argos

DEVELOPMENTAL BIOLOGY

Embryonic

argos is first expressed in the blastoderm in the dorsal ectoderm, forming a pair-rule type alternating striped pattern. After gastrulation, it is expressed in the head and the ventral midline ectoderm (Freeman, 1992). In the embryonic ventral ectoderm, argos is expressed in the ventralmost row of cells (Golembo, 1996).

In head midline structures, in particular the optic lobe and stomatogastric nervous system, there may be a late phase of EGFR signaling (as assayed by the expression of aos and activated ERK) whose significance is not yet known. EGFR signaling could be involved in modifying the inhibitory feed-back loop between neurogenic and proneural genes that exists in other neurectoderm cells. In the head midline neurectoderm, regulation of proneural and neurogenic genes has to be different. Thus, instead of a short burst of proneural gene expression in proneural clusters that is resolved into expression in individual neuroblasts, proneural genes are expressed for a long period of time; at the same time, the expression is never restricted to single neuroblasts. Since genes of the E(spl) complex are expressed in the same cells that express l’sc, the inhibitory loop between E(spl)-C and proneural genes must be interrupted at some level. It is possible that Egfr signaling is causing the interruption of this inhibitory loop. Based on genetic studies of Notch and Egfr signaling in the compound eye, it has been speculated that one of the consequences of Egfr activation (which ultimately is required for all ommatidial cell types to differentiate) is to inhibit N signaling, since constitutively active N inhibits ommatidial cell differentiation by preventing response to differentiative signals. However, the same effect could be achieved if Egfr signaling, similar to what is proposed here for the midline neurectoderm, interrupts the inhibition of proneural genes by E(spl). Although this would not prevent N signaling, it would cancel the effect of N signaling on downregulating proneural genes and thereby keep cells in a state of competency to respond to signals (Dumstrei, 1998).

The chordotonal (Ch) organ, an internal stretch receptor located in the subepidermal layer, is one of the major sensory organs in the peripheral nervous system of Drosophila. Clues as to Rhomboid's function are provided in an analysis of the role of Rhomboid in the determination of Ch organ precursor cells (COPs). The rhomboid gene and the activity of the Drosophila Epidermal growth factor receptor (Egf-R) signaling pathway are necessary to specifically induce three of the eight COPs in an embryonic abdominal hemisegment. The cell-lineage analysis of COPs indicates that each of the eight COPs originate from an individual undifferentiated ectodermal cell. The eight COPs in each abdominal hemisegment seem to be determined by a two-phase induction: first, five COPs are determined by the action of the proneural gene atonal and neurogenic genes. Subsequently, these five COPs start to express the rho gene, and rho activates the Efg-R-signaling pathway in neighboring cells and induces argos expression. Three of these argos-expressing cells differentiate into the three remaining COPs and they prevent neighboring cells from becoming extra COPs. In the five atonal dependent COPs, Egf-R signaling activity is required, but this signaling does not seem to involve the cell autonomous activity of Rho. In rho null mutants five chordotonal organs remain intact. However, rho expression is required to activate Egf-R in adjacent cells, and these three adjacent cells express the neuronal marker asense. Argos functions from the second wave of cells as a lateral inhibitor, restricting the number of recruited cells to the original three. As the rho-expressing first wave of COPs is adjacent to the three argos and asense expressing double postive COPs, Argos may function to prevent the continuance of Egf-R-signal activation in additional neighboring cells. A model is favored in which Rho protein is required for the activation of an Egf-R ligand, the Spitz transmembrane protein, by processing it into the functional soluble form. An alternative model, invalid at least in Ch organ determination but still valid for follicle cell determination in oogenesis, suggests that Rho protein is expressed in cells that require the activation of the Egf-R signaling pathway, and that Rho protein interacts with Egf-R protein directly or indirectly to amplify Egf-R signaling (Okabe, 1997).

The function of Atonal is best illustrated by its role in chordotonal organ development. A scolopidium, the basic unit of chordotonal organs, consists of four cells: a neuron with a single dendrite, the scolopale cell, cap cell and ligament cell. The scolopale cell (a glial cell) forms a sheath around the dendrite, while the cap cell and ligament cell mediate the attachment of the chordotonal organ to the body cell. Expression of atonal is restricted to a subset of atonal-requiring chordotonal precursors, called founder precursors (zur Lage, 1997).

The transcription factors encoding genes tailless (tll), atonal (ato), sine oculis (so), eyeless (ey) and eyes absent (eya), and Efgr signaling play a role in establishing the Drosophila embryonic visual system. The embryonic visual system consists of the optic lobe primordium, which, during later larval life, develops into the prominent optic lobe neuropiles, and the larval photoreceptor (Bolwig's organ). Both structures derive from a neurectodermal placode in the embryonic head. Expression of tll is normally confined to the optic lobe primordium, whereas ato appears in a subset of Bolwig’s organ cells that are called Bolwig’s organ founders. Phenotypic analysis of tll loss- and gain-of-function mutant embryos using specific markers for Bolwig’s organ and the optic lobe, reveals that tll functions to drive cells to an optic lobe fate, as opposed to a Bolwig’s organ fate. Similar experiments indicate that ato has the opposite effect, namely driving cells to a Bolwig’s organ fate. Since tll and ato do not regulate one another, a model is proposed wherein tll expression restricts the ability of cells to respond to signaling arising from ato-expressing Bolwig’s organ pioneers. The data further suggest that the Bolwig’s organ founder cells produce Spitz (the Drosophila TGFalpha homolog) signal, which is passed to the neighboring secondary Bolwig’s organ cells where it activates the Epidermal growth factor receptor signaling cascade and maintains the fate of these secondary cells. The regulators of tll expression in the embryonic visual system remain elusive, no evidence for regulation by the 'early eye genes' so, eya and ey, or by Egfr signaling is found (Daniel, 1999).

Epidermal growth factor receptor is activated in midline regions of the head neurectoderm, in particular in the anlage of the visual system. Moreover, increased and/or ectopic activation of Egfr results in a 'cyclops' phenotype very similar to what has been described for ectopic tll expression. Egfr signaling has been shown to be required in both chordotonal organs and compound eye for the inductive signaling triggered by ato expression. Two questions raised by these observations have been investigated: (1) is Egfr signaling required for tll expression in the optic lobe and (2) is Egfr signaling involved in the recruitment of the secondary (non-atonal-expressing) Bolwig’s organ cells? The answer to both these questions is no. tll expression is unaltered when levels of Egfr signaling are manipulated, suggesting that Egfr signaling is not required for tll expression. To investigate the second question, a test was performed for the presence of Egfr-relevant mRNAs or proteins: Rhomboid mRNA, which would be expected to be present only in the signaling cells, and phosphorylated MAPK, Pointed and Argos mRNAs, which would be expected to be expressed in all cells receiving an Egfr-mediated signal. In stage 12 embryos, rho is expressed only in the small group of Bolwig’s organ founder cells (the same cells expressing ato). In contrast, activated (phosphorylated) MAPK is present in a larger cluster of cells including the entire Bolwig’s organ and part of the adjacent optic lobe. Consistent with this, pnt and aos, both known to be switched on in cells receiving the Spi signal, are expressed at the same stage throughout the entire Bolwig’s organ primordium. These gene expression and MAPK activation patterns are consistent with the idea that the Spi signal is activated by rho in the Bolwig’s organ founders and passed to the neighboring secondary Bolwig’s organ cells where it activates the Egfr signaling cascade. Supporting this view, only 3-4 photoreceptor neurons are found in the Bolwig’s organ of embryos lacking rho or spi; furthermore, the size of the posterior lip of the optic lobe is reduced in such embryos. The fact that absence of secondary Bolwig’s organ cells in rho or spi mutant embryos can be rescued by blocking cell death in the background of a deficiency that takes out the reaper complex of genes indicates that the Spi signal is not necessary for the specification (recruitment) of secondary Bolwig’s organ cells, but rather, for their maintenance (Daniel, 1999).

