Dear Editor

The Notch signal transduction pathway has been implicated in a variety of cell fate decisions throughout development. In the developing Drosophila retina, the Notch receptor and its ligand Delta are required to organize the interommatidial lattice through selective programmed cell death (PCD).1,2 Another transmembrane molecule required for PCD in the Drosophila retina is the immunoglobulin superfamily member Irregular chiasmC-roughest (IrreC-rst).3,4,5 Loss-of-function mutations in irreC-rst lead to a block in PCD during retinal development, resulting in a rough eye phenotype in the adult. In addition, the irreC-rstCT mutation, which results in truncation of the IrreC-rst intracellular domain, is associated with reduced IrreC-rst protein at the plasma membrane and increased IrreC-rst within intracellular vesicles.4,5 Retinal PCD is lost, indicating that proper subcellular localization of IrreC-rst may be an important factor in permitting or promoting the death process. The mechanisms by which Notch signaling and IrreC-rst regulate PCD are unclear.

In a recent genetic screen (Tanenbaum et al, submitted) and in direct tests, Delta (Dl) mutations were identified as dominant enhancers of the rough eye phenotype conferred by the hypomorphic allele irreC-rst 3 (n=7 Dl mutant lines; Figure 1A-C). These observations demonstrate that Delta and irreC-rst interact genetically to execute PCD, and motivated us to examine further the relationship between Delta/Notch signaling and IrreC-rst.

Figure 1
figure 1

The rough eye phenotype of irreC-rst3 is enhanced by DeltaE58. (A) A wild-type adult eye contains approximately 750 ommatidia arranged in straight lines along multiple axes. (B) A y irreC-rst 3; dp ov; p p adult eye has a mildly altered pattern of ommatidia compared to wild-type; careful inspection reveals an occasional ‘jog’ in the ommatidial rows. (C) These alterations are enhanced significantly in an irreC-rst 3; Delta E58/+ adult eye. Six other Dl alleles were identified as dominant enhancers of irreC-rst 3: Dl J17, Dl J111, Dl K75 (Tanenbaum et al, submitted), Dl HD82, Dl S130403 and Dl S148504. In addition, Dl E58, Dl J17, Dl K75, Dl J111 and Dl HD82 were observed to enhance irreC-rst UB883 and irreC-rst IR34; the other alleles were not tested. (D-J) Localization of IrreC-rst protein in pupal retinae. In (E–H and J) flies received a heat shock (see below), were dissected immediately, and immunostained using the IrreC-rst-specific antibody mAb 24A5.1.11 (D) Schematic of the apical surface of a wild-type pupal retina. One central ommatidium is shown, and is labeled to indicate the non-neuronal cone cells (c), the optically-insulating primary pigment cells (1°), the optically-insulating secondary (2°) and tertiary (3°) pigment cells, and bristles (b). The photoreceptor core consisting of eight neurons is not visible in this view. Subcellular localization of IrreC-rst protein in 2°/3°s is indicated by white shading. For clarity, IrreC-rst expression in 1°s is omitted. (E) Wild-type control retina from an animal aged until 44 h after puparium formation (APF; 18°C), shifted to 32°C for 6 h and dissected immediately. One ommatidium and its associated 2°/3° lattice is traced (compare with (D)); the cone cells (c), 1°s, 2°/3°s (white asterisks), and bristles (filled asterisks) are labeled. Note the distinct localization of IrreC-rst protein between 1°s and 2°/3°s (e.g., arrows) but not, for example, between two 2°/3°s cells (arrowheads). An enlarged view of a single central 2°/3°s and three 2°/3°s radiating from it are shown in the inset; the boundaries between 2°/3°s cannot be seen due to a lack of IrreC-rst protein (compare with (G)). The vesicles within the 2°/3°s are likely multivesicular bodies, sites of protein internalization and turnover.12 (F) Dl RF/+ control retina. A 6 h heat pulse, performed as in (E), had no effect on IrreC-rst localization. Subcellular localization between two 1°s is more obvious than in (E) (arrows). (G) Dl RF/Dl 6B retina following a 6 h heat pulse as in (E). IrreC-rst protein is no longer localized primarily at the border between 1°s and 2°/3°s but is distributed equally throughout the 2°/3° cells. This can be seen most easily with the staining between two 2°/3°s (e.g., arrowheads). The inset contains an enlarged view of a central 2°/3°; its boundaries with three neighboring 2°/3°s can now easily be seen due to re-localization of IrreC-rst (compare with (E)). The 1° membranes that directly abut cone cells (arrows) also appear to have increased levels of IrreC-rst staining at the expense of regions that contact a neighboring 1° (compare with arrows in (F)). (H) N ts1 retinae that received an 8 h heat pulse demonstrated a similar redistribution of IrreC-rst as Dl RF/Dl 6B retinae; arrows indicate two examples. The additional 2°/3° cells – the result of a block in PCD – are particularly evident in this retina; they can be recognized as 2°/3°s based on their small apical profile and their position between ommatidia (see also1). Notch ts1 pupae were aged until 42 h APF at 18°C or 21 h APF at 25°C prior to the 8 h, 32°C heat pulse. (I) Expression of the caspase inhibitor p35 throughout the retina leads to a loss of programmed cell death, but had no effect on the localization of IrreC-rst protein. Shown is a retina from GMR-p35 at 28 h APF (25°C). (J) Retina from hs-dRas1 val12/Cy0 following a 1 h heat shock (37°C) at 26 h APF (25°C) and dissected immediately. Localization of IrreC-rst is similar to wild-type

