puckered, a gene involved in position-specific cell differentiation in the dorsal epidermis of the Drosophila larva

The patterning of the larval epidermis during Drosophila embryogenesis can be viewed as a sequence of three well defined periods. The first one takes place during the cleavage of the zygote and results in its regionalization from centers of activity located at each end of the egg (St. Johnston and Nüsslein Volhard. 1992). This phase culminates in the cellular blastoderm with the activation of some segment polarity genes in stripes which outline the metameric units of the larva (Ingham. 1988). A second phase covers the period of postblastoderm divisions and occupies most of what is known its the extended germ band stage. During this period the information laid down in the blastoderm is used to generate region-specific patterns of gene expression within each metameric unit (Ingham. 1991; Ingham and Martinez Arias. 1992; Martinez Arias. 1993). Finally, during the third phase cell division ceases and cells differentiate in a segment and position-specific manner (Martinez Arias, 1993). This view of epidermal patterning during embryogenesis has developed, in part, through the identification and functional characterization of genes involved in the first two phases (Ingham and Martinez Arias. 1992). These two phases arc characterized by a progressive increase in the number of cell identities which are organised in distinctive patterns. The progress of these processes is indicated by the patterns of expression of, and requirements for. the segment polarity genes (reviewed by Ingham and Martinez Arias, 1992: Peifer and Bejsovec, 1992: Hooper and Scott, 1992).

The third phase is characterized by two processes: the cessation of cell proliferation and the emergence of position specific cell shapes as a prelude to their overt differentiation. For example, ventrally in the anterior region of each segment the cells that will secrete denticles become elongated along the dorsoventral axis and acquire a shape that contrasts with that of cells in the posterior region of the segment which remain more isodiamctric and will differentiate smooth cuticle (Martinez Arias. 1993). This final phase in the patterning of the epidermal sheet reveals differences between cells that have been generated during the first two phases. The cellular and molecular mechanisms that underlie these late developmental events remain to be investigated.

In the larval epidermis, patterns of cellular differentiation cannot be directly correlated with known patterns of gene expression in cells prior to overt differentiation. Thus, although it is clear that segment polarity genes determine the pattern of the larval cuticle, it is not clear whether they do it directly or indirectly. Specifically, there is no known segment polarity gene whose expression is associated with the prospective denticle bell region or with regions of the cuticle with patterns of hairs. Instead, cells in these regions express overlapping patterns of segment polarity genes (sec e.g. Peifer and Bejsovec, 1992) suggesting that the information generated by these genes is integrated by the cells and somehow passed on to the ‘differentiation genes’.

Transformation of ‘positional’ to ‘differentiation’ information is exemplified by the evolution of Fasciclin III expression during germ band shortening. Fasciclin III is a cell adhesion molecule and although it is eventually found in all epidermal cells, this global distribution is achieved through onset of expression in a sequence of position specific patterns initialed after proliferation (Patel et al., 1987; Martine/ Arias. 1993). Thus the patterning information present in the form of overlapping patterns of. amongst others, segment polarity gene products results in position-specific downstream gene expression. This transfer of information could occur in one of two ways. Either the information, which undoubtedly exists in the spatial and temporal ox erlaps of the expression of segment polarity genes, is read directly by the differentiation genes or there arc intermediaries that integrate the activity of the former and act as links to the latter. In the first ease, the activity of segment polarity genes creates a code that would be interpreted by the promoters of the differentiation genes directly while in the latter, the integration would be performed by a class of intermediate genes. In either case, we believe the key moment for this transfer of information must be the end of proliferation, when cells stop dividing and begin to differentiate according to their position in the embryo.

In order to initiate the study of this process of ‘information transfer’ we describe here the pattern of expression of a β-galaclosida.se enhancer trap line in a small and conspicuous set of epidermal cells as they stop proliferating. The P element insertion responsible for this expression causes a mutation in a gene that appears to be required for the normal differentiation of the cells in which β-galactosidase is expressed. We describe the phenotype of mutations in this gene and discuss the implications of the expression pattern of the β-galaclosidase insertion, and of the mutation it produces, for the onset of position-specific cell differentiation.