EGF receptor signaling regulates pulses of cell delamination from the Drosophila ectoderm: argos is a high-threshold target of EgfR signaling

Many different intercellular signaling pathways are known but, for most, it is unclear whether they can generate oscillating cell behaviors. Time-lapse analysis of Drosophila embryogenesis has been used to show that oenocytes delaminate from the ectoderm in discrete bursts of three. This pulsatile process has a 1 hour period, occurs without cell division, and requires a localized EGF receptor (EGFR) response. High-threshold EGFR targets are sequentially activated in rings of three cells, prefiguring the temporal pattern of delamination. Surprisingly, widespread misexpression of the relevant activating ligand, Spitz, is compatible with robust delamination pulses. A single chordotonal organ precursor (called C1) and its progeny provide the source of secreted Spi relevant for oenocyte induction. Although Spitz ligand becomes limiting after only two pulses, artificially prolonging its secretion generates up to six additional cycles, revealing a rhythmic underlying mechanism. These findings illustrate how intercellular signaling and cell movements can generate multiple cycles of a cell behavior, despite individual cells experiencing only one cycle of receptor activation (Brodu, 2004).

The induction of larval oenocytes in Drosophila has been used as a simple model system for investigating the developmental regulation of EGFR signaling. Oenocytes are induced from the dorsal ectoderm of abdominal segments by a fixed and highly restricted source of Spi. This triggers a local EGFR response within a ring of overlying dorsal ectodermal cells, termed a whorl, leading to the upregulation of numerous oenocyte-specification genes and subsequent cell delamination. The simple cell geometry of the oenocyte whorl, together with time-lapse microscopy, was used to explore the timing of Spi secretion, EGFR-target activation, early cell induction, and later cell delamination. These studies reveal that oenocytes delaminate in bursts of three and identify the cell-counting mechanism as an EGFR-dependent pulse generator converting the time window of Spi secretion into final oenocyte number. This represents the first example of a rhythmic cell behavior other than the cell cycle to be reported in the Drosophila embryo (Brodu, 2004).
Rather than delaminating from the ectoderm in a continuous stream, oenocyte precursors segregate in discrete well-separated bursts of three cells. Genetic backgrounds affecting the pattern of cell segregation but not early fate specification were used to show how these pulses are regulated by EGFR signaling. The signaling parameters regulating the time of onset, time of cessation, and in particular, the cyclical nature of cell delamination of oenocytes are discussed (Brodu, 2004).

Spi secretion begins during stage 10, triggering weak activated Rolled/ERK but not the first morphological readout for oenocyte induction, the sickle-shape change, until 1 hr later. This early inhibition of EGFR induction occurs upstream of Pointed P1 and requires Delta-dependent Notch signaling. Although the supply of Spi ligand is not rate limiting for initiating induction, it does specify the final number of delamination pulses. In turn, this depends upon the duration of Rhomboid-1 expression by the C1 lineage, which is regulated by the Hox protein Abdominal-A. In this regard, it is interesting that oenocyte number is higher than six in many other winged insects. For example, in the parasitic wasp Phaenoserphus viator, oenocyte clusters of 'about 20 cells' have been reported, tempting speculation that this species may undergo seven rather than two delamination pulses (Brodu, 2004).

The sequence of events during wild-type oenocyte induction and delamination was identified using time-lapse movies. EGFR signaling initially induces all six precursors within a whorl to adopt a sickle-shape change within 10 min. There then follow two complete cycles of pulsatile delamination. Each 1 hr cycle comprises a 45 min pause, during which time no precursors leave the ectoderm, followed by a 15 min delamination phase, where three cells segregate rapidly, at 7.5 min intervals. Each cycle is reset when the inner-ring triplet delaminates and migrates away from the whorl site, allowing the remaining outer-ring cells to move into the inner position before they too delaminate (Brodu, 2004).

The mechanism involved in pulse generation was revealed, at least in part, by testing the roles of several different EGFR-signaling parameters. Surprisingly, although Spi ligand is essential for oenocyte induction and delamination, it plays only a permissive role in pulse generation. Thus, overexpression of Rhomboid-1 or secreted Spi in a widespread or prolonged manner does not suppress pulses of delamination nor alter their initial frequency, but it does produce up to six additional cycles. This leads to three main conclusions: (1) although only two pulses normally occur, the underlying mechanism is cyclical and has the potential to generate many more; (2) neither the frequency nor the number of cells per cycle are altered by increasing Spi-ligand levels; (3) pulses do not need Spi secretion to be pulsatile or even restricted to the Spitz normal source, C1. In addition, C1 does not provide any other essential rhythmic cue, because when it is eliminated, resupplying widespread Rhomboid-1 can rescue periodic delamination (Brodu, 2004).

In contrast to constitutive Spi secretion, widespread activation of the EGFR or its downstream effector, Ras1, disrupts delamination pulses. Loss of rhythmicity is also observed when the EGFR pathway is deregulated by removing the Yan or Argos inhibitors. Together, these functional data demonstrate that the spatiotemporal pattern and/or the levels of EGFR activation and downstream signal transduction are critical for pulse generation. For Ras1 overactivation or argos inactivation, it was also shown that some oenocytes fail to switch on a late differentiation marker at the appropriate time. Thus, one function of pulses may be to ensure cell-to-cell consistency in the duration or level of the oenocyte EGFR response, in turn promoting homogeneous cell differentiation (Brodu, 2004).

Using a panel of markers for double- and single-ring stages, it was possible to place gene expression 'snapshots' in temporal order with the cell movements recorded in movies. Three generic EGFR targets (activated Rolled/ERK, Yan, and argos) and three oenocyte-specific EGFR targets (Sal, svp^lacZ, and svp^lacZΔ18) were analyzed. In wild-type embryos, the high-threshold EGFR outputs of argos and svp^lacZ expression, detectable Rolled activation, and strong Yan downregulation are all inner ring specific, whereas lower-threshold outputs such as Sal upregulation and svp^lacZΔ18 expression are present in both precursor rings. Delamination itself also appears to be a high-threshold EGFR response and is thus confined to the inner ring (Brodu, 2004).

argos is a particularly interesting high-threshold target, as its expression is normally confined to the inner ring but its activity is required in the outer ring to tone down the EGFR response, as measured by Rolled activation. This remote inhibitor role is consistent with several previous studies, and real-time analysis shows that it promotes oenocyte pulses by preventing premature outer-ring delamination. During wild-type embryogenesis, such negative feedback would be transiently downregulated each time the inner-ring source of Argos is physically removed via delamination, thus facilitating upregulation of the EGFR response in the next triplet. In addition, Argos may play a more subtle autocrine role in ring-1, since loss of function not only eliminates a second 45 min pause phase completely but also partially reduces the first pause to 25 min (Brodu, 2004).

