746 nature neuroscience • volume 5 no 8 • august 2002

articles

As an axon navigates toward a target region during develop-ment, it alters its course based on attractive or repulsive mole-cular signals in its environment1,2. There are at least two phasesin the establishment of neuronal connections. First, axons pro-ject to and distinguish between regions or layers and second,once within the target layer, axons fine-tune their projections3.This second step involves precise interactions between growthcones and target cells and has been well studied, particularlyfor the participation of cell-surface molecules and their asso-ciated signal transduction machinery4–8. Here we investigatethe role of two transcription factors, Run and Bks, in the firstphase of target layer selection for differentiating R cells in theDrosophila optic lobe.

Each of ~800 individual ommatidia in a Drosophila eye con-tains an identical arrangement of eight R cells (R1–R8). The eyedevelops from an epithelial structure called the eye imaginal disc.During larval stages, a groove called the morphogenetic furrowproceeds across the disc, initiating cell differentiation and pat-tern formation9. As R cells differentiate, they project their axonsto the optic lobe, where they terminate in one of two optic gan-glia. Whereas R1–R6 axons terminate in the lamina, R7 and R8axons project through the lamina and terminate in the medulla.Many signaling pathways have been identified for their role indirecting R-cell axonal pathfinding1–7, but the mechanism bywhich individual R-cell axons differentially target the lamina ormedulla is not well understood.

Bks is a putative zinc-finger transcription factor that is involvedin R-cell axonal pathfinding and is required for axon terminationin the first optic ganglion, the lamina10,11. The Bks protein isexpressed in all R cells and, in its absence, axons from the six pho-toreceptor neurons R1–R6 (the ‘outer’ R cells) do not terminate in

Control of photoreceptor axontarget choice by transcriptionalrepression of Runt

Joshua S. Kaminker1,5, Jude Canon2,5, Iris Salecker4 and Utpal Banerjee1–3

1 Department of Molecular, Cell, and Developmental Biology, 2Department of Biological Chemistry and 3Department of Human Genetics, Molecular Biology Institute, University of California, Los Angeles, California 90095, USA

4 Division of Molecular Neurobiology, National Institute for Medical Research, London NW7 1AA, UK5 These authors contributed equally to this work.

Correspondence should be addressed to U.B. (banerjee@mbi.ucla.edu)

Published online: 15 July 2002, doi:10.1038/nn889

Drosophila photoreceptor neurons (R cells) project their axons to one of two layers in the optic lobe,the lamina or the medulla. The transcription factor Runt (Run) is normally expressed in the twoinner R cells (R7 and R8) that project their axons to the medulla. Here we examine the relationshipbetween Run and the ubiquitously expressed nuclear protein Brakeless (Bks), which has previouslybeen shown to be important for axon termination in the lamina. We report that Bks represses Run intwo of the outer R cells: R2 and R5. Expression of Run in R2 and R5 causes axonal mistargeting of allsix outer R cells (R1–R6) to the inappropriate layer, without altering expression of cell-specific devel-opmental markers.

the lamina. Instead, like R7 and R8 axons, they project into thesecond optic ganglion, the medulla10,11. Photoreceptors that aremutant for bks have normal morphology and show no alterationsin the expression of R cell–specific markers10,11.

Run, a member of the Runt domain family of transcriptionfactors12,13, which includes AML1 and other Runx proteins14–16,is specifically expressed in R7 and R8. Run is the first knownexample of an R7- and R8-specific transcription factor (Fig. 1aand b)17. R7 and R8 cells are the farthest apart in developmen-tal timing within an ommatidium, but they both extend theiraxons to the medulla. In R8, Run expression initiates immedi-ately posterior to the furrow, whereas Run expression in R7 is6–7 columns behind the furrow. Although Run has a very spe-cific expression pattern in the eye, mosaic clones of run mutanttissue in the developing visual system and adult eye did notshow defects in the expression of developmental markers, mor-phology or axonal projections of any R cell (data not shown).It is likely that Run function in the eye is redundant with that oftwo other Runt-related genes adjacent to the run locus18. Muta-tions in these two loci have not yet been identified. Here weshow that wild-type run expression is restricted by Bks, andrepression of run influences R-cell axons to terminate in eitherthe lamina or the medulla.

