ABSTRACT
We describe cell lineage of the tarsus of wild-type Drosophila. Large Minute+ clones were made to map the position of the antero-posterior compartment boundary in all three tarsi. The tarsus is mirror symmetric, but the compartment boundary does not coincide with the mirror plane. This boundary runs along the dorsal and the ventral rows of bristles which are immediately posterior to the mirror plane; elements in these rows being made by both anterior and posterior polyclones. The provenance of bristles and bracts suggests that the bristle cells move into their final positions. The homoeotic mutation engrailed affects only the posterior compartments of all three tarsi. The mutations bithorax and postbithorax affect only the anterior and posterior compartments of the third legs, respectively, transforming them into homologous compartments of the second leg. These results support the selector gene model of development (Garcia-Bellido, 1975) and emphasize that collaboration between polyclones is important in pattern formation.
INTRODUCTION
A compartment (Garcia-Bellido, Ripoll & Morata, 1973, 1976) is a specific region of the insect which is formed by a polyclone - all the descendants of a small group of founder cells (Crick & Lawrence, 1975). The boundary separating one compartment from another is precisely positioned: for example, the mesothoracic segment of Drosophila is subdivided into anterior and posterior compartments and the boundary bisects the wing running just anterior to vein IV (Garcia-Bellido et al. 1973, 1976). This boundary extends to the thorax and down the mesothoracic leg, dividing it longitudinally into anterior and posterior regions (Steiner, 1976). The position of the compartment boundary has been defined in the wing, the thorax and most of the leg, but not in the tarsus. Here we report an analysis of the cell lineage of the tarsus in which we find, unexpectedly, that both anterior and posterior polyclones can form common elements in the same bristle row. This illustrates that patterns can result from interactions between polyclones. The provenance of bristles and bracts in this row suggests that the bristle-forming cells move as the row develops.
The tarsus of the second leg is mirror symmetric, but the compartment boundary does not coincide with the mirror plane. This same boundary delimits the area affected by the homoeotic mutation engrailed, while in the third leg bithorax and postbithorax affect the anterior and posterior compartments respectively - results which support the selector gene model of development (Garcia-Bellido, 1975).
MATERIAL AND METHODS
We used the standard methods of clonal analysis (Becker, 1957; Bryant & Schneiderman, 1969; Garcia-Bellido & Merriam, 1971a) and the Minute technique (Morata & Ripoll, 1975). The tarsus is normally rather yellowish in colour, which makes the mutant yellow less useful as a marker. This problem is overcome when yellow clones are looked for in an ebony background. After dehydration in absolute alcohol the legs were mounted directly in a new mountant devised by Gary Struhl and John Sulston (dried canada balsam dissolved in a clearing agent, methyl salicylate). Standard methods of egg collection (the time is given as hours after egg laying, h AEL) and irradiation were used (see Lawrence & Morata, 1977). The genotypes (see Lindsley & Grell, 1968) used are listed with each section. Clones were classified as anterior or posterior on the basis of their position in the tarsus and in the proximal leg segments (Steiner, 1976).
RESULTS
Cell lineage of the tarsus of the second leg
(i) The Minute technique
[Flies were M(1)oSp/y wf36a and irradiated with 1000 R at 3 ± 1, 36+12 and 60 ± 12 h AEL.]
In the second leg all the tarsal segments are mirror symmetric, there being four well-defined rows of bristles in each half ; the claws and associated structures are also symmetrically arranged. In this study we have concentrated particularly on two rows of bristles, the anterior and posterior ventral rows (rows 1 and 8 of Hannah-Alava, 1958) which flank the mirror plane on the ventral face (see Figs. 1, 2).
Large clones were made during the embryonic and larval period and these were always either anterior or posterior. Posterior clones extended up to the posterior ventral row (Fig. lb) and marked bristles and bracts in it. None (0/31) of the clones completely filled this row and none crossed to the anterior ventral row, even to the extent of a single bristle or bract (Fig. lb). Anterior clones could completely mark the anterior ventral row, including every bristle and bract; such clones usually extended to the posterior ventral row although they never filled it (Fig. 1 a). Of 22 clones studied in detail, 17 crossed the mirror plane and marked bristles in the posterior ventral row. Anterior clones also crossed the dorsal midline of the tarsus, while posterior ones respected it. Accordingly, one anterior clone could mark bristles in six of the eight bristle rows on the tarsus (see Fig. 2b), while a posterior clone could not extend to more than 4/8.