Together, the real-time cell tracking, the expression analysis of the EGFR response, and the oenocyte counts in EGFR pathway altered backgrounds are consistent with the notion that pulses require at least some components of the high-threshold EGFR response to be more strongly expressed by inner- than outer-ring cells. It thus follows that one critical molecular transition underlying pulse generation occurs after each round of delamination, when cells of the outer triplet move centrally and upregulate a subset of EGFR-target genes (Brodu, 2004).

At least two distinct mechanisms ensure that strong expression of high-threshold EGFR targets is restricted to the dynamic population of inner-ring cells. The first of these arises from inner-ring cells being closer to C1 and therefore exposed to higher levels of secreted Spi. Hence, when Spi ligand is widely overexpressed, high-threshold EGFR readouts such as detectable activated Rolled/ERK and argos expand ectopically into the outer ring (Brodu, 2004).

A second mechanism that is not dependent on localized Spi-ligand secretion also enhances the inner-ring EGFR response. This was initially revealed in four different genetic backgrounds where Spi secretion is delocalized yet pulses remain. In UAS-rho1 embryos, real-time and EGFR-target analyses showed that, despite Spi secretion throughout the En stripe, oenocyte delamination and the full repertoire of inner-ring markers, including strong svp^lacZ expression and Yan downregulation, remain confined to the inner ring. It was not possible, however, to detect such a clear and consistent inner-versus-outer difference in levels with activated Rolled/ERK and argos expression, either reflecting technical limitations or indicating that some high-threshold EGFR targets remain more tightly restricted than others. Nevertheless, these studies provide evidence that, when exposed to the same Spi ligand concentration, inner-ring precursors express some components of the oenocyte EGFR response more strongly than their neighbors. One molecular explanation for this bias is revealed by the reduced sensitivity of inner-ring cells to the delamination-blocking effects of argos overexpression. Thus, the argos sensitivity difference may account for why pulses remain in UAS-rho1 embryos. In wild-type embryos, both this mechanism and graded Spi ligand would be expected to contribute to promoting robust pulses. The basis for differential argos sensitivity is not yet understood but it likely initiated independently of EGFR signaling. In addition, the parameters regulating whorl geometry and thus setting the size of delamination quanta to three cells remain unclear. In this regard, it is intriguing that among all the EGFR pathway components tested, only activated Ras1 produced oenocyte counts suggestive of an altered quantal size, in this case two cells (Brodu, 2004).

The EGFR-dependent pulse generator drives rhythmic clearance of cells from their induction site, one solution to the problem of how to induce a large number of cells using a point source of short-range signal. Coupling intercellular signaling to cell movement in this way also allows the generation of multiple output cycles, even though individual cells experience only one intracellular cycle of EGFR activation. This contrasts with the vertebrate segmentation clock, where cells undergo multiple intracellular oscillations of gene expression, in this case involving Notch signaling. One aspect that is shared with many oscillating systems, including the segmentation clock, is the essential contribution of negative feedback, which in the oenocyte context is mediated by Argos. The relative simplicity of the oenocyte oscillator may prove particularly amenable for constructing and testing future mathematical models of intercellular signaling rhythms. Similar real-time analyses of other inductive processes, especially those of a reiterative nature, should clarify whether pulsatile cell behaviors are commonly associated with EGFR and other intercellular signaling pathways (Brodu, 2004).

Larval

In the developing eye, argos is expressed behind the morphogenetic furrow. It is not expressed in mystery cells, extra non-neuronal cells present in normal development. In pupal retinae, argos is found in cone cells, in photoreceptor cells, and in primary pigment cells (Freeman, 1992). All cell types in the developing eye (except bristles) are sensitive to Argos concentration: over-expression leads to too few cells forming, the opposite phenotype of that seen in argos loss-of-function mutants (Freeman, 1994 and Sawamoto, 1994).

Use of a dominant-negative form of the EGF-R in the eye reveals that EGF-R is required for differentiation of all photoreceptor cell types (R1-R8), including R7 which is also subject to the Sevenless signal. DN-EGF-R is truncated in the 13 amino acids C-terminal to the transmembrane domain. Receptor tyrosine kinases dimerize and transphosphorylate each other upon activation. The removal of the intracellular domain produces a dominant-negative function because receptor molecules without the intracellular tyrosine kinase domain can dimerize with wild-type receptors, but the dimer is unable to signal. Expression of DN-EGF-R behind the morphogenetic furrow causes complete loss of the adult retina. As well as eight photoreceptors, each ommatidium comprises four cone cells and eight pigment cells. Expression of DN-EGF-R in the presumptive cone or pigment cells leads to them not differentiating. Overexpression of secreted Spitz the ligand of EGF-R causes overrecruitment of all cell types in the ommatidium. Spitz has extracellular protease cleavage sites that allow a fragment with an EGF repeat to be released. Overexpression of membrane-bound full-length Spitz has no effect on eye development. In all cases the source of the extra photoreceptors is the same: transformation into photoreceptors of the "mystery cells" (early members of the cluster, later destined to leave and apparently rejoin the surrounding undetermined cells) (Freeman, 1996).

Just as with EGF-R, overexpression of activated Sevenless recruits extra cells into the ommatidium. Sevenless is also able to recruit additional cone and pigment cells when expressed in the pupal retina. Sevenless can also replace EGF-R function in the wing. Finally, overexpression of secreted Spitz can replace the need for Sevenless. It is concluded that there is no significant difference in the intracellular effects of activation of these two RTKs, even in the R7 cell, where both receptors are required (Freeman, 1996).

A model is proposed for eye development based on these and other observations. First, Spitz activation of EGF-R can trigger all the cell types in the ommatidium, the choice of fate being dependent on when the activation occurs. Argos is an extracellular inhibitor of DER activation (Schweitzer, 1995). Third, the expression of Argos is dependent of EGF-R activation, establishing a negative feedback loop (Golembo, 1996). Fourth, Argos can diffuse further than Spitz. Fifth, the successive waves of induction of each cell type (photoreceptors, cone cells, primary pigment cells, and second/tertiary pigment cells) occur in concentric rings around the ommatidium: each cluster resembles a bullseye. In this model, Spitz is initially produced by the three central cells R8, R2 and R5 and that this recruits the immediately neighboring cells and photoreceptors. In R7, the later activation of Sevenless by its ligtand, Boss, is also required. As cells differentiate, they express Argos, which diffuses outwards, preventing more distal cells from responding to Spitz; Argos is unable to block cells that have already started to differentiate or cells that are exposed to high level of Spitz. Later, more cells start to produce Spitz, overcoming Argos inhibition in the nearest cells. This allows the next concentric ring of cells around the photoreceptors to be recruited, but now as a different cell type, cone cells. Again, Argos prevents more remote cells from responding by diffusing beyond the cone cells (now themselves producing it). Later still, the Spitz source expands again, now recruiting the pigment cells (Freeman, 1996).

