Positional signals provided by immediate neighbours appear to direct developmental decisions in the eye of Drosophila. By a combined genetic and molecular approach the biochemical bases of the signal and reception mechanisms are being systematically dissected. Three key gene products have now been identified, sevenless is a transmembrane tyrosine kinase probably transducing positional signals that direct the R7 cell to its fate. The bride of sevenless gene product is on the signalling side of the mechanism and is required in R8 for R7 to develop. The type of protein bride of sevenless encodes is not yet known. The rough gene encodes a transcription factor on the signalling side required in R2 and RS for positional signals to be transmitted to neighbouring cells.

Cells within a developing organism can be directed to their fate by positional cues. These can be long range in nature with cells being developmentally directed by signals from a distant source. Until recently few long-range diffusable signals had been identified, however some good examples have now been demonstrated. Perhaps the best example is the bicoid protein which is translated in the anterior region of the Drosophila embryo and redistributes posteriorly. A gradient of protein concentration is then established along the anterior-posterior axis of the embryo, to which the hunchback gene is thought to respond (Driever and Niisslein-Volhard, 1989). Many developmental phenomena are difficult to explain with models using long-range diffusable molecules and positional signals which operate over smaller distances local to the source itself are expected. Spatially restrained positional signalling could occur when cues are communicated between immediate neighbours. There have been few descriptions of this but there are some good examples in the vertebrate immune response where signals are passed through the MHC complexes of cells contacting each other. Strong evidence is accumulating from analyses of ommatidial development in the compound eye of the fruit fly for positional signals that are of shortrange nature and are presented by directly adjacent cells.

The compound eye is made from many hundred identical subunit ommatidia each of which is a simple assembly of 20 cells (Fig. 1). Each cell can be identified both by its position in the unit and its cell type, and the many hundred fold reiteration of the structure allows large numbers to be sampled in a single preparation. Cell lineage directed developmental decisions do no occur in assembling ommatidia (Ready et al. 1976; Lawrence and Green, 1979) and examinations of cellfate choices occurring in this system therefore focus upon the nature of the positional cues. The specific questions that are being addressed are the nature and transmission of the signals, the mechanisms the cells use to receive them, and the molecular events that follow reception and lead to developmental decisions.

Fig. 1.

(A) A smooth array of 700–800 ommatidia forms a Drosophila compound eye. Anterior is to the right. Small mechanosensory bristles project between ommatidia. (B) Tangential section of the eye; anterior is to the right. In any cross section, ommatidia present an asymmetric, trapezoidal pattern of seven rhabdomeres. (C) A schematic ommatidium, anterior is to the right. Below the corneal lens (cl) is a second lens element, the pseudocone (c), which is a refractile extracellular secretion of the four underlying cone cells (cc). The accessory cone cells meet in the centre occluding the principal cells from contact. The cone cells are collared by the two primary pigment cells (pp). Photoreceptors or retinuala cells (rc) are elongated sensory neurons that carry rhabdomeres (rh), dense stacks of rhodopsin-loaded microvilli. Rhabdomeres of photoreceptors R1–R6 extend the depth of the ommatidium. The rhabdomere of R7 lies above that of R8 on the central axis. A sheath of secondary and tertiary pigment cells (sp, tp) optically insulates each ommatidium. (D) A late third instar eye-antennal disc; anterior is to the top. The upper portion is the antennal disc which is folded into a series of concentric rings, and below it the briad slightly cupped eye disc. The morphogenetic furrow (indentation visible in eye disc) lies dorsoventrally and moves across the eye disc from posterior to anterior. Ommatidial patterning occurs posterior to the furrow. At the posterior of the eye disc is the optic stalk which carries the axons to the brain (not shown). Reprinted from Tomlinson and Ready (1987o), with permission.

Fig. 1.

(A) A smooth array of 700–800 ommatidia forms a Drosophila compound eye. Anterior is to the right. Small mechanosensory bristles project between ommatidia. (B) Tangential section of the eye; anterior is to the right. In any cross section, ommatidia present an asymmetric, trapezoidal pattern of seven rhabdomeres. (C) A schematic ommatidium, anterior is to the right. Below the corneal lens (cl) is a second lens element, the pseudocone (c), which is a refractile extracellular secretion of the four underlying cone cells (cc). The accessory cone cells meet in the centre occluding the principal cells from contact. The cone cells are collared by the two primary pigment cells (pp). Photoreceptors or retinuala cells (rc) are elongated sensory neurons that carry rhabdomeres (rh), dense stacks of rhodopsin-loaded microvilli. Rhabdomeres of photoreceptors R1–R6 extend the depth of the ommatidium. The rhabdomere of R7 lies above that of R8 on the central axis. A sheath of secondary and tertiary pigment cells (sp, tp) optically insulates each ommatidium. (D) A late third instar eye-antennal disc; anterior is to the top. The upper portion is the antennal disc which is folded into a series of concentric rings, and below it the briad slightly cupped eye disc. The morphogenetic furrow (indentation visible in eye disc) lies dorsoventrally and moves across the eye disc from posterior to anterior. Ommatidial patterning occurs posterior to the furrow. At the posterior of the eye disc is the optic stalk which carries the axons to the brain (not shown). Reprinted from Tomlinson and Ready (1987o), with permission.