P[ry⁺lac Z] E69 was generated in an enhancer trap screen (M. Bate. E. Rushton and A. Martinez Arias) using the starter strain carrying the ry⁵⁰⁶ P[ry⁺lac Z] C49 chromosome provided by Cahir O’Kane. l(3)84Eh^K19 was supplied by B. Baker. All other stocks were obtained from the Drosophila stock centre at the Department of Biology. Bloomington. Indiana. The following chromosomes were used in this study and are referred to by shortened names in this manuscript (in parentheses); D/(3R)dsv(Df(3R)ds\3). Df(3R)- dsx^M+10,° (Df(3R)dsx10), Df[3R)dsx^M+RI5(Dfl3R)dsx15) and Df(3R)dsx^M2l(Df(3R)dsx21. Balancer chromosomes are described in Lindsley and Zimm (1992).

Cuticle phenotypes examined were of embryos generated by adults heterozygous for a particular mutation and an Oregon R wild-type chromosome. These outcrossed flics were set to lay eggs overnight at 25 C on standard apple juice-agar plates supplemented with live yeast. For examination of unhatched embryos, development was allowed to proceed for a further 24 hours and unhatched embryos were collected. Cuticle preparations were done according to Wieschaus and Nüsslein-Volhard (1986; protocol based on the original technique described by van der Meer. 1977) except that embryos were not fixed before mounting.

Antibody stains were performed following previously described protocols (Ashburner. 1989) and after dehydration embryos were mounted in DPX. The simultaneous detection of βgaIactosidase activity and the expression of spectrin, as revealed by antibodies, also followed published protocols (Couso et al., 1993). The expression of β-galactosidase was detected cither as an enzymatic activity or with a monoclonal antibody that was a gilt from Dr C. Doe. Antibodies against spectrin (Pesacreta et al., 1989) were kindly provided by D. Branton and antibodies against Fasciclin III (Patel et al., 1987) were a gift from M. Wilcox.

After germ band shortening (end of stage 12) a sheet of epidermal cells covers the ventral and ventrolateral sides of the embryo while its dorsal surface is covered by the amnioserosa, a layer of large cells that do not divide after the blastoderm stage. Amnioserosa cells do not contribute to the larval cuticle since they become internalised and are eventually lysed in the final stages of embryogenesis (Hartenstein and Jan. 1992). We have identified a lacz. enhancer trap line P[ry⁺lacZ]E69 which marks the dor- salmost cells of the epidermis from just before germ band retraction to hatching (Fig. 1A-E). During stages 13–15. the epidermis from the two lateral sides of the embryo is joined into a single sheet of cells when the cells expressing lacZ in P[ry⁺lacZ]E69 meet at the dorsal midline. This event is known as dorsal closure (Campos-Ortega and Haricnstein. 1985) and can be divided into three processes; (a) the dorsal expansion of the epidermal sheet, (b) the concomitant expansion of the dorsal epidermis anteroposleriorly, and (c) the joining of the two sides of the epidermis at the dorsal midline.

A most conspicuous feature of the dorsal epidermis during dorsal closure is a row of elongated cells that define the dorsal limit of the epidermal sheet (Young et al., 1993; Fig. I). These cell correspond to the McZ-expressing cells in P[ry⁺lacZ]E69 and lead the dorsal ward movement of the epidermal sheet during dorsal closure (Fig. 21F-1). lacZ expression al the dorsal epidermal edge begins al stage 11 in clusters of cells (Fig. 1A) at about the lime they stop proliferating (unpublished observations). They are initially polygonal, as is typical of epidermal cells at this stage (Fig. 1F), but by the lime the germ band retracts they are seen to undergo changes specific to these cells: they align in a single, or sometimes double, row (Fig. 1B) and elongate along the dorsoventral axis (Fig. 1G). The dorsal side of these cells, adjacent to the amnioserosa. forms a straight edge to the moving front of the epidermal sheet (sec Fig. III). All leu Z-expressing cells contribute to this leading edge as it moves dorsally. Nuclei are restricted to the ventral side of these cells from stage 12 (Fig. 1G-I) highlighting their planar polarity. When the two sides of the epidermis meet at the dorsal midline the leading edges formed by these cells abut (Fig. 11). and shortly afterwards these cells change shape again to become more like Filerai epidermal cells. At this time the lacZ-expressing cells intercalate with each other over the dorsal midline (bracket in Fig. 11). Nuclei relocate to the centre of the cells after joining of the two sides takes place (bracket in Fig. 11). Al the end of dorsal closure the leu /-stained nuclei form two parallel rows along the length of the dorsal midline.