RESULTSPrevious studies have shown that the expression pattern of severalR cell–specific differentiation markers is normal in bks mutants10,11.We have confirmed these results. However, in striking contrast toother markers, we have found that Run is ectopically expressed intwo extra R cells per cluster in somatic loss-of-function clones ofbks mutant tissue. In bks clones, Run expression is expanded from

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://n

euro

sci.n

atu

re.c

om

articles

nature neuroscience • volume 5 no 8 • august 2002 747

R1–R6 photoreceptor axons misproject to the medulla inbks loss-of-function mutants10,11. To determine whether thisaxonal targeting defect is due to the relief of Run repression inR2 and R5, we used the GAL4/UAS system19 to express Run inthese cells. When we expressed Run in R2, R5 and R8 using theMT14–GAL4 driver20 (Fig. 1d), all innervating R-cell axonsbypassed the lamina and projected through to the medulla (Fig. 1g). When Run was misexpressed in R2 and R5 alone (Fig. 1g), the defect was as severe as when Run was misex-pressed in all R cells using the GMR–GAL4 driver (Fig. 1h).Run over-expression in R8 alone, where it is normallyexpressed, did not affect axonal projections (Fig. 1i). In addi-tion, misexpression of Run in R1, R6 and R7 using lz–GAL4(Fig. 1j) or in R3 and R4 using sal–GAL4 (Fig. 1k) did not giverise to a comparable axonal misprojection phenotype (Fig. 1g).We attribute the severe axonal mistargeting phenotype inMT14-GAL4/UAS-run flies to Run expression in R2 and R5.

Unlike the ordered wild-type array (Fig. 1f), thick bundles ofaxons were seen entering the medulla when Run was mis-expressed in R2 and R5 (Fig. 1g). The axons did not project intodeeper areas of the brain, but stopped within the medulla. Thephenotype observed in this genetic background is very similarto that for bks10,11. Therefore, in both bks loss-of-function andMT14-GAL4/UAS-run genetic backgrounds, Run expression inR2 and R5 results in the mistargeting of all retinal axon types tothe medulla. This also suggests that the targeting of R2 and R5

Fig. 2. Run expression in R2 and R5 does not alter developmentalmarkers in the eye disc. Expression of Boss (a, b, c), Pros (d, e, f), Bar(g, h, i) and Rough (j, k, l) are identical in wild type (a, d, g, j), bks2

clones (b, e, h, k) and MT14-GAL4/UAS-run (c, f, i, l). Arrows indicate R7cells in Prospero (Pros) panels, numbers indicate R1 and R6 in Bar pan-els, and arrowheads indicate morphogenetic furrow in Rough panels.

its normal R7/R8 pattern to also include R2 and R5 (Fig. 1c). Thissuggests that Bks represses Run in R2 and R5 cells. Cells along theedges of bks clones were analyzed for the expression of Run. In 196ommatidia counted along clone boundaries, R2/R5 expression ofRun was never seen in a cell that is wild type for bks (Fig. 1e). Weconclude that the repression of run by Bks is cell-autonomous.

