Using a specific antibody raised against the domain common to all Lola isoforms (Figure 1b), we found that Lola is expressed in all cells in the AL region at all stages of development from larva to adult (Figure 1c–g). These include expression in post-mitotic cells such as PNs, but also in neuroblasts, which we can identify by their size during larval development (Figure 1c). Antibody specificity was confirmed by loss of Lola staining in neuroblast and single-cell MARCM clones homozygous for a lola null allele (ore76, which introduces an early stop codon in the BTB domain of all Lola isoforms (Figure 1b), hereafter referred to as lola^-/-) in adult (Figure 1h,i) and in early pupa (data not shown). Note that Lola staining can be used to identify all mutant cells in a MARCM clone that are not necessarily labeled by UAS-mCD8::GFP because Lola is expressed ubiquitously throughout the brain, while Gal4 enhancer traps often have a more limited domain of expression (Figure 1h).

Since Lola is expressed in PNs during AL development, and mutant clones effectively eliminate Lola protein during the period of dendritic targeting, we proceeded to analyze the phenotype of lola^-/- PN MARCM clones. Using the Gal4-GH146 enhancer-trap line to label PNs, we observed wiring defects in all lineages of PN MARCM clones as well as anterodorsal single cell clones that normally target the DL1 glomerulus (Figure 2).

In wild type, Gal4-GH146 labels PNs of three neuroblast lineages. PNs from adPN and lPN lineages target their dendrites to stereotypical, intercalating but non-overlapping sets of glomeruli 2 (Figure 2a1,a2,c1). In general, phenotypes of lola^-/- adPNs and lPNs can be classified into two main categories: a loss of targeting to normal target glomeruli and a gain of targeting to off-target (ectopically targeted) glomeruli (Figure 2b1,b2). However, the phenotypes are highly variable between brains and there are few consistent 'shifts' where one off-target glomerulus is innervated in place of a normal target. The phenotypes also have variable penetrance, with innervation of both normal and ectopic targets ranging from partial to full innervation from brain to brain. Specifically, lola^-/- adPNs often invade DA1, DL3 and VA4, which are normal lPN targets. Likewise, mutant lPNs often innervate VA1d, VA1lm, DM6, D, DC2 and VM7, which are normal adPN targets (Figure 2c2). lola^-/- dendrites also appear more diffuse in general, and often 'wander' through the lobe, wrapping around, but not actually innervating, glomeruli. Additionally, lola^-/- dendrites are frequently not restricted to the AL and innervate regions of the suboesophageal ganglion (SOG; Figure 2b1,b3,c2). Wild-type vPNs strongly innervate DA1 and VA1lm, and one vPN (pan-AL PN) innervates the entire AL (Figure 2a3). lola^-/- vPNs show a strong loss of AL restriction and nearly 37% show ectopic dendritic extensions to the SOG (Figure 2b3,c2). Additionally, DA1 targeting in ventral clones is almost completely lost (6% versus 82% in wild type) and VA1lm targeting is reduced nearly 30% (Figure 2c2). We also analyzed an independent lola^ore5D2 allele of lola, which is predicted to be strongly hypomorphic due to a P-element insertion in the lola promoter (Figure 1b). lola^ore5D2 clones display many of the phenotypes observed in the ore76 null allele (see additional file 1), indicating that the phenotypes we observed are caused by loss of lola.

Given that Lola is ubiquitously expressed, dendritic defects in neuroblast clones can be attributed to its requirement in the neuroblast or post-mitotic cells or both; furthermore, one cannot determine cell autonomy with certainty. We therefore examined single cell clones that target to the DL1 glomerulus to test whether lola has a cell-autonomous and post-mitotic function. We found that nearly 80% of lola^-/- DL1 single cell clones show targeting defects (Figure 2b4,c2) and 23% completely fail to innervate DL1. Another 54% innervate DL1 (though often weakly or partially) and have additional extensions either wandering between glomeruli or, more frequently, innervating glomeruli anterior to DL1 in the AL (see Table S1 in additional file 2). Using the pan-Lola antibody, we can verify there are no unlabeled clones in the vicinity of the AL and be certain we are looking at cell-autonomous effects. At least in the DL1 PN, therefore, lola appears to have post-mitotic, cell intrinsic effects on dendritic targeting.

