Differential expression of gap junction mRNAs and proteins in the developing murine kidney and in experimentally induced nephric mesenchymes

Gap junctions contain transmembrane channels that directly link adjacent cells and provide pathways for the transfer of low molecular weight molecules and ions from one cell to another (Gilula et al., 1972; Bennett and Goodenough, 1978; Loewenstein, 1981). The junctional channels are bipartite structures formed by the association of two oligomeric structures or connexons, each connexon representing an oligomeric arrangement of six polypeptides. The gap junction (GJ) proteins have been derived from a multigene family, based on a conserved region of about 200 amino-terminal residues that includes four transmembrane and two extracellular domains (Zimmer et al., 1987; Beyer et al., 1987; Milks et al., 1988; Goodenough et al., 1988; Nicholson and Zhang, 1988; Hertzberg et al., 1988; Yancey et al., 1989). Fulllength sequences for several of these proteins have been deduced from cDNA analysis: a 32 ×10³M_r protein from mammalian liver (Paul, 1986; Kumar and Gilula, 1986), a 43 ×10³M_r protein from mammalian heart (Beyer et al., 1987), a 26 ×10³M_r protein from mammalian liver (Nicholson and Zhang, 1988), and a 31 ×10³M_r protein from several mammalian tissues, including placenta and skin (Hoh et al., 1991). A 46 ×10³M_r protein found in the lens also appears to be a member of this family because of its predicted structural motif (Kistler et al., 1988; Beyer et al., 1989). On the basis of predicted structural similarities, the multigene family has been divided into two classes, α and β (Risek et al., 1990). The α class contains α₁ (43 ×10³M_r), α₂ (38 ×10³M_r from amphibians) and α₃ (46 ×10³M_r); while the β class contains β₁ (32 ×10³M_r), β₂ (26 ×10³M_r) and β₃ (31 ×10³M_r).

Gap junctions mediate important developmental and physiological activities. In excitable tissues, gap junctions provide low-resistance coupling pathways for nerve conductance (Furshpan and Potter, 1968), myocardial contraction (Dreifuss et al., 1966) and the coordination of smooth muscle movement (for review, see Daniel, 1987). The specific function of each type of GJ protein remains largely unknown, although the dramatic elevation of α₁ mRNA seen in the myometrium the day before parturition is thought to be responsible for producing the junctions that synchronize the uterine contractions (Risek et al., 1990). During development, gap junctions have been shown to play important roles in compartmentalizing cells and in transmitting morphogenetic information (for review, see Guthrie and Gilula, 1989). Gap junctions connect the blastomeres of the 8-cell mouse embryo (Lo and Gilula, 1979) and, if junctional communication between cells of these embryos is blocked by treatment with antibodies that bind to GJ proteins, the communication-deficient cells are not retained by the compacted embryo (Lee et al.,1987). Drosophila imaginal discs and the gastrulating mouse embryo are both divided into communication compartments by GJ pathways that exist between certain cells and not others (Weir and Lo, 1982; Kalimi and Lo, 1988, 1989). Gap junctions also form communication compartments in developing frog embryos, which physiologically separate the presumptive neural ectoderm from presumptive epidermis (Warner, 1973). When antibodies to mammalian GJ protein were injected into blastomeres of 8-cell Xenopus embryos, electrical and ionic coupling was inhibited, and patterning defects (asymmetries) developed in those regions derived from the injected blastomere (Warner et al., 1984). Information transfer through gap junctions has also been found in developing invertebrates. The gradient of hydra head inhibitor was blocked by treatment with antibodies to mammalian GJ proteins (Fraser et al., 1987), and gap junctions appear at the specific time when the information required to form molluscan mesoderm is activated in the 3D macromere by contact with vegetal micromeres (de Laat et al., 1980).