Ommatidia develop in a monolayer epithelium, contained within the eye-antennal imaginal disc (Fig. 1D), in which cells extend from the apical surface to the basal membrane. Each ommatidium is built from a foundation precluster made from five cells destined to form photoreceptors R2, R3, R4, R5 and R8. The nuclei of these cells are positioned within the apical regions of the epithelium and the nuclei of all other cells destined to join the ommatidium are placed in the basal regions. Cells are systematically incorporated into the cellular unit and the ommatidium grows in a radial manner. As a cell joins the ommatidium it first establishes a precise set of contacts with cluster cells in the apical regions. Subsequently these cellular contacts extend throughout the entire apical/basal depth of the ommatidium and the nucleus of the cell rises from its basal position into the apical clustering. Following this the cell shows evidence of differentiation such as expression of neural antigens recognisable by specific antibodies and axon out-growth. The five cell unit grows to eight cells with the incorporation of the cells destined to become Rl, R6 and R7, completing the photoreceptor complement of the ommatidium. Lens-secreting cone cells are the next to be added and in this manner the ommatidium carries on growing. Using antibodies that recognise neural epitopes a developmental sequence can be detected in the five cells of the precluster. R8 is the first to differentiate followed by the pair R2 and R5 followed by the pair R3 and R4. R1 and R6, and R7 follow in the sequence as to be expected from the description above (Fig. 2; Ready et al. 1976; Tomlinson, 1985; Tomlinson and Ready, 1987a).

Fig. 2.

Sequential development of the ommatidium. Upper panel shows the differentiation sequence of the-eight photoreceptors. R8 differentiates first followed by the pair R2 and R5, next come R3 and R4, followed by R1 and R6 and lastly R7. The cells are shown in the positions they come to occupy rather than the position they hold at any particular stage. Below shows where mutations have identified gene products involved in the inductive sequence. rough (ro) breaks the sequence after the addition of R2 and R5 but before the determination of R3 and R4, the gene product is required in R2 and R5 to communicate a signal to R3 and R4. sevenless (sev) and bride of sevenless (boss) prevent the differentiation of R7. boss is required in R8 to signal and sev is required in R7 to receive.

Fig. 2.

Sequential development of the ommatidium. Upper panel shows the differentiation sequence of the-eight photoreceptors. R8 differentiates first followed by the pair R2 and R5, next come R3 and R4, followed by R1 and R6 and lastly R7. The cells are shown in the positions they come to occupy rather than the position they hold at any particular stage. Below shows where mutations have identified gene products involved in the inductive sequence. rough (ro) breaks the sequence after the addition of R2 and R5 but before the determination of R3 and R4, the gene product is required in R2 and R5 to communicate a signal to R3 and R4. sevenless (sev) and bride of sevenless (boss) prevent the differentiation of R7. boss is required in R8 to signal and sev is required in R7 to receive.

Mutants in which perfectly normal ommatidia form, even though they are surrounded by aberrantly patterned ones indicate the positional cues directing the cells to their fate in the developing ommatidium are local to the ommatidium itself. Since cells differentiate as pairs (R2/5, R3/4, Rl/6) then the positional cues directing the cells of a pair to their fate must be presented simultaneously on opposite sides of the ommatidium. A model was proposed to account for these features (Tomlinson and Ready, 1987a) in which undetermined cells are cued to their fate by the combination of differing cell types they contact. The model envisages that differentiating cells express cell-type-specific signals and undetermined cells, which occupy a precise position in the developing unit, would be in contact with a specific subset of differentiating types, the combination of which specify the developmental pathway of the undetermined cell (Tomlinson and Ready, 1987a). A prediction made from this model is that mutations should be recoverable that interfere with cells’ abilities to express signals and others that prevent cells from receiving the signals. Both types of mutation would be expected to cause cells to be developmentally misdirected. In any particular mutant the pattern formation should be normal up to the point when the particular gene product is used and then the developmental error should occur. Once mutations have been identified mosaic analysis can be used to assess whether the mutation is in a gene used on the signalling or reception side of the mechanism, and to determine in which cells of the ommatidium the gene product is required. Mosaic analysis is performed by inducing a patch of mutant tissue in an otherwise wild-type eye. Where the two tissue types meet the cells of the different genotypes mix freely and ommatidia containing cells of both types form. The questions that can be asked by mosaic and other analyses are: (1) Can surrounding wild-type cells rescue mutant cells from their inappropriate developmental pathway, indicating a role of the gene product on the signalling side of the mechanism? (2) Which specific cells must carry the gene product for the ommatidium to form correctly and do these correspond to the cells that behave wrongly in the mutant? After cloning the gene, its nucleotide sequence may indicate the type of protein encoded, and antibodies raised against the protein can be used to establish in which cells and where in these cells the protein is found. The developmental analysis of the mutant, the mosaic analysis, the nucleotide sequence of the gene and the spatial and temporal localisation of the protein can be collectively assessed for indications of the gene’s function in the communication mechanism.

