Squeeze involvement in the specification of Drosophila leucokinergic neurons: Different regulatory mechanisms endow the same neuropeptide selection

www.elsevier.com/locate/modo

Mechanisms of Development 124 (2007) 427–440

Squeeze involvement in the specification of Drosophilaleucokinergic neurons: Different regulatory mechanisms endow the

same neuropeptide selection

Pilar Herrero 1,*, Marta Magarinos 1, Isabel Molina, Jonathan Benito, Belen Dorado,Enrique Turiegano, Inmaculada Canal, Laura Torroja

Departamento de Biologıa, Universidad Autonoma de Madrid, Cantoblanco, E 28049 Madrid, Spain

Received 11 December 2006; received in revised form 14 February 2007; accepted 1 March 2007Available online 12 March 2007

Abstract

One of the most widely studied phenomena in the establishment of neuronal identity is the determination of neurosecretory pheno-type, in which cell-type-specific combinatorial codes direct distinct neurotransmitter or neuropeptide selection. However, neuronal typesfrom divergent lineages may adopt the same neurosecretory phenotype, and it is unclear whether different classes of neurons use differentor similar components to regulate shared features of neuronal identity. We have addressed this question by analyzing how differentiationof the Drosophila larval leucokinergic system, which is comprised of only four types of neurons, is regulated by factors known to affectexpression of the FMRFamide neuropeptide. We show that all leucokinergic cells express the transcription factor Squeeze (Sqz). How-ever, based on the effect on LK expression of loss- and gain-of-function mutations, we can describe three types of Lk regulation. In thebrain LHLK cells, both Sqz and Apterous (Ap) are required for LK expression, but surprisingly, high levels of either Sqz or Ap alone aresufficient to restore LK expression in these neurons. In the suboesophageal SELK cells, Sqz, but not Ap, is required for LK expression.In the abdominal ABLK neurons, inhibition of retrograde axonal transport reduces LK expression, and although sqz is dispensable forLK expression in these cells, it can induce ectopic leucokinergic ABLK-like cells when over-expressed. Thus, Sqz appears to be a regu-latory factor for neuropeptidergic identity common to all leucokinergic cells, whose function in different cell types is regulated by cell-specific factors.� 2007 Elsevier Ireland Ltd. All rights reserved.

Keywords: Leucokinin; Leucokinergic regulation; Squeeze transcription factor; Squeeze–Apterous interaction; Neuropeptidergic cell identity; Drosophila

1. Introduction

The accurate performance of the mature CNS relies onthe generation of a diversity of individual neuronal identi-ties at specific times and locations during development.Work on species from worms to humans has demonstratedthat this process involves multiple stages of regulatory geneactivation and repression orchestrated by intrinsic andextrinsic mechanisms. In the postmitotic neuron, it is coor-

0925-4773/$ - see front matter � 2007 Elsevier Ireland Ltd. All rights reserve

doi:10.1016/j.mod.2007.03.001

* Corresponding author. Tel.: +34 914978144; fax: +34 914978344.E-mail address: [email protected] (P. Herrero).

1 These authors contributed equally to this work.

dinated by the combined action of transcription factors,which defines a specific code conferring individual proper-ties on a given neuronal type. Transcription factors withbHLH, LIM and Zinc Finger domains, among others, havebeen reported as acting cell autonomously to establish spe-cific neuronal subpopulations (Edlund and Jessell, 1999;Lee and Pfaff, 2003). In addition, neuronal identity isconfigured extrinsically by external signals mediated bycross-repressive mechanisms (Broihier and Skeath, 2002)or retrograde transport (Ernsberger and Rohrer, 1999;Koo and Pfaff, 2002).

One of the most studied phenomena in the establishmentof neuronal identity is the determination of neurosecretory

d.

mailto:[email protected]

428 P. Herrero et al. / Mechanisms of Development 124 (2007) 427–440

phenotype. The initial analysis of the transcription regula-tory elements present in the Drosophila Fmrf gene sug-gested that different regulatory mechanisms controlledexpression of this neuropeptide in specific cell types(Schneider et al., 1993; Benveniste and Taghert, 1999). Sub-sequently, several elegant studies demonstrated the conver-gence of intrinsic and extrinsic mechanisms in the controlof the expression of the FMRFamide neuropeptide in theTv neuron of Drosophila melanogaster (Allan et al., 2003,2005; Marques et al., 2003). Expression of the FMRFa-mide gene in these cells requires the action of the LIM-HD transcription factor Apterous (Benveniste et al.,1998) in combination with the Zinc Finger transcriptionfactor Squeeze (Allan et al., 2003). Before neuropeptideexpression, coexpression of both transcription factorsenables Tv cells to respond to target-derived activation ofthe BMP pathway by the ligand Gbb (Allan et al., 2003;Marques et al., 2003). This leads to the transcriptional acti-vation of the FMRFamide gene, a process that alsorequires the action of Apterous and Squeeze, as well asthe bHLH transcription factor Dimmed (dimm) (Allanet al., 2003, 2005; Hewes et al., 2003).

Much of the work in this field has been directed towardsunderstanding how different combinatorial codes specifydifferent differentiation programmes. Studies in flies andworms have revealed the complexity of the combinatorialrelationships among transcriptional regulators. In a givencell, they can play both combinatorial and independentroles. Moreover, the apparently independent functions ofthese factors and their combinatorial relationships can dif-fer between cell types, so that, rather than following a gen-eral rule, they appear to depend on the context ofindividual cell types. This is clearly exemplified by the Dro-

sophila ap-lateral cell cluster, which is composed of fourneurons in the ventral ganglion (Tv, Tva, Tvb, and Tvc)that express ap but are phenotypically distinct. In this clus-ter, two combinatorial codes regulate expression of twoneurosecretory genes: ap, sqz, BMP signalling, and dimmcontrol FMRFamide transcription in the Tv cell, whileap and dimm (but not sqz or the BMP pathway) regulateFurin 1 expression in the Tvb cell (Allan et al., 2003,2005). However, while dimm action on FMRFamideexpression is mostly ap-dependent, ap and dimm also actsynergistically to regulate Furin 1 transcription. Finally,ap acts independently to regulate axon pathfinding by allap cells except for Tv cells. Similar examples of combinato-rial codes that control terminal differentiation of singleneuronal types and that show complex cell-type-specifichierarchical interactions have been described in C. elegans

(Altun-Gultekin et al., 2001; Wightman et al., 2005; Zhenget al., 2005) and vertebrates (reviewed in Goridis and Roh-rer, 2002; Howard, 2005).

In establishing cell identity, it is also important to under-stand how a variety of neuronal types of different lineagesshare a common phenotypic trait, such as a neurotransmit-ter. This has been extensively studied in vertebrate norad-renergic neuronal populations (reviewed in Goridis and

Rohrer, 2002; Howard, 2005), in which a conserved net-work of interacting factors seems to direct the neurotrans-mitter phenotype in neurons of very different origin.However, in this case, cell-type-specific variations in thisnetwork are also apparent. A completely different pictureemerges from studies of the serotonergic phenotype ininvertebrates, whereby expression of serotonin in neuronsderived from divergent lineages relies on different transcrip-tion factors, in both Drosophila (Thor and Thomas, 1997)and C. elegans (Sze et al., 2002; Zheng et al., 2005). In ver-tebrates, hindbrain rostral and caudal serotonergic neuronshave a similar origin, and share many components for theirterminal differentiation, such as Pet1 and Lmx1b (reviewedin Cordes, 2005). However, heterogeneity is also observedin this neuronal population, as reflected by the presenceof a small percentage of central serotonin-expressing cellsin Pet1 null mice, and the specific role of Gata3 only inthe caudal serotonergic neurons.

We have previously shown that, besides regulating Fmrf

transcription in the Tv cell (Benveniste et al., 1998), ap alsocontrols expression of the neuropeptide Leucokinin (LK)in the LHLK neuron (Herrero et al., 2003). In both cases,ap regulates neuropeptide levels in a small subset of theneurons that express that neuropeptide, supporting theneuron-type-specific model for the specification of neurop-eptidergic identity. Recently, Gauthier and Hewes (2006)have shown that Dimm contributes to the regulation ofLeucokinin expression through both transcriptional andpost-transcriptional mechanisms in a cell-type-specificfashion. We have taken advantage of the simplicity of theDrosophila larval leucokinergic neuronal system, which ischaracterized by only four types of LK-expressing cells(Herrero et al., 2003; this article), to analyze how differentfactors affect expression of the common neuropeptide LKin neurons of different origin. We show that all LK cellsexpress sqz. However, loss-and-gain-of-function experi-ments reveal that the different LK cell types have differentsqz requirements, and that cell-specific factors such as Apor an unidentified extrinsic retrograde signal, contributeto these differences.

