The wingless product is required for cell proliferation in the Malpighian tubule anlage of Drosophila melanogaster

The development of Malpighian tubules, a simple excretory epithelium associated with the hindgut in insects, results from a sequence of cell activities which includes the specification of primordial cells, their proliferation, rearrangement and final cell differentiation. The coordination of these processes, which are common to the generation of many epithelia (see Fleming, 1992), depends on intercellular communication. In the epidermis, the cellular interactions that regulate pattern formation occur concurrently with proliferation and morphogenetic movements of the cells (Campos-Ortega and Hartenstein, 1985; Foe, 1989). Thus the study of regulatory cell interactions is complicated by the interplay between these processes. In the Malpighian tubules, however, the patterning of the epithelium takes place in stages, thus allowing an analysis of the signalling events involved in the regulation of each step in the sequence (Skaer, 1992a). Further study of one stage has shown that proliferation of the primordial cells is controlled by an interaction between an identified cell (the tip cell) and its neighbours, since ablation of the tip cell prevents further cell division in the tubule of which it is a part (Skaer, 1989).

One approach towards a fuller understanding of the nature of this interaction is the isolation of mutants that dis-rupt intercellular signalling. The gene wingless is a member of the segment polarity class and encodes a secreted prod-uct (Rijsewick et al., 1987), which is thought to act as a signal for the establishment of cell identity in the epider-mis (Baker, 1988; Martinez Arias et al., 1988; Bejsovec and Martinez Arias, 1991; reviewed in Ingham, 1991). Here we demonstrate that mutants at the wingless locus show pro-found abnormalities in the development of the Malpighian tubules, including the absence of cell division in the primordia. This phenotype suggests that the wg product might also play a role in the signalling system underlying the regulation of cell division by the tip cell.

In this paper, we explore the temporal and spatial requirements for wingless to support normal embryogenesis of the Malpighian tubules. We demonstrate that wingless is required for all postblastodermal divisions in the proctodeum and Malpighian tubules and show that this requirement corresponds with the expression of wg in the tubule primordia and in a subset of the cells later in development. The importance of the wg product in regulating cell proliferation in the tubules is underlined by the demonstration that extra divisions can be induced in the tubules by extending the expression of wg under the control of a heat-shock promoter. These data allow us to analyse the possible role of the wg protein in the cell interactions that regulate the generation of Malpighian tubules.

The following strains were used: wild type, Oregon R; two enhancer trap lines, one in which the P element is regulated by the wg promoter (Kassis et al., 1992) and the other by the fasc II promoter (Grenningloh et al., 1991; A31, Ghysen and O’Kane, 1989), a null allele of wg, wg^cx4b pr/CyO (Baker, 1987); a ts allele of wg, wg^IL114cn bw sp/CyO (Nüsslein-Volhard et al., 1984), encoding a protein defective in secretion (Gonzalez et al., 1991); a strain in which wg is expressed under the control of a heat-shock promoter, HS-wg/TM3, hb lac z (Nordermeer et al., 1992); and stg^7B69/TM3 (Jürgens et al., 1984).

Eggs were collected on agar/apple juice plates (Wieschaus and Nüsslein-Volhard, 1986) and kept at 25°C. They were dechorion-ated in bleach and embryos were staged from gastrulation.

In situ hybridisation was carried out on sectioned embryos using ³⁵S probes from a wingless cDNA as described in Baker (1987) and Martinez Arias et al. (1988).

Immunostaining for wg was performed using two antibodies (van den Heuvel et al., 1989 and Gonzalez et al., 1991), which gave similar results. Antibodies against β-galactosidase were a gift of C. Doe and immunostaining of β-galactosidase as well as incu-bation for LacZ activity followed standard protocols (Ashburner, 1989).