To extend higher resolution phenotypic analysis to other PN classes, we examined adPN and lPN neuroblast clones using Gal4-Mz19, which labels a small subset of GH146-positive neurons that project to DA1, VA1d and DC3 4. Interestingly, dendritic targeting of lola^-/- DA1 and VA1d PNs exhibit phenotypes with much less severity and penetrance than other lola^-/- PNs. Normal target glomeruli were almost always strongly targeted, and defects mainly consisted of additional dendritic extensions into the AL (Figure 2b5,6). Consistent with this, we find that DA1 (innervated by lPNs) and VA1d (innervated by adPNs) are two of the most 'stable' glomeruli and show little phenotype in GH146 mutant neuroblast clones (Figure 2c2). These two glomeruli are also often ectopically innervated by lola^-/- PNs from the opposite lineage (that is, DA1 is ectopically targeted by lola^-/- adPNs while VA1d is ectopically targeted by lola^-/- lPNs). This experiment suggests that distinct PN classes differentially require lola.

We also tested available mutations that affect individual lola isoforms. The orc4 (affecting isoform K; Figure 1b) allele did not show obvious defects (see additional files 1, 4 and 8). The ore119 allele (affecting isoform L; Figure 1b) shows a small subset of the phenotypes observed in lola^-/- (see additional file 1). lola L^-/- adPN targeting appears normal, but lPNs show a loss of targeting to VM1 and DA2, while ventral clones show a loss of DA1 targeting. However, we cannot rule out the possibility that an additional mutation on the lola L mutant chromosome contributes to the phenotype, as our attempts to generate a second allele specific for lola L have not been successful and the ore119 phenotype could not be rescued by a UAS-lola L transgene (data not shown). In fact, MARCM expression of lola L in PNs results in severe phenotypes by itself (see below).

In addition to dendritic phenotypes, we observed defects in PN axonal wiring in lola^-/- MARCM clones. As with dendrites, defects were found in adPNs, lPNs and vPNs as well as DL1 single-cell clones (Figure 3). Phenotypes were highly variable both in penetrance and severity in neuroblast as well as single cell clones. In a semi-quantitative approach, we scored phenotypes as mild, medium and severe (see Figure 3 legend). A separate quantification of the targeting defects limited by specific phenotypes (ectopic branching, misrouting of axons, SOG targeting and a lack of MB/LH innervation) is presented in Table S2 (see additional file 2).

In lola^-/- adPNs and lPNs, the most common defect was ectopic branching outside of the region of the MB or the LH (28% and 63%, respectively; Figure 3b1,2; see Table S2 in additional file 2). Some clones showed a lack of innervation, where innervation of the MB failed to occur or branch extension in the LH was severely limited. A smaller percentage of lola^-/- clones (15% adPNs and 7% PNs) showed medium or severe axon defects (Figure 3d). Hypomorphic clones (ore5D2) show similar, but mostly mild, targeting defects (see Table S2 in additional file 2).

Axons of vPNs are the most severely affected: nearly 100% of clones exhibit defects in both lola^-/- and hypomorphic (ore5D2) alleles (Figure 3b3,d; see Table S2 and S4 in additional file 2). Often ventral axons show several defects in the same brain. Seventy-nine percent of lola^-/- clones show ectopic branching defects (often extensions from the LH towards the SOG or an increase in branches along the medial extension). Thirty-four percent of clones show misrouting defects; most often a single axon defasciculates from the bundle and projects directly to the medial-central region of the LH instead of the ventral region (compare Figure 3a3 to 3b3). Nine percent of null clones and 47% of hypomorph clones show clear axonal targeting to the SOG (this may be higher in the null but it can be difficult to distinguish between axon and dendrite mistargeting unless clear defasciculation from the main axon bundle is visible). We observe that all PN axons exit the AL in a single bundle, but then some axons defasciculate and project ventrally to the SOG (Figure 3c). It is interesting to note that axons misguided to the SOG typically project to similar locations in the SOG, suggesting they may be responding to some cues in this region.