2. Results

2.1. Squeeze is expressed in all Leucokinin expressing

neurons

The neuropeptide Leucokinin (LK) was detected in theCNS of Drosophila melanogaster larvae (Fig. 1A) andadults in neuronal subsets distributed in the cerebrumand ventral ganglion (Herrero et al., 2003). In third instarCNS, the LK pattern included one pair of prominent LKneurons in the brain, termed LHLK to reflect their locationin the lateral horn in adults (Fig. 1A). Two additional pairsof LK neurons were located in the suboesophageal gan-glion (SELK neurons), and one pair of LK-immunoposi-tive cells was found in each of the seven most anteriorabdominal neuromeres of the ventral ganglion (ABLK

Fig. 1. Leucokinin expression in CNS of third instar larvae. (A) Wild type larvae showing the different types of Leucokinin-immunoreactive cells in theCNS, namely LHLK (arrow), ALK (open arrow) and SELK (open arrowhead) cells in the brain, and seven pairs of cells in the ventral ganglion, namelyABLK cells (black arrowhead). The inset magnifies the weakly stained ALK cells, which are located close to the LHLK cell. (B) apUGO 35 mutant larvaeshowing the lack of expression of Leucokinin in LHLK cells, and normal anti-LK staining in SELK (open arrowhead) and ABLK (black arrowhead) cells.The inset magnifies the area containing the ALK cells (open arrow), where absence of the LHLK immunoreactive cells is apparent. (C) sqzlacZ mutantlarvae showing the lack of expression of Leucokinin in LHLK and SELK cells, but normal ABLK staining (black arrowhead). The inset shows expressionin ALK cells (open arrows) but not in LHLK cells. (D) sqzGAL4/sqzlacZ trans-heterozygote mutant. Individuals of this genotype always show theLeucokinin-immunoreactive LHLK (arrows) cells but not the SELK cells. The inset shows a magnification of the area in the brain lobe showing the ALKcells (open arrow) and two prominent leucokinergic cells with a shape and position characteristic of the LHLK cell (arrows). One of these cells is an ectopicLHLK-like Leucokinin-immunoreactive neuron. ABLK cells show normal anti-LK staining (black arrowhead). Scale bar = 100 lm in (A–D), 50 lm ininsets. (E) Graphical representation of the number of LHLK and SELK Leucokinin-immunoreactive cells in wild type and ap and sqz mutant larvae.*Different from wild type, p < 0.05. **Different from wild type, p < 0.005.

P. Herrero et al. / Mechanisms of Development 124 (2007) 427–440 429

neurons). A cluster of 2–4 weakly stained LK neurons wasoften observed in the anterior brain lobe (ALK neurons)(Fig. 1A), but started to fade by the late third instar, andwere absent from adults. Some variability was found inthe number of SELK and ABLK neurons, but none wasobserved in the LHLK cells (Table 1).

Studies in several organisms suggest that the neuropept-idergic identity of a given neuron is specified by a combina-

Table 1Summary statistics of leucokinergic cell number in third instars of different ge

Genotype N LHLK/brain lobeMean ± SD

% Lobes with doubLHLK

Wild type

(Canton S)50 1.00 ± 0.0 0

apUGO35 62 0.34 ± 0.48** 0sqzlacZ 66 0.06 ± 0.24** 0sqGAL4/sqzlacZ 50 1.06 ± 0.24 6.25sqzGAL4 68 1.09 ± 0.29 8.80elavGAL4; UAS:

GluedDN

66 0.98 ± 0.12 0

* P < 0.05, compared to wild type.** P < 0.005, compared to wild type.

torial code of transcription factors and signal transductioncascades (Jessell, 2000). In Drosophila, FMRFamideexpression in the Tv cells is controlled by the LIM-Homeo-domain protein Apterous (Ap), the Zinc Finger transcrip-tion factor Squeeze (Sqz), and the Wit signalling pathway(Allan et al., 2003, 2005), as well as by the transcriptionfactors Dimm, Eya, and Dac (Hewes et al., 2003; Miguel-Aliaga et al., 2004; Allan et al., 2005). We have previously

notypes

le SELK/hemineuromereMean ± SD

ABLK/abdominal hemiganglionMean ± SD

2.26 ± 0.26 6.88 ± 0.11

1.81 ± 0.81 6.50 ± 0.990.02 ± 0.12** 6.84 ± 0.71.60 ± 0.68* 6.55 ± 0.921.60 ± 0.74* 7.03 ± 1.152.23 ± 0.76 5.87 ± 1.44*


shown (Herrero et al., 2003) that Ap is required postmitot-ically for normal LK expression in LHLK cells (Fig. 1Band E; Table 1), the only LK-immunopositive neurons thatexpress apterous (ap) (Fig. 2A–B). We examined whetherLK expression in these cells was specified by additional fac-tors similar to those specifying FMRFamide, and, specifi-cally, whether squeeze (sqz) was involved in this process.

To determine whether the transcription factor Sqz playsa role in regulating LK expression, we performed colocal-ization studies using the sqzlacZ reporter and sqzGAL4 lines(Allan et al., 2003) to drive UAS: mCD8GFP expression.We found that all LK-positive neurons expressed sqz in lar-vae (see Fig. 2A–B and G for LHLK, SELK and ABLKcells; not shown for ALK). However, while sqzlacZ labelledall LK-positive neurons throughout development, expres-

Fig. 2. Colocalization of Ap, Leucokinin and sqz expression in the Leucokinin-in red, Leucokinin is shown in blue and sqzlacZ expression (Anti-b-Gal) is showcells. (B) sqz, but not Ap, colocalizes with Leucokinin in SELK neurons. (C–E)Ap and sqz. (F,G) When Sqz is over-expressed in elavGAL4;UAS-sqz/+ larvaabdominal hemineuromeres, in which the brightly stained cell is probably the Aectopic ABLK-like cell (open arrowhead). Neither ABLK nor ectopic ABLK-from single confocal planes showing colocalization of LK (blue) and sqz (greectopic leucokinergic brain neurons (arrowhead) located close to the oesophageeither sqz (H) or Ap (I, note that in this area there are no Ap-expressing cells

sion of the sqzGAL4 driver in LHLK neurons graduallydeclined during the third instar, and was completely lostin wandering third instars (not shown). Consistent withour previous results about ap expression (Herrero et al.,2003), triple-labelling experiments showed that the LHLKneuron is the only LK-positive neuron that coexpressesap and sqz (Fig. 2A), while all other LK-positive neuronsexpress sqz but not ap (Fig. 2B and F; not shown forALK cells).

2.2. Leucokinin expression in LHLK and SELK neurons

requires Squeeze

We next tested whether expression of LK was affected insqz mutants. We analyzed LK-immunostaining in larval

immunoreactive cells (arrowheads) of larval CNS. Anti-Apterous is shownn in green. (A) Leucokinin colocalizes with Apterous and sqz in the LHLKThe ectopic LHLK-like cell appearing in sqzGAL4/sqzlacZ larvae coexpressesl brains, pairs of leucokinergic ABLK-like cells are often found in someBLK neuron (white arrowheads) and the faintly stained cell is probably thelike cells expresses Ap (F), but both express sqz (G). (G1–G3) are imagesen) in the ABLK and ectopic ABLK neurons depicted in (G). (H,I) Theal hole (arrow) in elavGAL4; UAS-sqz/+ larval brains, do not colocalize with). Scale bar = 20 lm in (A–I).

Table 2LK expression in LHLK and SELK cells in first and second instars of wildtype and sqz mutants

Genotypes N LHLKMean ± SD

SELKMean ± SD

LI

Canton S 54 0.58 ± 0.99 1.03 ± 0.69sqzlacZ 56 0.28 ± 0.45** 0.60 + 0.61*

LII

Canton S 48 0.95 ± 0.2 1.54 ± 0.87sqzlacZ 66 0.13 ± 0.33** 0.08 ± 0.31**

* P < 0.05 compared to wild type.** P < 0.005 compared to wild type.


CNS of two sqz mutant alleles: sqzlacZ and sqzGAL4 (Perri-mon et al., 1996; Allan et al., 2003). Remarkably, sqzlacZ

third instars featured an almost total loss of LK-immuno-staining in the LHLK and SELK cells (Fig. 1C and E;Table 1), with a 94% and 99% reduction in the numberof LHLK and SELK LK-positive neurons, respectively(n = 33). In contrast, the ALK and ABLK neurons hadnormal levels of LK-immunostaining in soma (Fig. 1C).

