have been connected to the collecting ducts, all these structures will form the nephron.

In sagittal sections of day 15.5 p.c. rat embryos we demonstrate that both LFB1 and LFB3 transcripts are expressed in the epithelial component of several different organs; liver hepatocytes, gut and stomach lining epithelia (Fig. 2). In the developing lung, however, only LFB3 transcripts are detected (Fig. 2F); in these same structures, adjacent sections are negative for LFB1 transcripts (Fig. 2E). At this stage, LFB3 is strongly expressed in the developing kidney (Fig. 2C, F and I), whereas LFB1, although detectable in the same areas, is expressed at much lower levels (Fig. 2B, E and H). The lower urinary tracts (ureter, urinary bladder, urethra) are negative for both genes (Fig. 2H and I).

Higher magnifications of the metanephric blastema (Fig. 3) show that LFB3 is expressed in the ureter bud swellings and in the condensed clusters of induced mesenchymal cells from the earliest stages of differentiation, as well as in the progressively more differentiated comma and S-shaped structures (Fig. 3E, F, G and H). LFB1 transcripts are localized only in the S-shape structures (Fig. 3A, B) not in the ureter epithelium (Fig. 3C, D) nor in condensed or comma-shaped bodies. The distal extremity from which the glomeruli will originate is negative for both genes (Fig. 3A, B, E and F).

The rat kidney is still immature at birth and will continue to differentiate until 2 weeks post partum. In the newborn rat kidney, all stages of nephron development are thus detectable from early embryonic precursors through to mature functional nephrons. Nephrogenesis takes place at the periphery of the kidney where continuous branching of the ureter and the proliferation of mesenchymal cells still occurs. As a result of the reiteration of this induction/proliferation process, differentiated nephrons will be gradually displaced from the periphery, and move towards the inner parts of the kidney (Saxén, 1987; Mugrauer et al., 1988).

In situ hybridization analysis of adjacent sections of new bom rat kidneys revealed that both LFB1 and LFB3 transcripts are present in the renal cortex and medulla (Fig. 4). However, their expression differs in the central papillary area and in the nephrogenic cortex, where only LFB3 can be detected. These differences are better observed at higher magnifications (Fig. 5A). The different zones of tubule differentiation are schematically marked: 1 = early differentiation (nephrogenic zone, n.z.); 2 = late differentiation; 3 = terminal differentiation. In order to localize the fully developed proximal and distal tubules (Henle’s loop), adjacent sections from the same samples were immunolabelled with antibodies raised against the brushborder and the Tamm-Horsfall antigens respectively (Fig. 5D) (Ronco et al., 1987; Ekblom et al., 1981). The peripheral nephrogenic zone 1 (n.z., arrowheads) is negative for both antigens, and staining is clear only in (3) differentiation zones. In the nephrogenic zone (1) at the extreme periphery of the organ, we could only detect LFB3 (Fig. 5C, L) and not LFB1 transcripts (Fig. 5B, I). LFB1 and LFB3 expression in the proximal and distal convoluted tubules and the Henle’s loop, as well as their absence in the glomeruli, confirm the observations made at day 15.5 p.c. (see also Table 1).

In the kidney papilla (Pa) (Fig. 5A), which is mainly composed of collecting ducts (90%), vasa recta and loose mesenchyme, we could detect only LFB3 mRNA (Fig. 5C, G). Adjacent sections of the same area are negative when hybridized with LFB1 cRNA probes (Fig. 5B, F). Higher magnifications of the papillary area show that LFB3 expression is clearly restricted to the collecting tubules; both the loose mesenchyme and the vascular endothelium are negative (not shown). In a transverse section through the renal pelvis (Fig. 5E), we show that LFB1 and LFB3 transcripts are undetectable in the renal pelvis (P) and in the ureter (Ur) (Fig. 5F, G).