To investigate the provenance of bristles and bracts in the posterior ventral row one would ideally need tarsi in which all the cells of either one or the other compartment were genetically marked. Large Minute+ clones often almost fill a compartment (Garcia-Bellido et al. 1973). Fourteen large anterior and 14 large posterior clones were chosen and together these clones marked 369 bristles of the posterior ventral rows in the 28 legs. If these clones had completely filled their respective compartments then the total number of bristles in 14 posterior ventral rows would have been marked; as there are about 30 bristles in each, posterior ventral row this would have been some 420 bristles. As these 28 clones filled nearly 90% (369/420) of the row bristles they were used to investigate the provenance of bristles and bracts. The 14 large anterior clones marked 156 bristles and 82 bracts in the posterior ventral row, while the 14 posterior clones marked 213 bristles and 272 bracts in the same row. Clearly, the posterior ventral row is made by the two polyclones together, although there is a greater contribution by the posterior cells. Scrutiny of the clones (for example, Fig. 1 a, b) shows there is variation in the origin of particular bristles and bracts. Although the elements of the posterior ventral row cannot be divided into two sets, one always being made by the anterior, and one by the posterior polyclone there is some consistency in the formation of the row, for example, the central bristles in the basitarsus are only rarely made by anterior cells, while the bristles at each end are frequently anterior (e.g. Fig. 1b).
Anterior clones marked more bristles than bracts in the posterior ventral row [156 bristles:82 bracts]; which shows that the bristles are more likely to be formed by anterior cells than bracts. If this is so then posterior clones should mark more bracts than bristles, which they do [213 bristles:272 bracts]. We can conclude that, on average, the bristles of the posterior ventral row have a more anterior origin than the bracts.
(ii) Conventional clonal analysis
[Flies were y/ + and irradiated with 1000 R at 36 ± 12 h AEL.]
There was the possibility that Minute+ clones, growing disproportionately in a Minute background, might not behave normally with respect to bristle pattern. We therefore made clones that were marked only with yellow, which we can assume to be a gratuitous marker. A large number of second legs (640) were studied and 13 clones that marked the anterior or posterior ventral rows were found in the tarsi. All these clones, although small, were subsets of the larger clones studied earlier. For example, some clones were confined to the posterior ventral row and were presumably posterior (they marked more bracts than bristles, Fig. 2 b), others crossed between the two rows (Fig. 2 a, c, d) and behaved like small parts of the anterior Minute+ clones. We can conclude that the Minute+ clones express the normal cell lineage of the tarsus.
Cell lineage of the first and third leg
[Flies were M(I)oSp/y w/36ft and irradiated with 1000 R at 3 ± 1, 36 ± 12 and 60±12hAEL.]
In the first and third legs Minute+ clones in the tarsal segments 11, III, IV and V behaved as in the second leg in all respects listed above (Fig. 1 c, d). These segments have anterior and posterior ventral rows although, in addition, there are transverse rows present in segment II (Fig. 1c).
In the basitarsus of the first leg of females the anterior ventral row is missing, instead there are transverse rows (Fig. Id). These bristles can be completely marked by anterior clones which, as in the second leg, can cross to the posterior ventral row (Fig. 1 d). Posterior clones in the basitarsus also mark the posterior ventral row but not the anterior transverse rows.
In the basitarsus of the third leg transverse rows are present and the posterior ventral row is missing - there is a large gap in the medial region (Fig. 1 c). Anterior clones behave as in the second leg, they fill the anterior ventral row and cross to some bristles which are more medial. They do not extend across the gap to the transverse rows proper (Fig. 1c). Posterior clones can completely fill the transverse rows.
In summary. In all three thoracic legs the compartment boundary coincides approximately with the posterior ventral and dorsal rows, and not with the mirror plane. The posterior ventral row is made by both anterior and posterior polyclones, but anterior cells contribute more bristles than bracts to it.
Selector genes and tarsal pattern
(i) engrailed
We have examined tarsi of flies mutant for engrailed and wild-type flies bearing engrailed Minute+ clones. We find that engrailed clones in the anterior compartment of the second tarsus behave like wild-type clones and do not affect the pattern. However, posterior clones form numbers of extra bristles (compare Fig. 3a and b). In the first tarsus of engrailed flies there are extra bristles in the posterior compartment of males and females, although they do not form clear transverse rows. In males there is an additional sex comb in the posterior compartment (Tokunaga, 1961). In the third tarsus there are extra transverse rows in the posterior compartment, they have more bristles but are less well-ordered than in wildtype. There are no effects on the anterior compartments of either leg.
ii) bithorax and postbithorax
We have looked at these mutants in the tarsi of third legs. In legs mutant for postbithorax, the transverse rows of the basitarsus and second tarsal segment disappear and are replaced by four rows of bristles characteristic of the posterior second leg, including a posterior ventral row. The same phenotype is found in legs carrying large posterior clones of postbithorax (Fig. 3d). Smaller clones behave autonomously; they do not contribute to the transverse rows, although transverse rows are formed by wild-type cells outside the clone (Fig. 4a).