Fig. 1. Runt is misexpressed in bks mutant eye discs caus-ing axonal mistargeting. (a, b) In wild type, Run expressionis limited to the R7 cell (a) and the R8 cell (b) in eachommatidial cluster. (c) Somatic clone of bks2/bks2. In eachcluster, two additional cells, R2 and R5, express Run. Thewild-type expression in R7 and R8 is maintained. (d) MT14-GAL4/UAS-run. This GAL4 line20 causes expression of Runin R8, R2 and R5. Rare examples of leaky and non specificexpression in other R cells can be seen (arrowhead). (e) Eye disc mosaic for bks2, showing Run expression (red)in wild-type (green) and bks mutant tissue (non-green). Inommatidia along clone borders (n = 196), R2 and R5 cellsexpressing Run are never Bks-positive (never yellow).Thus, Bks control of Run is cell-autonomous. Individualexamples of ommatidia along the clone boundary aremarked. When R2 and R5 are mutant for bks (arrowhead)they express Run (red). In a mosaic ommatidium in whichR2 and R5 are wild type for bks (small arrow), they do notexpress Run, whereas the R8 cell in the same cluster does(yellow). And in a mosaic ommatidium (large arrow) whereR8 and R5 are mutant for bks and express Run (red), the R2cell in the same ommatidium is wild type for bks and doesnot express Run (green). (f–k) R-cell axonal projections inthe optic lobe. In wild type (f), the R1–R6 axons terminatein the lamina and their growth cones expand to form the lamina plexus, which is seen as a dark region of staining in the brain (arrow). The R7 and R8axons project past the lamina and terminate in the medulla (me). (g) Run misexpression in R2 and R5 causes all R-cell axons to project to the medulla(me). No lamina plexus can be seen. (h) Ectopic expression of Run in all R cells using the GMR–GAL4 driver causes targeting defects similar to thatseen when Run is misexpressed in only R2 and R5 (g). (i) GAL4109–68/UAS-run. This GAL4 line35 causes Run overexpression in R8 alone, where it is nor-mally expressed. The axonal projections remain wild type. (j, k) Run misexpression in R1, R6 and R7 using lz–GAL4 (j), or in R3 and R4 using sal–GAL4(k), does not result in the same axonal misprojection phenotype as seen when Run is misexpressed in R2 and R5 (g).

a b c d

e

f g h

i kj

a b c

d e f

g h i

j k l

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://n

euro

sci.n

atu

re.c

om

748 nature neuroscience • volume 5 no 8 • august 2002

articles

axons affects axonal pathfinding of other outer R cells. For tech-nical reasons, we were unable to generate marked, double-lethalclones of run and bks.

The mistargeting of R-cell axons could, in principle, resultfrom the conversion of all R-cell fates to R7 and R8. We there-fore analyzed markers for every R-cell type and for cone cells,both in mosaic clones of null bks mutant tissue and in the con-text of MT14-GAL4/UAS-run. In each of these backgrounds, theRun-expressing R2 and R5 cells did not express the R8-specificantigen, Bride of Sevenless (Boss, Fig. 2b and c) or the R7 marker, Prospero (Pros, Fig. 2e and f). Therefore, the projectionphenotype of R cells to the medulla in these backgrounds didnot result from the conversion of these R cells to the R7/R8 typeduring their development. The R1/R6 marker Bar (Fig. 2h and i)and the R2/R5/R3/R4 marker Rough (Fig. 2k and l) are alsounaffected in these backgrounds. We conclude that Run repro-grams the projection pattern of outer R cells without affectingthe expression of developmental markers of cell identity.

Consistent with the R cell marker expression in larval tissue,plastic sections of adult eyes showed that Run misexpression in R2and R5 during development did not perturb adult R cells orommatidial structure. We found the correct complement andarrangement of R cells (Fig. 3a and b). Strikingly, these seem-ingly normal adult R cells misproject their axons to the inappro-priate optic layer (Fig. 3c and d). Axon termination in the laminaregion was virtually absent and all R-cell axons projected to themedulla. This adult phenotype is also identical to that reported inbks mutant clones in which R cells are unchanged in the expres-sion pattern of Rhodopsins10,11. These data provide strong evi-

dence that Run expression in R2 and R5 causes mistargeting ofall outer R-cell axons without changing their individual R-cellfates, as determined by developmental markers and by the adultmorphology of rhabdomeres. In spalt (sal) mutants, cell fate isaltered without changes in axonal connectivities21. Similarly, insal-GAL4/UAS-run flies, some change in R3/R4 fate to R7 celltype was evident (data not shown) without a significant pertur-bation of outer R-cell projections (Fig. 1k). Hence, transcrip-tional events that control cell identity are separable from thosethat control axonal targeting.

