Evidence for a two-step mechanism operating during in vitro mouse kidney tubulogenesis

Many organs form as the result of developmentally significant interaction between tissues brought into association by morphogenetic movements that occur during embryogenesis.

Kidney differentiation requires an association between ureteric epithelium and metanephrogenic mesenchyme (Fraser, 1950; Gruenwald, 1952). The developmental necessity for this tissue interaction may be illustrated by considering the Sd-strain of mouse. In mutants of this Danforth’s short-tailed strain, failure of the ureteric bud to branch off the Wolffian duct and grow into the metanephrogenic mesenchyme results in the formation of anephrogenic embryos (Gluecksohn-Schoenheimer, 1945).

The metanephrogenic system of the mouse is an excellent model for the tissue culture analysis of an inductive process. Metanephrogenic mesenchyme forms tubules when combined in vitro with ureteric bud, spinal cord, or embryonic brain (Grobstein, 1954; Saxen et al. 1968; Lombard & Grobstein, 1969). The inductive influence of the spinal cord is transmitted across a Millipore membrane which is believed to exclude cellular contact (Grobstein, 1955, 1956; Grobstein & Dalton, 1957). Experiments involving radioactive labelling of the spinal cord with tritiated amino acids have implicated the transfer of large-molecular-weight materials during inductive interaction (Koch & Grobstein, 1963). These types of in vitro tissue separation and recombination experiments have provided evidence to support the concept that cytodifferentiation is controlled by factors extrinsic to the cell (Grobstein, 1964). What is not certain is whether these extrinsic factors, or inducers, are non-specific and merely initiate, or trigger, a permissive differentiation of an already determined organ primordium, or whether they both trigger and direct the subsequent organogenesis.

Certain organ rudiments show a strong differentiative bias when transferred to in vitro culture, and express this bias by differentiating upon exposure to cell-free tissue extracts or ‘nutrient’ factors. For example, somites will undergo chondrogenesis if maintained continuously in the presence of medium supplemented with fetal calf serum and chick embryo extract (Lash, 1968; Ellison & Lash, 1971); and pancreatic epithelium will differentiate if cultured in the presence of a particle fraction from chick embryo extract (Rutter, Wessells & Grobstein, 1964; Rutter et al. 1968). In these particular systems the ‘natural’ inducers, notochord or spinal cord and pancreatic mesenchyme, respectively, are considered not to supply an essential differentiative stimulus to the responding tissue, which differentiates, in a non-directed or ‘permissive’ manner (Holtzer, 1968; Ellison & Lash, 1971). However, this is not universally the case, and many organ systems are considered to respond to the influence of specific inducers, which both initiate and direct the differentiative course (Wolff, 1968). Examples of these latter systems include: the dependence of epidermal differentiation on the nature of the underlying dermis (Saunders, 1958; McLoughlin, 1961; Billingham & Silvers, 1968); differentiation of the liver (Le Douarin, 1967, 1968); and the ability of certain epithelia to show altered morphology when associated with different mesenchymes; gastric epithelium (David, 1967); mammary epithelium (Kratochwil, 1969); thymus epithelium (Auerbach, 1960b); tracheal epithelium (Wessells, 1970); and ureteric epithelium (Bishop-Calame, 1966).

When considering inductive tissue interaction, it appears to be conceptually beneficial to recognize that organogenetic interaction is not an instantaneous event, but a complex, extended process (Grobstein, 1967). The morphogenetic sequence characterizing tubulogenesis in mouse metanephrogenic mesenchyme, cultured in transfilter contact with spinal cord, extends over 6 days (Grobstein, 1956; Saxen et al. 1968), although chemodifferentiation continues until about the tenth day in culture (Koskimies, 1967). A morphologically ‘silent’ period of some 24 h precedes the aggregation of mesenchymal cells into condensates that are apparent after 30 h culture and well established by 48 h. The condensates form whorls of cells which gradually transform into tubule rudiments. S-shaped tubules are abundant by day 4 and these structures elongate and differentiate during the next 2 days of culture to form the thin, convoluted structures characteristic of 6-day tubules. The mechanism by which this series of events is controlled is obscure; however, unpublished observations indicate that limiting the time of inductive contact may lead to the formation of ‘smaller, less complex’ tubules (Auerbach, 1960 a; Grobstein, 1967). We designed a series of experiments to confirm and extend this undocumented report. It appeared to be of particular significance to determine whether these tubules were small because they were arrested at an early developmental stage, and whether an inductive stimulus was obligatory for their complete tubulogenesis.