More than 50% of DL1 single cell clones show axon defects (Figure 3b4,d, see Table S3 in additional file 2), suggesting that Lola acts cell-autonomously in axon targeting. Twenty-six percent of DL1 lola^-/- axons have ectopic branches, often with additional branches in the LH region (see Table S2 in additional file 2). Fifteen percent of clones fail to extend branches into the MB (see Table S2 in additional file 2). Some DL1 axons clearly bifurcate just prior to entering the MB. It is interesting to note that many DL1 clones that fail to target the DL1 glomerulus have relatively normal axonal projections. A severe dendrite phenotype does not predict the most severe axon phenotypes (or vice versa), suggesting that these two processes are separable, in agreement with recent observations of independence between axon and dendrite targeting 14. These results suggest that lola has post-mitotic, cell-autonomous functions in PN axon targeting.

Our MARCM analysis suggests that lola has an important function for PN dendritic and axonal targeting. Given that lola encodes 20 alternatively spliced isoforms, several scenarios may account for lola function (see additional file 3). In scenario I, different isoforms may be expressed in different neuroblast lineages and thereby specify their lineage-specific wiring analogous to POU transcription factors 13. In scenario II, individual PNs may express a unique isoform or combination of isoforms independent of lineage to specify PN identity. In scenario III, all cells may express most or all lola isoforms.

To distinguish between these possibilities, we performed RNA in situ hybridization with probes against 16 of the 20 lola isoforms at 0 h after puparium formation (APF), a developmental time when PNs start to elaborate dendrites in the AL 4. We first verified our in situ protocol by using probes against drifter (dfr), a TF that has previously been shown to be expressed in GH146 labeled lPNs but not adPNs 13. As expected, we found that signal from anti-sense probe against dfr mRNA coincides with green fluorescent protein (GFP)-labeled lPNs but not adPNs (see additional file 3). A sense control probe shows little signal coinciding with the GFP-positive cell bodies (see additional file 3). We then performed in situ analysis with lola probes. A probe to the lola common region shows expression throughout the brain (Figure 4a1,a3) compared with sense control (Figure 4a2), consistent with our Lola antibody staining results. For individual isoform mRNAs, we detect signal in the region of the AL for all lola isoforms tested, as well as other brain regions. Some isoforms have narrower domains of expression, such as isoforms N (see additional file 3) and T (Figure 4d1,d3), which have much stronger signal in the central brain and weaker staining in the optic lobe. Other isoforms appear to be expressed more uniformly throughout the region of the AL (Figure 4; see additional file 3). It should be noted that several isoforms have more punctate signal that co-localizes with the nucleus (see additional file 3f, k, r), perhaps indicative of pre-mRNA. Another caveat is that mRNA production may not translate to protein expression, as recently shown by the wide mRNA but narrow protein expression domains of nerfin 31. We did not find PN lineage-specific expression for any of the lola isoforms, nor is there evidence that any isoforms are expressed in a subset of PNs within a lineage. Rather, most lola isoform mRNAs appear to be expressed in all PNs, consistent with scenario III (see additional file 3).

As an independent and more quantitative method of assessing expression of different isoforms in PNs, we performed laser-dissection microscopy (LCM) to capture GH146-positive PNs from frozen sections of 0 h APF tissue, followed by quantitative RT-PCR to detect the expression level of individual isoforms. Although we do not have single-cell resolution, laser-dissection allows us to assay mRNA expression in a small subset of neurons. Consistent with the in situ hybridization data, at 0 h APF we detected most lola isoforms by RT-PCR, although several isoforms are near the RT-PCR detection limit (Figure 4e). We find marked differences (up to 100-fold) in the level of RNA expression of individual isoforms in PNs at 0 h APF (Figure 4e). We additionally analyzed AL samples from different developmental timepoints (third instar larvae, 0 h APF, 25 h APF and adult) and find that expression levels of several lola isoforms, for example isoform E, H, I, L, M and Q, appear to be developmentally regulated (see additional file 4). Our in situ results also suggested that lola isoforms are expressed at different levels in the optic lobe. Indeed, RT-PCR on LCM captured optic lobe at 0 h APF confirmed that mRNAs of different lola isoforms are expressed at different levels and revealed differences in the pattern of isoform expression between AL and optic lobe samples (see additional file 4). Together, these data suggest that the level of individual lola isoform expression is dynamically regulated and that individual isoforms may function at different stages in development.

