Proper information processing in neural circuits requires establishment of specific connections between pre- and postsynaptic neurons. Targeting specificity of neurons is instructed by cell-surface receptors on the growth cones of axons and dendrites, which confer responses to external guidance cues. Expression of cell-surface receptors is in turn regulated by neuron-intrinsic transcriptional programs. In the Drosophila olfactory system, each projection neuron (PN) achieves precise dendritic targeting to one of 50 glomeruli in the antennal lobe. PN dendritic targeting is specified by lineage and birth order, and their initial targeting occurs prior to contact with axons of their presynaptic partners, olfactory receptor neurons. A search was performed for transcription factors (TFs) that control PN-intrinsic mechanisms of dendritic targeting. Two POU-domain TFs, acj6 and drifter have been identified as essential players. After testing 13 additional candidates, four TFs were identified, (LIM-homeodomain TFs islet and lim1, the homeodomain TF cut, and the zinc-finger TF squeeze) and the LIM cofactor Chip, that are required for PN dendritic targeting. These results begin to provide insights into the global strategy of how an ensemble of TFs regulates wiring specificity of a large number of neurons constituting a neural circuit (Komiyama, 2007).

For technical simplicity, larval born GH146-Gal4-positive PNs, originating from three neuroblast lineages, anterodorsal (adPNs), lateral (lPNs), and ventral (vPNs), were studied. Out of ~25 classes defined by their glomerular targets, focus was placed on 17 classes whose target glomeruli are reliably recognized across different animals. The MARCM technique allows visualization and genetic manipulation of PNs in neuroblast and single-cell clones in otherwise heterozygous animals, so PN-intrinsic programs can be studied for dendritic targeting. GH146 is expressed only in postmitotic PNs (Komiyama, 2007).

acj6 and drifter have been identified as lineage-specific regulators of PN dendritic targeting. To identify additional transcription factors (TFs) that regulate dendritic targeting of different PN classes, candidates were tested that have been shown to regulate neuronal subtype specification and targeting specificity and have available loss-of-function mutants. The following was tested; (1) the expression of candidate genes in PNs at 18 hr after puparium formation (APF) when PN dendrites are in the process of completing their initial targeting, and/or (2) their requirement in PNs by examining dendritic targeting in homozygous mutant MARCM clones (Komiyama, 2007).

DL1 adPN expresses Acj6, an adPN lineage factor, but not Drifter or Cut. acj6^−/− DL1 PNs typically have diffuse dendrites that always innervate, but are not limited to, DL1. drifter misexpression alone did not affect their dendritic targeting. However, when loss of acj6 and gain of drifter were combined, the dendrites completely missed DL1 and targeted anterior glomeruli (Komiyama, 2007).

Misexpression of Cut alone caused DL1 PNs to target part of DL1 and the vicinity, similar to acj6^−/−. Notably, this diffuse phenotype was directional, because most mistargeted dendrites targeted medially to DL1 (Komiyama, 2007).

cut misexpression combined with loss of acj6 caused severe mistargeting of DL1 adPNs. The dendrites completely missed DL1 and occupied the medial to dorsomedial AL, typically VM2, DM6, and DC1. Interestingly, these glomeruli are all adPN targets near DM1 and DM2, the two glomeruli that most frequently fail to be innervated by cut^−/− lPNs. One interpretation is that loss of acj6 made the DL1 adPN more sensitive to the instructive information of cut to target the medial AL, but the remaining lineage information kept the dendrites within the adPN glomeruli in the area. If this were true, adding a lPN lineage factor drifter may bring the dendrites to DM1 or DM2, since this might recreate, based on partial knowledge of the TF code, a code for targeting these glomeruli. Loss of acj6 and misexpression of cut and drifter were combined simultaneously in DL1 adPNs. Under this condition, the dendrites again mostly targeted the medial to dorsomedial AL. However, glomerular preferences were strikingly different: they frequently innervated 1, DM2, and DA2. Notably, DA2 and DM2 are lPN targets (Komiyama, 2007).

These results suggest that cut and drifter have qualitatively different instructive information, with cut controlling global targeting and drifter controlling local glomerular choice according to their lineage (Komiyama, 2007).