All imaginal discs in Drosophila are made up of a layer of columnar epithelium or the disc proper and a layer of squamous epithelium called the peripodial membrane. Although the developmental and molecular events in columnar epithelium or the disc proper are well understood, the peripodial membrane has gained attention only recently. Using the technique of lineage tracing, it has been shown that peripodial and disc proper cells arise from a common set of precursors cells in the embryo, and that these cells diverge in the early larval stages. However, peripodial and disc proper cells maintain a spatial relationship even after the separation of their lineages. The peripodial membrane plays a significant role during the regional subdivision of the wing disc into presumptive wing, notum and hinge. The Egfr/Ras pathway mediates this function of the peripodial membrane. These results on signaling between squamous and columnar epithelia are particularly significant in the context of in vitro studies using human cell lines that suggest a role for the Egfr/Ras pathway in metastasis and tumor progression (Pallavi, 2003).

A significant finding of this study is the role of the peripodal membrane (PM) in wing/notum/hinge decision. The wing disc is initially divided into anterior and posterior compartments by virtue of En expression only in a subset of disc cells. Later, it is subdivided into three distinct groups of cells: wing, notum and the hinge. This is marked by the expression of Wg in the presumptive wing region -- Pnr in the presumptive notum and Tsh in the presumptive hinge. PM cells over the notum and the pouch may provide positional cues for notum/hinge-wing decision. The Egfr pathway functions in the PM to specify notum-specific genes and/or to inhibit wing-specific genes. Mis-expression of dominant negative Egfr (DNDER), DN-Ras, DN-Raf or Argos (Aos; a negative regulator of the Egfr pathway) in the PM is enough to induce notum/hinge-to-wing transformations. The dynamic expression pattern of Aos marks the spatial and temporal pattern of Egfr activation. In the second instar larval stage, when wing-notum decision is made, Aos is expressed specifically in those PM cells that overlay the posterior notum. Once the wing-notum decision is made, Aos expression recedes from PM cells and it starts expressing in the notum-associated myoblasts and in the pouch. Although, at this stage. the possibility cannot be ruled out that the Egfr pathway is required in both PM and DP cells to specify notum fate, the results suggest that the Egfr pathway mediates interactions between PM and disc proper (DP) cells during the notum/hinge specification (Pallavi, 2003).

Oogenesis

Intercellular signaling through the EGF receptor (EGFR) patterns the Drosophila egg. The TGF alpha-like ligand Gurken signals from the oocyte to the receptor in the overlying somatic follicle cells. In the dorsal follicle cells, this initial paracrine signaling event triggers an autocrine amplification by two other EGFR ligands: Spitz and Vein. Spitz becomes an effective ligand only in the presence of the multitransmembrane domain protein Rhomboid. Consequent high-level EGFR activation leads to localized expression of the diffusible inhibitor Argos, which alters the profile of signaling. This sequential activation, amplification, and local inhibition of the EGFR forms an autoregulatory cascade that leads to the splitting in two of an initial single peak of signaling, thereby patterning the egg (Wasserman, 1998).

In other tissues Rhomboid appears to activate Spitz/Egfr signaling, leading to the suspicion that Rhomboid might mediate autocrine Spitz signaling in the follicle cells. Consistent with this idea, the phenotype caused by loss of Spitz from the follicle cells is similar to that caused by loss of Rhomboid. Expression of antisense rhomboid causes loss of dorsal tissue and fusion of the appendages in eggs from heat-shocked females expressing HS-as-rho. Unmarked follicle cell clones of a rhomboid null mutation also give fused appendage phenotypes; as with spitz clones, these range from mild to severe fusions. Like Spitz and the Egfr, Rhomboid is not needed in the oocyte, implying that it, too, is only required in the follicle cells (Wasserman, 1998).

In the absence of Egfr signaling, rhomboid expression is lost and, conversely, it is ectopically expressed in fs(1)K10 egg chambers. These expression profiles of spitz and rhomboid are consistent with Gurken signaling from the oocyte activating the expression of rhomboid in the follicle cells. This may in turn allow Spitz to become an autocrine ligand in the follicle cells and thus establish an autocrine amplification of the initial paracrine signal. The expression of the neuregulin-like Egfr ligand vein was also examined. It is also expressed in two stripes of follicle cells at stage 10b. Interestingly, vein expression is dependent on Egfr signaling: it is ectopically expressed in fs(1)K10 eggs and absent from gurken null eggs, establishing another potentially important feedback mechanism. This suggests that the autocrine amplification of Egfr signaling also involves Vein, although in this case the feedback occurs by direct transcriptional activation of the ligand (Wasserman, 1998). vein expression has also been found to be dependent on Egfr signaling during embryogenesis (T. Volk, personal communication to Wasserman, 1998).

The expression of the secreted Egfr inhibitor, Argos, is dependent on Egfr signaling in many tissues. Consistent with this, argos is expressed in the dorsal-anterior follicle cells at the time when Egfr signaling occurs. At stage 11 the RNA is detectable in a single, T-shaped group of cells centered on the dorsal midline, and by stage 13, argos, like rhomboid and vein, is found in two groups of cells: one on either side of the midline. As elsewhere, argos expression is dependent on Egfr activation: in gurken mutant egg chambers it is lost, and it is ectopically expressed in fs(1)K10 egg chambers. Is argos expression dependent on Spitz amplification of Egfr signaling? An examination was performed to see if Spitz contributes to a signaling threshold required to induce argos expression. argos expression is normal in eggs from mothers with reduced Ras1, but when Spitz is halved, dorsal-anterior argos expression is abolished in most egg chambers. Therefore, there is indeed a threshold of Egfr signaling required to switch on argos, and both Gurken and Spitz participate in reaching this threshold (Wasserman, 1998).

The initial expression of argos at the dorsal midline led to the speculation that it might cause a reduction of Egfr signaling near the midline, thereby splitting the single signaling peak in two. The resulting twin peaks of Egfr activation would then specify the location of the dorsal appendages. A prediction of this model is that loss of Argos should remove inhibition of the Egfr at the midline and produce a single peak of signaling, leading to the formation of a fused appendage phenotype. The eggs from females with hypomorphic argos mutations were examined. A significant proportion of these eggs have a partially or, in the most severe cases, fully fused phenotype. The same fused appendage phenotype is observed in follicle cell clones of an argos null mutation. These data imply that there is a requirement for Argos in eggshell patterning and that, as with Spitz, Rhomboid, and the Egfr, this requirement is confined to the follicle cells (Wasserman, 1998).