Using a transfilter method, we established that limiting the period of association of mouse metanephrogenic mesenchyme and brain to 30 h led to the formation of small tubules. These small tubules were similar in size to structures formed after 3 or 4 days of continuous contact with brain. The small tubules completed their tubulogenesis upon association with certain mesenchymal tissues or chick embryo extract. Evidence is presented to support the concept that at least a two-step process is operating during kidney tubulogenesis. The initial step, cell aggregation, requires association of metanephrogenic mesenchyme with inductively active tissue; the second step, tubule elongation and morphogenesis, is dependent upon association of the cellular aggregates with non-induced mesenchymal cells, or factor(s).

Embryos were obtained from randomly mated CF strain Swiss white mice. The day of discovery of a vaginal plug was considered day zero of gestation. Kidney rudiment dissection, salivary mesenchyme dissection, tissue culture methods and composition of the nutrient medium were standard procedures adequately described elsewhere (Lombard & Grobstein, 1969; Unsworth & Grobstein, 1970). The transfilter method was essentially that originally described by Grobstein (1956), except that the modified filter assembly of Auerbach (1960b) was utilized for greater convenience. Brain tissue was clotted on the well side of the filter (Millipore 22 ± 3 μm thick, 0·45 μm pore diameter), with a mixture of chilled chick plasma and embryo extract (2:1). Kidney mesenchyme from one rudiment (except where otherwise indicated in the text) was placed without a clot, transfilter to the brain.

Tissues were routinely maintained in culture medium (Gibco Co.) supplemented with 3% chick embryo extract, although in some experiments explants were supplied with medium supplemented with 20% embryo extract.

The cultures were maintained in a high-humidity incubator gassed with 5% CO₂ in air at 37°C, and their progress was followed daily with a binocular microscope. Photographic recordings were made with a Zeiss Ultraphot II camera microscope.

At the conclusion of the culture period the cultures were washed three times in Tyrode’s salt solution, fixed with 2·5% glutaraldehyde for 30 min and stored in 70% ethanol at 4°C. All tissues were stained with Mayer’s hematoxylin and eosin, and whole mounts were routinely prepared.

The morphological stages characterizing tubulogenesis in metanephrogenic mesenchyme interacting with brain were found to be identical to those described in the introduction, for metanephrogenic mesenchyme in transfilter contact with dorsal half of spinal cord(cf. Grobstein, 1956). Mesenchyme cultured without brain, spread out on the filter and showed no morphological signs of differentiation. The later stages of tubule formation (days 4–6) are illustrated in Figs. 1–3. This control series was established for the purpose of comparing the differentiation obtained when the time of association with brain was varied.

Metanephrogenic mesenchyme, from one rudiment, was cultured transfilter to brain for either 24, 30 or 48 h. The kidney mesenchyme was reincubated, after removal of the brain, for a total of 6 days. Tubules were compared morphologically with the control series and the effects of limited inductive interaction are summarized in Table 1.

When the brain was removed after 24 h transfilter contact, the majority of the kidney mesenchymes failed to differentiate, the cells remained randomly oriented and spread over the filter surface. Extending the period of transfilter interaction by 6 h increased the frequency of tubulogenesis. The structures that differentiated after 30 h transfilter contact with brain were mainly small in size and limited in number (Fig. 4). These incompletely differentiated tubules are referred to as ‘small tubules’ in the text, and may be compared in size with the stage of tubulogenesis observed after 3 or 4 days, in kidney mesenchyme cultured continuously in the presence of brain (Table 1, controls; and Fig. 1). After 30 h of transfilter interaction, over 50% of the positive cultures differentiated only relatively few small tubules, and these small tubules became surrounded by undifferentiated mesenchymal tissue by the sixth day of culture (Fig. 5). Mesenchymal differentiation usually appeared to be normal through day 2 of culture, at which time condensates appeared (Fig. 6 A). However, as incubation continued, following removal of the brain at 30 h, the small tubules that formed ‘regressed’ during the next 2 or 3 days in culture (Fig. 6B).

The most noticeable result of restricting the period of transfilter interaction with brain to 30 h was a high incidence of small tubule formation (Table 1). However, many mesenchymes failed to differentiate and the incidence of tubule elongation was no greater than that observed after 24 h transfilter interaction.