Flat preparations of embryos were made as described in Bate (1990). Toluidine blue staining is described in Truman and Bate (1988). Incubation in BrdU and subsequent staining of embryos was carried out as previously described (Skaer, 1989). Counts of Malpighian tubule cell numbers were made from toluidine-blue-stained flat preparations of embryos, which were selected as gas-trulae and aged at 25°C before dissection.

Embryos laid overnight by wg^IL114cn bw sp/CyO flies were dechorionated, gastrulae selected and maintained on damp blot-ting paper for known periods of time before shifting into Ringer (Roberts, 1986) prewarmed at 25°C. All operations prior to the temperature shift were carried out at 17.5°C. Temperature-shifted embryos were left to develop at 25°C for a further 7-8 hours before dissection to make flat preparations and staining with toluidine blue as described above. In each experiment, some embryos were allowed to hatch to check for the viability of heterozygous and homozygous balancer embryos.

Downshift experiments were carried out in exactly the same way except that the laying of eggs and preparation of embryos was carried out at 25°C before shifting to 17.5°C and the period of incubation before subsequent dissection was 18 –25 hours. As before some embryos were left to hatch as controls.

Embryogenesis at the permissive temperature (17.5°C) takes twice as long as at 25°C (Wieschaus and Nüsslein-Volhard, 1986). In the description of temperature-shift experiments, all develop-mental times are given, for the sake of comparison, as those for 25°C.

The numbers of cells in each tubule were counted using a 63× objective on a Zeiss Standard microscope.

A heterozygous HS-wg/TM3, hb lac z stock (made by P. John-ston from a stock donated by G. Struhl) was used and the homozy-gous HS-wg embryos were identified by their failure to stain for LacZ activity. Dechorionated embryos from an overnight lay at 25°C were selected at the beginning of germ band shortening (7.75-8 hours AEL). They were shifted from room temperature (22-25°C) into prewarmed PBS at 36°C and incubated at that tem-perature for 20 minutes every 2 hours, the intervening periods being spent at room temperature. This treatment was repeated 3 times. The heat-shock regime was found to slow development so that after 6 hours, embryos had reached stage 14-15 (approx. 12 hours AEL). Some embryos were dissected just before cuticle deposition (14-15 hours AEL) and stained for LacZ activity (Ash-burner, 1989) in order to identify those homozygous for HS-wg. Those embryos that stained for LacZ activity were discarded as subsequent staining with toluidine blue can obscure the LacZ reaction product. The remaining preparations were then stained with toluidine blue, as described above, in order to count the number of cells in their Malpighian tubules. Counts were made separately for the anterior and posterior pairs of tubules. The remaining embryos were left to complete embryogenesis in order to examine their cuticle morphology to confirm the effectiveness of the heat-shock regime. Cuticle preparations were made as described in van der Meer (1977).

Control embryos were genetically identical to the experimental stock except that they did not carry the heat-shock construct or the hb lac z insert.

The Malpighian tubules arise as four separate primordia from the posterior ectodermal intucking of the gut, the proc-todeum. Immunostaining of embryos from the enhancer trap line A31 (Ghysen and O’Kane, 1989, in which LacZ is regulated by the fasc II promoter (Grenningloh et al., 1991)), reveals β-galactosidase in cells of the Malpighian tubules throughout their development and can therefore be used to highlight the Malpighian tubules during embryogenesis (Fig. 1). Staining can first be observed in the extended germ band, initially in a subset of cells on the ventral side of the proctodeum (Fig. 1A) and by 5 hours after egg laying (AEL) (stage 10; Campos-Ortega and Hartenstein, 1985) as two groups of cells spreading out from the ventral midline of the proctodeum. From 5.30 hours AEL two protruber-ances can be seen pushing out from the proctodeum (Campos-Oretega and Hartenstein, 1985), which subdivide immediately after they evert so that by 5.75 –6 hours AEL cells labelled for β-galactosidase can be resolved as four separate buds: the Malpighian tubule primordia (Fig. 1B). The number of cells in each primordium increases by cell division during the next 4 hours (Figs 1C, 2). Cells moving through the cell cycle can be labelled during S phase by the incorporation of BrdU, a thymidine analogue, which can then be stained immunocytochemically. Staining specific to the Malpighian tubules rather than the hindgut can be identified only after the tubule primordia have everted, approx. 6 –6.5 hours AEL. From this time until the end of the extended germ band stage (7.75 hours AEL), cells in S phase are restricted to a lateral patch on the posterior side of each Malpighian tubule (Fig. 3C).