The cell-type-specific reduction in LK expression insqzlacZ mutants gradually increased during larval develop-ment. In first instar mutants, LK-immunostaining was lostin 51.7% and 41.7% of LHLK and SELK neurons, respec-tively, while this reduction increased to 86.3% and 94.8% insecond instars (Fig. 3 and Table 2).

Surprisingly, sqzGAL4 (n = 34) and sqzGAL4/sqzlacZ

(n = 25) mutant larvae had normal levels of LK expressionin the LHLK cells. Moreover, ectopic LK-positive cellswith shape, location, and axonal projections similar tothose of the LHLK neuron, were found in 8.8% and6.2% of sqzGAL4 and sqzGAL4/sqzlacZ larval brain lobes,respectively (Fig. 1D and E; Table 1). These ectopic cellsalso expressed Ap and the sqz reporter (Fig. 2C–E). In con-trast, LK-immunoreactivity in sqzGAL4 and sqzGAL4/sqzlacZ

mutant larvae was reduced in most SELK cells, and com-pletely absent from 29.2% of SELK neurons (Fig. 1Dand E; Table 1). The ALK and ABLK neurons displayednormal LK-immunoreactivity in sqzGAL4 and sqzGAL4/

sqzlacZ mutants.The striking difference in the phenotype of LHLK neu-

rons in sqzGAL4 and sqzlacZ mutants prompted us to inves-tigate the molecular nature of these two sqz alleles by RealTime RT-PCR. Although, the sqzGAL4 allele was obtainedfrom sqzlacZ by P-element replacement (Allan et al., 2003),the amount of sqz transcript in the two alleles was dramat-ically different. While sqz mRNA levels in sqzlacZ thirdinstar mutants were reduced to almost undetectable levels

Fig. 3. Leucokinin expression in first and second instar larval CNS of wild typinstar (C) and one LHLK in the second instar (D) are visible in sqzlacZ mutantsfirst instar sqzlacZ mutant larva (C, open arrowhead), and most of them disappneuron present in the mutant ganglion). Scale bar = 50 lm in (A–D). (E) Grimmunoreactive cells in the three larval instars in wild type and sqzlacZ. *Diffe

(0.007-fold that of controls), sqzGAL4 mutant larvae haddetectable, although reduced levels of mRNA (0.243-foldthat of controls). These results are consistent with our pre-vious finding that sqzlacZ mutants have a much more severeLK phenotype than do sqzGAL4 mutant larvae.

To further demonstrate the requirement of sqz for LKexpression we attempted to rescue the sqzlacZ LK-pheno-type by expressing a UAS:sqz transgene under the postmi-totic driver elavGAL4. Normal LK expression wascompletely restored in LHLK neurons (96% rescue) withthe elavGAL4 driver, but only partially rescued (51%) inSELK neurons (Fig. 4A and Table 3). Thus, postmitoticexpression of sqz is sufficient to restore LK expression,although higher, or earlier, Sqz expression may be requiredfor total LK rescue in SELK cells.

The LHLK and SELK neurons are both located withinlarge groups of sqz-expressing cells (Fig. 2). This location,and the lack of sqz-independent markers with which tolabel these cells specifically, prevented us from establishingwhether these cells are present in sqz mutants. However,two sets of observations support the hypothesis that sqz

is not necessary for the birth of these neurons. First, LKexpression was totally or partially restored by postmitotic

e (A,B) and sqzlacZ mutants (C,D). Notice that two LHLK cells in the first(arrows). SELK cells are weakly stained with anti-Leucokinin antibody in

ear in second instar mutant CNS (D, open arrowhead points to one SELKaphical representation of the number of LHLK and SELK Leucokinin-rent from wild type, p < 0.05. **Different from wild type, p < 0.005.

Fig. 4. Rescue and cross-rescue of LK expression in sqz and ap mutants, and expression pattern of Leucokinin in sqz and ap double mutants. (A) CNS ofelavGAL4; UAS-sqz; sqzlacZ third instar. sqz mutants show fully rescued expression of LK in LHLK (arrows) and SELK (open arrowheads) neurons whenSqz is supplied under the control of the pan-neuronal driver elavGAL4. (B) CNS of elavGAL4;UAS-ap;sqzlacZ third instars. sqz mutants show rescuedexpression of LK in LHLK neurons (arrows) when Ap is supplied under the control of the pan-neuronal driver elavGAL4. Note that SELK neurons are notrescued by the pan-neuronal expression of Ap in sqzlacZ mutants. (C) CNS of elavGAL4;UASsqz apUGO35 third instars. apUGO35 mutants show rescuedexpression of LK in LHLK neurons (arrows) when Sqz is expressed under the control of the elavGAL4 driver. The open arrowhead points to SELKneurons, which are unaffected in apUGO35 mutants. (D) Expression of LK in CNS of double mutant apUGO35; sqzGAL4 third instars. This phenotype is thecombination of two single phenotypes: the loss of LHLK immunostaining, as in apUGO35 mutants, and the partial loss of SELK cells in sqzGAL4 mutants.Open arrowhead points to the SELK neurons that are still present in these mutants. (E) Expression of LK in CNS of double mutant aprk; sqzlacZ thirdinstars. The figure shows the loss of LK immunoreactivity in the LHLK and in the SELK neurons. (F) Expression of LK in CNS of aprk third instarsshowing LK-immunoreactivity in the SELK cells (open arrowheads) and absence of LHLK LK-immunoreactive cells. Scale bar = 75 lm in (A–F). (G)Graphical representation of the number of LHLK and SELK Leucokinin-immunoreactive cells in ap and sqz mutant larvae that express either Ap or Sqzunder the control of the pan-neuronal driver elavGAL4. ns, no significant difference. *Different from indicated genotype, p < 0.05. **Different from indicatedgenotype, p < 0.005.


expression of sqz (Fig. 4A and Table 3), and second, LKexpression in sqzlacZ mutants was gradually lost during lar-val development in the LHLK and SELK neurons (Fig. 3and Table 2).

These results demonstrate that sqz is necessary for nor-mal LK expression in a subpopulation of the LK-express-ing neurons, i.e., the LHLK and SELK neurons.Similarly, we have previously shown that ap controls LKlevels in the LHLK neurons, but not in other LK-express-ing cells (Herrero et al., 2003). Thus, LK expression is reg-ulated in a cell-type-specific fashion that involvesregulation by ap and sqz in the LHLK cells, regulation

by sqz but not ap in the SELK cells, and regulation by fac-tors other than ap and sqz in the ABLK and ALK neurons.

2.3. When over-expressed, Apterous and Squeeze can

substitute for each other to partially restore LK expression

LK expression in the LHLK cells is regulated by at leasttwo transcription factors, ap and sqz. These factors act in acombinatorial manner to drive FMRFamide expression inTv cells (Allan et al., 2003, 2005). We wished to establishwhether ap and sqz act in a combinatorial, hierarchical orindependent fashion to regulate LK transcription in LHLK

Table 3Quantification of LK rescue in sqz and ap mutants


SELKMean ± SD

Wild type (Canton S) 50 1.00 ± 0.00 2.26 ± 0.26elavGAL4;UASsqz/+;sqzlacZ/sqzlacZ 34 0.96 ± 0.19# 1.17 ± 0.69**,#

elavGAL4;UASap/CyO;sqzlacZ/sqzlacZ 26 0.67 ± 0.56**,# 0.00 ± 0.00**

elavGAL4;UASsqz apUGO35/apUGO35 92 0.76 ± 0.42*,+ 2.42 ± 0.67

* P < 0.05 compared to wild type.** P < 0.005 compared to wild type.+ P < 0.05, compared to apUGO35.# P < .005, compared to sqzlacZ.


cells. To do so, we measured LK expression in various ap

and sqz mutant combinations and performed cross-rescueexperiments in which we investigated whether Apterouscould rescue the sqz LK phenotype and vice versa.