It is proposed that Argos modifies the initial Egfr activation profile in the follicle cells, producing twin peaks of activity displaced from the midline. These specify the position of the dorsal appendages. Direct evidence for a transition from one to two peaks of signaling was obtained with an antibody that recognises only the activated, diphosphorylated form of MAP kinase, a key member of the signal transduction pathway downstream of the receptor. At stages 9-10, there is a single domain of activated MAP kinase in the follicle cells, centered on the dorsal midline. By stage 11, two domains, are observed: one on each side of the dorsal midline. From their position, these cells correspond to the cells that will form the dorsal appendages. In Egfr hypomorphs, which have a fused appendage phenotype, the single peak of activated MAP kinase does not split in two. These results clearly demonstrate that Egfr signaling does indeed evolve from a single peak into twin peaks of activation. This is supported by examining the expression pattern of known Egfr target genes in the follicle cells. By stage 11 these targets (pointed, rhomboid, argos, vein, and Broad) are expressed in two dorsal anterior domains, one on each side of the midline. This is taken as additional evidence for twin peaks of Egfr activation. Earlier, pointed, rhomboid, and argos are all also detectable in a single peak at the dorsal midline (Wasserman, 1998).

Egfr signaling specifies the dorsoventral axis and patterns the eggshell. It is suggested that these two functions are controlled by temporally separate phases of Egfr activation. When amplification and splitting of Egfr signaling do not occur, eggs have only a single, fused appendage. Surprisingly, larvae emerge from these eggs at the frequency predicted by Mendelian principles, and those that emerge have no apparent dorsoventral defects. When follicle cell clones of a spitz null are induced, the hatching rate of eggs with fused appendages os 82% of the predicted number. Similarly, all of the predicted number of eggs with a single fused appendage hatch from mutant females. The same is true of eggs with fused appendages caused by follicle cell clones of argos null mutations. Therefore, disruption of the amplifying and splitting process does not perturb dorsoventral axis specification, implying that the initial Gurken signal to the Egfr is sufficient to specify the axis. The subsequent cascade of amplification and splitting then patterns the eggshell (Wasserman, 1998).

Autocrine signaling through the Epidermal growth factor receptor operates at various stages of development across species. A recent hypothesis has suggested that a distributed network of Egfr autocrine loops is capable of spatially modulating a simple single-peaked input into a more complex two-peaked signaling pattern, specifying the formation of a pair organ in Drosophila oogenesis (two respiratory appendages on the eggshell) (Wasserman and Freeman, 1998). To test this hypothesis, genetic and biochemical information about the Egfr network were integrated into a mechanistic model of transport and signaling. The model allows the relative spatial ranges and time scales of the relevant feedback loops to be estimated, the phenotypic transitions in eggshell morphology to be interpreted and the effects of new genetic manipulations to be predicted. It has been found that the proposed mechanism with a single diffusing inhibitor is sufficient to convert a single-peaked extracellular input into a two-peaked pattern of intracellular signaling. Based on extensive computational analysis, it is predicted that the same mechanism is capable of generating more complex patterns. At least indirectly, this can be used to account for more complex eggshell morphologies observed in related fly species. It is proposed that versatility in signaling mediated by autocrine loops can be systematically explored using experiment-based mechanistic models and their analyses (Shvartsman, 2002).

The mechanism that converts a single-peaked Gurken input into a more complex pattern of MAPK signaling in the responding epithelial cells has been analyzed. Interest in this particular event in Drosophila oogenesis mechanism is twofold. First, it provides an excellent example of versatility in signaling, demonstrating how simple stimuli define complex patterns in development. More importantly, this mechanism is arguably one of the best studied at the genetic and biochemical level. It therefore provides a good target for the development of modeling methodologies and experimental tests of modeling predictions (Shvartsman, 2002).

The oocyte-derived signal is modulated by a network of feedback loops in the follicular epithelium. One positive feedback loop is established when the Egfr, which acts through the Ras/MAPK pathway, induces the expression of rhomboid. Rhomboid is an intracellular protease that, together with intracellular protein Star, is responsible for processing and secretion of Spitz, another TGF-beta like ligand of the Egfr. The secreted Spitz directly interacts with the Egfr, further stimulating the intracellular MAPK. Thus, Rhomboid acts as a positive regulator of EGFR signaling. This positive feedback, together with another one mediated by a secreted EGFR ligand Vein, both amplifies and spatially expands MAPK signaling induced by Gurken (Shvartsman, 2002 and references therein).

The amplified and expanded signal is then downregulated by a number of negative feedback loops in the Drosophila Egfr system. Although there are large numbers of negative regulators of Egfr signaling in Drosophila, the original mechanism invokes a single endogenous inhibitor, Argos (Wasserman, 1998). The expression of argos is induced at high levels of MAPK signaling. Argos is a secreted protein that directly interacts with the extracellular domain of the Drosophila Egfr, and inhibits intracellular MAPK signaling induced by Gurken, Vein and Spitz. According to Wasserman, the amplification of the Gurken signal by Spitz and Vein, and its inhibition by Argos first expands the domain of MAPK signaling and then splits it into two smaller domains, establishing in this way the two groups of appendage-producing cells. In this mechanism, the two peaks detected in the MAPK signaling pattern (Wasserman, 1998) are closely linked and co-localized with the two stripes that have been repeatedly observed in the pattern or rhomboid expression (Shvartsman, 2002).

The mechanism of Wasserman links gurken, Egfr, spitz, rhomboid, vein and argos into a system of interconnected feedback loops that are jointly regulated by the EGFR and intracellular MAPK signaling (Wasserman, 1998). Although the mechanism is supported by genetic and biochemical data, several important questions remain unanswered. (1) Is the proposed network actually capable of converting single-peaked inputs into persistent two-peaked outputs? An independent measurement of the MAPK dynamics in oogenesis reports patterns that are more complex, and cannot be as straightforwardly correlated with the number of appendages of the mature egg. (2) Although the necessary role of argos is supported by genetic approaches, the role of other known Egfr inhibitors remains to be clarified. Argos is the only secreted inhibitor of Egfr in Drosophila. If it is the only necessary inhibitor, does this mean that the other negative regulators merely serve to modulate the basic pattern established by the secreted Argos? This must be reconciled with the phenotypic transitions in eggshell morphology induced by changing the levels of other inhibitors. (3) Several papers have indicated that the expression of argos is detected later than the relevant changes in the rhomboid expression. Can this be used to argue against the mechanism with a single secreted inhibitor? To summarize, the sufficiency of the proposed mechanism with a single diffusing inhibitor and its consistency with the large body of genetic data remains to be established (Shvartsman, 2002).

Based on the computational analysis of this model, the proposed mechanism of pattern formation by peak splitting (Wasserman, 1998) has been validated. The proposed network of positive- and negative-feedback loops in the Egfr system can indeed convert a quasistatic, single-peaked input, into stable, two-peaked outputs. Analysis of the parametric dependence of the stationary patterns in the model indicates that the underlying mechanism accounts for a number of experimentally observed phenotypic transitions (Shvartsman, 2002).