To date there are three genes identified that have been shown to have a role in the communication of the developmental directives between the cells, sevenless and bride of sevenless are genes used in determining the R7 cell type and rough is used to establish R3 and R4.

sevenless

sevenless is a mutation that causes each ommatidium to specifically lack R7 (Harris et al. 1976). Analysis of the developmental phenotype showed that, although a cell occupies the position in the developing ommatidium that normally generates R7, it fails to differentiate as that cell type, becoming instead a lens-secreting cone cell. Occupation of the correct developmental position by a cell which then fails to differentiate as the type normal to that position is the phenotype expected of mutations in genes used in the communication mechanism. From mosaic analysis, it has been shown that genetically sevenless cells in the R7 position cannot be rescued from transformation to the cone cell by surrounding wild-type cells (Harris et al. 1976; Campos-Ortega et al. 1979; Tomlinson and Ready, 1987b). This indicates that the cell in the R7 position is supplied with the correct positional signals but is incapable of receiving or interpreting them. The nucleotide sequence of the gene correlates well with this, showing that the gene encodes a putative trans membrane protein with a large extracellular domain and an intracellular tyrosine kinase, similar in general structure to hormone receptors such as the EGF receptor and insulin receptor (Hafen et al. 1987). This led to the proposal that the sevenless protein transduces signals for the R7 developmental pathway by binding of a signalling ligand to the large extracellular domain and subsequent modulation of the tyrosine kinase activity internally within the cell. A single amino acid substitution within the tyrosine kinase domain of the protein (known in similar proteins to abolish kinase activity) eliminates sevenless gene function, indicating that the signal transduction operates through the kinase domain (Basler and Hafen, 1988). Localisation of the protein indicates it is expressed in many cells (including the presumptive R7) in their apical plasma membranes (Bannerjee et al. 1987; Tomlinson et al. 1987). Expression of a receptor protein in more cells than those in which it is required is expected of a positional signalling system since prior to occupying a specific position a cell must be able to differentiate as one of many cell types, including the R7 type, which requires the sevenless protein. In cells that express the sevenless protein and contact R8, an accumulation of the protein is seen where the cells meet R8 in the apical cell junctions (Tomlinson et al. 1987). This suggested that a ligand for the sevenless protein is expressed by R8, and a requirement for signals from R8 for R7 to develop will be described below. By reintroducing the sevenless gene under an inducible (heat shock) promoter it has been demonstrated that sevenless protein is required in a short temporal window of ommatidial development, correlating well with the few hours that the protein is detected in the presumptive R7 by antibody analysis (Basler and Hafen, 1989). Blanket expression of the protein using heat shock does not interfere with the R7 developmental decision, neither does it affect the development of the rest of the fly, indicating that sevenless activation is achieved by a precise spatial restriction of its ligand(s).

bride of sevenless

bride of sevenless is a mutation with a similar phenotype to sevenless in that each ommatidium specifically lacks R7. Mosaic analysis has shown that unlike sevenless, R7 can be rescued by surrounding wild-type cells. More specifically, the bride of sevenless gene product is required in R8, and only R8, for the ommatidium to generate R7 (Reinke and Zipursky, 1988). This clearly indicates developmental communication occurring between R8 and R7, and correlates well with the interaction between these two cells indicated by the sevenless antibody analysis. The bride of sevenless gene is not yet cloned so the type of protein it encodes remains to be established.

rough

In developing rough ommatidia, the initial stages appear identical to wild type. R8 begins to differentiate normally, followed by R2 and R5 with cells correctly positioned to differentiate as R3 and R4. However, the cells in the R3/4 positions fail to behave correctly and ommatidial development passed this point is aberrant. Again, this is the expected phenotype of a mutation in a gene involved in the communication mechanism, with cells correctly positioned to become specific cell types but failing to do so. Mosaic analysis indicates that the rough gene product is required only in R2 and R5 for the ommatidium to develop normally. All other cells in a wide variety of combinations can be mutant without affecting ommatidial development. However, the developmental analysis indicates that R2 and R5 probably develop normally in the mutant, R3 and R4 are the cells that clearly behave inappropriately. This then indicates a developmental communication between R2/5 and R3/4. After R3 and R4 misbehave the ommatidia carry on building in variable and uninterpretable ways. Cells joining the ommatidia later than R3 and R4 may also need R2 and R5 to have the rough gene product, but from the analyses performed these requirements would remain undetected.