LK expression was normal in the transheterozygous com-binations apUGO35/+; sqzGAL4/+, apGAL4/+; sqzlacZ/+ andaprK/+;sqzlacZ/+ (not shown). We next examined LK immu-noreactivity in double mutants with different allelic combi-nations: apGAL4/apGAL4; sqzlacZ/sqzlacZ, apUGO35/apUGO35;sqzGAL4/sqzGAL4, and aprK/aprK; sqzlacZ/sqzlacZ (Fig. 4D–Fand Table 4). The sqzlacZ allele was epistatic over every ap

allele tested, yielding an almost complete loss of LK-immu-noreactivity in the LHLK and the SELK neurons (irrespec-tive of the ap allele) (Table 4). However, we were able toobtain information from double mutants carrying thesqzGAL4 allele, since LHLK neurons are largely unaffectedin single sqzGAL4 mutants, whereas the number of SELK-immunopositive cells is reduced. The phenotype found inthese cases (apUGO35/apUGO35; sqzGAL4/sqzGAL4) was theresult of the addition of the two strongest individual pheno-types: LHLK cells behaved as in apUGO35 mutants, while theSELK neurons behaved as in sqzGAL4 mutants (Fig. 4D).

We then attempted to rescue LK expression by over-expressing Ap in sqz homozygous mutants and vice versa,using the pan-neural driver elavGAL4. Surprisingly, we foundthat, when over-expressed, each transcription factor couldsubstitute for the other, although the degree of rescueobtained differed (Fig. 4B, C, and G; Table 3). When Apwas pan-neurally over-expressed in sqzlacZ mutants, normalLK expression was restored in 67% of LHLK cells, whilethere was no rescue whatsoever in SELK neurons (Fig. 4Band G). A similar number of LK-immunoreactive LHLKcells was found in apUGO35 mutants in which Sqz was over-expressed with elavGAL4 (76% rescue, Fig. 4C and G), butthe level of expression in the rescued LHLK cells was weakerthan in wild type or elavGAL4;UAS: ap; sqzlacZ/sqzlacZ larval

Table 4LK expression in ap;sqz double mutants

Genotypes N

aprK/aprK; sqzlacZ/sqzlacZ 20apGAL4/apGAL4; sqzlacZ/sqzlacZ 10apUGO35/apUGO35; sqzGAL4/sqzGAL4 6

brains. Moreover, given that apUGO35 had a weaker LHLKphenotype than does sqzlacZ (Table 1), we can conclude thatthe degree of rescue was higher in sqz mutants over-express-ing Ap (from 0.06 cells/lobe in sqz mutants to 0.67 cells/lobein Ap-rescued sqz mutants) than in ap mutants over-express-ing Sqz (from 0.34 cells/lobe in ap mutants to 0.76 cells/lobein Sqz-rescued ap mutants).

2.4. Ectopic LK cells can be induced by misexpression of

Squeeze but not of Apterous

We previously reported that ectopic expression ofApterous in all neurons does not generate ectopic LK neu-rons (Herrero et al., 2003). Hence, Apterous by itself is notsufficient to induce LK expression in cells other thanLHLK neuron. To determine whether sqz is able to triggerectopic LK expression, we over-expressed sqz in an other-wise wild type background (Fig. 5 and Table 5). Strikingly,we detected one extra LHLK cell in 8% and 7% of the brainlobes that over-expressed Sqz with elavGAL4 or apGAL4,respectively (Fig. 5A). This extra cell expressed bothendogenous ap and sqz (not shown). These ectopic LHLKcells were never observed in larval brains over-expressingAp, even when two doses of UASap were used (Table 5).

No other ectopic LK-positive cell was detected whenapGAL4 was used to drive Sqz expression. However, usingelavGAL4 we often detected ectopic LK-immunoreactivityin a small brain neuron close to the oesophageal hole,and in ABLK-like abdominal neurons (1–4 extra neu-rons/hemiganglion) (Figs. 2F, G and 5B; Table 5). Theextra ABLK-like neurons were often found adjacent toABLK cells. Except for the extra LHLK cell, none of theectopic LK cells expressed endogenous ap (Fig. 2F andI). The ectopic oesophageal LK neurons did not expresssqz (Fig. 2H), and thus were induced by ectopic Sqz expres-sion, while the ABLK-like neurons expressed endogenous

LHLKMean ± SD

SELKMean ± SD

0.03 ± 0.18 0.08 ± 0.870.02 ± 0.24 0.07 ± 0.050.35 ± 0.67 1.57 ± 0.74

Fig. 5. Mis-/over-expression of sqz triggers ectopic expression of Leucokinin. (A) Leucokinin-immunoreactive cells in CNS of apGAL4/+; UAS-sqz/+ LIIImutants. The over-expression of Sqz in the ap expression domain gives rise to two ectopic LK-immunoreactive cells in the brain, while the number ofSELK and ABLK cells remains unchanged. The inset shows a higher-magnification image of the brain area containing the left two neighbouring LHLK-like neurons. Note that both cells have the shape and location characteristic of LHLK cells, but one of them is an ectopic leucokinergic cell. The processesin this ectopic cell also seem to project to the same region as does the LHLK cell. (B) Leucokinin-immunoreactive cells in CNS of elavGAL4/+; UAS-sqz/+

LIII. Mis-/over-expression of sqz under the control of the postmitotic driver elavGAL4 produces ectopic cells in the brain and in the ventral ganglion. (b 0)The inset shows ectopic LK cells in the brain (black arrowheads) and normal SELK cells (open arrowhead). (b 0 0) The inset shows a more detailed image ofthe ectopic LK neurons in the ventral ganglion. In the abdominal hemineuromeres there are nine LK-immunoreactive cells instead of the seven ABLK cellspresent in wild type ganglia. (C) Ectopic leucokinergic ABLK-like cells are also generated when Sqz is over-expressed in an ap-null background(elavGAL4;UAS-sqz apUGO35/apUGO35). The arrow points to a faintly stained LHLK rescued leucokinergic cell. In this larval brain, 10 ABLK-like cells arefound in each abdominal hemiganglion, as depicted in the magnified image (c 0). Scale bar = 100 lm in A–C, 50 lm in insets.

Table 5Quantification of ectopic cells by mis-/over-expression of Sqz and Ap


% Lobes with double LHLK SELKMean ± SD

Ectopic LK-cells

% Brain lobes % Ventral ganglia

elavGAL4;; UASap/+ 20 1.00 ± 0.00 0 2.56 ± 0.84 0 0elavGAL4;UASsqz/CyO 64 1.07 ± 0.20 8 1.90 ± 0.86 52 42elavGAL4;UASsqz/+;UASap/+ 70 0.97 ± 0.32 0 1.79 ± 0.90 48 35apGAL4/+;UASap/+ 28 0.98 ± 0.42 0 2.20 ± 0.67 0 0apGAL4/+; UASap/UASap 59 0.95 ± 0.2 0 2.41 ± 0.67 0 0apGAL4/UASsqz 40 1.06 ± 0.15 7 1.89 ± 0.63 0 0apGAL4/UASsqz; UASap/+ 30 1.00 ± 0.00 0 2.07 ± 0.44 0 0UASsqz/+; sqzGAL4/+ 48 1.00 ± 0.26 2 2.09 ± 0.47 0 0sqzGAL4/UASap 20 1.00 ± 0.00 0 2.48 ± 0.91 0 0UASsqz/+; sqzGAL4/UASap 38 1.00 ± 0.00 0 1.85 ± 0.92 0 0scaGAL4/UASap 26 1.00 ± 0.00 0 1.93 ± 0.82 0 0UASsqz/+; scaGAL4/+ 32 1.00 ± 0.00 0 2.30 ± 0.79 0 0


sqz (Fig. 2G), suggesting that these and the ectopic LHLKcells were probably the result of over-expression, ratherthan ectopic expression of sqz. These data, and the findingthat extra ABLK-like leucokinergic cells were alsoobserved when Sqz was over-expressed in ap-null mutants(Fig. 5C), indicate that ectopic expression or over-expres-sion of Sqz can induce ectopic LK expression throughap-independent mechanisms.

It has been shown that Ap and Sqz can act synergisti-cally to induce ectopic FMRFamide expression (Allanet al., 2003). However, when Ap and Sqz were coexpressedusing either elavGAL4 or apGAL4, the number of LK ectopiccells was similar to that obtained in larvae over-expressingSqz alone (Table 5). A remarkable exception was the extraLHLK-like cells, which were never found in larval brainscoexpressing both transcription factors. These findings,and the observation of the LHLK phenotype in sqzGAL4

mutants, suggest that the relative dose of Sqz and Ap isimportant in determining LHLK properties (see Section 3).

In order to establish whether either Apterous or Squeezeis needed in previous cellular stages to induce LK expres-sion, we misexpressed these factors with the scabrousGAL4

driver, which initiates expression in stage 8 embryos in allneuroblasts (Mlodzik et al., 1990). No ectopic expressionwas observed with early expression of either ap or sqz

(scaGAL4; UAS:ap (n = 13) and scaGAL4; UAS:sqz (n = 16),Table 5), thereby corroborating the postmitotic effect ofboth factors in LK regulation.