Genetic analysis using available isoform-specific alleles (see additional file 1) indicates that at least two isoforms (K and L) are not required for dendritic targeting of the majority of PN classes. To test further potential functions of different isoforms, we examined the overexpression phenotypes of three lola isoforms. Previous reports suggested differences between the overexpression phenotypes of lola A and lola T in motor neurons 25. We have generated a third transgene, UAS-lola L. We found that Gal4-GH146 expression of any of these three UAS transgenes results in lethality, but we bypassed this lethality by MARCM mediated transgene expression in specific neuroblast lineages or single cells (Figure 5; see additional file 5). We tested one, three and two insertions for UAS-lola A, UAS-lola L and UAS-lola T, respectively (see Table S1 in additional file 2). Levels of transgene expression were verified in our attempted rescue experiments (see below). Because Gal4-GH146 drives transgene expression in post-mitotic PNs (J Liu, MS, LL, unpublished observation), the overexpression phenotypes we describe below are caused by post-mitotic expression of lola transgenes.

Neuroblast or single cell clones overexpressing UAS-lola A (which lacks a Zn-finger) do not exhibit detectable defects (Figure 5b; see additional file 5). However, UAS-lola L and UAS-lola T (which both have C₂HC-C₂H₂-type Zn-fingers) show strong targeting defects in adPN and lPN neuroblast clones (see additional file 5). Phenotypes often include a large increase in wandering dendrites in the AL and some degree of loss of targeting and ectopic targeting. Often dendrites appear to be largely restricted to one region of the AL such as dorsal or lateral, and have wandering dendrites through this region in many glomeruli, but fail to innervate normal targets or extend any dendrites into other regions of the AL (see Additional file 5c, e, f). UAS-lola T expression also results in a decrease in the number of labeled PNs, in part likely due to cytotoxicity based on the observation of clones with very punctate GFP staining patterns indicative of cell and process degeneration, although we were unable to reliably stain brain tissues for activated caspase-3 (V Trunova and E Giniger, unpublished data). Due to the difficulty in quantifying phenotypes in adPN and lPN neuroblast clones because of the magnitude and variability, we focused our analysis on vPN neuroblast clones and single-cell DL1 clones.

vPN neuroblast clones in animals raised at 25°C show no defects when overexpressing lola A (Figure 5b1) and weak phenotypes when overexpressing lola L (Figure 5c1), but vPNs are strongly affected by lola T overexpression (Figure 5e1,f1). Targeting to DA1 is often lost in vPN clones overexpressing lola T (Figure 5e1,f1,g). vPN dendrites frequently extend outside of the AL boundary, most often in the ventral region projecting towards the SOG. Processes of the pan-AL vPN are not elaborated properly and only partially innervate the AL.

DL1 single-cell clones were used to examine cell-autonomous overexpression effects on both dendritic and axonal targeting. lola A overexpression did not affect targeting of either DL1 axons or dendrites (Figure 5b2,b3). The DL1 dendrites targeted correctly with lola L overexpression, but frequently showed a large increase in dendritic mass outside of DL1, wandering throughout the AL (Figure 5c2,h). In axons, overexpression of lola L often resulted in an increase in branching in MB and LH regions (Figure 5c3,i). In the LH, these additional branches resulted in a loss of the stereotypic 'L' projection and a failure of branches to fully extend into normal target areas. lola T overexpression resulted in variable DL1 phenotypes, ranging from a complete loss of DL1 innervation and targeting to other regions of the AL to DL1 targeting with dendritic extensions outside of the DL1 glomerular region (Figure 5e2,f2,h). lola T overexpression also affected axon targeting, most often resulting in a high degree of ectopic branching, axon bifurcation and incorrect LH innervation patterns (Figure 5e3,f3,i). It is interesting to note that isoforms L and T are normally expressed at low levels based on RT-PCR results, while isoform A is expressed roughly 10-fold higher (Figure 4e). Differences in overexpression phenotype may be related to the presence or absence of a Zn-finger domain, and/or to the expression level relative to the endogenous level of expression.