In the tarsi of third legs mutant for bithorax (bx^/Ubx130) the anterior ventral row bristles are similar to those found in the second leg. In the basitarsus the posterior transverse rows are unaffected, but in the gap there are always some ‘adventitious’ bristles which form a partial posterior ventral row. This row is sometimes almost complete, but usually consists of three to eight bristles. Similar bristles are almost never found in the gap in normal or Ubxls0 third legs. The cell lineage of bithorax flies was studied to determine the polyclonal origin of this ‘adventitious’ row. We found that both anterior and posterior polyclones formed bristles and bracts in the row (Fig. 4b, c), so that in this respect it behaved like a typical posterior ventral row of a second leg.
We also made fast-growing clones of cells that were mutant for bithorax in a wild-type background. A typical anterior clone is illustrated in Fig. 3 a. Note that the transverse rows are not marked and that all the adventitious bristles are of anterior origin - although their bracts can come from either anterior or posterior polyclones. Seven large anterior clones which made one or more ‘adventitious’ bristles were examined. In every case all the ‘adventitious’ bristles (n = 20) were marked, although only half the bracts belonged to the clones (10/20). Some posterior clones filled the transverse rows completely, but none of these made any ‘adventitious’ bristles (n = 11).
DISCUSSION
Cell lineage of the wild-type tarsus
Clonal analysis of the tarsus of the second leg has shown that although the pattern is mirror symmetric the compartment boundary does not coincide with the mirror plane. This example illustrates that formation of some patterns can be independent of compartments; the tarsal cylinder decorated with the mirror symmetric bristle rows conceals an underlying asymmetry of the two polyclones which construct it (Fig. 6).
Two bristle rows each contain elements formed by both anterior and posterior polyclones and there is variation in the provenance of specific bristles and bracts. The same observation has been made independently by Held (personal communication) who has also shown that in the other rows of the tarsus there is relative movement by the bristle forming cells. A similar situation has been described in the antenna, where two bristles, identifiable by their size and position, can be made either by the anterior or by the posterior polyclone (Morata & Lawrence, 1978, 1979).
The posterior ventral row was studied in detail; anterior cells form twice as many bristles as bracts in the row; while, because every bristle has a bract, posterior cells necessarily form more bracts than bristles. If the variable origin of bracts and bristles were simply due to an irregular compartment boundary along a straight presumptive bristle row, one would expect bracts and bristles to be of anterior origin equally often. The results show that, on average, the bracts have a more posterior origin than the bristles. As bracts and bristles are finally positioned in a straight row which is parallel to the antero-posterior axis it follows that there must be relative movement of the cells forming them. The bracts are formed from nearby epidermal cells which are induced by the bristleforming cells (Tobler, 1966; Tobler, Rothenbuhler & Nöthiger, 1973) so it is most likely they develop later than the bristle forming cells, and in situ. We therefore envisage the bristle rows forming as shown in Fig. 5.
It is clear from Fig. 5 that the polyclonal origin of the cells plays little part in the process; the area competent to form the bristle rows is determined according to the pattern of the limb as a whole. This situation can be contrasted with the anterior wing margin. Here the asymmetric triple row is formed by the dorsal polyclone always and only forming the two dorsal rows, while the ventral polyclone forms the ventral row (Bryant, 1970; Garcia-Bellido & Merriam, 1971 a). Tt would appear that in the wing, unlike the tarsus, pattern elements form as a consequence of the juxtaposition of two polyclones (Santamaria & Garcia-Bellido, 1975).
The idea that cells forming organules can move is not new: in the formation of the sex comb a group of cells is thought to rotate (Tokunaga, 1962) and in the abdomen marked bristles are frequently found separated from a patch of epidermal cells of the same clone (Garcia-Bellido & Merriam, 1971 b). In the triple row of the wing small clones become split into separate regions suggesting that the bristle-forming cells shuffle into position (Garcia-Bellido & Merriam, 1971 a; unpublished observations). Moreover, during development of hairs in Oncopeltus (Lawrence, 1966) poorly arranged cells have been seen to form into precisely oriented rows. Local cell movement seems therefore to be a general mechanism of pattern formation in insects.