The incidence of tubule elongation was increased to the level observed in control cultures (Table 1, controls) by extending the period of transfilter interaction to 48 h. The majority of the cultures differentiated to the coiled, elongated tubule stage, and the incidence of small tubule formation was not significantly above control values. It was concluded that 48 h transfilter interaction with brain was the minimum period of time necessary to ensure complete tubulogenesis in a single metanephrogenic mesenchyme.

It was established in the previous section that limited heterotypic interaction resulted in the incomplete differentiation of kidney tubules. Small tubules predominantly differentiated when the period of transfilter interaction with brain was restricted to 30 h. An attempt was therefore made to determine whether mesenchymal tissue could substitute for the brain in promoting tubule elongation.

Kidney mesenchyme from one rudiment was cultured transfilter to brain for either 24 or 30 h. The brain was removed and mesenchyme was added in direct contact with the induced kidney mesenchyme. Either kidney mesenchyme from two freshly dissected rudiments or condensed capsular mesenchyme from two 13-day salivary rudiments was added, according to Table 2.

Conditions for tubulogenesis were improved by the addition of either kidney or salivary mesenchyme, to kidney mesenchyme that had received short periods of exposure to brain.

Salivary mesenchyme proved to be more effective in promoting tubule differentiation than was kidney mesenchyme, and delaying the addition of salivary mesenchyme for 42 h after the removal of the transfilter brain did not significantly alter its tubule elongating ability. The morphological appearance of the tubules, formed under the influence of salivary mesenchyme, was consistently different from day-6 controls. Although a certain degree of elongation occurred, the tubules were about five cells in diameter (Fig. 7), as compared with the tubules of 3-cell diameter of day-6 controls (Fig. 3). Extending the culture to 8 days produced neither thinner tubules nor greater elongation, the tubules remained bulbous, unlike controls.

The addition of kidney mesenchyme significantly enhanced tubule differentiation. However, the number of differentiated tubules per culture and the total differentiated area per culture (Fig. 8) was generally considerably less than that observed in mesenchyme cultured continuously with brain for 6 days (Fig 3). Also, the elongated tubules that formed in this series of experiments usually resembled the 5-day controls (Fig. 2), even when the cultured period was extended to 8 days.

Since both kidney mesenchyme and salivary mesenchyme enhanced tubulogenesis, a possible growth-stimulating effect was postulated. Several years ago a chick-embryo fraction extracted from predominantly mesenchymal tissue was reported to promote differentiation and growth of embryonic pancreatic epithelia in vitro (Rutter et al. 1964). More recently it was reported that this extract may act by promoting DNA synthesis and cell division (Rutter et al. 1968). Chick embryo extract alone does not promote tubule differentiation in kidney mesenchyme (Rutter et al. 1964); however, it was postulated that the extract might supply ‘growth’ or ‘environmental’ factor(s) essential for tubulogenesis.

Cultures were set up in which kidney mesenchyme was incubated transfilter to brain tissue, for either 24 or 30 h. These cultures were fed standard culture medium, supplemented with 3% embryo extract (E.E.), during the period of transfilter association. After the brain was removed, this medium was replaced by medium enriched with 20% E.E. The results of addition of 20% E.E. are summarized in Table 3.

Addition of 20% E.E. after 24 h transfilter culture with brain caused no significant improvement in tubulogenesis. Transfilter culturing for 30 h, followed by the addition of 20% E.E. at the time of brain removal, reduced the number of mesenchymes differentiating to the small tubule stage and significantly increased the frequency of tubule elongation. The elongated tubules (Fig. 9) morphologically resembled 5- and 6-day control tubules (Figs. 2, 3) but the number of tubules per culture was less than in the controls. Removing the brain at 30 h and delaying the addition of E.E. until 72 h decreased the incidence of tubule elongation and supported the differentiation of relatively few elongated tubules per positive culture (Fig. 10).

The present study demonstrated that conditions necessary for tubule elongation were created by the addition of embryonic mesenchymal tissues, or factor, at the time that the transfilter inductive brain was removed from culture. This suggests the possibility that the concentration of nutritional or ‘environmental’ factors may be critical for the later stages of tubule differentiation, and such factors may be supplied if a larger mass of kidney mesenchymal tissue were initially present. Therefore mesenchyme dissected from three kidney rudiments was placed transfilter to brain, instead of the single rudiments used in the previously reported experiments. The heterotypic inducer was removed from the system after either 24 or 30 h, and the stage of tubule differentiation recorded at the end of the 6-day culture period (Table 4).