At 7.75 hours AEL (early in stage 12), just as germ band retraction begins, a prominent cell becomes apparent at the tip of each tubule and remains in this characteristic position as the tubule grows by further division (Fig. 1C). By 8 hours AEL, cells throughout the length of the developing tubules incorporate BrdU. The number of cells in S phase diminishes soon after this, cells ceasing to cycle earlier in the proximal region of the tubules, while more dis-tally, cells closer to the tip cell continue to incorporate BrdU (Skaer, 1989). This pattern of staining is seen earlier in two of the tubules, at 8.25 –8.5 hours AEL, while cells in the other pair continue to cycle for a further 30 minutes. Cell division follows S phase so that the mature number of cells is reached in all four tubules by the end of stage 13 (10 –10.5 hours AEL; Fig. 2); the cells at the distal end of the tubules being the last to divide. The difference in the cycling pattern of the two pairs of tubules is reflected in the final cell numbers; the posterior pair contain approx. 105 cells and the anterior pair approx. 140 cells (Fig. 2 and Janning et al., 1986). Thus it appears that the greater number of cells in the anterior tubules results from an extension of the period over which cells divide in the anterior compared with the posterior tubules (see Fig. 2). Following cell proliferation, the cells enter the first of a series of endomitotic cycles, which spread through the tubules from the proximal to the distal cells and result in a progressive increase in polyteny and consequently in cell size (Poul-son, 1950; Maddrell et al., 1985; Smith and Orr-Weaver, 1991). From 10 –11 hours AEL (early stage 14), rearrange-ment of the cells produces rapid elongation of the tubules, with the tip cells remaining in their prominent position (Fig. 1D).

The first easily identified sign of cell differentiation in the tubules is the appearance during stage 17 (from 16-17 hours AEL) of the excretory product, uric acid, in the lumen of the posterior pair of tubules before hatching (Fig. 3A) and a little later in the anterior tubules, in the first instar larva.

We were not able to use the P element transformant A31 to examine tubule morphology in wingless mutant embryos, as there is no expression of β-galactosidase in the tubules of wg embryos carrying the A31 construct (Fig. 3F). How-ever, wg mutant embryos dissected as flat preparations (Bate, 1990) reveal two major defects in Malpighian tubule development: two instead of four primordia appear and these two primordia remain as clusters of about 20 –25 small cells (Fig. 3F). Later in embryogenesis, these primordia elongate to produce two rudimentary tubules with a prominent cell at the distal tip. Small patches of uric acid appear in the mutant Malpighian tubules towards the end of embryogenesis (Fig. 3B).

Labelling embryos mutant for wg at the beginning of stage 12 (7.5-8 hours AEL) with a 30 minute pulse of BrdU reveals that cells in the Malpighian tubule primordia are not replicating their DNA (cf. Fig. 3C and D). Further, in situ hybridisation using a probe for string, a gene whose product is required for entry into mitosis and accumulates in cells as they pass through G₂ (Edgar and O’Farrell, 1989), shows that in wg mutants expression of stg in the proctodeum is absent (data not shown), indicating that there is no cell division in this region of the mutant embryo after gastrulation. In line with this observation, there are fewer cells in the hindgut of embryos mutant for wg (120-140 cf. 820 (Harbecke and Janning, 1989)).