2.5. ABLK neurons are the only LK neurons affected by

inhibition of retrograde axonal transport

We have shown that, similarly to FMRFamide expres-sion in the Tv cells, LK expression in the LHLK requiresboth Ap and Sqz. Obviously, since most of the neuronsthat express both transcription factors do not express LKor FMRFamide, other components must be involved inthe specification of their neuropeptidergic identity. In the

Table 6LK expression in TGFb (wit) and Activin (babo) receptor mutants, andeyes absent and dachshund mutants


SELKMean ± SD

witB11/witA12 16 1.00 ± 0.0 2.23 ± 0.79babo32/babo32 50 1.00 ± 0.0 2.05 ± 0.59dac3/dac4 52 1.04 ± 0.19 2.36 ± 0.40eya10/eyaCli 48 1.00 ± 0.0 1.98 ± 0.82


Tv cells, FMRFamide expression is also regulated byextrinsic signals mediated by the BMP signalling pathway(Marques et al., 2003; Allan et al., 2003), and by additionaltranscription factors such as dac and eya (Miguel-Aliagaet al., 2004) or dimm (Hewes et al., 2003; Allan et al.,2005). These factors can act either individually to conferspecific characteristics of the cell, or in combination todetermine the neuropeptidergic identity of the Tv cell(Allan et al., 2005).

Therefore, to determine whether there are specific codesfor each LK cell type, we examined whether other genesrequired for FMRFamide expression in the Tv cells werealso involved in specifying LK identity. Dimm has alreadybeen shown to regulate LK expression in Drosophila larvaethrough cell-type-specific mechanisms (Gauthier andHewes, 2006), by upregulating Lk transcription in ABLKcells and by post-transcriptionally sustaining high levelsof LK expression in LHLK cells.

In Tv cells, dac is required for maintaining high-levelFMRF expression, while eya is required in axon pathfindingand BMP signalling (Miguel-Aliaga et al., 2004). In contrast,we did not find any significant reduction of LK immunoreac-tivity in dac or eya mutant third instars (Table 6). However, asmall percentage (3.8%, n = 52) of dac mutant brain lobesshowed one ectopic LHLK-like cell (Fig. 6E).

FMRFamide expression in the Tv neuron is triggered by aretrograde BMP signal, which is initiated by the binding of theextracellular ligand Glass bottom boat (Gbb) to the Wishfulthinking (Wit) receptor at the nerve terminal. Activation ofthe Wit receptor leads to the phosphorylation and subsequentnuclear translocation of Mad. To establish whether LKexpression is regulated by extrinsic signals, we studied thepresence of phosphorylated Mad (pMad) in the LK neurons.We found that all LK neurons except the LHLK cells wereindeed pMad-positive (Fig. 6A–B). However, although pMadexpression in wit mutants (witB11/witA12) disappeared in allventral ganglion neurons (Marques et al., 2003) and in theSELK and ALK neurons, LK was normally expressed inthese neurons (Fig. 6C–D and Table 6).

These results demonstrate that the Wit signalling path-way is not involved in the regulation of LK expression inlarval Drosophila CNS. However, two BMP signallingpathways have been described in Drosophila: one involvesWit (type-II BMP receptor) and Gbb; the other is mediatedby the Baboon (Babo) type I receptor, and its ligand Acti-vin. Babo is expressed in the CNS of Drosophila and hasbeen described as playing a role in axonal remodelling in

mushroom bodies (Zheng et al., 2003). To determinewhether this pathway had any role in regulating LKexpression, we analyzed LK immunoreactivity in babo

mutants, but found no difference from wild type (Fig. 6Fand Table 6). Thus, BMP signalling does not seem to playa major role in regulating LK neuronal expression.

Finally, we examined whether other, unidentified,extrinsic retrograde signals may still be necessary for LKexpression, by blocking anterograde and retrograde axonaltransport with a dominant-negative version of the P150/Glued Dynactin/Dynein motor complex component(UAS:GluedDN). Flies expressing GluedDN under theelavGAL4 driver did not eclose, but larvae had the normalpattern of LK distribution in the brain and suboesophagealganglion, and the morphology of the LK cells in theseregions was unaffected (Fig. 6G and Table 1). However,in the ventral ganglion there was a significant reductionin the number of ABLK cells, from 6.88 ± 0.11 cells/abdominal hemiganglion in wild type larvae to5.87 ± 1.44 cells/abdominal hemiganglion in larvae pan-neurally expressing GluedDN (14.6% reduction, p < 0.01,Student’s t-test) (Fig. 6G and Table 1). The number andlocation of ABLK cells expressing LK varied from just 2cells to the 7 cells found in wild type abdominal hemigan-glia, and leucokinergic varicosities were often visible in theabdominal neuropile of late third instar larvae expressingGluedDN (see inset in Fig. 6G). This later phenotype wasnever observed in wild type CNS, which suggests that inhi-bition of axonal transport may alter the synaptic connectiv-ity of ABLK cells. Thus, these data suggest another class ofLK regulation in the ABLK, which may depend on an axo-nal transport signal and is consistent with these cells beingLK neurons that project outside the CNS (see Section 3).

3. Discussion

Understanding how neuronal diversity is achievedremains one of the central topics in neurobiology. Postmi-totic specification of individual neuronal traits requires spe-cific combinations of both intrinsic and extrinsic signallingcascades. However, do neurons from different lineages usesimilar or different pathways to specify the same pheno-typic trait, such as a neurotransmitter? In this report, wehave exploited the Drosophila larval leucokinergic neuronalsystem as a model to address this question from a geneticperspective. Based on neuronal morphology and location,the leucokinergic system consists of four types of neurons,including the ALK and LHLK neurons in the brain, theSELK neurons in the suboesophageal ganglion, andthe ABLK neurons in the abdominal ganglion. Althoughthe cell lineage of these neurons has not been specificallystudied, their disparate locations indicate that they arisefrom very different precursors. Our results demonstratethat Sqz is a common factor that controls leucokinergicdifferentiation, but the requirements for Sqz in differentneuronal types differ, and seem to rely on specific factorspresent in each cell type.

Fig. 6. LK expression does not depend on the BMP or Activin pathways, but is affected by dac and inhibition of retrograde transport. (A–B)Colocalization of pMad (in red) and LK (in green) in the CNS of third instars. (A) While anti-pMad staining is evident in ALK neurons (open arrows), itis absent from LHLK neurons (arrows). (B) pMad is found in the SELK neurons (open arrowheads). (C–D) pMad immunoreactivity is completely lost inwitA12 mutants, while LK expression in LHLK and ALK neurons (C) and in SELK neurons (D) remains unaltered. (E) dac3/dac4 mutant CNS showingtwo leucokinergic LHLK cells in the right brain lobe (arrows). The ABLK cells (arrowhead) and SELK cells (out of focus) are normal. (F) babo32 mutantsshow wild type LK immunoreactivity in the CNS of third instars in LHLK (arrows), SELK (open arrowhead) and ABLK (black arrowhead) cells. (G) LKexpression in the CNS of elavGAL4;UAS-P150-Glued third instars. Notice that LK expression in LHLK (arrows) and SELK (open arrowhead) neurons isnot affected by the pan-neuronal disruption of retrograde transport. However, disruption of retrograde transport causes a variable, but significantreduction in the number of LK-immunoreactive ABLK cells. In this ventral ganglion, the left hemineuromeres show the normal complement of ABLKcells (seven cells, black arrowhead), while the right hemineuromeres contain only three ABLK cells. The inset shows a higher magnification of the boxedarea where only three ABLK cells are present, and leucokinergic varicosities are observed in the neuropile (thin arrows). Scale bar = 20 lm in (A–D),100 lm in (E,G), 80 lm in F, 50 lm in inset.