R-cell axons provide critical anterograde signals to the laminatarget region to induce the proliferation and differentiation oflamina neurons, and to induce the differentiation and migrationof glial cells to their correct position adjacent to the laminaplexus22–24. In turn, the glial cells provide positional informa-tion that directs R1–R6 axons to terminate in the lamina, and inthe absence of glia these axons project into the medulla24,25. Weassessed the development of the lamina target region using neu-ronal and glial cell differentiation markers. Wild-type brainsstained with α-Dachshund (Dac) antibody showed a large areaof labeled cells corresponding to maturing lamina precursor cells(LPCs) and differentiated lamina neurons (Fig. 4a). Thisremained unchanged in GMR-GAL4/UAS-run brains, althoughthe R-cell axons did not target properly (Fig. 4b). Additionally,the three rows of glial cells that delineate the lamina plexus inwild type (Fig. 4c) also remained unaltered when Run was mis-

a b

c d

Fig. 3. Run misexpression in R2 and R5 causes axonal mistargeting tothe medulla without affecting ommatidial organization in adult animals.(a, b) Sections of wild-type (a) and MT14-GAL4/UAS-run (b) adult eyes.The ommatidia have a normal complement and arrangement of R cellsas shown by the regular trapezoidal pattern of R-cell rhabdomeres. Rareexamples of abnormal outer R cells (arrowhead in b) are seen, presum-ably due to leaky driver expression (see Fig. 1d). (c, d) R-cell axonalprojection in adults. In wild type (c), R1–R6 axons terminate in the lam-ina, while R7 and R8 project to the medulla (dashed box magnified ininset). When Run is misexpressed in R2 and R5 (d), virtually none of theaxons terminate in the lamina, while thick disorganized axon bundlesmisproject through to the medulla. Once in the medulla, axons termi-nate in both the R7 (arrow in inset) and R8 (arrowhead in inset) layers.

a b

c d

Fig. 4. Mistargeting of R1–R6 axons to the medulla is not due to a dis-ruption of the lamina target region. (a, b) In wild-type tissue (a), Dac (red)is localized to a large pool of differentiating lamina neurons (brackets in aand b). This remains unchanged when Run is misexpressed (b), suggestingthat the lamina neurons are able to differentiate normally. (c, d) Threerows of glial cells (red), the epithelial glia (eg), marginal glia (mg) andmedulla glia (meg) surround the lamina plexus (la). As the R1–R6 axons(green) innervate the lamina, they expand their growth cones betweenthe eg and mg. When Run is misexpressed (d), all axons project to themedulla (me), but the three rows of glia (arrowheads) are unperturbed.

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://n

euro

sci.n

atu

re.c

om

expressed (Fig. 4d). We conclude that the abnormal pathfindingof R cells is due to defects intrinsic to R cells and not to a globaldisruption of lamina target neurons or glia.

DISCUSSIONThis study highlights two important characteristics of neuronalpathfinding in the optic lobe. First, a mechanism involving Bksexists in wild-type flies for repressing Run expression specifi-cally in R2 and R5 cells. The bks loss-of-function causes reliefof run repression in these two cells, which entirely abolishes thedistinct targeting of R cells to the two optic ganglia. Perhapsadditional genes are responsible for repressing run in the otherR cells. Second, the mistargeting of R2 and R5 alone is suffi-cient for all outer R cells to project to the medulla. It appearsthat R2 and R5 cells, the first two outer R cells to be specified,provide pioneering axons whose tracts the other axons follow. Inrough loss-of-function, R2 and R5 are converted to R8 cells26,27,resulting in disrupted axon targeting (data not shown). Theresulting phenotype, however, is not as severe as that describedhere. Presumably, a residual pioneering function is maintained.The mechanism underlying the interaction between R2/R5 andR3/R4/R1/R6 axons in regulating target layer selection isunclear, although interaction between R1–R6 axons regulatesa later step in axonal pathfinding: the fine-tuning of R-cell con-nections in the lamina28.