Furthermore, the phenotype of the Malpighian tubules in embryos mutant for stg, in which all postblastodermal cell divisions fail (Edgar and O’Farrell, 1989), resembles their appearance in wg mutants: the tubules remain as clusters of about 20 cells (Fig. 3E). However, in mutants for stg, all four tubule primordia appear. As in embryos mutant for wg, the rudimentary tubules elongate with an apparent tip cell at the distal end and uric acid is produced later in embryogenesis.

In summary, it appears that wingless function is required for two distinct processes in tubule development; for the cellular activities underlying normal eversion of the four primordia from the proctodeum and for cell proliferation in the tubules.

Localisation of wg mRNA by in situ hybridisation on sections of Drosophila embryos shows that expression in the proctodeum appears initially during extension of the germ band, on the ventral side (which at this stage of development lies closer to the dorsal side of the embryo, Fig. 4A) and can be clearly observed in the primordia of the Malpighian tubules as they evert from the proctodeum (Fig. 4C). As the tubules grow, wg expression is associated with the posterior side of each tubule (Fig. 4E), a pattern that persists until about 8 hours AEL.

The wg protein can be detected in the Malpighian tubules, although at a lower level than in the epidermis, CNS or in other parts of the gut. It is present in the proctodeum of early extended germ band embryos (4.5 hours AEL) but, like the RNA, is restricted to the ventral side. As the tubules evert from the proctodeum, all the cells in the primordia contain the protein, with a greater concentration on the luminal side of the cells (Fig. 4B). After eversion, this luminal pattern of staining persists but the protein is found only in cells on the posterior side of the tubules, extending along the whole length of each tubule (Fig. 4D,F). Like the RNA, wg protein disappears from one pair of tubules approx. 8.5 hours AEL, persisting in the other pair for a further 30 minutes. After approximately 9 hours, AEL wg protein cannot be found in the Malpighian tubules.

Further details of the expression of wg in the Malpighian tubules can be deduced by preparing flat preparations of embryos from a P-element insertion line in which β-galac-tosidase is expressed under the control of the wg promoter (Kassis et al., 1992). The pattern revealed by staining embryos from this line shows that, while all the Malpighian tubule cells express β-galactosidase as they evert (Fig. 5A), the staining becomes faint on the anterior side of the tubules (Fig. 5B) and in the tip cells (Fig. 5C) during germ band shortening. Taking into account that in such constructs the perdurance of β-galactosidase is often longer than that of the normal gene products (Hiromi and Gehring, 1987), this pattern of staining suggests that the expression of both wg and β-galactosidase becomes restricted to cells on the posterior side of the tubules as development proceeds and, further, that wg expression in the tip cells stops earlier than in this posterior group of cells.

In order to establish the period during embryogenesis when functional wg product is required for normal development of the Malpighian tubules, we have used the temperature-sensitive (ts) allele wg^IL114 (Nüsslein-Volhard et al., 1984). Embryos homozygous for this allele exhibit a completely wild-type phenotype when raised at 17.5°C, producing viable larvae, whereas at 25°C they show a wg null pheno-type (Baker, 1988; Bejsovec and Martinez Arias, 1991). Thus the Malpighian tubules of embryos raised at 25°C appear as two clumps of 20 –25 small cells protruding from the hindgut.

Experiments in which embryos were raised at the per-missive temperature until timed stages of development before transfer to the restrictive temperature (upshift; see Materials and Methods), show that the later the shift, the milder the resulting phenotype in the Malpighian tubules (Fig. 6). Shifting to the restrictive temperature prior to 3.75 hours AEL results in the null wg phenotype. In mutant embryos shifted after 3.75 hours, all four primordia are present. However, shifting to the restrictive temperature between 3.75 and 7.5 hours AEL limits the final number of cells in the tubules. As the shift in temperature is performed later in development, the number of cells in each tubule approaches closer to the wild-type number. In embryos shifted after 7.5 hours AEL, the number of cells approxi-mates to the wild type.

