Many complex organs like the lung, the salivary gland, and the mammalian kidney develop through mesenchymal–epithelial interactions and transitions. During nephrogenesis, the ureteric bud and the surrounding mesenchyme reciprocally interact to form the nephrons and the collecting system, which are embedded in a mesenchymal interstitium. In the past, several factors involved in the inductive interaction between the ureteric bud and the metanephrogenic mesenchyme have been identified and analyzed. However, the origin and development of stroma cells, which form the renal interstitium, are poorly characterized.
The interstitial cells surround the essential components of the kidney, i.e., the collecting ducts, nephrons, and blood vessels. They secrete the extracellular matrix and growth factors essential to support the development of these components. The primary interstitium in the center of the embryonic metanephros is composed of fibroblastic cells arranged in layers around ureteric bud branches (Alcorn et al., 1999). Stromal cells are additionally found in the kidney periphery surrounding the nephron progenitors. During development, the primary interstitium changes its phenotype, gene expression, and function to form the outer cortical and inner medullary interstitium. The adult interstitium takes part in the production and remodeling of the extracellular matrix, of regulatory substances and in immune responses (Kaissling et al., 1996). It is generally assumed that stromal cells derive from mesenchymal cells that do not form nephrons. However, it is unclear whether interstitial cells develop as a remnant of the metanephrogenic cells or undergo active differentiation from metanephrogenic progenitor cells. Additionally, DiI labeling and grafting experiments in birds have suggested that neural crest cells contribute to the renal stroma (Le Douarin and Teillet, 1974; Bronner-Fraser and Fraser, 1989). Stromal cells are very abundant during the fetal period but are subsequently reduced through massive apoptosis (Koseki et al., 1992).
A variety of mouse knockout and in vitro culture studies have demonstrated that the embryonic stroma cells are essential for the development of both the ureteric bud and the metanephrogenic mesenchyme. Targeted disruption of the transcription factors Foxd1 (BF2), Pod1, or Pbx1 that are expressed in renal stroma cells show defects in ureter branching and nephron differentiation (Hatini et al., 1996; Quaggin et al., 1999; Schnabel et al., 2003). Similarly, Rarα/β2 compound mutant mice display impaired ureter branching (Mendelsohn et al., 1999) and FGF7 knockout kidneys are smaller with fewer collecting ducts and nephrons (Qiao et al., 1999). In vitro antibody inhibition of Lamininα or Nidogen1 likewise affects both processes (Ekblom et al., 1994; Kadoya et al., 1995). Thus, the cortical stroma surrounding ureteric bud branches and metanephrogenic mesenchyme as well as the medullary stroma synthesize transcription factors and secrete growth factors and extracellular matrix proteins that help to direct the formation of nephrons and collecting system.
Here, we describe the isolation and expression of a novel gene called Snep (stromal nidogen extracellular matrix protein), which is interesting in this respect for several reasons: Snep is specifically expressed in stroma cells of the kidney and other organs. It is additionally found at sites of massive apoptosis in the embryo. The Snep gene encodes a secreted extracellular matrix protein that contains a nidogen domain, found only in Nidogen1, Nidogen2, Tectorin α, and Mucin4 before. Comparative expression analysis of Snep with other stroma markers and with Nidogen1/2 shows that Snep transcripts are localized in a specific subpopulation of stromal cells. Moreover, Snep is coexpressed with Nidogen1/2 in the medullary kidney stroma.
Identification of the Murine Snep Gene
In a screen for genes involved in early kidney development, we had compared mesenchyme from embryonic day (E) 11 mouse kidney anlagen with mesenchyme induced with spinal cord for up to 3 days by using differential display polymerase chain reaction (ddPCR; Leimeister et al., 1999). One of the genes up-regulated in the induced kidney mesenchyme was found to be expressed in the renal stroma but not in the nephrogenic mesenchyme. The 3′ differential display cDNA sequence was extended by expressed sequence tag (EST) walking, followed by screening of an E11 mouse cDNA library and reverse transcriptase (RT) -PCR using degenerate primers based on gene predictions in the corresponding human genome sequence. Our cDNA sequence is expected to represent the full-length transcript, as we found upstream in-frame stop codons and a typical CpG island structure that likely represents part of the promoter. The 9.2-kb mouse cDNA encodes a putative secreted extracellular matrix protein, designated Snep.