Stationary patterns predicted by the model fall into several qualitatively different classes characterized by the number of peaks in the signaling profiles. The transitions between the classes are discontinuous; this might explain why numerous experiments can be classified in terms of a small number of phenotypes. Stable two-peaked patterns not only exist in this model, but are also kinetically accessible from the state of zero stimulation and are realized through inputs with a single maximum. These stable two-peaked patterns are robust for a wide range of network inputs and strengths of the positive and negative feedback loops. The existence of these stationary patterns requires intermediate input amplitudes and widths, intermediate strengths of the positive feedback, as well as a sufficiently strong and long-ranged inhibition. The fact that large parameter changes induce transitions to qualitatively different patterns is consistent with a large body of genetic data (Shvartsman, 2002).

Several of the predictions of the original mechanism are at variance with experiments. The first quantitative disagreement is observed in the behavior of the two-peaked patterns upon increases in the doses of the stimulatory signal or the strength of the positive feedback. In the model, these changes produce a discontinuous transition to the pattern with a single broad peak; this collapse of the two peaks is preceded by their slight separation. In experiments, eggs laid by the females with extra copies of gurken, heat-shock activated rhomboid, or deficient in Cbl (a negative Egfr regulator) have significantly increased inter-appendage distance. However, a broad appendage phenotype has been reported in a recent experiment that had used tissue-specific gene expression to increase the level of oocyte-derived Gurken (Shvartsman, 2002 and references therein).

A more serious problem is related to the role of argos. First, it is unclear why the onset of argos expression is detected much later than the major changes in the patterns of rhomboid and MAPK activity. The second question, based on the analysis of the model, is related to the relative position in the maxima of rhomboid and argos gene expression patterns. According to the model, the maxima in the expression of the two genes should be co-localized; this is an immediate consequence of the fact that both genes require receptor activity for their production. At the same time, several independent measurements find that for a while argos is expressed between the two regions of high rhomboid expression. If the resulting two-peaked signaling patterns are quasi-stationary, as suggested by the analysis of the original mechanism, then it is unclear what maintains argos expression in the region of decreased MAPK activity. One possibility is that the rate of argos expression is much slower than that of rhomboid, and that the experimentally observed patterns should be interpreted as a transient in the model. This mechanism has been computationally explored and rejected; making the generation of Argos much slower than that of Rhomboid generates pathologic oscillatory instabilities of the two-peaked patterns. Another possible explanation for the observed relative location of the maxima in the gene expression patterns might involve an additional feedback loop. In this extended mechanism, a yet undiscovered positive feedback regulates the expression of argos, making it insensitive to decreases in Egfr signaling after peak splitting (Shvartsman, 2002).

In order for the peak-splitting mechanism to work, the differences in the thresholds of Argos and Rhomboid production must not be too large. In the simulations, the difference in these thresholds is 25%. It has been found that in the case when Argos generation is characterized by a significantly higher threshold than that of Rhomboid, only the one-peaked patterns are realized and peak splitting does not occur. In fact, peak splitting requires a rather delicate balance between the spatially distributed stimulation by Gurken and inhibition by Argos (Shvartsman, 2002).

The analysis supports the original hypothesis about the differences in the spatial ranges of Argos and Spitz. The cause of this separation of length scales is still unclear. Argos and Spitz are secreted into the gap between the oocyte and the follicle cells, where their transport is accompanied by binding to Egfrs in the follicular epithelium. The strength of ligand/receptor binding can regulate the range of the secreted signal in this situation. If this is the case, then the binding constant characterizing the Argos/Egfr interaction should be less than that of Spitz. Surprisingly, it was found that Argos has a higher affinity for the Egfr (but, at the same time, an alternative Argos-like EGF mutant has a lower Egfr-binding affinity). Another mechanism regulating the range of secreted ligands relies on their receptor-mediated endocytosis. Based on the fact that all ligands of mammalian Egfrs are rapidly internalized, it is believed that this mechanism can control spatial ranges of Egfr ligands in Drosophila oogenesis. Furthermore, it is known that Argos interaction with Egfr prevents receptor dimerization and phosphorylation. Since these processes are required to initiate receptor-mediated endocytosis of receptor tyrosine kinases, this further supports the mechanism in which the ranges of Argos and Spitz are controlled by the rates of their internalization (Shvartsman, 2002).

The model does not rely on the difference between the diffusivities of Spitz and Argos. This is in contrast to the classical activator-inhibitor mechanism for morphogenesis proposed by Turing (1952). Instead, it relies on differences of ranges of these molecules, which depend on a combination of their diffusivities and rates of degradation. This is also true for the Turing mechanism; however, in the latter, the difference between the time constants of the activator and the inhibitor leads to oscillatory instabilities. Moreover, the model devised in this study is different from the activator-inhibitor models: while Argos plays the role of a long-range inhibitor, the positive feedback is operated by an autocrine switch with a non-diffusing component (Rhomboid) and a short-ranged diffusing messenger (Spitz). Here, it is the time constant of Spitz degradation that determines the range of the positive feedback; however, the existence of the oscillatory instability is determined by the ratio of the time scales of Rhomboid and Argos. Furthermore, it is not difficult to show that with the given relationship between the thresholds of Rhomboid and Argos production the Turing-like instability (as a result of the homogeneous increase of the level of Gurken input) is not realized at all in the current model. Another difference between this model and that of Turing is that in this model, pattern formation occurs as a result of the instabilities, leading to the abrupt formation and transitions between large-amplitude localized patterns. Large amplitude localized patterns, often referred to as autosolitons, are frequently encountered in different nonlinear systems (Shvartsman, 2002).

In addition to patterns with one and two peaks, the mechanism supports more complex patterns. If the number of peaks in the profile of the receptor activity determines the number of respiratory appendages, then this finding predicts that more complex eggshell phenotypes are to be expected upon quantitative variation of the parameters of the Egfr network. At this point there is a single published observation of eggs with four appendages in mutants of D. melanogaster. Occasionally, eggs with three appendages are laid by the mutants with defects in Gurken accumulation. Notably, eggs with four and even six appendages represent wild-type phenotypes of several related fly species: the eggs of D. virilis have four appendages, while the eggs laid by the flies of subgenus Pholadoris have six appendages. According to the current model, eggs with multiple appendages can be generated by increases in the width and amplitude of the stimulatory signal. Further experimental, modeling and computational studies are required to check whether these more complex phenotypes can be realized in oogenesis or whether they manifest a pathological feature of the mechanism with a single diffusing inhibitor (Shvartsman, 2002).

In Drosophila melanogaster, the patterning of dorsal appendages on the eggshell is strictly controlled by EGFR signaling. However, the number of dorsal appendages is remarkably diverse among Drosophila species. For example, D. melanogaster and D. virilis have two and four dorsal appendages, respectively. During oogenesis, the expression patterns of rhomboid (rho) and argos (aos), positive and negative regulators of EGFR signaling, respectively, are substantially different between D. melanogaster and D. virilis. Importantly, the number of locations and position of both rho expression and MAPK activation are consistent with those of the dorsal appendages in each species. Despite the differences in spatial expression, these results suggest that the function of EGFR signaling in dorsal appendage formation is largely conserved between these two species. Thus, these results link the species-specific activation of EGFR signaling and the evolution of eggshell morphology in Drosophila (Nakamura, 2003).