The mammalian kidney provides an opportunity to analyze whether three of the well-characterized GJ proteins (α₁, β₁ and β₂) are utilized differentially during organ development. The kidney develops as a result of interactions between two tissues, the ureter bud and the metanephrogenic mesenchyme (reviewed in Saxén, 1987). When the bud enters the mesenchymal blastema on day 11 of mouse development, the mesenchymal cells induce it to branch. Conversely, the epithelium induces the mesenchyme to form the secretory nephrons first visualized as condensations around the tips of the ureter. The condensed cells then form a renal vesicle which develops a central cavity, and, subsequently, an S-shaped body is formed. The portion of this tube closest to the ureter becomes the distal tubule cells, while the rest develops into proximal tubules and the epithelium of the glomerulus. The continuous branching of the ureter bud and its induction of nephron formation in the mesenchyme generate the metanephric kidney.

Using antibodies and cDNA probes to three GJ proteins and their messengers, we analyzed the spatial and temporal distribution of these GJ proteins during mouse renal development. Results from this study demonstrate that the expression of these GJ proteins correlates with specific cell types.

Embryonic kidneys at different developmental stages were obtained by timed matings of Balb/c × C57BL/6 F₁ mice. The appearance of the vaginal plug was noted as day 0. The embryonic kidneys were removed at appropriate post-implantation stages in order to prepare RNA for analysis. For extraction of total RNA from the kidney samples, an acid phenol-guanidinium thiocyanate procedure was applied with RNAzol (Cinna/Biotex). The extracted RNA from the different stages was analyzed by using a mixture of GJ probes (for α₁, β₁ and β₂) in an S1 nuclease protection assay as described previously (Davis et al., 1986; Nishi et al., 1991).

The kidneys were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.3, for 1 hour at room temperature, mounted in Epon, and 1 μm sections were cut for staining with toluidine blue and examined in light microscopy.

Sections from whole embryonic kidneys, transfilter explants and reaggregated mesenchymal cell cultures were analyzed by immunohistochemistry. Affinity-purified rabbit antibodies were prepared against synthetic peptides corresponding to the S epitopes of GJ proteins α₁ and β₁ and the J epitope of β₂ (Milks et al., 1988; Risek et al., 1990). Mouse embryos from matings of CBA males with NMRI females and Spraque-Dowley rat embryos, used for doublestaining immunofluorescence experiments, were obtained from the breeding colony of the Department of Pathology, University of Helsinki. For indirect immunohistochemistry, whole embryonic kidneys were rapidly frozen in OCT and sectioned on a Leitz 1720 cryostat without prior fixation. Sections of 5–6 μm were incubated with 3% BSA, 3% goat serum (Vector Laboratories, Burlingame, CA) in PBS (10 mM sodium phosphate, pH 7.5, 0.9% NaCl) for 1 hour at room temperature to reduce non-specific binding. Incubation with preimmune serum or antibodies was performed overnight at 4°C in the BSA- and goat serum-containing buffer, followed by three washes in PBS. FITC- and TRITC-conjugated donkey antirabbit IgG (Jackson ImmunoResearch, West Grove, PA) diluted 1:200 in PBS was added to the slides for 1 hour at room temperature. The slides were then washed three times in PBS and mounted in PBS-containing glycerol or in Elvanol (Klein et al., 1988). Podocalyxin, a rat podocyte- and endothelium-specific antibody (Schnabel et al., 1989, Miettinen et al., 1990), dipeptidylpeptidase IV, a rat proximal-tubule-specific antibody (Miettinen et al., 1990), as well as brush border and Tamm-Horsfall glycoprotein antibodies (Miettinen and Linder, 1976; Sikri et al., 1979; Dawnay et al., 1980; Ekblom et al., 1980) were generous gifts of Dr A. Miettinen (Department of Bacteriology and Immunology, University of Helsinki). Double-staining immunofluorescence also utilized TRITC-conjugated goat anti-mouse IgG (Jackson ImmunoResearch) diluted 1:100 and FITC-conjugated goat anti-rat IgG (Cappel Laboratory, Cochranville, PA) diluted 1:500. Immunofluorescence was analyzed using a Zeiss Axiophot microscope with epifluorescence. Photographs were taken with Fuji Neopan 1600 ASA black-and-white professional print film.