3.1. Three neuronal leucokinergic subpopulations can be

described based on the factors that control LK expression

We have previously shown that Ap was required for LKexpression only in LHLK cells (Herrero et al., 2003). Ap isalso necessary for proper transcription of the Fmrf gene inthe thoracic Tv neurons (Benveniste et al., 1998; Allanet al., 2003, 2005). In our attempt to understand the mech-anisms underlying leucokinergic differentiation, we askedwhether other factors known to control expression of theFMRFamide neuropeptide, i.e., Sqz and the BMP signal-ling pathway (Allan et al., 2003; Marques et al., 2003),affected LK expression. Indeed, we have seen that the num-ber of LK-immunopositive cells is strongly reduced insqzlacZ mutant larvae. We have previously proposed thatApterous is not necessary for the emergence and mainte-nance of LHLK cells (Herrero et al., 2003). Earlier reportshave established that the Tv cells are present in sqz

mutants, although they do not express Fmrf (Allan et al.,2003). Are the LK cells present in sqz mutants? We couldnot directly address this question due to the lack of inde-

pendent markers for following the fate of the LK cells,but the results presented here indicate that in sqz mutants,leucokinergic cells are born, but fail to express LK. First,LK expression is restored by postmitotic expression ofSqz. Second, LK-immunoreactive cells are detected in sqz

mutant brains during early larval stages, and later disap-pear. Finally, the few LK-expressing cells detected in sqz

mutants often show very faint immunostaining, which sug-gests that reduction of Sqz protein decreases Lk transcrip-tion but does not affect cell survival. The reduction in thenumber of leucokinergic cells in sqz mutants even in firstinstar larvae indicates that sqz is necessary for inductionof LK expression, and the weaker phenotype observed inearly larval stages may reflect perdurance of the wild typeproduct supplied by the mother (Perrimon et al., 1996;Tomancak et al., 2002), or a requirement for sqz also formaintenance of LK expression.

Although all leucokinergic neurons express sqz, theydiffer in how this transcription factor affects LKexpression. In this study, we have identified at least threeneuronal types based on the components that regulate


LK expression. First, in the LHLK neurons, LK expres-sion is controlled by the transcription factors Ap andSqz. Second, in the SELK neurons, Sqz, but not Ap, is nec-essary for wild type Lk transcription. Finally, in the ABLKneurons, Sqz is dispensable for LK expression even thoughit can induce leucokinergic ABLK-like cells, and LKexpression is affected by inhibition of retrograde axonaltransport only in the ABLK cells. Regarding the ALK neu-rons, the number of this cell type is highly variable, andthus was not further analyzed in this study.

By analyzing two different sqz alleles, we have shownthat, besides differences in the components that controlLK expression, the amount of Sqz protein required toachieve wild type LK expression also varies. Our Real-Time PCR analysis indicates that the sqzlacZ allele hasundetectable levels of sqz transcripts, while the sqzGAL4

allele has reduced, but measurable sqz transcription. Con-sistent with these results, sqzlacZ mutant larvae show alarge reduction in the number of LHLK and SELK cellsdetected by anti-LK immunostaining (0.06 and 0.02 cells/hemineuromere, respectively, compared with 1 and 2.26cells/hemineuromere in wild type), while sqzGAL4 mutantsdo not affect LK expression in the LHLK neuron, but doshow a significant decrease in the number of LK-immuno-positive SELK cells (1.60 cells/hemineuromere). More-over, restoring neuronal Sqz protein with the elavGAL4

driver can completely rescue LK expression in LHLKneurons in sqzlacZ mutants, but only partially rescue LKin SELK neurons (51% rescue). The differential effect ofthese two sqz alleles suggests that there is a threshold ofSqz protein below which Lk transcription is prevented,and that this threshold is higher in SELK cells than inLHLK cells.

3.2. Sqz and Ap cooperate to control Lk expression in LHLK

neurons

We have demonstrated that the reduced LK expressionobserved in the LHLK cells of sqz mutants is fully rescuedwhen over-expressing the lost protein, indicating that thisprotein is indeed responsible for the observed phenotype.In addition, our data show that LK expression in these cellsdoes not depend on the Gbb/Wit or Babo/Activin signal-ling cascades, or any extrinsic retrograde signal. Thus, neu-ropeptidergic identity of the LHLK neurons is controlledby Ap (Herrero et al., 2003) and Sqz, but not BMP. Thesecells also express the bHLH transcription factor Dimm,which has been shown to act post-tanscriptionally on theregulation of LK expression in these cells (Hewes et al.,2003; Gauthier and Hewes, 2006). This combination offactors (i.e., Ap, Sqz, Dimm, but not BMP signalling) isdifferent from any of the known codes that regulate theneuropeptidergic differentiation of the peptidergic neuronsin the thoracic ap cluster, which includes the Furin1-expressing Ap-let cells that do not express Sqz, and theFmrf-expressing Tv cell, which requires extrinsic signallingthrough Gbb/Wit (Allan et al., 2005).

Our results demonstrate that high levels of either Ap orSqz protein alone are sufficient to induce LK expression inthe LHLK cells. Thus, based on the cross-rescue experi-ments, we can infer that Sqz can form transcriptionallyactive complexes with proteins other than Ap to promoteLk expression. Similarly, Ap is able to promote Lk expres-sion when it complexes with proteins other than Sqz. How-ever, wild type levels of LK expression are only achievedwhen both proteins are present. Allan et al. (2005) showedthat Sqz and Ap can physically interact, and that both pro-teins can form separate complexes with Chip. We have evi-dence that Chip is involved in Lk regulation, because LKexpression in LHLK cells is affected when dLMO, a cofac-tor that binds Chip and prevents Ap action (Milan et al.,1998), is expressed in ap-expressing cells (our unpublishedresults). Thus, Ap and Sqz could weakly activate Lk tran-scription independently by interacting with Chip and/orother transcription factors, but strong activation wouldrequire a synergistic interaction mediated by complexescontaining both proteins. A similar mechanism is exertedby Brn2 and Otp to stimulate transcription of the cortico-trophin-releasing hormone gene in the neuroendocrinehypothalamic neurons (Burbach et al., 2001), and byCdx-2 and Brn-4 to activate expression of the proglucagongene in pancreatic B-cell lines (Wang et al., 2006).

3.3. Distinct Lk regulation in ABLK cells

Gauthier and Hewes (2006) recently reported that Lk

transcription in dimm mutants is downregulated in ABLKcells, while LHLK and SELK cells show slightly upregu-lated Lk transcription but reduced levels of LK peptide.We have found that the ABLK neurons also differ fromLHLK and SELK neurons in that Sqz appears to be dis-pensable for LK expression in these cells. However, neuro-nal over-expression of Sqz can frequently induce ectopicLK expression in ABLK-like cells in an ap-independentfashion. Likewise, misexpression of the mammalianhomeodomain transcription factor Phox2a in the sympa-thetic ganglion induces ectopic noradrenergic neurons,even though in the absence of Phox2a, sympathetic devel-opment is largely normal (reviewed in Goridis and Rohrer,2002).

Another peculiarity of the ABLK cells is that LKexpression in these cells seems to depend on a retrogradesignal, different from BMP, because the number of ABLKLK-expressing cells is reduced when axonal transport isinhibited. Although the degree of reduction is variable, itis indeed highly significant, because as few as two cellscan be found when GluedDN is pan-neurally expressed.This variability could be due to partial inhibition of retro-grade transport, as suggested by the late pupal lethality ofGluedDN-expressing flies, instead of the earlier embryoniclethality of Glued amorphic mutations. In the thoracic ap-cluster, projection of the Tv cell outside the CNS is essen-tial for its response to Gbb and subsequent Fmrf expression(Allan et al., 2003; Marques et al., 2003). Likewise, ABLK


cells are known to project outside the CNS (Cantera andNassel, 1992), and this may be essential for them to reactto an extrinsic signal important for induction and/or main-tenance of LK expression. Alternatively, extrinsic signal-ling may control guidance and/or targeting of ABLKaxons, as suggested by the ectopic leucokinergic varicositiesfound in the abdominal neuropile of GluedDN-expressingganglia, and proper axonal targeting may be an importantfactor in the regulation of LK expression. Further experi-ments will be necessary to test these hypotheses, and ruleout a possible deleterious effect of GluedDN on ABLKviability.

3.4. Sqz can induce ectopic LK expression

Ectopic leucokinergic cells can only be induced by over-expressing Sqz, but not by Ap. Moreover, ectopic LK neu-rons can also be classified into three different groups: first,the ectopic LHLK-like cells, which appear to depend onthe relative amount of Sqz and Ap (see below); second,the ABLK-like cells, which require high levels of Sqzexpression, but not ap expression; and finally, the ectopicbrain cells, which are the only leucokinergic ectopic cellsgenerated by ectopic expression of Sqz. These latter cellsmust have unidentified components present in other leu-cokinergic cells that enable them to activate Lk transcrip-tion when Sqz is present.