The N-terminal signal peptide of Snep is indicative of a secreted protein (Fig. 1A). The nidogen domain is only found in the basement membrane proteins Nidogen1/2, in Tectorin α and Mucin4. However, a specific function of this domain is not known to date. Fibronectin type III domains and calcium-binding epidermal growth factor (EGF) repeats are often found in extracellular matrix proteins. The latter are responsible for protein–protein interactions. For the central part of the Snep coding region, two different classes of transcripts were found, which result in protein products with or without a CCP (complement control protein) domain. CCP modules, also known as short consensus repeats or SUSHI repeats, have been identified in a wide variety of complement and adhesion proteins. The presence of two different Snep splice variants has been verified by RT-PCR using flanking primers to amplify and sequence the respective cDNA sequences from different mouse tissues. There was no evidence for tissue-specific differences in the ratios of both isoforms with the longer form being the most abundant (not shown).
The mouse Snep gene is encoded by 31 exons, spread over 66 kb of genomic sequence on chromosome 1. Exon 16 is subject to alternative splicing (CCP domain). The human gene has a similar exon– intron organization and maps to chromosome 2q37.3. The expression of Snep in adult mouse tissues is shown in Figure 1B. Note that Snep is not expressed in the adult kidney.
Expression of Snep During Mouse Embryogenesis
In situ hybridization of whole-mount mouse embryos, showed no clear staining at E9.5 (data not shown). At E10.5, Snep transcripts are detected in skeletal precursors of the limb buds, in two stripes along the body axis ventral to the spinal cord, and in stripes parallel to the developing somites (Fig. 2A).The latter stripes are faint at the dorsolateral end but strong at the ventral end. Vibratome sections of these embryos showed no staining in the dermomyotome but more medial, most likely in the sclerotome (Fig. 2B). At E12.5, Snep transcripts are detected again in stripes in the tail, in the head mesenchyme, and in the developing limb buds (Fig. 2C). In the latter, the staining is between the mesenchymal condensations forming the digits, i.e., in cells that will be lost through apoptosis.
Expression of Snep during later stages of embryogenesis was analyzed by in situ hybridization of tissue sections. Figure 2D gives an overview of Snep expression sites at E14.5. Specific examples of Snep expression are magnified in Figure 2E–L. Snep transcripts are detected in mesenchymal cells throughout the body: surrounding the brain and the spinal cord (Fig. 2D), in choroidal structures at the developing cerebellum (Fig. 2E), in the mesenchyme of the nose (Fig. 2G), the upper and lower jaw (Fig. 2D), in stromal cells of the salivary gland (Fig. 2H), the ear (Fig. 2F), the lung (Fig. 2K), the kidney (Fig. 2J), the pancreas (Fig. 2D), and surrounding the adrenal gland (Fig. 2D). Snep is also expressed at sites of bone formation, in the head and trunk (Fig. 2D). In the developing limbs, Snep is weakly expressed in the cartilage, but strongly in the mesenchyme between the developing digits and phalanges (Fig. 2L). Snep transcripts are also found in mesothelia and surrounding developing organs like thymus (Fig. 2I), adrenal gland, salivary gland, and kidney. In the peripheral nervous system, Snep is expressed in dispersed cells of the ganglia and nerves (Fig. 2M,N).