It is clear that only 30 h transfilter interaction with brain was sufficient to ensure complete tubulogenesis in three kidney mesenchymes; whereas, a single kidney mesenchyme required 48 h transfilter interaction to achieve a similar incidence of tubule elongation. The resulting elongated 6-day tubules (Fig. 11) were long, thin (three cells in diameter), noticeably convoluted, and abundant; morphologically resembling 6-day control cultures (Fig. 3). Interestingly, the temporal sequence of tubulogenesis in three kidney mesenchymes cultured transfilter to brain was indistinguishable from the control series in which one mesenchyme was used.

Our results confirm unpublished observations, indicating that if the period of heterotypic interaction is restricted, tubulogenesis is incomplete (cited in Auerbach, 1960a; Grobstein, 1967). However, contrary to these earlier reports we found that 30 h inductive interaction was not adequate to ensure stability of tubules in the mesenchyme. The predominantly small tubules that formed as a result of 30 h transfilter contact with brain were unstable structures that regressed during the 6-day culture period. Stability of tubules required 48 h of transfilter interaction. This discrepancy may be explained by the greater mesenchymal tissue mass and the higher embryo extract concentration used by the earlier workers (cf. Auerbach, 1960 a); we have now shown that both these factors are of critical importance in tubulogenesis. The progressive stabilization of tubules as differentiation proceeds is also reflected in the results of tubule dissociation and reaggregation studies (Auerbach & Grobstein, 1958; Auerbach, 1960a).

It is clear from the present study that the condensates formed after 30 h transfilter interaction generally differentiate only to the small tubule stage in the absence of additional mesenchyme or factor(s). A small number of explants undergo elongation after 30 h transfilter interaction, and this is probably due to the fact that condensate formation is not a strictly synchronous event (Koskimies, 1967). The cells that are first induced may receive a quantitatively sufficient stimulus from the brain, in 30 h, to differentiate fully. This differen-tiative process may be contrasted with pancreatic epithelium, where the association of cells into a ‘package’, the proacinus, requires a certain minimum time of inductive association. Once this package is formed and stabilized by association with inductive mesenchyme, secretion of zymogen granules occurs. Removal of the inducer prevents new proacinar formation, and the number of acini differentiating is directly proportional to the duration of the inductive association (Grobstein, 1962), i.e. proacini, once formed, complete their differentiation.

The inductive mechanisms operating during kidney differentiation are but poorly understood. In vitro observations of cultured metanephrogenic primordia suggest that only the tips of the ureteric bud are inductively active (Wolff, 1968; Saxen & Kohonen, 1969). Time-lapse photography of entire rudiments, together with tissue separation and recombination experiments (Saxen & Wartiovaara, 1966), encouraged the concept that the initial cellular condensation (or aggregation) reaction is induced by the tips of the ureteric bud. Tubule differentiation was considered to follow autonomously. However, the results presented in this paper suggest that the capacity for autonomous differentiation from condensate stage to elongated, coiled tubule stage is critically dependent upon the mass of available non-induced mesenchymal tissue, or factor(s).

The ability of various mesenchymal tissues or embryo extract to support tubule elongation is of considerable interest. These observations indicate that the fastidiousness of tissue requirement decreases with increasing degree of differentiation. This phenomenon appears to be a basic property of organogenetic tissue interaction, and is reflected in the varying degrees of specificity shown by different epithelia in their mesenchymal requirements for differentiation (Grobstein, 1967).

The fact that a cell-free extract can support later stages of tubulogenesis suggests that mesenchymal factors may play an important role during tubule elongation. However, embryo extract cannot mimick the inductive event responsible for condensate formation. This indicates that specificity for tubule induction may reside in the chemistry of the inducer, a concept of importance when considering mechanisms of kidney tubulogenesis.

A mesenchymal contribution appears to be obligatory for complete tubulogenesis, indicating that induction is not a one-time triggering event. Moreover, condensates exposed to salivary mesenchyme differentiate morphologically different tubules (larger in diameter, less convoluted), from condensates receiving either embryo extract or kidney mesenchyme. The formation of larger-diameter tubules in the presence of salivary mesenchyme may, in part, be due to the reported growth-promoting ability of salivary mesenchyme (Attardi, Monticalini & Wenger, 1965). Salivary mesenchyme has also been shown to possess tubule-inducing properties; however, the time course of tubulogenesis is extended by several days when salivary mesenchyme is the inducing tissue (Unsworth & Grobstein, 1970). In the present study no delay in the elongation response was observed when salivary mesenchyme was added to the small tubules, thereby suggesting that the induction properties of salivary mesenchyme may be separate from the elongation properties.