The repression of Run by Bks in R2 and R5 cells could reflectimportant features about the evolutionary history of the devel-opment of the Drosophila visual system. During Drosophiladevelopment, the lamina and the outer medulla, the targets ofR-cell axons, both arise from the same anlagen29. It has beenproposed that these two layers evolved from a single neuropil towhich the axons of all R cells projected30. In this context, it isreasonable to propose that in a primitive organism, a single pro-gram directed the leading R-cell axons to innervate one neu-ropil. A system involving transcriptional repression could havesubsequently evolved to create two programs for targeting axonsspecifically to the lamina or outer medulla.

Several signal transduction components1–8 and transcrip-tion factors31–33 controlling axon guidance have been identi-fied and found to be conserved across species, from worms tomammals. This study highlights that transcriptional eventsinvolving Run can control axon target selection without affect-ing the identity of a neuron. In this context, it is particularlyencouraging that mammalian Runx3, a Runt domain proteinwith homology to run, has been shown to function in sensoryneurons of the dorsal root ganglia (ref. 34 and Inoue, K., Ozaki, S. and Ito, Y., pers. comm.).

METHODSGenetics. The Oregon-R Drosophila strain was used as wild type. Mosaic clones of bks were generated by crossing yw/Y; bks2 FRT42D/SM6awith yw ey-flp/yw ey-flp; P[w+] (M) cl2R11 FRT42D/SM6a. The Minute(M) mutation caused the clones to cover almost the entire eye field.Mutant tissue was recognized either by the presence of Run-positive R2and R5 cells or by the expression of a lacZ counter marker. For Fig. 1e,mosaic clones of bks2 were generated without the Minute mutation, andthe green fluorescent protein (GFP) counter marker ubi–GFP was used toidentify mutant versus nonmutant tissue.

Histology. The following primary antibodies were used for immuno-histochemistry: monoclonal antibody 24B10 to visualize photoreceptoraxons (1:100), mouse anti-β-galactosidase (1:100; Promega), rabbit α-β-galactosidase (Cappel, Costa Mesa, California; 1:1000), rabbit anti-Runt (1:250; gifts from A. Brand and P. Gergen), mouse anti-Boss(1:500), mouse anti-Prospero (1:100), rabbit anti-Bar (gift from K.

Saigo), mouse anti-Rough (1:100), mouse anti-Dachshund (1:25) andrabbit anti-Repo (1:500). Secondary antibodies (Jackson Laboratories,West Grove, Pennsylvania) were HRP-conjugated goat anti-mouse(1:100), HRP-conjugated goat anti-rabbit (1:400) and Cy3-conjugatedgoat anti-mouse and goat anti-rabbit (1:200). GFP labeling was by aUAS–GFP or ubi–GFP countermarker (Bloomington Stock Center,Bloomington, Indiana). Fluorescent samples were visualized using aBio-Rad MRC 1024 laser scanning confocal microscope (Hercules, Cal-ifornia). Eye imaginal discs were dissected from late third-instar larvaand labeled as described17. Whole-mount eye–brain preparations of latethird-instar larva, tangential sectioning of adult eyes, and adult headcryostat sectioning were performed as described25.

AcknowledgmentsWe are grateful to A. Brand, R. Carthew, B. Dickson, P. Gergen, S. L. Zipursky, N.

Perrimon, K. Saigo, R. Stocker, G. Rubin, K. Matthews and the Bloomington

Stock Center for fly stocks and antibodies. We thank T. Clandinin, C-H. Lee, T.

