Either Canton S (CS) or w¹¹¹⁸, back-crossed for seven generations to CS, were used as wild-type control flies. Brain DA-deficient Drosophila (dTH def.) were DTH^FS+/− BAC ple² and rescue controls (dTH resc.) were DTH BAC ple² (as per Cichewicz et al., 2016). Animals used for immunohistochemistry were 3–5 days post eclosion. For reconstitution of splitGFP experiments TrH-Gal4 flies (Cassar et al., 2015) were crossed with flies carrying a combination of UAS:splitGFP1-10 (Pech et al., 2013), DsRed (Riemensperger et al., 2005) and splitGFP11 under the control of the mb247 promotor (Pech et al., 2013). Flies were raised under standard conditions at a 12:12 h light-dark schedule at 25°C and 60% relative humidity.

If not otherwise indicated, brains of 3- to 5-day old female flies were dissected in ice-cold Ringer’s solution containing 5 mM HEPES-NaOH (pH = 7.4), 130 mM NaCl, 5 mM KCl, 2 mM MgCl₂, 2 mM CaCl₂, and 36 mM sucrose, and fixed for 2 h on ice in 4% paraformaldehyde dissolved in phosphate-buffered saline (PBS), and subsequently washed three times in PBS containing 0.6% Triton X-100. Samples were incubated overnight at 4°C in PBT containing 2% bovine serum albumin (BSA). If not otherwise indicated, the samples were subsequently incubated with mouse anti-TH (Immunostar, dilution 1:100) with rabbit anti-5-HT (Sigma-Aldrich, dilution 1:500) or for corroboration of the results, with other 5-HT antibody (Supplementary Figure S1; rat anti-5HT, Merck, 1:100) diluted in block solution at 18°C for at least 6 h. After washing the samples at least three times for 20 min each with PBS containing 0.6% Triton X-100, the brains were incubated at 4°C overnight in secondary antibodies diluted in PBS containing 0.6% Triton X-100. Secondary antibodies were anti-mouse Alexa 488-conjugated (Invitrogen, 1:300) or anti-mouse Cy3-conjugated (Jackson, 1:300). Samples were then washed again three times in PBS containing 0.6% Triton X-100, incubated for at least 6 h in PBS, and mounted in Vectashield (Vector Laboratories). Images were taken using a Leica SP8 confocal microscope equipped with a Leica Apochromat 20×/0.7 water immersion objective. The brains were scanned at 1.0 μm steps in the z-axis with a resolution of 1.0 μm/pixel. Images were analyzed using ImageJ. To determine fluorescence intensities, scans were transformed to z-projections and analyzed as described in Neckameyer and Bhatt (2012).

Flies were incubated for 5–10 days at 25°C on standard fly food containing 1 mg/mL L-DOPA (D9628, Sigma-Aldrich). Test and control flies were transferred to fresh food of the according regimen every second day.

To visualize the differential innervation of serotoninergic and dopaminergic neurons in the Drosophila central brain, we stained with antisera to 5-HT, and to TH, in conjunction with TH-Gal4- (Friggi-Grelin et al., 2003) and Trh-Gal4- (Cassar et al., 2015) driven GFP in DA and 5-HT neurons, respectively (Figures 1A,B). In agreement with previously published observations (Monastirioti, 1999), the anterior dopaminergic system of the adult brain mainly consists of three neuronal clusters, including neurons situated laterally in the anterior protocerebrum (PAL), a small group of neurons located in the lateral and medial parts of the subesophagal zone (SEZ) and about 100 neurons in the medial protocerebrum (PAM). In several aspects, the neurons of the PAM cluster differ from the other clusters of the DA system. PAM cluster neurons are the last born DA producing neurons and develop only during pupation. They have much smaller somata and they are the only DA producing neurons of the CNS that are for most (~85%) of them not included in the expression pattern of the TH-Gal4 line, indicating that these neurons also differ at the level of gene regulation. Being the only neurons of the anterior DA system innervating the mushroom body (MB), these neurons send their projections towards the tips of the γ-lobes, the β’-lobes and the shaft of the β-lobes (Figure 1A1, yellow arrow heads; see also Pech et al., 2013). The posterior DA system of the adult brain consists of the PPL1, PPL2, PPM1/2, PPM3, a group of small neurons located lateral to the SEZ, and a group of large neurons in the medial part of the SEZ. The neurons of the posterior cluster, positioned lateral to the MB calyx (PPL1), densely innervate the heel and the vertical lobes of the MB. The cluster located between the CNS and the optic lobes (OL) innervate the lobula plate at the ipsilateral side and send their projections to the lobula plate of the contralateral OL (Figure 1A2).

The dopaminergic system of the adult thoracic nerve cord consists of dense innervations deriving partially from the CNS and projecting to all three segments, as well as from groups of dopaminergic neurons positioned laterally (Figure 1A3, ThL) or medially between the first and second (Figure 1A3, Th) and between the second and the third segment, as well as between the third thoracal segment and the abdominal ganglion (AG; Figure 1A3, ThL). The dopaminergic neurons of the AG are positioned laterally to the ganglion and send their projection to the tip of the AG (Figure 1A3, AbL), where a second group of DA producing neuron in the AG is positioned (Figure 1A3, AbU). The central DA system of third-instar (LIII) larvae is largely comparable to the one of adult flies, with the exception of the PAM cluster that consists of only four neurons per hemisphere in LIII-larvae and counts about 100 neurons in adults. Despite the much simpler anatomy of the larval DA system, similar functions of these neurons were described for adults and larvae (Rohwedder et al., 2016). In the larval ventral ganglion, lateral DA neurons send long projections to the neuropil where they form lateral, longitudinal bundles and from whence they project towards the medial part of the ventral ganglion, whereas medially positioned DA neurons with short projections form a medial, longitudinal bundle, projecting to the lateral parts of the ventral ganglion (Figure 1A4).