Downshift experiments, in which embryos are raised at the restrictive temperature before being shifted to the per-missive temperature at timed stages in development, show the reverse of the upshift experiments: the later the tem-perature shift, the more extreme the wg phenotype (Fig. 6). Embryos shifted at 3.75 hours AEL, or earlier, show a wild-type Malpighian tubule morphology. Shifting after this stage results in only partial rescue of the mutant phenotype, so that, although four tubules evert, there is a reduction in the final number of cells in the tubules; the later the time of the shift, the smaller the number of cells. Embryos shifted early after gastrulation and through germ band extension have four Malpighian tubules. Embryos with two rather than four tubules are found only when the tempera-ture shift is later than 6.5 hours AEL.

The effects of shifting temperature on the levels and dis-tribution of the wg product are not immediate. The delay in loss of wg function after raising the temperature depends on the half life of the functional protein that remains. Studies using inhibitors to protein synthesis show that during embryogenesis the half life of the wg product is approx. 20 minutes (A. M. A. and F. Gonzalez, unpublished observations). The gain of function can be studied by anti-body staining: the normal distribution of wg product has been shown to be re-established 20-30 minutes after embryos are shifted down to the permissive temperature (Bejsovec and Martinez-Arias, 1991). Taking this informa-tion into account, upshift experiments indicate that the period during which wg function is required for normal tubule development is until at least 8 hours AEL. Similarly, downshift experiments show that rescue of the phenotype is not discernible if the production of a functional wg product is initiated later than 7.5 –8 hours AEL.

The temperature shifts of wg^IL114 embryos suggest that the wg protein is involved in the regulation of cell division in the Malpighian tubules. To test this hypothesis further, we have made use of a wg gene under heat-shock control (Nor-dermeer et al., 1992) to express the wg product in all Malpighian tubule cells at a time when, in the wild type,these cells have ceased to express wg. The appearance of supernumerary cells in the tubules of these embryos would confirm a role for wg in the regulation of cell division.

The timing of heat shock was arranged so that embryos at the beginning of germ band shortening (7.75-8 hours AEL) were induced to express wg continuously for a further 6 hours. Since the time taken to induce the expression of wg under the heat-shock promoter is approx. 20 minutes (Nordermeer et al., 1992), embryos actually expressed wg ectopically from 8.3 hours AEL. This was confirmed by staining heat-shocked embryos immunocytochemically (data not shown, see Nordermeer et al., 1992).

Table 1 shows the number of Malpighian tubule cells in embryos carrying the heat-shock construct and in a control stock. In embryos carrying the heat-shock construct, the anterior Malpighian tubules are made up of significantly more cells than in control embryos (P<0.01%) or in the wild type (Fig. 2 and Janning et al., 1986). In contrast, the number of cells in the posterior tubules shows no signifi-cant difference between heat-shock, control (Table 1) or wild-type stocks (Fig. 2 and Janning et al., 1986).

During normal development, the Malpighian tubules arise from the proctodeum with which they share a blastodermal anlage (Technau and Campos-Ortega, 1985), so that the early postblastodermal divisions contribute cells both to the Malpighian tubules and to the hindgut. The tubule primordia evert from the proctodeum, initially as two outpocketings which very soon subdivide to produce four buds of cells, the tubule primordia (Campos-Ortega and Harten-stein, 1985). The absence of wingless interferes with this process so that two rather than four primordia appear. We have shown that cells in the Malpighian tubule anlage do not proliferate in the absence of wg. However, the failure of cell division per se is insufficient to explain the wg phenotype, since four primordia appear in embryos mutant for stg, in which there are no postblastodermal divisions. Instead the loss of two tubules in the absence of wg must result from the mis-specification of cells or from abnormalities in tubule eversion.