The twin-peaks model for D. melanogaster dorsal appendage induction suggests that initial EGFR activation is triggered by Grk localized at the dorsal midline of the oocyte at stage 8/9. This leads to the expression of rho in the overlying follicle cells that receive the Grk signal. In cooperation with Spi, Rho triggers the positive feedback loop of EGFR signaling at stage 10. Consequently, the EGFR signal reaches its highest level at the dorsal midline, leading to the subsequent induction of aos at the dorsal midline at stage 11. The resulting signaling profile has twin peaks that maintain rho expression as a stripe of cells on either side of the dorsal midline (Nakamura, 2003 and references therein).

Surprisingly, although MAPK activation was detected in the midline in D. virilis, the initial expression of rho was detected in two groups of follicle cells on the dorsolateral sides. Notably, no rho expression was detected in the dorsal midline. Furthermore, aos expression was not observed in the dorsal midline from stage 10 to 11, which suggests that Aos is not involved in inhibiting the EGFR signal in the dorsal midline. This observation suggests that in D. virilis the stripe of rho-expressing cells on either side of the midline is not specified by the sequential induction of rho and aos, in contrast to the twin-peaks model in D. melanogaster. Based on these results, it is speculated that a negative regulator of EGFR signaling, other than Aos, is induced primarily in response to the Grk signal, and leads to the expression of rho in two domains in each lateral half. It is speculated that a low level of EGFR signaling is sufficient to induce this putative inhibitor(s), and that this inhibitor(s) is induced in the broad area of the dorsoanterior region, because MAPK activation and rho expression are excluded from a broader area of this region in D. virilis than in D. melanogaster. This putative negative regulator(s) of EGFR signaling is probably induced prior to the induction of aos. Thus, aos may never be induced in the dorsal midline in D. virilis, because EGFR signaling never reaches a threshold that is high enough to induce it in this scenario. From studies in D. melanogaster, kekkon-1 and sprouty, two negative regulators of EGFR signaling, are strong candidates, although this issue has to be addressed in D. virilis. Alternatively, an enhancer of rho may gain a novel characteristic to be repressed in response to the high level of EGFR signaling (Nakamura, 2003).

A combinatorial code for pattern formation in Drosophila oogenesis

Two-dimensional patterning of the follicular epithelium in Drosophila oogenesis is required for the formation of three-dimensional eggshell structures. Analysis of a large number of published gene expression patterns in the follicle cells suggests that they follow a simple combinatorial code based on six spatial building blocks and the operations of union, difference, intersection, and addition. The building blocks are related to the distribution of inductive signals, provided by the highly conserved epidermal growth factor receptor and bone morphogenetic protein signaling pathways. The validity of the code is demonstrated by testing it against a set of patterns obtained in a large-scale transcriptional profiling experiment. Using the proposed code, 36 distinct patterns were distinguished for 81 genes expressed in the follicular epithelium, and their joint dynamics were characterize over four stages of oogenesis. The proposed combinatorial framework allows systematic analysis of the diversity and dynamics of two-dimensional transcriptional patterns and guides future studies of gene regulation (Yakoby, 2008b).

Drosophila eggshell is a highly patterned three-dimensional structure that is derived from the follicular epithelium in the developing egg chamber. The dorsal-anterior structures of the eggshell, including the dorsal appendages and operculum, are formed by the region of the follicular epithelium, which is patterned by the highly conserved epidermal growth factor receptor (EGFR) and bone morphogenetic protein (BMP) signaling pathways. The EGFR pathway is activated by Gurken (GRK), a transforming growth factor α-like ligand secreted by the oocyte. The BMP pathway is activated by Decapentaplegic (DPP), a BMP2/4-type ligand secreted by the follicle cells stretched over the nurse cells (Yakoby, 2008b).

Acting through their uniformly expressed receptors, these ligands establish the dorsoventral and anteroposterior gradients of EGFR and DPP signaling and control the expression of multiple genes in the follicular epithelium. Under their action, the expression of a Zn finger transcription factor, Broad (BR), evolves into a pattern with two patches on either side of the dorsal midline. The BR-expressing cells form the roof (upper part) of the dorsal appendages. Adjacent to the BR-expressing cells are two stripes of cells that express rhomboid (rho), a gene that is directly repressed by BR and encodes ligand-processing protease in the EGFR pathway. These cells form the floor (lower part) of the appendages (Yakoby, 2008b).

The patterns of genes expressed during the stages of egg development that correspond to appendage morphogenesis are very diverse. At the same time, inspection of a large number of published patterns suggests that they can be 'constructed' from a small number of building blocks. For instance, the T-shaped pattern of CG3074 is similar to the domain 'missing' in the early pattern of br, while the two patches in the late pattern of br appear to correspond to the two 'holes' in the expression of 18w. Based on a number of similar observations, it was hypothesized that all of the published patterns could be constructed from just six basic shapes, or primitives, which reflect the anatomy of the egg chamber and the spatial structure of the patterning signals (Yakoby, 2008b).

In computer graphics, representation of geometrical objects in terms of a small number of building blocks is known under the name of constructive solid geometry, which provides a way to describe complex shapes in terms of just a few parameters -- the types of the building blocks, such as cylinders, spheres, and cubes, their sizes, and operations, such as difference, union, and intersection. Thus, information about a large number of structures can be stored in a compact form of statements that contain information about the types of the building blocks and the operations from which these structures were assembled. This study describes a similar approach for two-dimensional patterns and demonstrate how it enables the synthesis, comparison, and analysis of gene expression at the tissue scale (Yakoby, 2008b).

The six building blocks used in the annotation system can be related to the structure of the egg chamber and the spatial distribution of the EGFR and DPP signals. The first primitive, M (for 'midline'), is related to the EGFR signal. It reflects high levels of EGFR activation and has a concave boundary, which can be related to the spatial pattern of GRK secretion from the oocyte. The second primitive, denoted by D (for 'dorsal'), reflects the intermediate levels of EGFR signaling during the early phase of EGFR activation by GRK, and is defined as a region of the follicular epithelium that is bounded by a level set (line of constant value) of the dorsoventral (DV) profile of EGFR activation. The boundary of this shape is convex and can be extracted from the experimentally validated computational model of the GRK gradient. The third primitive, denoted by A (for 'anterior'), is an anterior stripe which is obtained from a level set of the early pattern of DPP signaling in the follicular epithelium. This pattern is uniform along the DV axis, as visualized by the spatial pattern of phosphorylated MAD (P-MAD). Thus, the D, M, and A primitives represent the spatial distribution of the inductive signals at the stage of eggshell patterning when the EGFR and DPP pathways act as independent AP and DV gradients (Yakoby, 2008b).

Each of the next two primitives, denoted by R (for 'roof') and F (for 'floor'), is composed of two identical regions, shaped as the respective expression domains of br and rho, and reflect spatial and temporal integration of the EGFR and DPP pathways in later stages of eggshell patterning. The mechanisms responsible for the emergence of the F and R domains are not fully understood. It has been shown that the R domain is established as a result of sequential action of the feedforward and feedback loops within the EGFR and DPP pathways. The formation of the F domain requires the activating EGFR signal and repressive BR signal, expressed in the R domain. Thus, at the current level of understanding, the R and F domains should be viewed as just two of the shapes that are commonly seen in the two-dimensional expression patterns in the follicular epithelium. The sixth primitive, U (for 'uniform'), is spatially uniform and will be used in combination with other primitives to generate more complex patterns (Yakoby, 2008b).