In the developing kidney at least four different lineages are present: ureter epithelium (inducer), nephric mesenchyme (responder), vascular endothelium, and stromagenic mesenchyme. The developmentally regulated interplay of two of these four lineages (the nephric and stromagenic mesenchymes) and conversion of the nephric mesenchyme into an epithelium can be reproduced in vitro by the mesenchyme transfilter culture technique (Grobstein, 1955). Explants of mouse presumptive nephrogenic mesenchymes can be cultivated in vitro on synthetic filters. These tissues do not express any kidney-specific markers, but they can be induced to undergo nephric differentiation by cocultivation with inducer spinal cord Explants. In this experimental system, the first morphological changes (cell condensation) are detectable 36–48 h after induction. At 120 h after induction fully developed tubules are observed (Saxén, 1987).

Total RNA was extracted from transfilter cultured kidney mesenchyme at 42 h and 120 h after induction. The RNAs were analyzed by RNAse mapping using cRNA probes specific for LFB1 and LFB3 transcripts, along with RNA extracted from uninduced kidney mesenchymes. The results (Fig. 6) indicate that there is no expression of LFB1 and/or LFB3 in the uninduced mesenchyme. At 42 h after induction LFB3 is expressed simultaneously to the first morphologic changes, while LFB1 is still undetectable. At 120 h after induction LFB1 can be detected also.

‘This study reports the expression pattern in the developing rat kidney of transcripts encoding two related transcription factors, LFBl and LFB3 (De Simone et al., 1991; Mendel et al., 1991). During fetal development LFB3 transcripts are detected at the earliest stages of mesenchyme induction (day 12.5 p.c.) in the branching ureters and throughout the differentiation pathway of mesenchyme-derived polarized epithelial cells. LFBl transcripts appear 2.5 days later in S-shaped bodies, during the initial phase of tubule formation. LFB3 early expression during nephron differentiation was confirmed in the newborn kidney, where the developing nephrons are distributed according to their stage of differentiation. LFB3 is expressed at the periphery of the organ in the early differentiation zone (nephrogenic zone) where LFBl is undetectable. Conversely, both genes are expressed in the late and terminal differentiation zones and LFB3 is also expressed in the collecting ducts.

Interactive cellular induction is a mechanism fundamental to vertebrate development (for a review see Gurdon, 1987). Kidney morphogenesis largely depends on this inductive mechanism (Montesano et al., 1991), based on the interplay between inducing and responding cell populations. The early inductory phase in the developing metanephric blastema is a heterotypic inductive interaction between the epithelium of the ureter bud (inducer) and the condensed mesenchymal cells (responder), whereas the postinductory phase is characterized by homotypic interactions between the induced cells (Grobstein, 1962). LFB3 and LFB1 appear to be sequentially expressed in the metanephric blastema; LFB3 in the early inductory phase, LFB1 in the post inductory phase and only in cells committed towards a “tubulogenic field” (Saxén and Saksela, 1971).

One intriguing explanation for this could be that LFB3 expression is triggered by the heterotypic inductive signal, while LFB1 transcription is activated later as a consequence of the “homotypic” program of cellular interactions. This hypothesis would be in line with previous observations from our and other groups, that in F9 embryonal carcinoma cells, only LFB3 responds to retinoic acid after short-term treatment (De Simone et al., 1991), whereas LFB1 induction is much delayed (Kuo et al., 1991). The presence of LFB3 but not LFB1 in the kidney-derived A1251 and MDCK (data not shown), and in the de-differentiated H5 and C2 hepatoma cell line (De Simone et al., 1991) suggests that these cells represent an intermediate stage during the sequential activation of differentiation programs.

The temporal and spatial expression patterns of LFB1 and LFB3 are similar to those observed for some polarized epithelia-specific markers e.g. uvomorulin (Vestweber and Kemler, 1985; Vestweber et al., 1985). However, a functional relation between these two transcriptional activator genes and any of the known molecules considered landmarks of the polarized phenotype, has still not been established. The only LFB1-(and LFB3-) dependent genes that have been identified so far are liver-specific or liver-enriched.