Experiments with the ts allele (wg^IL114) show that wg is required before 4 hours AEL to form four Malpighian tubules. Interestingly, restoring wg function after this time can rescue the appearance of four Malpighian tubules (Fig. 6). Since the allocation of cells to a tubule fate must occur between 3 hours 10 minutes (Technau and Campos-Ortega, 1985) and tubule eversion at 5 hours 30 minutes, these results indicate that wg could be involved initially either in the allocation of cells to make one pair of tubules or in the subdivision of the initial protruberances. Since rescue of the four Malpighian tubules is possible over a period that extends for an hour beyond the time when tubule eversion can first be seen in wild-type embryos, it is more likely that wg is required to mediate the changes in cell adhesion, shape or motility that normally subdivide the initial primordia.

A further striking characteristic of the Malpighian tubules of embryos lacking wg function is that the cells of the tubule primordia fail to proliferate. In line with these observations, wg is expressed in the proctodeum, in the everting tubule primordia and in the growing tubules, in a pattern that is consistent with a requirement in or close to cycling cells. Further evidence that wg is involved in promoting cell divi-sion is provided by the demonstration that over-expression of wg, under the control of a heat-shock promoter, can result in extra divisions to produce supernumerary cells in the tubules.

Temperature-shift experiments indicate that wg is continuously required for cell proliferation from shortly after gastrulation until 8 hours AEL and that rescue of the mutant phenotype is possible over a similar period of embryogenesis (Fig. 6). This represents a minimum estimate for the period of wg requirement as the normal variation in the final number of cells in the tubules (s.d. ±20) means that the division of the last cells to proliferate (15 cells in the last hour; see Fig. 2) cannot be discerned simply by counting cells. However, as shown in Fig, 6, the periods of both wg requirement (8 hours) and expression (to 9 hours AEL) are shorter than the observed period of cell division in the tubules (to 10.25 hours AEL; see also Fig. 2). This suggests that wg activity is required early in the cell cycle to stimulate the next cell division.

Temperature-shift experiments also show that the effects of removing or restoring wg function in the tubules depend on the age of the embryo at the time of the shift. A change in wg expression has a more profound effect on the pheno-type early after gastrulation, when many or all of the cells divide in the proctodeum in the wild-type embryo. Later, when only a subset of tubule cells divides during normal development (Fig. 2 and Janning et al., 1986), the removal or restoration of functional wg has only a slight effect on the final number of cells in the tubules (Fig. 6). Thus, even in the absence of functional wingless protein, with the con-sequent arrest of cell cycling in the proctodeum and the tubules, the normal programme for cell division in these tissues is retained. The restoration of functional wingless protein triggers the number of cell divisions appropriate to the age of the embryo.

In a similar way, the normal programme of cell proliferation limits the effects of extending the expression of wg. Although expression beyond 8 hours AEL results in super-numerary cells in the anterior tubules, there are no extra divisions in the posterior tubules. Even in the anterior pair, the production of supernumerary cells is limited; only approx. 26 out of a possible approx. 138 cells (Table 1) enter an additional round of division, indicating that fac-tors other than the absence of wingless protein restrict the ability of cells to continue dividing. In normal development, cell proliferation stops earlier in one pair of tubules than the other. A similar discrepancy in response to the extended expression of wg suggests that there is a limited period after normal proliferation is complete when the presence of wingless protein promotes entry into a further division cycle.

These observations are in line with those of Edgar and O’Farrell (1990), who found that the ectopic expression of stg under the control of a heat-shock promoter triggered mitosis in all cells during cycles 14-16. However, following mitosis 16, cells became unresponsive to the presence of string, suggesting that other regulatory factors become limiting at this stage of development.