Full size image

In a wild-type retina, differentiation of the primary pigment cells (1°s) in the young pupa is followed by selective removal of approximately one-third of the neighboring interommatidial precursor cells by PCD.3,6 The remaining interommatidial precursor cells are organized concurrently into a hexagonal array and differentiate as optically-insulating secondary/tertiary pigment cells (2°/3°s; Figure 1D). During the state of PCD, IrreC-rst protein is expressed at high levels in the 2°/3°s and at low levels in the 1°s. It is localized subcellularly along the border between the 2°/3°s and the 1°s, and between the two 1°s of each ommatidium5 (Figure 1D-F). This subcellular localization of IrreC-rst is lost in two Notch alleles (N fa-g2, N fa-swb) that reduce Notch activity specifically in the young pupal retina during pigment cell differentiation; however, these alleles also result in a loss of 1°s and a block in PCD.1,5 Mislocalization of IrreC-rst could represent an indirect effect due to defects in 1° differentiation or PCD. Alternatively, Notch protein or signaling could affect more directly the localization of IrreC-rst.

To distinguish between these possibilities, we utilized temperature-sensitive alleles to reduce Delta and Notch function specifically during the stage of PCD, subsequent to formation of the 1°s.1,2,7 Flies heterozygous for the temperature sensitive allelic combination Dl RF/Dl 6B were shifted to the non-permissive temperature. A temperature shift of 6 h duration resulted in a loss of IrreC-rst localization: the IrreC-rst protein was redistributed throughout the apical surfaces, revealing the boundaries between 2°s and 3°s (Figure 1G). Similarly, shifting N ts1 pupae to the non-permissive temperature for 8 h resulted in a redistribution of IrreC-rst protein (Figure 1H). In parallel control experiments, redistribution of IrreC-rst protein was not observed in temperature shift experiments with wild-type or with either Dl RF or Dl 6B in trans to a normal copy of Delta (Figure 1E,F), in Dl RF/Dl 6B flies shifted for intervals ranging from 0 to 4 h, or in unshifted N ts1 flies (data not shown).

The observed redistribution of the IrreC-rst protein was not an indirect effect of a loss of PCD in the Dl RF/Dl 6B and N ts1 backgrounds: when PCD was blocked by directed expression of the activated dRas1 isoform dRas1val12 8,9 or the baculovirus caspase inhibitor p35,10 IrreC-rst localization was unaffected (Figure 1I,J). This indicates that the redistribution of IrreC-rst protein is a specific effect of the reduction of Notch or Delta activity or protein and not a result of incorrect 2°/3° patterning. Correct localization of IrreC-rst is likely important for its normal function,5 and it will be interesting to determine the mechanism by which it is regulated through Delta and Notch.

We thank K Fischbach, B Hay, M Muskavitch, the Szeged Stock Center and the Umea Stock Center for supplying fly stocks, and K Fischbach for providing mAb 24A5.1. We also thank R Kopan and M Marra for comments on the manuscript. This work was supported by grant NIH R01EY11495; C Baker Brachmann was supported by a fellowship from Damon Runyon.