The ‘mesenchymal effect’, observed in the present study, is corroborated by recent studies indicating that induced kidney mesenchymal cells, which remain on the filter after the non-induced mesenchyme is removed by stripping, require the addition of mesenchyme for their subsequent tubulogenesis (Saxen, 1970). The added mesenchyme need not be species-specific, as heterologous chick mesenchyme supports elongation, but the added cells are not incorporated into the differentiating tubules (Saxen, 1970). Other studies also indicate that tubule elongation is not achieved by incorporating surrounding mesenchymal cells into the pretubular structures (Saxen & Sakselsa, 1971). Rapid mitotic growth of the induced cells is a logical alternative mechanism to explain tubule elongation, and this mitotic activity may be promoted by embryo extract (Rutter et al. 1968). Further work is required to establish whether the molecular identity and mechanisms of action of the extract is identical in both tubule elongation and pancreatic acinar formation.

Recently, relatively complex differentiation has been achieved by culturing embryonic cells in improved media, such as Ham’s F12 with fetal calf serum (Ellison & Lash, 1971). This raises the question of whether improved methods of cultivation would allow the complete expression of the tubulogenic properties of kidney mesenchyme, making the inductive event unnecessary. This concept receives some support from the report that kidney mesenchyme, transplanted to the anterior chamber of the eye, can undergo tubulogenesis (Grobstein & Parker, 1958). To date, all efforts to achieve tubulogenesis in isolated kidney mesenchyme cultured in improved media have proved unsuccessful (Unsworth, unpublished observations), and increasing the tissue mass is alone unable to promote tubulogenesis (Unsworth & Grobstein, 1970).

Culturing three kidney mesenchymes transfilter to brain for 30 h ensured an incidence of tubule elongation equivalent to that observed in a single mesenchyme cultured continuously in the presence of brain. However, the addition of two kidney mesenchymes at the time of brain removal, although creating a similar tissue mass, resulted in only limited improvement in elongation. This effect might be explained if there were an enhanced ability of mesenchyme retained in culture for 30 h, to protect the cells from loss of tubule-elongating factor(s) due to diffusion. At the present time we have no results that would support this hypothesis. A possible alternative explanation would be to assume that the brain not only induces mesenchymal cell aggregation, but also stimulates the synthesis of tubule elongation factor(s) in the non-induced mesenchymal cells. Further work is required to distinguish between these two possible mechanisms of tubule elongation.

Support for a two-step process, operating during kidney tubulogenesis, is provided by several observations. Isolated induced cells are morphologically inactive in the presence of inductively active tissue, and require the addition of mesenchyme to complete tubulogenesis (Saxen, 1970). The morphology of the differentiated tubules can be altered by the type of mesenchyme with which the induced cells are associated, and limiting the duration of inductive contact allows differentiation to only a small tubule stage, which is unstable. It therefore would appear that the inductive interaction that initiates tubulogenesis is followed by interaction between the induced mesenchymal cells and the surrounding non-induced mesenchymal cells. This latter reaction both stabilizes and elongates the partially differentiated tubules. This may be considered as an example of continued inductive tissue interaction, a phenomenon that has been postulated to be of importance in the development and maintenance of many organs (Auerbach, 1964; Tarin, 1972).

In summary, kidney induction is clearly not a one-step triggering process. Based on the available experimental evidence, it appears appropriate to offer an hypothesis for the mechanisms operating during kidney tubulogenesis. Transfilter contact with brain results in the formation of cellular aggregates or condensates. These aggregates require continued association with mesenchymal tissue (or factor) to autonomously differentiate into elongated, coiled tubules, characteristic of the fully differentiated kidney. Tubule elongation can be supported by a cell-free extract from chick embryo, suggesting that the factor is not species-specific. It appears to be mechanistically beneficial to consider kidney tubulogenesis as a two-step process. The first step, cell aggregation and pretubular formation; the second step, tubule elongation. Both steps may be either directly, or indirectly, under the control of the inductively active tissue, allowing the possibility that the inductive tissue interaction may be responsible for both initiating and directing the course of tubulogenesis.

We wish to express our sincere thanks to Drs Robert Auerbach and Alapati Krishnakumaran for their critical comments during the preparation of this manuscript. The data in this paper are taken from a thesis submitted in partial fulfillment of the requirements for the Master of Science degree at Marquette University. This investigation was supported by grants from the Marquette University Committee on Research, and the American Cancer Society, Milwaukee Branch.