At E17.5, Snep continues to be expressed in the same types of cells as described for E14.5. Figure 2M and N shows transverse sections through the trunk and head, respectively. Snep is weakly expressed in cartilage and also at sites of ossification, but strongly in stromal cells throughout the trunk and skull. In the trunk, a very prominent staining can be seen in the spleen, in the kidney stroma, in several mesothelia, surrounding the neural tube and in its center and in dispersed cells of the dorsal root ganglia. Snep is weakly expressed in the cartilage of the ribs and in smooth muscle cells of the stomach. In the head, Snep expression is detected in many loose mesenchymal cells, in a layer surrounding the brain (e.g., pons here), in single cells of the pituitary gland and the trigeminal ganglia, the vomeronasal organ, the teeth, and the whisker follicles. In summary, Snep transcripts are only observed in mesenchymal or stromal cells throughout murine embryogenesis. Although expression is widespread, it is very specific. In this regard, it is interesting to note that Nidogen1 is also expressed in mesenchymal derivatives throughout the entire embryo (not shown), but there is only partial overlap with Snep expression sites and both genes have unique sites. Overall Snep tends to exhibit more restricted and specific patterns.
Comparative Expression Analysis of Kidney Stroma Markers
As described above, we detected Snep transcripts within the stroma cells of the embryonic kidney. Because the renal stroma consists of several cell types, characterized by the expression of different marker genes, we compared the expression of Snep with that of other stroma markers.
Foxd1 is selectively expressed in cortical stroma cells (Fig. 3A). Pod1 is complementary expressed in nephron progenitors, podocytes, cells at the corticomedullary border, and in the medullary stroma (Fig. 3C,I). TenascinC mRNA is detected in the dense mesenchyme that immediately surrounds the budding ureter and in a subset of stromal cells at the boundary of cortical and medullary stroma (Fig. 3B). Rarβ2 is expressed in cortical and medullary stroma (Fig. 3D). Snep is likewise expressed in the cortical and the medullary stroma in a pattern similar to Rarβ2. Compared with Foxd1, Snep transcripts are detected in fewer cells of the cortical stroma (Fig. 3A,E,G,J,K). In the medulla, Snep expression overlaps with that of Pod1 and Rarβ2 but not TenascinC. Of interest, Nidogen1 is also coexpressed with Snep in the kidney medulla (Fig. 3F). Even Nidogen2 is expressed in a similar pattern but at barely detectable levels (not shown). We observed only a few dispersed cells expressing Nidogen1/2 in the kidney periphery, but it is unclear whether these cells also express Snep.
The extracellular matrix (ECM) appears in two different forms: the basement membranes of epithelial cells and the stromal matrix between mesenchymal cells. It is composed of glycoproteins and proteoglycans and was underestimated for many years as an inert ground substance. However, it has become clear that the ECM also controls cell behavior in that it participates in cell migration, isolation of tissues from each other, and signal transduction. During embryonic development, the ECM can act as a guide or a barrier for cell movement. It plays a role in the folding of epithelial sheets (e.g., salivary gland duct branching), in epithelial–mesenchymal interactions/ transitions, in conversion of mesenchymal into epithelial cell types, and in the differentiation of cells (e.g., kidney development; Wallner et al., 1998; Perris and Perissinotto, 2000; Lelongt and Ronco, 2003).
In the present study, we describe Snep, a novel putative ECM molecule with an interesting protein domain composition and a highly specific embryonic expression pattern. The Snep protein contains a nidogen domain, followed by calcium-binding EGF repeats, fibronectin type III repeats, and an alternatively spliced CCP motif in the middle. Whereas the latter protein modules are widely distributed among ECM proteins, the nidogen domain is only present at the N-terminus of the ECM molecules Nidogen1/2, Tectorin α, and Mucin4, and its function is unknown to date. Nidogen proteins are essential components of epithelial basement membranes, where they serve to crosslink perlecan, collagen IV, and laminin. In early kidney development, nidogen mRNA is expressed by mesenchymal cells, and the protein is detected in the ureteric bud basement membranes as well as the interphase between the condensed and loose mesenchyme. Although targeted disruption of Nidogen1 does not lead to an obvious phenotype, mice deficient for the nidogen-binding site of laminin die after birth due to kidney agenesis and impaired lung development (Murshed et al., 2000; Willem et al., 2002). There is evidence of Nidogen2 compensating Nidogen1 deficiency in transgenic mice (Miosge et al., 2002). A second Nidogen1 knockout mouse showed neurological defects and selective disruption of basement membranes (Dong et al., 2002), but the difference in phenotypes has yet to be explained. In our expression studies, we found a colocalization of Snep and Nidogen1/2 transcripts in the majority of embryonic kidney stroma cells and in several other, mostly mesenchymal tissues. It is too early to speculate about a potential interaction or redundancy between these proteins without further knowledge on the protein localization and functional role of Snep and the specific function of the nidogen domain, but future loss of function studies may have to take three genes into account.