Confocal microscopy was performed with the confocal scanning laser beam fluorescence microscope developed at the European Molecular Biology Laboratories, Heidelberg. The design and operating principles of this microscope have been described previously (Bacallao et al., 1989; Stelzer et al., 1989). FITC was excited at 488 nm by an argon laser (2020-05 SpectraPhysics, Inc., Mountain View, CA), and serial optical sections were made at 0.3-μm or 0.5–μm intervals.

Metanephric kidney rudiments were dissected from 11-day mouse and 13-day rat embryos. To separate the mesenchyme from the epithelial ureter bud, the explants were incubated in 0.75% pancreatin-2.25% trypsin for 1.5 minutes at 0°C. The manual separation was performed under a stereomicroscope at room temperature in Eagle’s Minimum Essential Medium (MEM) supplemented with 10% fetal calf serum (FCS; Myoclone Plus, GIBCO, Paisley, Scotland). Fragments of spinal cord from the same embryos were used as inducers in the transfilter experiments (Saxén and Lehtonen, 1978). In some experiments, the transfilter contact was interrupted after 22 hours and 48 hours to follow the development of the mesenchyme in prolonged cultures after a short induction pulse.

The isolated mesenchymes were transferred onto Nuclepore filters (General Electron Co., Pleasanton, CA) with an average pore size of 1.0 μm. A piece of spinal cord was glued beforehand on the opposite side of the filter, using agarose (Grobstein, 1956; Saxén and Lehtonen, 1987). The transfilter explants were cultured in MEM with 10% FCS and harvested at different intervals for immunohistochemistry.

The expression of GJ mRNA from three genes (α₁, β₁, β₂) during kidney development was studied by applying an S1 nuclease protection assay to RNA that was isolated at different developmental stages.

For this analysis, probes for all three transcripts were added together so that all three products could be analyzed simultaneously (Fig. 1). In the 13-day mouse embryonic kidney, the α₁ transcript was seen to be present at high levels. These levels remained high throughout development, but declined around birth. The β₁ transcript was readily detected by day 15, and the expression of this mRNA increased throughout development. On day 17, transcription of the β₂ message was detected, and the accumulation of this transcript increased dramatically after birth. The overall developmental pattern of expression differed for α₁ versus β₁ and β₂The α₁ transcript was expressed at high abundance during early development, but the expression decreased significantly after birth. Conversely, the expression of the β₁ and β₂ transcripts increased significantly during development and after birth.

In summary, the data from this analysis provides evidence that all three GJ genes are expressed during the development of the kidney, that these genes are developmentally regulated, and that there is a change in the relative abundance of these transcripts during kidney organogenesis.

Spatial and temporal immunolocalization of GJ proteins in the developing kidney

Stages of the early development of the metanephric kidney are illustrated in Fig. 2.

By immunohistochemistry, the α₁ GJ protein was detectable in the 12-day embryo. At this early stage of kidney development, the ureteric bud has branched only two or three times and patches of immunofluorescence were observed in the mesenchyme near these branches. This immunofluorescence in the mesenchyme was not seen in samples stained with preimmune sera or with the antibodies for other GJ proteins. By day 14, the expression of α₁ was localized to a particular subset of kidney mesenchymal cells: it was detected only on the cells that had formed the S-shaped bodies. The cells expressing α₁ in the 14-day mouse kidney were predominantly found in a subcortical zone containing early S-shaped bodies (Fig. 3A). Higher magnification of one of these regions (Fig. 3B) shows the relationship between the ureter bud and the S-shaped bodies on either side of it. The highest staining intensity of the α₁ antigen was in the crevice furthest from the ureter-derived collecting duct. The epithelial cells of this region are the presumptive glomerular podocytes, but when they mature, the α₁ antigen is concomitantly downregulated; double immunostaining with α₁ antibody and antipodocalyxin reveals no overlapping (Fig. 3C-F).

The β₂ GJ protein could not be detected until around day 17.

No β₂ antigen was detected in day 12 and day 15 kidneys. At day 17, β₂ GJ protein was detected in only a few tubular cells in each section of kidney where dots of fluorescence were observed (Fig. 4). After day 17, more cells expressed this antigen, and it was abundant in the tubules of newborn and adult kidneys.