After eversion of the tubule primordia, cell division in the tubules depends not only on the normal expression of wg but also on the presence of a tip cell in each tubule (Skaer, 1989). Although these cells become apparent some time after the primordia form in Drosophila, they are evident from the time that the tubules evert in another species, Rhodnius prolixus, in which their role in regulating cell division in the tubules has also been demonstrated (Skaer, 1992b). One model suggested for the activity of the tip cell is that it secretes a signal, stimulating cells close to it to cycle and divide (Skaer, 1989). Several lines of evidence suggest that wingless is secreted from cells expressing it (Morata and Lawrence, 1977; Baker, 1987; Rijsewijk et al., 1987; van den Heuvel et al., 1989; Gonzalez et al., 1991) raising the possibility that wg is secreted by the tip cells and regulates division in neighbouring cells in the tubules. However, two aspects of the expression pattern of wg argue against this suggestion. Firstly, wg expression can be shown to decline in the tip cells before other tubule cells and, sec-ondly, the protein is found in many cells of the tubules in a pattern indicating that wg might be required in or close to the cells that are moving through the cell cycle. Fur-thermore, there is a difference in the effects of removing either the tip cell or wg function on the final number of cells in the Malpighian tubules (tip cell ablation at 7.75 hours AEL, 75±4 cells (Skaer, 1989); temperature shift to remove wg function at the same stage, 123±4 cells). This discrepancy might be explained if the activities of wg and the tip cell are required at different stages in the cycle. An early requirement for wg (e.g. to initiate S phase) and a later requirement for the tip cell (e.g. to initiate mitosis) would result in immediate arrest of cell division after tip cell ablation, while the effects of removing wg function would be felt only after cells had progressed through the cycle to re-enter S phase. It should be noted that this would not preclude a more direct interaction between the tip cell and wg, as the early consequences of tip cell ablation would mask the later effects of perturbing this interaction.

Our studies have therefore identified two factors that are involved in regulating cell division in Malpighian tubule primordia; the expression and normal secretion of wg (reported here) and the activity of the tip cells (Skaer, 1989; 1992b). The response of the Malpighian tubules to HS wg expression suggests an interplay between the wg gene product and the activity of the tip cells; supernumerary cells can be induced in the tubules for only a limited period by extending the expression of wg. The end of this period coincides with the proposed decline in tip cell activity (Skaer, 1989), which suggests that the continued activity of the tip cell is required for the presence of wg to be effective. Indeed the observation that it is possible to stimulate the production of supernumerary cells in the tubules, by extending wg expression, indicates that in normal development it is the decline of wg expression, at a time when the tip cells are still active, that stops cells cycling mitotically in the tubules.

The precise nature of the interaction between wg and the tip cells is not yet known. However, the ultrastructure of tip cells suggests that they have a secretory function (Skaer, 1989; 1992b). This observation raises the possibility that products secreted by the tip cells interact synergistically with wg to stimulate a cellular response. Such interactions have been demonstrated in the activation of other growth factors such as basic fibroblastic growth factor, which stim-ulates a response only after it has bound to components of the extracellular matrix (Yayon et al., 1991). Analysis of products secreted specifically by the tip cells is required to understand the interaction with wg at a molecular level.

The protein encoded by the wg gene is a member of the Wnt family (Rijsewijk et al., 1987; Nusse et al., 1991), which includes int-1, a mammalian proto-oncogene (Nusse et al., 1984). When expressed ectopically, int-1 stimulates cell division in mammary tissue, which can result in the production of tumours (Tsukamoto et al., 1988). We find an interesting parallel between the effects of int-1 in mammary epithelia and those of wg in embryonic tissues of Drosophila: in both cases, expression of the gene stimulates cell division but only in certain tissues and under certain circumstances. We are currently investigating the possibility that further parallels in the modes of secretion and activity of these molecules might underlie similarities in the consequences in their expression.

We are grateful to Michael Bate, Peter Lawrence, Javier San-pedro, Tony Brown, Michael Taylor, Amy Bejsovec and Mary Baylies for many enlightening and stimulating discussions and to Adrian Friday for his advice on statistical analysis. We thank C. Doe and J. M. van den Heuvel for antibodies and C. O’Kane, G. Struhl and N. Perrimon for fly stocks. We are especially indebted to P. Lawrence for allowing H. S. to use the HS-wg flies (Nor-dermeer et al., 1992) and to Paul Johnston for his help and guid-ance during the conduct of these experiments. This work is sup-ported by the Wellcome Trust.