Although the Drosophila visual system has been described extensively, little is known about the Drosophila olfactory system. A major reason for this discrepancy has been the lack of simple, reliable means of measuring response to airborne chemicals. This paper describes a jump response elicited by exposing Drosophila to chemical vapors. This behavior provides the basis for a single-fly chemosensory assay. The behavior exhibits dose dependence and chemical specificity: it is stimulated by exposure to ethyl acetate, benzaldehyde, and propionic acid but not ethanol. Animals can respond repeatedly at short intervals to ethyl acetate and propionic acid. The response relies on the third antennal segments. To illustrate the use of this behavior in genetic analysis of chemosensory response, nine acj (abnormal chemosensory jump) mutants defective in chemosensory response were isolated, and their responses to two chemicals were characterized. All of the acj mutants are normal in giant fiber system physiology, and two exhibit defects in visual system physiology (McKenna, 1989).

Mutations affecting olfactory behavior provide material for use in molecular studies of olfaction in Drosophila. Using the electroantennogram (EAG), a measure of antennal physiology, an adult antennal defect has been found in the olfactory behavioral mutant abnormal chemosensory jump 6 (acj6). The acj6 EAG defect was mapped to a single locus. The same mutation is responsible for both reduction in EAG amplitude and diminished behavioral response, as if reduced antennal responsiveness to odorant is responsible for abnormal chemosensory behavior in the mutant. acj6 larval olfactory behavior is also abnormal; the mutation seems to alter cellular processes necessary for olfaction at both developmental stages. The acj6 mutation exhibits specificity in that visual system function appears normal in larvae and adults. These experiments provide evidence that the acj6 gene encodes a product required for olfactory signal transduction (Ayer, 1991).

This article provides characterization of the electrical response to odorants in the Drosophila antenna and provides physiological evidence that a second organ, the maxillary palp, also has olfactory function in Drosophila. The acj6 mutation, previously isolated by virtue of defective olfactory behavior, affects olfactory physiology in the maxillary palp as well as in the antenna. Interestingly, abnormal chemosensory jump 6 (acj6) reduces response in the maxillary palp to all odorants tested except benzaldehyde (odor of almond), as though response to benzaldehyde is mediated through a different type of odorant pathway from the other odorants. In other experiments, different parts of the antenna are shown to differ with respect to odorant sensitivity. Evidence is also provided that antennal response to odorants varies with age, and that odorants differ in their age dependence (Ayer, 1992).

Mutations in the Drosophila class IV POU domain gene abnormal chemosensory jump 6 (acj6) cause physiological deficits in odor sensitivity. However, loss of Acj6 function also has a severe detrimental effect on coordinated larval and adult movement that cannot be explained by the simple loss in odorant detection. In addition to olfactory sensory neurons, Acj6 is expressed in a distinct subset of postmitotic interneurons in the central nervous system from late embryonic to adult stages. In the larval and adult brain, Acj6 is highly expressed in central brain, optic and antennal lobe neurons. Loss of Acj6 function in larval optic lobe neurons results in disorganized retinal axon targeting and synapse selection. Furthermore, the lamina neurons themselves exhibit disorganized synaptic arbors in the medulla of acj6 mutant pupal brains, suggesting that Acj6 may play a role in regulating synaptic connections or structure. To further test this hypothesis, two Acj6 isoforms were misexpressed in motor neurons where they are not normally found. The two Acj6 isoforms are produced from alternatively spliced acj6 transcripts, resulting in significant structural differences in the amino-terminal POU IV box. Acj6 misexpression causes marked alterations at the neuromuscular junction, with contrasting effects on nerve terminal branching and synapse formation associated with specific Acj6 isoforms. These results suggest that the class IV POU domain factor, Acj6, may play an important role in regulating synaptic target selection by central neurons and that the amino-terminal POU IV box is important for regulation of Acj6 activity (Certel, 2000).