The cells expressing the β₂ GJ protein were seen only in those areas of the kidney that contained proximal tubules. The identification of the β₂-positive cells with proximal tubules was made by immunofluorescence. First, adjacent sections of 20-day embryonic mouse kidney were stained with antibodies to the β₂ GJ protein and to proximal tubule brush border antigen and antibodies to distal tubule Tamm-Horsfall glycoprotein. These studies demonstrated that the β₂ GJ antigen was located specifically on the brush-borderpositive proximal tubule cells (data not shown). Second, double immunofluorescent staining of sections of 20-day rat embryonic kidney (Fig. 4E, F) localized the β₂ antigen only to those tubules that expressed dipeptidylpeptidase IV, a marker for proximal tubule cells. As the kidney matured, more proximal tubules in each section were positive for the β₂ antigen and, by two weeks after birth, all brush border antigen-positive tubules were also positive for the β₂ antigen. This suggests that, in the kidney, the β₂ GJ protein is a specialized product of the proximal tubule cells.

The β₁ antigen was expressed in the embryonic kidney in the same way as β₂: it appeared late during organogenesis, around day 17 in the mouse, and became localized first in a subset of proximal tubules (Fig. 5A). Towards the end of kidney development, new β₁-positive tubules were detectable until most (if not all) proximal tubules expressed the antigen in the newborn kidney (Fig. 5B).

To explore further the appearance of the GJ antigens and their localization, we used the transfilter technique by which isolated kidney mesenchymes can be experimentally triggered to develop into advanced tubular structures (see Methods). In the rat, the uncommitted mesenchyme is dissected from 13-day embryos and brought into transfilter contact with a fragment of spinal cord. On day 4 after setting up the culture, the α₁ GJ protein could be detected in many tubular structures (Fig. 6A-C). Immunostaining with the β₁ and β₂ GJ antigens yielded invariably negative results up to 7 days in vitro.

In the mouse, the uncommitted nephric mesenchyme is dissected from 11-day embryos and brought into contact with the inducer. We have previously shown that a short transfilter induction “pulse” of 24 to 28 hours is sufficient to program the mesenchyme into epithelial transformation and tubule formation (Wartiovaara et al., 1974; Saxén and Lehtonen, 1978). Using markers for the various segments of the secretory nephron, we have shown that all three main segments, the distal and proximal tubules and the glomerular podocytes, will differentiate after this pulse (Ekblom et al., 1980, 1981; Lehtonen et al., 1983). Here, however, a 22-hour induction was not long enough to trigger the α₁-GJ antigen in the observed, cytokeratin-positive tubules, but an additional 26 hours of transfilter contact yielded positive tubules in mesenchymes subcultivated for 3 days (Fig. 6D). Interestingly, as in vivo, this expression was transient and no α₁-positive cells were detected in mesenchymes cultivated for a total of 7 days.

As in the rat, immunostaining with the β₁ and β₂ GJ antibodies yielded invariably negative results (Fig. 6E) despite the presence of welldeveloped proximal tubules expressing the brush-border antigen (Fig. 6F).

This study examined the expression of mRNA and protein from three gap junction (GJ) genes during murine kidney development. Two non-overlapping patterns of expression were observed. The mRNA for the α₁ GJ protein was already prevalent in the mouse kidney by embryonic day 13, and it continued to be expressed throughout embryonic development. Immunofluorescence microscopy showed that this GJ protein was expressed as early as day 12 in the metanephric kidney, as also recently reported by Yancey et al. (1992). The α₁ GJ protein becomes localized to the transient S-shaped bodies produced by the mesenchymal aggregates.

The expression of the mRNAs and antigens for both β₁ and β₂ GJ proteins follow similar patterns. By S1 nuclease protection assay, this mRNA is detected around day 17 of mouse development, and soon thereafter the protein can be localized to a subset of proximal tubules as verified with the brush border (mouse) and the dipeptidylpeptidase (rat) antigens. By birth, the kidney already expresses an adult pattern wherein all proximal tubules can be decorated by the two gap junction antibodies. This is consistent with the mRNA data, which provides evidence for an increased expression of the mRNAs after their initial appearance during late intrauterine development.