Another example of the planar polarity of the dorsal-most cells is the accumulation of nonmuscle myosin al their leading edge (Young et al., 1993). Furthermore. Fasciclin III is found on all surfaces of epidermal cells, but in the dorsal- most cells it is excluded from the side that is adjacent to the amnioscrosa. dining dorsal closure (Fig. 5A). Once the two sides of the epidermis have met at the dorsal midline Fasciclin III is visible on all lateral surfaces of these cells (Fig. 5B). This distribution of Fasciclin III in the dorsal most cells during dorsal closure represents another aspect of their polarity in the plane of the epidermis.

The P[ry⁺lacZ]E69 enhancer trap chromosome is homozygous embryonic lethal and produces a mutant cuticle phenotype in the dorsal epidermis (Fig. 3). We have named the mutated gene in this chromosome puckered and the puckered mutation in the P[ry⁺lacZ]E69 enhancer trap chromosome is referred to as puc^E69. In situ hybridisation of P element DNA to polytene chromosomes from the P[ry⁺lacZ]E69 line (not shown) indicates that it contains a unique P element al distal 84E(3R), a region that has been extensively characterised genetically (Baker et al., 1991; see Fig. 2). When in trans to deficiencies for this region puc^E69 is also lethal and the resulting phenotype is similar to. but more severe than, that of homozygous puc¹’^ embryos (not shown). Thus puc^E69 is not a null mutation. Furthermore, loss of the P element by dysgenesis results in reversion of the puckered mutant phenotype to wild type (unpublished observations) indicating that the P-lac/ element is the cause of the puckered mutation. Deficiency mapping indicated that puckered is located between the proximal breakpoints of Df(3R)p¹³ and Df(3R)p⁴⁰ (Fig. 2). Two lethal complementation groups hase been identified in this region, one of which. l(3)84E. is allelic lo puckered (Fig. 2). The single existing EMS allele of l(3)84E. Eh^K19 (Baker et al., 1991). is embryonic lethal and displays a mild puckered phenotype (Fig. 3). We have renamed the Eh^K19 allele of puckered. puc^K19.

Analysis of the puckered mutant phenotype in embryos homozygous for puc^{1 69} reveals that the development of the puckered cells is aberrant (Fig. 4). In mutant embryos after germ band shortening the lacZ-expressing cells are not restricted to a straight line at the edge of the epidermis, but arc seen lo extend several cell diameters away from the edge (compare Fig. 4A and 4D). As dorsal closure proceeds these cells become increasingly disorganised (Fig. 4E) so that by completion of dorsal closure they arc haphazardly arranged in clusters al the dorsal midline (Fig. 4F) rather than forming a single row on either side of the midline as in wild type (Fig. 4C).

The shape of the dorsal-most cells is also abnormal in puckered mutants. During germ band shortening these cells do not elongate as in wild type, but retain the polygonal shape observed in the extended germ band phase (Fig. 5C). Consequently, they do not generate a straight edge to the moving front of the epidermis during dorsal closure. When the two sides of the epidermis meet at the dorsal midline they do not form two parallel rows (Figs 5D, 4F). The pattern of the cuticle secreted by these cells is also abnormal (Fig. 3D.F).

The subcellular distribution of Fasciclin III shows that in mutant embryos the dorsal-most cells do not show the planar polarity normally seen in these cells (Fig. 5). In the majority of the dorsal-most cells. Fasciclin III protein is not excluded from the dorsal side, adjacent to the amnioserosa (Fig. 5C). Moreover, nuclei are not restricted to ventral areas of the cell (Fig. 5C,D). Indeed, none of the specialisations that we have described as characteristic of the dorsal-most cells is adopted in embryos mutant for puckered.