Snep is a novel ECM molecule expressed in stromal cells during early and late kidney development. Compared with other stroma markers, Snep mRNA is present in a specific subset of stromal cells, supporting the idea of at least partly nonredundant functions. Of interest, a similarly significant and essential role of mesenchymal cells for organ development is known for the salivary gland and lung—both tissues where Snep is also expressed. Targeted disruption of Pod1 or indirect inhibition of Nidogen1 function affect development not only of the kidney, but again of the lung and salivary gland, providing a unifying theme for stromal contribution (Kadoya et al., 1997; Quaggin et al., 1999; Willem et al., 2002).
During late kidney development, stromal cells are lost through massive apoptosis (Koseki et al., 1992). In this respect, it is noteworthy that Snep is not detectable in the adult kidney by Northern blot or in situ hybridization analyses, but it is found at sites of apoptosis in other tissues, e.g., flanking to the digital rays. A recent analysis of apoptosis in the developing mouse kidney came to the conclusion that the apoptosis rate in kidney is not different from tissues whose development does not require cell death (Foley and Bard, 2002). However, the authors state that the level of apoptosis is highest in the stromal cell and stem cell compartment, regions of Snep expression. Very recently, a partial cDNA clone of the human Snep protein was isolated in a screen for stromal cell-derived membrane or secreted proteins that support hematopoietic stem cells (Ueno et al., 2003). Although Snep (there SST-3) function was not analyzed any further, this observation supports the notion of Snep representing a functionally important stroma marker.
Cloning and Sequence Analysis
The PCR fragment J6-3 obtained by differential display PCR was sequenced and compared with the mouse EST database (Leimeister et al., 1999). Overlapping EST sequences were used to walk toward the 5′ end. The radiolabeled insert of EST 1054398 (GenBank accession no. AA611111) was used to screen the E11 mouse embryo cDNA library 69631-1 (Novagen). Several positive clones with up to 3-kb insert size were isolated. Inserts were sequenced and compared with EST and genomic sequence databases by using the Web-accessible programs NCBI Blast, NIX, and GCG. These clones covered the nontranslated 3′ end of Snep. Based on computer predictions and comparison with human sequences, we designed standard and degenerate primer sets to successively amplify and sequence the entire coding region of the mouse Snep gene. The mouse Snep nucleotide sequence was submitted to the EMBL database (accession no. AJ584850). The protein domain structure was analyzed with the SMART program. The presence and tissue distribution of the two different Snep splice variants was tested by RT-PCR with cDNA from different mouse organs by using standard procedures (Leimeister et al., 1999).
For Northern blot hybridization, 10 μg of total RNA from each tissue was separated on a 1.2% agarose/formaldehyde gel, transferred to a nylon membrane, and hybridized with the radiolabeled insert from EST 1054398. In situ hybridization of whole-mount embryos and tissue sections was done on CD1 outbred embryos as described previously (Leimeister et al., 1999). EST clone 1054398 was used to generate the Snep riboprobe. Partial cDNA sequences of the Pod1, Foxd1 (BF2), Nidogen1, and Nidogen2 were amplified by PCR from E12 cDNA and cloned into the pCS2 vector (primer sequences available upon request). Plasmids containing the cDNA sequences of Rarα and Rarβ2 were generously provided by Cathy Mendelsohn. EST clone 736372 (GenBank accession no. AA270625) was used for the TenascinC riboprobe. Tissue sections were counterstained with nuclear Fast Red (Sigma). Stained whole-mount embryos were embedded in albumin and sectioned to 50 μm with a Polyscience Vibratome.
We thank Cathy Mendelsohn for providing the Rar cDNA. We also thank Barbara Klamt for technical assistance and sequence analysis and Miriam Mannefeld for Snep isoform analysis.