The different expression of the α and β GJ proteins during nephrogenesis suggests that they have different functions. The transient expression of the α₁ GJ protein in vivo coincides with a unique stage of development of the secretory nephron. At this stage, the primitive nephric vesicle undergoes a transformation into the S-shaped body, an event involving cleft formation and invagination followed by differential cytodifferentiation of the main segments (Jokelainen, 1963; Saxén and Wartiovaara, 1966). In the experimental in vitro model system, the appearance of the α₁ GJ antigen coincides temporally with this crucial step of tubulogenesis although the shaping process remains incomplete. In such transfilter cultures where development was followed after a short induction pulse, the antigen was not expressed. This may lend further support to the morphogenetic role of the α₁ GJ protein. Gossens and Unsworth (1972) have provided experimental evidence for a two-step process in tubule induction: an initial, epithelializing stimulus is followed by further interactions between the epithelium of the renal vesicle and the mesenchymal stroma leading to the shaping and coiling of the tubule. This second step might be impaired in our short-term induction pulse experiments in which the uninduced mesenchyme is soon lost in prolonged cultures. When the segments of the secretory nephron have segregated (in vivo and in vitro), the α₁ GJ protein is downregulated. All of this suggests that the protein is involved in a specific stage of development, a suggestion consistent with many previous experimental results and observations on various systems. The findings in the kidney might be analogous to that described by Yancey et al. (1992) in the developing limb bud, where α₁ GJ protein was seen to interconnect the polarizing cells within the apical ectodermal ridge rather than in the epithelial-mesenchymal interphase. The role of gap junctional communication during development has been shown in snail mesoderm formation (de Laat et al., 1980), in mammalian oocyte maturation (Anderson and Albertini, 1976; Gilula et al., 1978), in preimplantation mouse embryos (Lo and Gilula, 1979), and also suggested by observations on mutant Drosophila (Jurnisch et al., 1990). The transient expression of this GJ antigen and its appearance in the avascular, nonfunctional isolated mesenchymes in vitro speak against its functioning in the physiological processes of the newborn or mature kidney.

The expression of the mRNAs and antigens for the β₁ and β₂ GJ proteins follows a pattern different from that above, and it suggests a different function. Both appear rather late during development in vivo, and the proteins can be localized first to a subset of maturing proximal tubules. Around birth, apparently all proximal tubules express both β proteins. In vitro, despite the prolonged culture period and the appearance of well-developed, non-functioning proximal tubules, the two β GJ proteins were not detected. Both findings can be best interpreted as suggesting that the β₁ and β₂ GJ proteins are not directly involved in the process of tubulogenesis, but rather are connected to proximal tubule function.

The genes for these GJ proteins show different patterns of regulation. This has also been observed for other embryonic organs (Nishi et al., 1991). In the uterus, the day prior to parturition is characterized by a dramatic increase in α₁ gene expression and a corresponding decrease in β₂ expression (Risek et al., 1990). However, in heart and liver, no changes in GJ protein gene transcription are seen around parturition. Thus, different GJ proteins can be regulated differently within the same organ, and the same GJ gene can be regulated differently in different organs in the same organism.

This study of GJ proteins and their transcripts demonstrates that different GJ proteins are utilized in different portions of the renal nephron. As a biological model, the developing kidney offers a good opportunity to analyze the formation of physiological compartments during the development of a complex mammalian organ. The major aspects of renal development are well characterized, and the nephron forms an array of distinct anatomical and physiological compartments. The formation of these physiological units may be structured by GJs which are used to compartmentalize information and to inform cells of their neighbors.

We wish to thank Ms Anja Tuomi, Ms Birgitta Tjäder and Ms Ulla Kiiski for their excellent assistance in the isolation, culturing and staining of the embryonic organs. This work was supported by a grant from the Sigrid Jusélius Foundation and NSF grant INT-8922469 (S.G.), grants from the University of Helsinki and the Medical Research Council of the Academy of Finland (E. L.) and by NIH grants GM 37904 (N. B. G.), GM 37907 (N. B. G. and N. M. K.), and EY O6884 (N. M. K).