We have found that an ectopic LHLK-like cell is presentin a small percentage of the sqzGAL4 mutants. Appearanceof this ectopic cell requires Ap, because ap is expressed inthis cell and its appearance is prevented in the apUGO35 nullallele. In the thoracic ganglion, sqz determines the numberand identities of cells of the ap cluster, so that in szq nullmutants an extra Dimm and Furin-expressing cell was gen-erated in every ap cluster (Allan et al., 2005). Our attemptto analyze changes in the number of ap and dimm express-ing cells in sqz mutants was precluded by the large numberof ap cells surrounding the LHLK cell, and the inconsistentexpression of the dimm-GAL4 driver c929 in the LHLKcells, in both wild type and mutant brains (data notshown). However, the results obtained when Sqz and/orAp were over-expressed in a wild type background suggesta more complex scenario. First, we obtained a phenotypeequivalent to that observed in sqzGAL4 mutants, i.e., theappearance of ectopic LHLK-like cells, when Sqz wasover-expressed with sqzGAL4 and with the postmitotic ap-and elav- GAL4 drivers. Moreover, these ectopic LK cellsdisappeared if Ap was coexpressed with Sqz, even thoughover-expression of Ap alone had no effect whatsoever onLK expression. Based on these data, and on the differentlevels of Ap protein in cells surrounding the LHLK neuron(see Fig. 2A), we hypothesize that the relative dose of Apand Sqz, rather than the absolute amount of Sqz, is essen-tial for specifying the correct number of LHLK cells withdetectable LK, and that this process is controlled postmi-totically. According to this model, when Sqz is reduced,those cells with low Ap levels would reach an optimum

stoichiometric Sqz/Ap ratio leading to ectopic LK expres-sion, while when Sqz is increased, the correct ratio will bepresent in cells with high Ap levels. In this last case, furtherincreasing Ap would drive the Sqz/Ap ratio away from theoptimum for LK expression, and ectopic cells would not beproduced.

Two other situations lead to ectopic leucokinergicLHLK-like cells. First, we have detected a small percentageof ectopic LHLK cells in dac mutants. Because Ap hasbeen shown to repress dac expression in the thoracic Tvbcell (Miguel-Aliaga et al., 2004), it is possible that repres-sion of dac is required to restrict LK expression to theLHLK neuron. Second, over-expression of the transcrip-tion factor Dimm produces ectopic LHLK-like cells(Hewes et al., 2003; 2006), albeit at much higher frequency.Ap also regulates dimm transcription in most cells (Allanet al., 2005). Thus, Dimm over-expression might overcomethe requirement for a Sqz/Ap optimum ratio to induce LKexpression, provided additional necessary factors are alsopresent in the cell. More experiments will be needed tounderstand the role of Dac and Dimm, and their interac-tion with Ap and Sqz, in regulating LK expression in theLHLK cell.

Deciphering how cells of different origin acquire thesame neurosecretory identity is one of the major challengesin neuronal development. Much of the progress in this fieldhas been achieved by studying the vertebrate catecholamin-ergic system, in which central and peripheral noradrenergicneurons use a similar combination of factors to specifytheir neurotransmitter phenotype, but they differ in thehierarchical relations between these factors, and recruitcomponents specific to each neuronal type. It is temptingto speculate that analogous mechanisms may control Lk

expression in leucokinergic cells in Drosophila. Consistentwith this hypothesis, Sqz, and probably Dimm, appear tobe regulatory factors for neuropeptidergic identity com-mon to all leucokinergic cells. However, sqz affects LKexpression in each cell subclass in very different ways,and cell-type-specific regulatory modes on Lk have alsobeen described for Dimm (Gauthier and Hewes, 2006).Moreover, additional cell-type-specific factors regulateLK expression in different leucokinergic cells, such as Apin the LHLK cells, and an unidentified retrograde signalin the ABLK cells. Thus, the leucokinergic neurons com-prise a simple, genetically amenable system for understand-ing how complex regulatory networks confer a similarneurosecretory identity on cells of different origins.

4. Experimental procedures

4.1. Fly strains and cultures

Wild type Canton-S flies were used as control strain in all experiments.We used apUGO35 as an ap null allele (Cohen et al., 1992). apMD544 (referredto as apGAL4, a pGAL4 insertion in the ap locus) (Calleja et al., 1996) andaprK568 (referred to as aprK) (Cohen et al., 1992) are ap hypomorph alleles.l(3)0210202102 (referred to as sqzlacZ) and sqzGAL4 are hypomorph alleles(Allan et al., 2003). For rescue and over-expression experiments elavGAL4


strain (c155 line), apGAL4, sqzlacZ, UAS::mcD8GFP, UAS:sqz (Allan et al.,2003) and UAS:ap were used as transgenic lines. The mutant alleles dac3,dac4, eya10, eyaCli (Miguel-Aliaga et al., 2004), witA12, witB11, and babo32

(Brummel et al., 1999) were used to study the influence of other intrinsicand extrinsic signals on LK expression. Transgenic line UAS:P150-Glued

(referred to as UAS:GluedDN ) was used to inhibit axonal transport. Allexperiments, including those that involved the use of the GAL4/UAS sys-tem, were carried out at 25 �C.

4.2. Immunohistochemistry and confocal images

Immunolabelling was carried out as previously described (Herreroet al., 2003). Primary antibodies used were rat anti-Apterous (Lungrenet al., 1995; 1:200), mouse anti-b-gal (PROMEGA; 1:150) and rabbitanti-pMad (Tanimoto et al., 2000; 1:500). For detection of Leucokinin,we used a new aliquot of the rabbit anti-LK IV antibody (Chen et al.,1994; 1:1000), kindly provided by Dr. J. Veenstra. Under the same exper-imental conditions, this aliquot showed greater sensitivity than the onepreviously reported (Herrero et al., 2003), as demonstrated by the detec-tion of the faint ALK leucokinergic cells, and the fact that the numberof LK-immunoreactive cells detected in apUGO35 mutants was slightly lar-ger than that found by Herrero et al. (2003).

For confocal imaging we used a Zeiss Meta microscope. Z series weremanaged using the accompanying software. For optical microscopy weused LEICA Wild MPS52 adapted for use with a Leica digital camera.Figures were generated using Adobe Photoshop v. 6.

4.3. Quantitative real-time PCR (QRT-PCR)

Total RNA was extracted from freshly collected whole third instarsusing the RNeasy Mini Kit from Qiagen. RNAs were quantified by spectro-photometry and retro-transcribed using the first-strand cDNA synthesis kitfrom Amersham Biosciences (Piscataway, NJ, USA) following the manufac-turer’s instructions. Amplifications were performed using the DrosophilaProbeLibrary technology (Exiqon) and analyzed using an ABI PRISM7700 sequence detection system (Applied Biosystems). All experiments wererepeated three times. Thermocycler conditions were: step 1, 50 �C for 2 min;step 2, 95 �C for 10 min; step 3, 95 �C for 15 s; step 4, 60 �C for 1 min; cycleback to step 3, 40 times. The primers for sqz were as follows: forward 5 0-CGAGCAGCAGGAACAGAAGT-3 0; reverse, 5 0-CGCTAGCGCTGCTCATCT-3 0 and ProbeLibrary probe #65. Expression of the ribosomal gene18S was used as a control for RNA loading and to test integrity. Primer andprobes were designed using ProbeFinder software.

Acknowledgements

We are most grateful to J. Veenstra for the LeucokininIV antibody and to H. Tanimoto for the pMad antibody.We thank W. Allan, G. Marques and Bloomington StockCenter for sharing fly lines, and J. de la Torre for statisticalassistance. Confocal imaging was performed at the Centrode Biologıa Molecular, assisted by M.A. Munoz, T. Villal-ba, and C. Sanchez. We thank R. Ramos (PCM, MadridScientific Park) for the analysis of sqz expression by RealTime RT-PCR. We appreciate Dr. P Mason’s assistancewith the English language as well as P Martınez’ commentson the manuscript.

This work was supported by grants from SpanishMinistry of Science and Technology (Grant No. BFU-2004-03894) and from the Comunidad Autonoma deMadrid-Universidad Autonoma de Madrid (Grant No.11BCB/009).

References

Allan, D.W., Pierre, S.E., Miguel-Aliaga, I., Thor, S., 2003. Specification ofneuropeptide cell identity by the integration of retrograde BMPsignaling and a combinatorial transcription factor code. Cell 113, 73–86.

Allan, D.W., Park, D., St Pierre, S.E., Taghert, P.H., Thor, S., 2005.Regulators acting in combinatorial codes also act independently insingle differentiating neurons. Neuron 45, 689–700.