Herman and members of the Banerjee laboratory for thoughtful suggestions and

comments. We thank L. Zipursky for helpful discussions and for supporting part

of this work in his laboratory. We thank Y. Ito and his collaborators and Y. Groner

and his collaborators, for communicating results prior to publication. The Rough

and Dac antibodies were obtained from the University of Iowa Developmental

Studies Hybridoma Bank, developed under the auspices of the National Institute

of Child Health & Human Development. This work was supported by a National

Institutes of Health grant (U.B.), U.S.P.H.S. National Research Service Awards

(J.K. and J.C.) and the Medical Research Council (I.S.).

Competing interests statementThe authors declare that they have no competing financial interests.

RECEIVED 1 APRIL; ACCEPTED 20 JUNE 2002

1. Chisholm, A. & Tessier-Lavigne, M. Conservation and divergence of axonguidance mechanisms. Curr. Opin. Neurobiol. 9, 603–615 (1999).

2. Tessier-Lavigne, M. & Goodman, C. S. The molecular biology of axonguidance. Science 274, 1123–1133 (1996).

3. Goodman, C. S. & Shatz, C. J. Developmental mechanisms that generateprecise patterns of neuronal connectivity. Cell 72 (Suppl.), 77–98 (1993).

4. Schmucker, D. & Zipursky, S. L. Signaling downstream of Eph receptors andephrin ligands. Cell 105, 701–704 (2001).

5. Stoker, A. W. Receptor tyrosine phosphatases in axon growth and guidance.Curr. Opin. Neurobiol. 11, 95–102 (2001).

6. Newsome, T. P. et al. Trio combines with dock to regulate Pak activity duringphotoreceptor axon pathfinding in Drosophila. Cell 101, 283–294 (2000).

7. Hing, H., Xiao, J., Harden, N., Lim, L. & Zipursky, S. L. Pak functionsdownstream of Dock to regulate photoreceptor axon guidance in Drosophila.Cell 97, 853–863. (1999).

8. Klein, R. Excitatory Eph receptors and adhesive ephrin ligands. Curr. Opin.Cell Biol. 13, 196–203 (2001).

9. Ready, D. F., Hanson, T. E. & Benzer, S. Development of the Drosophila retina,a neurocrystalline lattice. Dev. Biol. 53, 217–240 (1976).

10. Senti, K., Keleman, K., Eisenhaber, F. & Dickson, B. J. Brakeless is required forlamina targeting of R1-R6 axons in the Drosophila visual system.Development 127, 2291–2301 (2000).

11. Rao, Y., Pang, P., Ruan, W., Gunning, D. & Zipursky, S. L. Brakeless isrequired for photoreceptor growth-cone targeting in Drosophila. Proc. Natl.Acad. Sci. USA 97, 5966–5971 (2000).

12. Kania, M. A., Bonner, A. S., Duffy, J. B. & Gergen, J. P. The Drosophilasegmentation gene runt encodes a novel nuclear regulatory protein that isalso expressed in the developing nervous system. Genes Dev. 4, 1701–1713(1990).

13. Canon, J. & Banerjee, U. Runt and Lozenge function in Drosophiladevelopment. Semin. Cell Dev. Biol. 11, 327–336 (2000).

14. Speck, N. A. et al. Core-binding factor: a central player in hematopoiesis andleukemia. Cancer Res. 59, 1789s–1793s (1999).

15. Wijmenga, C. et al. Identification of a new murine runt domain-containinggene, Cbfa3, and localization of the human homolog, CBFA3, tochromosome 1p35-pter. Genomics 26, 611–614 (1995).

16. Downing, J. R. AML1/CBFbeta transcription complex: its role in normalhematopoiesis and leukemia. Leukemia 15, 664–665 (2001).