The central brain neuropils include the antennal lobes (the first central region for processing olfactory information), the mushroom bodies (higher order structures involved in complex behaviors) and the central complex (a region necessary for the coordination and modulation of motor activities). A large subset of central brain interneurons express Acj6 including central complex neurons, which function to receive, process and convey information from one site within the nervous system to another. A loss of Acj6 function could therefore affect the ability to carry commands necessary to direct motor activity and thus serve to provide a reasonable hypothesis for the quantifiable reduction in coordinated movement exhibited by acj6 mutant flies. Acj6 is also expressed in the optic lobes in regions functioning as hierarchical processing sites for visual information. No Acj6 expression is detected in photoreceptor cells but instead Acj6 optic lobe expression is initiated in differentiated neurons that receive R cell synaptic input. Differentiated lamina and medulla neurons as well as neurons in the lobula express Acj6 through adult stages (Certel, 2000).

Although motor activity defects are probably due to Acj6 function in the central complex, as an initial step in determining whether Acj6 is required for events following lineage determination and initial axon guidance, focus was placed on the easily visible axonal projections and organized synapses found in the optic lobe. In the wild-type larval eye-brain complex, R cell axons project from the eye disc, through the optic stalk and into the optic ganglia. R1-R6 photoreceptors send their axons to the lamina ganglion layer of the brain where their growth cones form an array of postsynaptic 'cartridge' units. Each cartridge unit contains the set of R1-R6 axons, five lamina neurons (L1-5) and several glial cells. Acj6 expression is observed in the synaptic partners of the R cells, the lamina and medulla neurons, but not in the R cells themselves. To first analyze any effects that loss of Acj6 function might have on synapses as a post-synaptic target, the organization of R cell projections was assessed in acj6 mutants using the R cell-specific antibody mAb 24B10. The array of expanded R1-R6 growth cones appears as a continuous line of immunoreactivity in wild-type larvae. In acj6 mutants, R cell axons project into the brain in a wild-type manner; however, these fibers do not uniformly form the lamina neuropil. In a acj6 heteroallelic combination, occasional gaps are observed in the lamina plexus. In the acj6 null larvae, the entire lamina plexus is of variable thickness generating irregular small breaks. Therefore, the loss of Acj6 function in the R cell synaptic partners affects the ability of these R cell growth cones to establish connections with the appropriate lamina neuron column (Certel, 2000).

Defects in R cell connectivity could be due to a loss of neurons or abnormal lamina neuron specification. In acj6 mutants, lamina precursor cell (LPC) proliferation and initial lamina neuron differentiation are wild type, as assessed using anti-Dachshund and anti-Elav staining. At the level of antibody labeling, the organization of lamina neurons into columns also appears largely normal. Possible explanations for the connectivity defects may be the inability of the R cell growth cones to recognize their synaptic partners or to adhere correctly and form a stable synaptic cartridge (Certel, 2000).

To investigate possible pre-synaptic changes in Acj6- expressing lamina monopolar neurons, mAb 1D4, which is directed against the cell adhesion molecule Fasciclin II, was used. The axons and synapses of a subset of lamina neurons, L1 and L3, express Fasciclin II during selected stages of pupal development as they project in a highly structured pattern into the distal medulla. In acj6 mutants, the L1 and L3 arborizations and terminals are disorganized and unevenly spaced. In addition, the structure of the synaptic terminals is diffuse and overlaps are observed. These results provide further evidence that Acj6 is not required for initial differentiation steps such as axon pathfinding, but instead may be necessary for target cell selection and the establishment of synaptic connections (Certel, 2000).

To further test the hypothesis that Acj6 may regulate the formation of synaptic connections, Acj6 was misexpressed in motor neurons in regions where it is not normally found in order to observe any distinct morphological effects upon the well-characterized neuromuscular junction (NMJ). However, multiple acj6 transcripts have been identified and it is not clear whether different Acj6 isoforms are capable of unique functions. The Acj6 protein contains two domains with extensive homology to the vertebrate class IV members: Brn-3a, Brn-3b and Brn-3c. In addition to the DNA-binding POU domain, members of this group contain a class IV-specific 40-amino-acid POU IV box at the N-terminal end. Four of the five acj6 transcripts differ only in the use of the four small exons encoding the N-terminal POU IV box (amino acids 88-119). Neither the POU IV box nor the POU domain are affected in the fifth alternatively spliced transcript, which utilizes an alternative consensus acceptor site to generate a short form of exon 5 (Certel, 2000).