In the course of our overexpression study, we tested two different insertions of UAS-lola T. We noticed that the insertion on the third chromosome gives stronger and more penetrant phenotypes than the insertion on the X chromosome, and hypothesized that this was due to differences in transgene expression levels. To test this hypothesis, we altered the levels of UAS-lola T (X insertion) transgene expression by raising animals at 18°C, 25°C or 29°C, as levels of Gal4-induced transgene expression increase with increasing temperatures. In neuroblast clones we observed a qualitative increase of phenotypic severity in both dendritic and axonal targeting with increasing temperature (Figure 5d,e,f). This dosage sensitivity can be quantified with the DL1 dendritic phenotypes (Figure 5h). Flies raised at 18°C have few dendritic phenotypes, while at 25°C and 29°C dendrites mistarget and have ectopic innervations and axons have additional branches and bifurcations (Figure 5h,i; see Table S1 in additional file 2). Additionally, dosage sensitivity is evident in vNB DA1 and SOG extension phenotypes (Figure 5g). This experiment indicates that PN dendritic targeting is sensitive to the expression level of a specific lola isoform.

To test if individual lola isoforms can rescue any of the lola^-/- phenotype, we performed MARCM analysis on clones that are simultaneously lola^-/- and express either UAS-lola A, UAS-lola L or UAS-lola T transgenes. We verified all transgenes are expressed based on the presence of Lola antibody staining in these otherwise lola^-/- clones. Expression of all transgenes was qualitatively similar at a level roughly half as intense as neighboring non-mutant cells expressing endogenous levels of Lola (data not shown). Instead of rescuing the lola^-/- phenotype, overexpression of transgenes in mutant PNs resulted in additive defects (Figure 6; see additional file 6). In adPN and lPN neuroblast clones, simultaneous loss of endogenous lola and expression of any of the single isoforms caused striking additive effects in cell loss and lack of dendritic extension, which are described in detail in additional file 6. Below we focus on vPN neuroblast clones and DL1 single cell clones.

Additive effects of transgene expression were most striking in vPNs. lola^-/- vPNs expressing UAS-lola A, UAS-lola L or UAS-lola T resulted in strong reduction of VA1lm and DA1 targeting (Figure 6f). Additionally, the pan-AL neuron often showed little dendritic process elaboration and rarely innervated much of the AL. Dendrites and axons of ventral cells often looked punctate, indicative of degeneration. These data indicate that expression of UAS transgenes in lola^-/- clones fails to rescue lola phenotypes and further disrupts dendritic wiring. Additionally, expression of certain lola isoforms alone may be toxic, perhaps due to dominant-negative interactions with other proteins or lola dimerization.

DL1 clones were difficult to obtain for all UAS-lola transgenes expressed in lola^-/- clones, possibly due to toxicity that we observe in neuroblast clones. Phenotypes did not always appear additive for DL1 clones, perhaps due to low levels of MARCM-based expression in single cell clones during the critical time of dendritic and axonal targeting. The strongest dendritic phenotypes were observed in lola^-/- cells expressing UAS-lola T, where every clone either failed to innervate DL1 or had extensions into other regions of the AL (Figure 6e,g). lola^-/- cells expressing UAS-lola L innervated DL1 in addition to having extensive dendritic innervation in the rest of the AL, while lola^-/- cells expressing UAS-lola A were variable, with half appearing normal (Figure 6c,d,g). All lola^-/- cells expressing a UAS-lola transgene showed axonal defects, ranging from a failure to extend or elaborate branches in the LH to an increase in ectopic branching (Figure 6h). These phenotypes suggest that individual lola isoforms can function post-mitotically to disrupt dendrite and axon targeting in DL1, and suggest that Lola interacts either as homomers or with other BTB domain proteins. Additionally, expression of single lola isoforms, especially lola T, appears to be toxic to many cells, so it is unclear if phenotypes are directly related to targeting or perhaps secondary effects due to disruption of cell viability.

Taken together, these results suggest that isoform diversity of the lola locus is important. Post-mitotic expression of a single isoform is not sufficient to rescue the lola phenotype, and indeed is more disruptive to a cell than simple loss of lola. This does not, however, rule out an additional function of Lola in the neuroblast. Lastly, some aspects of Lola function are likely mediated through the BTB domain alone, as additive phenotypes can be caused by expressing isoforms with or without Zn-fingers.