Altun-Gultekin, Z., Andachi, Y., Tsalik, E.L., Pilgrim, D., Kohara, Y.,Hobert, O., 2001. A regulatory cascade of three homeobox genes, ceh-

10, ttx-3 and ceh-23, controls cell fate specification of a definedinterneuron class in C. elegans. Development 128, 1951–1969.

Benveniste, R.J., Taghert, P.H., 1999. Cell type-specific regulatorysequences control expression of the Drosophila FMRF-NH2 neuro-peptide gene. J. Neurobiol. 38, 507–520.

Benveniste, R.J., Thor, S., Thomas, J.B., Taghert, P.H., 1998. Cell typespecific-regulation of the Drosophila FMRF-NH2 neuropeptide geneby Apterous, a LIM homeodomain transcription factor. Development125, 4757–4765.

Broihier, H.T., Skeath, J.B., 2002. Drosophila homeodomain proteindHb9 directs neuronal fate via crossrepressive and cell-nonautono-mous mechanisms. Neuron 35, 39–50.

Brummel, T., Abdollah, S., Haerry, T.E., Shimell, M.J., Merriam, J.,Raftery, L., Wrana, J.L., O’Connor, M.B., 1999. The Drosophila

activin receptor Baboon signals through d Smad2 and controls cellproliferation but not patterning during larval development. GenesDev. 13, 98–111.

Burbach, J.H.P., Luckman, S.M., Murphy, D., Gainer, H., 2001. Generegulation in the magnocellular hypothalamo-neurohypophyseal sys-tem. Physiol. Rev. 81, 1197–1267.

Calleja, M., Moreno, E., Pelaz, S., Morata, G., 1996. Visualization of geneexpression in living adult Drosophila. Science 274, 252–255.

Cantera, R., Nassel, D.R., 1992. Segmental peptidergic innervation ofabdominal targets in larval and adult dipteran insects revealed with anantiserum against leucokinin I. Cell Tissue Res. 269, 459–471.

Chen, Y., Veenstra, J.A., Davis, N.T., Hagedorn, H.H., 1994. Acomparative study of leucokinin-immunoreactive neurons in insects.Cell Tissue Res. 276, 69–83.

Cohen, B., McGuffin, M.E., Pfeifle, C., Segal, D., Cohen, S.M., 1992.apterous, a gene required for imaginal disc development in Drosophila,encodes a member of the LIM family of developmental regulatoryproteins. Genes Dev. 6, 715–729.

Cordes, S.P., 2005. Molecular genetics of the early development ofhindbrain serotonergic neurons. Clin. Genet. 68, 487–494.

Edlund, T., Jessell, T.M., 1999. Progression from extrinsic to intrinsicsignaling in cell fate specification: a view from the nervous system. Cell96, 211–224.

Ernsberger, U., Rohrer, H., 1999. Development of the cholinergicneurotransmitter phenotype in postganglionic sympathetic neurons.Cell Tissue Res. 297, 339–361.

Gauthier, S.A., Hewes, R.S., 2006. Transcriptional regulation of neuro-peptide and peptide hormone expression by Drosophila dimmed andcrytocephal genes. J. Exp. Biol. 209, 1803–1815.

Goridis, C., Rohrer, H., 2002. Specification of catecholaminergic andserotonergic neurons. Nat. Rev. Neurosci. 3, 531–541.

Herrero, P., Magarinos, M., Torroja, L., Canal, I., 2003. Neurosecretoryidentity conferred by the apterous gene: lateral Horn leucokininneurons in Drosophila. J. Comp. Neurol. 457, 123–132.

Hewes, R.S., Park, D., Gauthier, S.A., Schafer, A.M., Taghert, P.H.,2003. The bHLH protein Dimmed controls neuroendocrine celldifferentiation in Drosophila. Development 130, 1771–1781.

Hewes, R.S., Gu, T., Brewster, J.A., Qu, C., Zhao, T., 2006. Regulation ofsecretory protein expression in mature cells by DIMM, a basic helix-loop-helix neuroendocrine differentiation factor. J. Neurosci. 26,7860–7869.

Howard, M.J., 2005. Mechanisms and perspectives on differentiation ofautonomic neurons. Dev. Biol. 277, 271–286.


Jessell, T.M., 2000. Neuronal specification in the spinal cord:inductive signals and transcriptional codes. Nat. Rev. Genet. 1,20–29.

Koo, S.J., Pfaff, S.L., 2002. Fine-tuning motor neuron properties:signaling from the periphery. Neuron 35, 823–826.

Lee, S.K., Pfaff, S.L., 2003. Synchronization of neurogenesis and motorneuron specification by direct coupling of bHLH and homeodomaintranscription factors. Neuron 38, 731–745.

Lungren, S.E., Callahan, C.A., Thor, S., Thomas, J.B., 1995. Control ofneuronal pathway selection by the Drosophila LIM homeodomaingene apterous. Development 121, 1769–1773.

Marques, G., Haerry, T.E., Crotty, M.L., Xue, M., Zhang, B., O’Connor,M.B., 2003. Retrograde Gbb signaling through the Bmp type 2receptor Wishful Thinking regulates systemic FMRFa expression inDrosophila. Development 130, 5457–5470.

Milan, M., Diaz-Benjumea, F.J., Cohen, S.M., 1998. Beadex encodes anLMO protein that regulates Apterous LIM-homeodomain activity inDrosophila wing development: a model for LMO oncogene function.Genes Dev. 12, 2912–2920.

Miguel-Aliaga, I., Allan, D.W., Thor, S., 2004. Independent roles ofthe dachshund and eyes absent genes in BMP signaling, axonpathfinding and neuronal specification. Development 131,5837–5848.

Mlodzik, M., Baker, N.E., Rubin, G.M., 1990. Isolation and expression ofscabrous, a gene regulating neurogenesis in Drosophila. Genes Dev. 4,1848–1861.

Perrimon, N., Lanjuin, A., Arnold, C., Noll, E., 1996. Zygotic lethalmutations with maternal effect phenotypes in Drosophila melanogaster.

II. Loci on the second and third chromosomes identified by P-element-induced mutations. Genetics 144, 1681–1692.

Schneider, L.E., Roberts, M.S., Taghert, P.H., 1993. Cell type-specifictranscriptional regulation of the Drosophila FMRFamide neuropep-tide gene. Neuron 10, 279–291.

Sze, J.Y., Zhang, S., Li, J., Ruvkun, G., 2002. The C. elegans POU-domain transcription factor UNC-86 regulates the tph-1 tryptophanhydroxylase gene and neurite outgrowth in specific serotonergicneurons. Development 129, 3901–3911.

Tanimoto, H., Itoh, S., ten Dijke, P., Tabata, T., 2000. Hedgehog creates agradient of DPP activity in Drosophila wing imaginal. Mol. Cell 5, 59–71.

Tomancak, P., Beaton, A., Weiszmann, R., Kwan, E., Shu, S., Lewis, S.E.,Richards, S., Ashburner, M., Hartenstein, V., Celniker, S.E., Rubin,G.M., 2002. Systematic determination of patterns of gene expressionduring Drosophila embryogenesis. Genome Biol. 3, research0088.

Thor, S., Thomas, J., 1997. The Drosophila islet gene governs axonpathfinding and neurotransmitter identity. Neuron 18, 397–409.

Wang, P., Liu, T., Li, Z., Ma, X., Jin, T., 2006. Redundant and synergisticeffect of Cdx-2 and Brn-4 on regulating proglucagon gene expression.Endocrinology 147, 1950–1958.

Wightman, B., Ebert, B., Carmean, N., Weber, K., Clever, S., 2005. TheC. elegans nuclear receptor gene fax-1 and homeobox gene unc-42

coordinate interneuron identity by regulating the expression ofglutamate receptor subunits and other neuron-specific genes. Dev.Biol. 287, 74–85.

Zheng, X., Chung, S., Tanabe, T., Sze, J.Y., 2005. Cell-type specificregulation of serotonergic identity by the C. elegans LIM-homeodo-main factor LIM-4. Dev. Biol. 286, 618–628.

Zheng, X., Wang, J., Haerry, T.E., Wu, A.Y., Martin, J., O’Connor,M.B., Lee, C.J., Lee, T., 2003. TGF-b signaling activates steroidhormone receptor expression during neuronal remodeling in theDrosophila brain. Cell 112, 303–315.

Documents

Squeeze involvement in the specification of Drosophila leucokinergic neurons: Different regulatory mechanisms endow the same neuropeptide selection