The nucleotide sequence of the rough gene shows it contains a homeobox which suggests a role for the protein in DNA binding and transcriptional regulation. Since the cells in which the protein is required appear normal in the mutant, neighbouring cells being the ones affected, then it is suggested that the rough protein acts as a transcription factor regulating the expression of signals by R2 and R5 communicated to the cells destined to become R3 and R4 (Tomlinson et al. 1987).

From these analyses it is becoming clear that developmental cues can be presented by directly adjacent cells. Whether these signals are held on the expressing cells’ plasma membranes or whether they are shortrange diffusable molecules is not clear. The question to be addressed is how applicable very short-range communication is to development generally. There are a large number of developmental phenomena that could be explained by positional signals being passed between directly neighbouring cells, but in general this has yet to be demonstrated. The power of the genetic and mosaic analyses in Drosophila has allowed this short range communication to be detected, but finding similar signalling mechanisms in animals lacking experimental advantages will be difficult.

Banerjee.
U.
, RENFRANZ. P. J., HINTON. D. R.,
Rabin
,
B. A.
and BENZER. S
. (
1987b
).
The sevenless protein is expressed apically in cell membranes of developing Drosophila retina, it is not restricted to cell R7
.
Cell
51
,
151
158
.
Basler
,
K.
and
Hafen
,
E.
(
1988
).
Control of photoreceptor cell fate by the sevenless protein requires a functional tyrosine kinase domain
.
Cell
54
.
299
311
.
Basler
,
K.
and
Hafen
,
E.
(
1989
).
Ubiquitous Expression of sevenless: Position dependent specification of cell fate
.
Science
243
.
931
934
.
CAMPOS-ORTEGA.
J. A.
,
JURGENS.
G.
and
Hofbauer
,
A.
(
1979
).
Ceil clones and pattern formation: Studies on sevenless a mutant of Drosophila melanogaster
.
Wilhelm Roux’ Arch devl Biol
186
,
27
50
.
Driever
,
W.
and
Nüsslein-volhard
,
C.
(
1989
).
The bicoid protein is a positive regulator of hunchback transcription in the early Drosophila embryo
.
Nature (Land)
337
,
138
143
.
Hafen
,
E.
,
BASLER.
K.
,
Edstrom
,
J. E.
and
Rubin
,
G. M.
(
1987
).
sevenless, a cell-specific homeotic gene of Drosophila, encodes a putative transmembrane receptor with a tyrosine kinase domain
.
Science
236
,
55
63
.
Harris
,
W. A.
, STARK. W. S. and
Walker
,
J. A
(
1976
).
Genetic dissection of the photoreceptor system in the compound eye Drosophila melanogaster
.
J. Physiol. (Land)
256
,
415
439
Lawrence
,
P. A.
and
Green
,
S. M.
(
1979
).
Cell lineage in the developing retina of Drosophila
.
Devi Biol
.
71
,
142
152
.
Ready
,
D. F.
,
Hanson
,
T. E.
and
Benzer
,
S.
(
1976
).
Development of the Drosophila retina, a neurocrystalline lattice
.
Devl Biol
.
53
,
217
240
.
Reinke
,
R.
and
Zipursky
,
S. L.
(
1988
).
Cell-cell interactions in the Drosophila retina: The bride of sevenless gene product is required in photoreceptor cell R8 for R7 cell development
.
Cell
55
,
321
330
.
Tomlinson
,
A.
(
1985
).
The cellular dynamics of pattern formation in the eye of Drosophila
.
J. Embryol. exp. Morph
.
89
,
313
331
.
Tomlinson
,
A.
and
Ready
,
D. F.
(
1986
).
Sevenless: A cell specific homeotic mutation of the Drosophila eye
.
Science
231
.
400
402
.
Tomlinson
,
A.
and
Ready
,
D. F.
(
1987a
).
Neuronal differentiation in the Drosophila ommatidium
.
Devi Biol
.
120
,
366
376
.
Tomlinson
,
A.
and
Ready
,
D. F.
(
1987b
).
Cell fate in the Drosophila ommatidium
.
Devl Biol
.
123
,
264
275
.
Tomlinson
,
A.
,
Bowtell
,
D. D. L.
,
Hafen
,
E.
and
Rubin
,
G. M.
(
1987
).
Localization of the sevenless protein, a putative receptor of positional information in the eye imaginal disc of Drosophila
.
Cell
51
,
143
150
.