The dendritic and axonal phenotypes in lola^-/-clones we described so far could be caused by the action of Lola on PN targeting directly (for instance, by regulating the expression of cell surface receptors), or Lola could regulate PN identity with axonal and dendritic mistargeting as a secondary consequence. Although these models are not mutually exclusive, we provide some evidence in lola^-/-MARCM clones using several Gal4 lines, including Mz19, GH146 and NP3529, to support the second hypothesis.

MARCM with several mutant alleles suggests that lola regulates the expression of Gal4-Mz19. Ten percent of lola^ore5D2 (lola hypomorph) and lola^-/- clones show a 30% to 50% increase in the number of labeled PNs and targeted glomeruli (Figure 7a2,a3; MARCM control Figure 2a5,a6). Strikingly, we observed (in 64% of all adPN clones) that lola L^-/-(lola^ore119) clones frequently target the DL1 glomerulus (Figure 7a5), which is normally targeted by GH146 positive, but never Mz19 positive, PNs. As our heat-shock timing in these experiments should not result in any Mz19 single cell clones (they are born later) but coincides exclusively with the birth of the DL1 PN 2, the simplest interpretation is that Mz19 is now misexpressed in DL1 PNs. This would suggest that lola normally represses Mz19 expression in DL1 PNs. As additional evidence that lola regulates gene expression, 25% of clones of both lola^-/- and lola L^-/-visualized by Gal4-Mz19 appear to label local interneurons (LNs), an additional cell-type that innervates a large number of glomeruli in the AL and lacks axonal projections out of the AL region (Figure 7a4,a6). The similarity of isoform L specific phenotypes to phenotypes in other lola alleles suggests that lola^ore119 phenotypes in PNs might be caused by the disruption in isoform L instead of a second site mutation on the same chromosomal arm. Thus, it appears that in the absence of Lola (particularly the Lola L isoform), Gal4-Mz19 is ectopically expressed in neuronal types that normally do not express this Gal4 enhancer trap line.

Gal4-GH146, used extensively in this study, labels a large subset of PNs in the AL, roughly 50 adPNs, 35 lPNs and 5 or 6 vPNs (Figure 7b1; MARCM controls Figure 2a1–3). However, in lola^-/- neuroblast clones we observe labeling of a new cell type that innervates a small region of the AL and sends projections to the medial axonal lobe of the MB as well as the central body complex (Figure 7b2–3). These cells are never labeled in control experiments, but make up 22% of all observed clones in lola^-/- experiments (see Table S5 in additional file 2). Interestingly, we observe similar projection patterns in wild-type neurons labeled with other Gal4 drivers, suggesting that central body complex targeted lola^-/- clones are a different cell type where Gal4-GH146 expression is normally repressed by lola.

Gal4-NP3529 labels two dorsal PNs that project to DL1 (Figure 7c1). Clones generated with both lola^-/- and lola L^-/- label much larger subsets of cells. These clones often contain PNs that innervate typical GH146 target glomeruli, as well as LNs and other PN classes that do not express GH146. lola^-/- clones label many additional cells in the adPN lineage as well as lPNs and LNs (Figure 7c2). lola L^-/- clones show similar types of mislabeling (Figure 7c3). This strongly suggests that lola function is normally necessary to repress the expression of NP3529 in multiple cell types.

Lola also appears to positively regulate gene expression. Lim1 is expressed in a subset of PNs, including most (at least four) vPNs labeled by GH146 (Figure 7d1,d2) 14. In lola^-/- mutant cells, Lim1 expression is lost (Figure 7e1,e2). Lim1 expression cannot be rescued by expression of either UAS-lola A or UAS-lola T in lola^-/- clones (n = 2 each for UAS-lola A and UAS-lola T; data not shown). Lim1 expression is also lost in vPNs mutant for the TF cut 14. However, Cut expression is not altered in lola^-/-clones, and Lola expression is not altered in cut^-/-clones (data not shown), suggesting these TFs coordinately regulate Lim1 expression through parallel pathways. Intriguingly, PNs mutant for lola, lim1 or cut all fail to target DA1, suggesting that cut and lola may be necessary for regulation of other TFs as well as wiring. Antibody staining patterns of other TFs previously identified to be important for PN targeting, including acj6, drifter, cut, chip and islet, are not disrupted in lola^-/- clones (data not shown). Taken together, these data suggest that lola can positively and negatively regulate gene expression of other genes known to label subsets of PNs or to play a role in PN wiring specificity.