While a number of patterns, such as those of jar and Dad, can be described with just a single primitive, more complex patterns are constructed combinatorially, using the operations of intersection (∩), difference ( ), and union (∪) For example, the dorsal anterior stripe of argos expression is obtained as an intersection of the A and D primitives (A∩D). The ventral pattern of pip is obtained as a difference of the U and D primitives (U D). The pattern of 18w is constructed from the A, D, and R primitives, joined by the operations of union and difference (A∪D R). For a small number of published patterns, the annotations reflect the experimentally demonstrated regulatory connections. For example, the U D annotation for pip reflects that actual repression of pip by the dorsal gradient of EGFR activation. For a majority of genes, the annotations should be viewed as a way to schematically represent a two-dimensional pattern and as a hypothetical description of regulation (Yakoby, 2008b).

The geometric operations of intersection, difference, and union can be implemented by the Boolean operations performed at the regulatory regions of individual genes. Boolean operations evaluate expression at each point and assign a value of 0 (off) or 1 (on). As an example, consider a regulatory module, hypothesized for argos, that performs a logical AND operation on two inputs: the output of the module is 1 only when both inputs are present. When both of the inputs are spatially distributed, the output is nonzero only in those regions of space where both inputs are present, leading to an output that corresponds to the intersection of the two inputs. Similarly, a spatial difference of the two inputs can be realized by a regulatory module that performs the ANDN (ANDNOT) operation. This is the case for pip, repressed by the DV gradient of GRK signaling and activated by a still unknown uniform signal. Finally, a regulatory module that performs an OR operation is nonzero when at least one of the inputs is nonzero. When the inputs are spatially distributed, the output is their spatial union (Yakoby, 2008b).

Boolean operations on primitives lead to patterns with just two levels of expression (the gene is either expressed or not). In addition to Boolean logic, developmental cis-regulatory modules and systems for posttranscriptional control of gene expression can perform analog operations, leading to multiple nonzero levels of output. Consider a module that adds the two binary inputs, shaped as the primitives. The output is nonzero in the domain shaped as the union of the two primitives, but is characterized by two nonzero levels of expression. This type of annotation is reserved only for those cases where the application of Boolean operations would lead to a loss of the spatial structure of the pattern (such as the A + U expression pattern of mia at stage 11 of oogenesis. For example, the union of the A and U primitives is a U primitive, whereas the sum of these primitives is an anterior band superimposed on top of a spatially uniform background (Yakoby, 2008b).

Signaling pathways guide organogenesis through the spatial and temporal control of gene expression. While the identities of genes controlled by any given signal can be identified using a combination of genetic and transcriptional profiling techniques, systematic analysis of the diversity of induced patterns requires a formal approach for pattern quantification, categorization, and comparison. Multiplex detection of gene expression, which has a potential to convert images of the spatial distribution of transcripts into a vector format preferred by a majority of statistical methods, is currently feasible only for a small number of genes and systems with simple anatomies. This paper presents an alternative approach based on the combinatorial construction of patterns from simple building blocks (Yakoby, 2008b).

In general, the building blocks can be identified as shapes that are overrepresented in a large set of experimentally collected gene expression patterns. This approach can be potentially pursued in systems where mechanisms of pattern formation are yet to be explored. At the same time, in well-studied systems, the building blocks can be linked to identified patterning mechanisms. This study chose six primitives based on the features that are commonly observed in real patterns and related to the structure of the tissue as well as the spatial distribution of the inductive signals. A similar approach will be useful whenever a two-dimensional cellular layer is patterned by a small number of signals, when cells can convert smoothly varying signals into spatial patterns with sharp boundaries, and when the regulatory regions of target genes have the ability to combinatorially process the inductive signals. One system in which this approach could be feasible is the wing imaginal disc, which is patterned by the spatially orthogonal wingless and DPP morphogens (Yakoby, 2008b).

The six primitives are sufficient to describe the experimentally observed patterns during stages 10-12 of oogenesis. A natural question is whether it is possible to accomplish this with a smaller number of primitives. Two of the primitives, R and F, could be potentially constructed from the D, M, and A primitives, which are related to the patterns EGFR and DPP activation during the earlier stages of eggshell patterning. Specifically, recent studies of br regulation suggest that the R domain is formed as a difference of the D, A, and M patterns (Yakoby, 2008a). Furthermore, the formation of the F domain requires repressive action in the adjacent R domain. With the R and F domains related to the other four primitives, the size of the spatial alphabet will be reduced even further (from six to four), but at the expense of increasing the complexity of the expressions used to describe various spatial patterns (Yakoby, 2008b).

Previously, the question of the diversity of the spatial patterns has been addressed only in one-dimensional systems. For example, transcriptional responses to the Dorsal morphogen gradient in the early Drosophila embryo give rise to three types of patterns in the form of the dorsal, lateral, and ventral bands. This work provides an attempt to characterize the diversity and dynamics of two-dimensional patterns. Thirty-six qualitatively different patterns were constructed, and it is proposed that each of them can be constructed using a compact combinatorial code. The sizes of the data sets from the literature and from transcriptional profiling experiments are approximately the same (117 and 96 patterns, respectively. Based on this observation, it is expected that discovered patterns will be readily described using this annotation system (Yakoby, 2008b).

A gene expressed in more than one stage of oogenesis is more likely to appear in different patterns, and it was found that groups of genes sharing the same pattern at one time point are more likely to scatter in the future than to stay together. More detailed understanding of the dynamics of the spatial patterns of the EGFR and DPP pathway activation is crucial for explaining these trends and the two observed scenarios for the emergence of complex patterns. A gene that makes its first appearance as a complex pattern, such as the A∩D pattern of argos at stage 10B, can be a direct target of the EGFR and DPP signal integration. In contrast, a gene such as Cct1, which changes from the A to the R pattern, can be a dedicated target of DPP signaling alone, and changes as a consequence of change in the spatial pattern of DPP signaling. Future tests of such hypotheses require analysis of cis-regulatory modules responsible for gene regulation in the follicular epithelium. While only a few enhancers have been identified at this time, this categorization of patterns should accelerate the identification of enhancers for a large number of genes (Yakoby, 2008b).

Proposed for the spatial patterns of transcripts, these annotations can also describe patterns of protein expression, modification, and subcellular localization. For example, the stage 10A patterns of MAD phosphorylation and Capicua nuclear localization can be accurately described using the A and U D annotations, respectively. The ultimate challenge is to use the information about the patterning of the follicular epithelium to explore how it is transformed into the three-dimensional eggshell. A number of genes in the assembled database encode cytoskeleton and cell adhesion molecules, suggesting that they provide a link between patterning and morphogenesis. It is hypothesized that the highly correlated expression patterns of these genes give rise to the spatial patterns of force generation and mechanical properties of cells that eventually transform the follicular epithelium into a three-dimensional eggshell (Yakoby, 2008b).

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.