It was therefore imperative to determine whether the predicted Acj6 isoforms are expressed in vivo so that the significance of any differences in functional capability could be evaluated. Alternatively spliced Acj6 transcripts should produce proteins with predicted molecular masses of 40.0, 41.2, 42.2 and 43.5 kDa. A cluster of bands corresponding to proteins of the predicted size was detected using the Acj6 antibody on Western blots. The cluster of anti-Acj6 immunoreactive bands is absent in extracts from acj6 null flies demonstrating that multiple Acj6 isoforms are expressed (Certel, 2000).

Based upon previous experiments with other members of the POU domain family, it was not expected that alterations in the Acj6 POU IV box would have significant effects on DNA-binding activity. To verify that amino-terminal changes do not affect the ability of Acj6 isoforms to bind DNA, fusion proteins were generated and used in gel mobility-shift assays. All of the Acj6 isoforms differing in the highly conserved POU IV box are capable of binding octamer and neuronal DNA recognition elements with affinities comparable to wild type. Although the POU IV box does not appear to be important for DNA-binding, in vitro studies indicate that this region is necessary for both the transforming activity of Brn-3a and activation of specific promoters. The Gal4/UAS system was used to analyze the functional capabilities of two Acj6 isoforms, Acj6(1,4) and Acj6(1,3,4), through misexpression studies (the numbers in parentheses represent the protein primary structure in terms of the contributions of the first four exons) (Certel, 2000).

Transgenic strains carrying a UAS-acj6(1,4) or UAS-acj6(1,3,4) transposon were mated with either scabrous-GAL4 (sca-GAL4) or elav-GAL4 flies to express the Acj6 isoforms at high levels in all neurons. Acj6(1,4) and Acj6(1,3,4) differ only in the absence or presence of exon 3 encoding a portion of the conserved POU IV box. Multiple transgenic lines were tested to eliminate the possibility that differences in phenotypes might be due to the position of insertion. In addition, expression of Acj6 proteins at comparable levels, from each of the UAS-acj6 transposons, was confirmed by labeling with Acj6 antibody. In each abdominal hemisegment of the Drosophila embryo and larva, the axons of approximately 40 motor neurons exit the ventral nerve cord and specifically synapse with 30 identified muscle fibers. The analysis focused on the well described motor axons of the ISNb fascicle visualized using mAb 1D4. The ISNb fascicle contains motor axons from at least four motor neurons innervating the ventral muscles 6, 7, 12 and 13. The ISNb fascicles of sca-GAL4/UAS-acj6(1,3,4) embryos correctly leave the nerve cord, defasciculate from the ISN and project to target muscle clefts in nearly all late stage 16/stage 17 hemisegments examined. However, in many of the hemisegments with normal initial axon pathfinding, a striking increase in the number of nerve terminal branches and processes arising from each motor axon was observed. In addition, some of the ectopic terminal branches are able to extend and form connections onto inappropriate muscle fibers. To quantitate the observed increase in ectopic boutons, transgenic sca-GAL4/UAS-acj6(1,3,4) embryos were allowed to develop to crawling third instar larvae. Analysis of the larger NMJs in these larvae indicates that a subset of the ectopic terminal branches generated in the embryo were maintained, increasing the number of boutons by 27% at muscles 6 and 7 and 22% at muscles 12 and 13. The ectopic boutons also expressed the synaptic markers Synaptotagmin and Cysteine string protein, consistent with functional synapses. In sharp contrast, misexpression of the Acj6(1,4) isoform causes a failure of motor axons of the ISNb fascicle to defasciculate from the ISN fascicle, resulting in defective innervation of the ventrolateral muscle field in approximately 65% of embryonic hemisegments examined. In approximately 15% of the hemisegments, the ISNb motor axons stop at the correct location of their target muscles but do not branch out to innervate the ventral muscle groups. The remaining hemisegments (20%) show a wild-type pattern of muscle innervation. These studies suggest that distinct Acj6 isoforms can influence specific aspects of nerve terminal branching and synapse formation. In addition, this activity appears to be mediated by differential activities of the N-terminal POU IV box (Certel, 2000).