The role of selector genes
Selector genes are one class of homoeotic genes. Their products are continuously required for the development of a polyclone so that without them development is switched to another pathway (Lewis, 1963; Garcia-Bellido, 1975; Morata & Lawrence, 1977). We have studied three mutations in selector genes to see their effect on tarsal pattern.
The gene engrailed is thought to determine the posterior developmental pathway in at least the three thoracic segments (Garcia-Bellido & Santamaria, 1972; Morata & Lawrence, 1975; Lawrence & Morata, 1976) and the eye- antennal segment (Morata & Lawrence, 1978). The first leg of engrailed flies has been described in detail (Brasted, 1941 ; Tokunaga, 1961) while some effect was noted on the posterior part of the third leg (Garcia-Bellido & Santamaria, 1972). No effect on the second leg had been reported, but one would be predicted on the selector gene model. Accordingly we have studied the mirror symmetric tarsus of the second leg. As in the wing, and as expected, engrailed does not alter the anterior compartment. However, when the posterior compartment is mutant for engrailed numerous extra bristles are formed. The wild-type posterior compartment is smaller and contains fewer bristles (Fig. 6), so if it were to be transformed into an anterior compartment one would expect extra bristles, as is observed (Figs. 3, 6).
In the metathoracic segment the anterior compartment depends on the bithorax selector gene, and the posterior on postbithorax. Without these genes the anterior and posterior polyclones of the metathorax develop into the corresponding compartments of the mesothorax (Lewis, 1963; Garcia-Bellido, 1975; Morata & Garcia-Bellido, 1976).
In the tarsus the mutation postbithorax behaves as expected, the transverse bristle rows characteristic of the posterior compartment of the third segment are eliminated and replaced with ordinary rows of bristles. In postbithorax clones the transverse rows are removed autonomously, clones were seen to contribute bracts but not bristles to the transverse rows, suggesting that transverse row bristles are exclusively part of the ‘repertoire’ of the posterior metathoracic polyclone. In the case of bithorax, where the anterior metathoracic compartment becomes mesothoracic, the picture is complicated by the partial development of a posterior ventral row - a row which is missing in normal third legs. These ‘adventitious’ bristles would be readily explicable if they came exclusively from the anterior polyclone (which, in second legs, contributes to a posterior ventral row). This is indeed the case when large bx3/ – clones are induced in a wild-type background (Fig. 3c). However, cell lineage analysis of bx3/Ubx130 legs shows that some are formed by the posterior polyclone. This could be due to slight pbx phenotype (pbx130 is deficient for pbx, and bx3 has some polarity effect on pbx+Lewis, 1963).
CONCLUSION
These experiments cast some light on the relationship between compartments and pattern. Some elements of the pattern belong to the repertoire of a specific compartment; for example, the sex comb teeth are only made by male cells belonging to the anterior first leg. These structures can be made in other places on the fly but only when they have the same genetic address as male, anterior first leg. They can be found in the posterior compartment of the first leg when it is mutant for engrailed (Tokunaga, 1961) or in the anterior compartment of second and third legs when mutant for Polycomb (Lindsley & Grell, 1968). engrailed partially transforms posterior into anterior, Polycomb second and third legs into first. Sex combs cannot be formed by female cells even when there are only a few such cells in the area where the sex comb teeth normally form (Tokunaga, 1962). Another example is those transverse rows which characterize posterior third legs and are locally removed by clones of pbx cells. The formation of such elements is cell autonomous, depending only on the genotype of the responding cells (see Stern, 1968).
The arrangement of pattern elements which one or more compartments have in common may involve co-operation between different polyclones. In the tarsus, a prepattern is established without regard to the compartment boundary and the bristle-forming cells taken indiscriminately from the two polyclones. Similarly, in the wing, the first cross vein extends smoothly over the anteroposterior boundary. These are further illustrations that compartment boundaries are no barrier to intercellular communication and co-operation (Lawrence, 1975) even though these boundaries may coincide with the limits of gradients of positional information (Crick & Lawrence, 1975).
ACKNOWLEDGEMENTS
We thank Paul Johnston and Sheila Green for assistance and EMBO for a travel grant. Gary Struhl is supported by a Graduate Fellowship from the National Science Foundation. We thank Rolf Nothiger for his comments on the manuscript.