In point of fact, some of the liver-specific genes regulated by LFB1 and LFB3, such as albumin and alpha-fetoprotein, are also expressed in various non-hepatic tissues, including the developing kidney (Nahon et al., 1988); however, their amount in kidney is at least two-fold less than in the liver (Putnam, 1975). LFB1 and LFB3 proteins and mRNAs, on the other hand, are even more abundant in kidney than in liver (De Simone et al., 1991). Nevertheless, recent in situ hybridization studies showed that albumin and alpha-fetoprotein expression in the developing kidney coincides with LFB1 and LFB3 expression domains (Poliard et a]., 1988), suggesting that these transacting factors could also regulate the expression of their target genes in extrahepatic domains.

The genetic control of kidney development is poorly understood (Montesano et al., 1991). So far a number of genes encoding transcriptional regulators have been reported as being expressed in the developing kidney. Most, if not all, homeobox-containing (Hox) genes are expressed to a varying extent in the meso- and metanephric mesoderm, although their expression is not restricted to polarized epithelia, as in the case of LFB1 and LFB3; in fact they are more widely expressed in the neuroectoderm and/or in other mesoderm derivatives (Holland and Hogan, 1988a; Dollé and Duboule, 1989; Kress et al., 1990). The expression pattern of all these Hox genes suggests they play a role in the regional specification of the kidney along the anteroposterior axis (Holland and Hogan, 1988b) rather than in the differentiation of specific cell lineages, as we propose for LFB1 and LFB3 (De Simone et al., 1991).

There is growing evidence that tissue-specific transcriptional activators are expressed in various cell types (Sharp and Cao, 1990; Cockerill and Klinken, 1990). The tissue- or cell-specificity of their target genes seems more likely to be dictated by unique combinations of transcriptional activators (Kuo et al., 1990), which, as a consequence, provokes multiple interactions. According to this model, the tissue distribution of a transcriptional activator does not have a definite topological “meaning” per se, and its functional role can only be understood if considered in combination with that of other factors and finked to a specific set of target genes (Xanthopoulos et al., 1991; Costa and Grayson, 1991; Struhl, 1991). An alternative possibility is that the expression domains of transcriptional activators coincide with the distribution of cells that share a similar fate in different organs; this could be the case of LFB1 and LFB3, found in the epithelial component of various organs.

The sequential pattern of expression of LFB1 and LFB3 described here for the kidney, is in accordance with previous morphological and more recent functional studies (Gossens and Unsworth, 1972; Bard and Ross, 1991) and it is also observed in differentiated F9 and hepatoma cell lines (De Simone et al., 1991; Kuo et al., 1991): LFB3mRNÁ appears first, during the earlier stages of differentiation; LFB1 appears at a later stage, simultaneously to the acquisition of a more differentiated state. The distinctive and/or interactive transcriptional functions of these two homeoproteins seem to be, at least in part, modulated by translational or post-translational regulatory mechanisms affecting the relative amount of the two proteins in the different tissues (Mendel et al., 1991), however in the case of the adult kidney the good correlation between the amount of LFB1 and LFB3 proteins with the respective messages suggests that they are probably not subject to such controls in this tissue (Mendel et al., 1991) The sequential activation of LFB3 and LFB1 could be required by the kidney, as well as by other organs, to implement the “epithelial” genetic program and express a set of functions that are specific to epithelial cells. This phenomenon is reminiscent of the sequential activation of the Hox-4 complex of homeobox-containing genes during mouse limb development (Dollé and Duboule, 1989). A recent analysis suggested a correlation between the order of these genes in the cluster and their temporal activation in the posterior part of the mouse embryo (Izpisua-Belmonte et al., 1991). In the case of the genes coding for LFB3 and LFB1, the mechanism responsible for their apparent temporal regulation is probably different because they are not clustered and in fact they map on different chromosomes (Abbott et al., 1990; Bach et al., 1990).

We are indebted to K. Simmons, I. Mattaj and P. Dollé for constructive discussions and critical reading of the manuscript. D.L. is recipient of an A.I.R.C. fellowship.