MARCM stocks used were as previously published 35. Gal4-GH146 and Gal4-Mz19 are available from the Bloomington Stock Center, and Gal4-NP3529 is from the Kyoto Stock Center. lola^ore76 (early stop codon in BTB domain predicted to be protein null for all Lola isoforms), lola^ore119 (Pro712 to Leu mutation in the Zn-finger linker region of isoform L predicted to reduce DNA binding by 95%), lola^orc4 (C to T transition introduces an early stop in amino acid 771 of isoform K), lola^5D2 (P-element insertion into the lola promoter predicted to be strongly hypomorphic allele and eliminate most lola transcription), UAS-lola A (X), UAS-lola T (X) and UAS-lola T (III) have been described previously 162425. We generated three new UAS-lola L lines with insertions on chromosome III. The lola L coding region was cloned into p [UAST] 36 and injected into embryos. Single G₀ males were crossed to yw; Pin/CyO virgins. Single F₁ males with red eyes were crossed to yw; Pin/CyO virgins to establish stocks and identify the insertion chromosome.

Genotypes of flies generated in experiments are as follows: (Lola protein expression time course and Lim1 WT staining) yw; Gal4-GH146, UAS-mCD8a::GFP (GH146 staining and LOF analyses) hsFlpCD8; FRT42D, Gal80/FRT42D, lola^x, GH146, UAS-mCD8a::GFP [where x denotes any lola allele] (control GH146 MARCM) hsFlpCD8; FRT42D, Gal80/FRT42D, GH146, UAS-mCD8a::GFP (lola^ORC4 LOF analysis) hsFlpCD8; FRT42D, Gal80/FRT42D, lola^orc4, GH146, UAS-mCD8a::GFP (control Mz19 MARCM) hsFlpCD8; Gal4-Mz19, UAS-mCD8a::GFP, FRT42D, Gal80/FRT42D (Mz19 LOF analysis) hsFlpCD8; Gal4-Mz19, UAS-mCD8a::GFP, FRT42D, Gal80/FRT42D, lola^x (UAS-lola A misexpression) UAS-lola A/hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, Gal4-GH146, UAS-mCD8a::GFP (UAS-lola L (X) misexpression) UAS-lola T/hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, Gal4-GH146, UAS-mCD8a::GFP (UAS-lola T (III) misexpression) hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, Gal4-GH146, UAS-mCD8a::GFP; UAS-lola T (UAS-lola L misexpression) hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, Gal4-GH146, UAS-mCD8a::GFP; UAS-lola L (UAS-lola A rescue) UAS-lola A/hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, lola^ORE76, Gal4-GH146, UAS-mCD8a::GFP (UAS-lola T rescue) UAS-lola T/hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, lola^ORE76, Gal4-GH146, UAS-mCD8a::GFP (UAS-lola L rescue) hsFlp¹²², UAS-mCD8a::GFP; FRT42D, Gal80/FRT42D, lola^ORE76, Gal4-GH146, UAS-mCD8a::GFP; UAS-lola L (NP3529 LOF analysis) hsFlp¹²², UAS-mCD8a::GFP;FRT42D, Gal80/FRT42D, lola^x;Gal4-NP3529.

MARCM was performed as previously described 3035. Clones were induced at 24 h APF for all Gal4 driver lines. Brains were dissected, fixed in 4% paraformaldehyde for 20 minutes at room temperature and stained. Brains were mounted in Slow-Fade Gold or BioMedia Mounting Medium and imaged on a Zeiss LSM Meta 510 system. Z-series were captured in 1 μm sections. Fluorophores were imaged using band-pass filters to remove cross-detection between channels and pseudocolored for ease of viewing. Images were processed and scored using Image J 37 or Zeiss software. Data were analyzed using FileMaker Pro and Microsoft Excel. In all, we scored 48 distinct glomeruli in the AL, based on published models 38. Glomeruli were considered innervated when dendrites entered the nc82 stained synaptic dense region and elaborated extensions within the glomerulus (that is, a single extension passing through a glomerulus was not enough to warrant scoring as positively innervated).