Despite the abnormalities of the dorsalmost epidermal cells in puckered mutant embryos, dorsal closure does occur to completion. This demonstrates that the shape and arrangement of the dorsal-most cells is not required for dorsal closure to occur. Il is possible that rather than depending on forces generated by shape changes in these cells (Young et al., 1993), dorsal closure draws its mechanical force from a coordinated change in cell shape of all epidermal cells, especially in the lateral regions of the epidermis.

The final step during pattern formation within cell assemblies is the activation of the process of cell differentiation which leads to the final arrangement and appearance of the different cell types specified during embryogenesis and. through their specializations, to the shaping and final appearance of organs and tissues and hence the organism (see e.g. Edelman. 1988). For the larval epidermis of Drosophila this phase is initialed between the cesation of cell division in stage I 1 and the onset of cuticle secretion in stage 15. a period during which cells neither cycle nor divide and undergo position-specific changes in shape which prelude their specializations during cuticle secretion (Martinez Arias. 1993).

By the end of stage 11. epidermal cells have acquired position- and segment-specific identities through the concerted activities of the segment polarity and the homeotic genes. The segment polarity genes encode a class of molecules that are involved in the generation of positional information within every segment (Ingham. 1991; Hooper and Scott. 1992; Martinez Arias. 1989; Peifer and Bejsovec. 1992) whereas the homeotic genes encode a family of trancriptional regulators which allow the interpretation of this information in a segment-specific manner (Akam et al., 1988). While mutations in either homeotic or segment polarity genes result in embryos with dramatic changes of pattern, in each case all cells undergo normal differentiation but differentiate in a manner appropriate to positions elsewhere in the embryo. This means that while the activities of these genes arc involved in the assignation of the type of differentiation a cell will ultimately undergo, they do not directly bring about the differentiation of cells. Il has been accepted that the homeotic genes act through an intermediary set of genes that in turn control the expression of genes involved directly in differentiation. Although this has always been accepted for the homeotic genes (reviewed by Andrew and Scott. 1992). little is known about the genes that arc downstream of the segment polarity genes and which are related to the process of cell differentiation.

One way of thinking about the control of epidermal differentiation is to imagine that information encoded by the combined activities of segment polarity and homeotic genes is read by intermediate genes w hose products would thus integrate and transduce positional information into position-specific differentiation. By way of example, genes of the achaete-scute complex (AS-C) can be regarded as performing an analogous role to the genes we propose link patterning information to differentiation, but in the linking of patterning information to subsequent patterning processes in the nervous system (Fig. 6). The emergence of neuronal precursors in the central and peripheral nervous systems is closely associated with the patterning of the ectoderm since neuroblasts and sensory mother cells arise al specific places in (he ectoderm with identities that reflect the molecular information that exists at that position. The expression of transcripts from the AS-C is linked to the segregation of these neuronal precursors (Cabrera et al., 1987: Romani et al., 1987). closely follows the generation of cell diversity in the ectoderm, and is under the control of genes that establish positional information (Marlin Bermudo et al., 1991: Skeath et al., 1992). In addition, this expression is required for the correct pattern of the nervous system (reviewed by Cabrera. 1992). In this sense the AS-C performs a role similar to that of the genes that we are proposing exist between the segment polarity genes and the process of cell differentiation. Such genes would perform a role of integration of information and transfer to a particular process: the patterning of the nervous system in the case of the AS-C. and cell differentiation in the case we have discussed here (Fig. 6).

Here we have described a P-lacZ enhancer trap line in which the puckered gene is mutated. In this line lacZ expression is initiated in the most dorsal epidermal cells as they finish proliferating, a pattern that reflects the expression of the puckered transcript (unpublished observations). In embryos mutant for puckered, the cells that express lacZ fail to differentiate properly: they do not adopt shape changes normally seen in these cells during dorsal closure and secrete cuticle with an abnormal pattern. These observations suggest that puckered might be involved in integrating positional information prior to cell differentiation. However, it is also possible that puckered acts at a later point to implement the differentiation of these cells. Determination of the precise role of puckered in the differentiation of the dorsalmost epidermal ceils will require further genetic and molecular studies which are currently in progress.

We want to thank A. Gampel and II. Skaer for useful discussions and continents on the manuscript. This research was supported by a Wellcome Trust Senior Fellowship to A. M. A. and a Commonwealth Scholarship to J. M. R.