17. Kaminker, J. S., Singh, R., Lebestky, T., Yan, H. & Banerjee, U. Redundantfunction of runt domain binding partners, big-brother and brother, duringDrosophila development. Development 128, 2639–2648 (2001).

18. Adams, M. D. et al. The genome sequence of Drosophila melanogaster. Science287, 2185–2195 (2000).

articles

nature neuroscience • volume 5 no 8 • august 2002 749

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://n

euro

sci.n

atu

re.c

om

750 nature neuroscience • volume 5 no 8 • august 2002

articles

19. Brand, A. H. & Perrimon, N. Targeted gene expression as a means of alteringcell fates and generating dominant phenotypes. Development 118, 401–415(1993).

20. Tissot, M., Gendre, N., Hawken, A., Störtkuhl, K. F. & Stocker, R. F. Larvalchemosensory projections and invasion of adult afferents in the antennal lobeof Drosophila. J. Neurobiol. 32, 281–297 (1997).

21. Mollereau, B. et al. Two-step process for photoreceptor formation inDrosophila. Nature 412, 911–913 (2001).

22. Huang, Z. & Kunes, S. Hedgehog, transmitted along retinal axons, triggersneurogenesis in the developing visual centers of the Drosophila brain. Cell 86,411–422 (1996).

23. Huang, Z., Shilo, B. Z. & Kunes, S. A retinal axon fascicle uses spitz, an EGFreceptor ligand, to construct a synaptic cartridge in the brain of Drosophila.Cell 95, 693–703 (1998).

24. Perez, S. E. & Steller, H. Migration of glial cells into retinal axon target field inDrosophila melanogaster. J. Neurobiol. 30, 359–373 (1996).

25. Poeck, B., Fischer, S., Gunning, D., Zipursky, S. L. & Salecker, I. Glial cellsmediate target layer selection of retinal axons in the developing visual systemof Drosophila. Neuron 29, 99–113 (2001).

26. Dokucu, M. E., Zipursky, S. L. & Cagan, R. L. Atonal, rough and theresolution of proneural clusters in the developing Drosophila retina.Development 122, 4139–4147 (1996).

27. Frankfort, B. J., Nolo, R., Zhang, Z., Bellen, H. & Mardon, G. Senseless

repression of rough is required for R8 photoreceptor differentiation in thedeveloping Drosophila eye. Neuron 32, 403–414 (2001).

28. Clandinin, T. R. & Zipursky, S. L. Afferent growth cone interactions controlsynaptic specificity in the Drosophila visual system. Neuron 28, 427–436(2000). [Comment in Neuron 28, 310–313 (2000).]

29. Campos-Ortega, J. A. & Hartenstein, V. The Embryonic Development ofDrosophila melanogaster (Springer-Verlag, New York, 1997).

30. Meinertzhagen, I. A. in Evolution of the Eye and Visual System (eds. Cronly-Dillon, J. R. & Gregory, R. L.) 341–363 (Macmillan, London, 1991).

31. Erkman, L. et al. A POU domain transcription factor-dependent programregulates axon pathfinding in the vertebrate visual system. Neuron 28,779–792 (2000).

32. Thor, S., Andersson, S. G., Tomlinson, A. & Thomas, J. B. A LIM-homeodomain combinatorial code for motor-neuron pathway selection.Nature 397, 76–80 (1999).

33. Kania, A., Johnson, R. L. & Jessell, T. M. Coordinate roles for LIM homeoboxgenes in directing the dorsoventral trajectory of motor axons in the vertebratelimb. Cell 102, 161–173 (2000).

34. Levanon, D. et al. The Runx3 transcription factor regulates development andsurvival of TrkC dorsal root ganglia neurons. EMBO J. (in press).

35. White, N. M. & Jarman, A. P. Drosophila atonal controls photoreceptor R8-specific properties and modulates both receptor tyrosine kinase andHedgehog signaling. Development 127, 1681–1689 (2000).

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://n

euro

sci.n

atu

re.c

om