Antibody staining was performed as previously described 1439. The following primary antibodies were used; rat anti-mCD8a (Caltag [Burlingame, California, USA]), 1:100; monoclonal antibody (mAb) nc82 (developed by E Buchner and obtained from the Developmental Studies Hybridoma Bank (DSHB) [Iowa City, Iowa, USA]), 1:30; rabbit anti-Lola common region 24, 1:100; mAb anti-Cut (DSHB), 1:20; guinea pig anti-Lim1 (a gift from J Botas), 1:500; rat anti-Islet (a gift from J Skeath), 1:1,000; rabbit anti-Chip (a gift from D Dorsett), 1:500; mouse monoclonal anti-Acj6 40, 1:5; rat anti-Drifter 41, 1:3,000; mouse monoclonal 7F1 anti-lola zf5 21. Nuclei were counterstained with DAPI (5 mg/ml stock 1:1,000).

White y, w; Gal4-GH146, UAS-CD8::GFP pre-pupa (considered 0 h APF) were collected, covered with optimal cutting temperature compound (Tissue Tek O.C.T. #4583 [Sakura Finetek, Torrance, California, USA]), frozen on dry ice and stored at -80°C not longer than 1 month. Blocks were sectioned on a cryostat and sections containing ALs were mounted on PEN membrane slides (Leica #11505158) and stored on dry ice. Slides were put through an EtOH dehydration series (75%, 95%, 100% EtOH 5 minutes each), treated for 2 minutes with xylenes, dried and mounted for LCM capture. PN enriched samples as identified by GFP fluorescence were immediately captured into PCR tube caps on a Leica LCM microscope (model ASLMD) using Laser Microdissection software version 4.4. Tubes were frozen on dry ice and stored at -80°C up to one day. RNA was extracted using the Arcturus PicoPure™ RNA Isolation kit (#KIT0204, Mountain View, California, USA) as per the manufacturer's instructions.

The cRNA samples were used as the template for another round of reverse transcription (#18080-051, Invitrogen [Carlsbad, California, USA]), producing a high yield of cDNA. The cDNA samples were aliquoted and stored at -20°C before use. Unique primer pairs and Taqman probes were designed for individual lola isoforms. The Taqman probes were synthesized by Integrated DNA Technologies, Inc. (Coralville, Iowa, USA). Microfluidic matrix chips (48 × 48 arrays, Fluidigm (South San Francisco, California, USA) were used to perform real time PCR assays in a high throughput fashion. Each chip allows complete combinatorial tests (48 × 48 = 2,304, N = 48) between N independent cDNA samples and N distinct primer pairs 42. A separate paper is to be published with more technical details.

Probes for lola were designed to the common region and isoform specific exons (for primers see Table S6 in additional file 2). Isoforms are referred to based on the previously described naming scheme 16. Probes were amplified from adult genomic preps or from embryo RNA via RT-PCR, topo-cloned and sequenced to verify identity. DIG-labeled RNA probes were generated from plasmid DNA using the Roche DIG-RNA Labeling Kit (#11175025910, Indianapolis, Indiana, USA) following the manufacturer's protocol. Sense and antisense probes were generated from the same plasmid, taking advantage of the SP6 and T7 dual promoter flanked PCR2.1-Topo (# K460001, Invitrogen) cloning site. Probes were diluted to 10 ng/μl and stored at -80°C for up to 6 months. We collected 0 h APF pre-pupa, which were frozen in OCT on dry ice, stored at -80°C not more than 1 month and sectioned on a cryostat. An in-depth protocol for in situ hybridization is available in additional file 7. Briefly, slides were fixed in 4% PFA for 15 minutes, washed, acetylated, washed, and pre-hybridized at room temperature. Probes were added at concentrations ranging from 5 ng to 100 ng and slides were incubated 18 h overnight at 55°C. Slides were washed, native biotin blocked and blocked in normal goat serum. Sheep anti-DIG or rabbit anti-DIG and chicken anti-GFP primaries were added and incubated overnight at 4°C. Slides were washed, treated with a tyramide signal amplification kit for DIG signal detection (Dako #K0620 [Carpinteria, California, USA]), then incubated with secondary anti-chicken antibody. Slides were washed, mounted and imaged.