The CNS of adult Drosophila is densely innervated by ~90 5-HT producing interneurons that can be subdivided in 10 clusters (Vallés and White, 1988; Sitaraman et al., 2008; Alekseyenko et al., 2010; Sadaf et al., 2012; Pech et al., 2013; Pooryasin and Fiala, 2015; Figures 1B1,B2). Out of these ~90 5-HT producing neurons about 80 neurons can be targeted by TrH-Gal4 (Sitaraman et al., 2008; Cassar et al., 2015) that however drives ectopic expression in nearly 170 non-serotoninergic neurons (Figure 1B; see also Pooryasin and Fiala, 2015). Similar to vertebrates, where neuropils that are innervated by DA producing neurons are also innervated by 5-HT neurons (Niederkofler et al., 2015), with the exceptions of the antennal lobes (AL) that are innervated by 5-HT producing neurons but only sparsely at the outer rim by DA neurons (Figures 2A1–3) and the protocerebral bridge (PB) that is innervated by DA neurons but not by 5-HT producing neurons (Figure 2E1), both aminergic systems innervate the same target regions, but differ in their density on local innervation patterns. At the anterior side, the serotoninergic system can be subdivided in a medial (AMP) and a dorsomedial (ADMP—anterior dorso medial protocerebrum) cluster and a small group of neurons positioned in the anterior lateral protocerebrum (ALP) and a large of group of neurons positioned between the OL and the CNS in the lateral protocerebrum (LP). The SEZ is innervated at the anterior side by a cluster of large lateral neurons (SEL) and a group of small medial neurons (SEM), whereas at the dorsal side a group of small neurons is positioned laterally, and a group of three neurons is located at the medial part (SEM) in direct vicinity to a group of large DA producing neurons (Pooryasin and Fiala, 2015; Figure 1B1). At the posterior side, the 5-HT producing neurons can be classified in a medially positioned cluster that can be further subdivided in the posterior medial dorsal (PMPd), posterior medial (PMPm) and posterior ventral cluster (PMPv). In the LP a group of two neurons with strong 5-HT immunoreactivity (IR) forms the posterior lateral protocerebral cluster (LP; Pooryasin and Fiala, 2015; Figure 1B2).

The thoracic ganglion (TG) is strongly innervated by 5-HT neurons partially deriving from the CNS and projecting to all three segments as well as from groups of neurons positioned medially between the first and second (Figure 1B3, PR) and between the second and the third segment Figure 1B3, MS). The 5-HT neurons of the AG are positioned laterally and medially in the ganglion sending their projection to the tip of the AG (Figure 1B3, AB, MT). The 5-HT system of LIII larvae has been described in detail (Vallés and White, 1988; Huser et al., 2012) and consists mainly of four clusters in the brain hemispheres, and four clusters in the SEZ. In the larval ventral ganglion 5-HT neurons with short projections innervate the neuropil at the ipsilateral side and 5-HT neurons with long projections innervate the contralateral side or both sides of the ganglion (Figure 1B4; Huser et al., 2012).

Both the AL (Figures 2A1–3) and MB (Figures 2A1–3,B1,B2) are contacted by fine 5-HT terminals. In agreement with Vallés and White (1988), who used a different antibody, we find that in the CC, a structure composed of the ellipsoid body (Figures 2B1,B2, EB), the fan-shaped body (FB; Figures 2C1,C2, FB), the noduli (Figures 2D1,D2, NO) and the PB (Figures 2E1,E2, PB), 5-HT-producing neurons predominantly send their projections to the inner rim of the EB (Figures 2E1,E2) and to the superior arch of the FB (Figures 2D1,D2), but only sparsely into the noduli (Figures 2D1,D2). With the exception of the AL, which is only sparsely innervated by TH immunoreactive neurons at the outer rim and the PB that is exclusively innervated by DA producing neurons (Figures 2E1,E2), DA and 5-HT producing neurons project to the same neuropils, but differ in their exact innervation characteristics within their target neuropils. Whereas the MB is sparsely, but homogeneously innervated by the 5-HT producing dorsal pair medial (DPM) neuron, DA producing neurons group in different sub-populations that form dense and spatially restricted innervations onto characteristic sub-regions within the MB lobes (Figures 2B1,B2; see also Pech et al., 2013), consistent with immunocytochemical pattern observed with an antibody to DA (Cichewicz et al., 2016). The terminal projections of DA-producing neurons into the CC are mainly complementary to those of 5-HT producing neurons. TH immunoreactive neurons innervate densely the outer rim of the EB (Figures 2C1,C2) and the medial parts of the FB (Figures 2D1,D2), whereas 5-HT projections are concentrated to the inner and anterior part of the EB (Figures 2C1,C2) and the dorsal FB (Figures 2D1,D2).

In summary, we find in the logic underlying the innervation pattern of the DA and 5-HT system similarities to the innervations characteristics of aminergic systems in vertebrates. DA- and 5-HT-producing neurons innervate large portions of the CNS, with largely overlapping target neuropils. With the exception of the AL and the PB, both 5-HT- and TH-immunoreactive neurons send their projections to the same neuropils, but differ mainly in the densities of their innervation within sub-region of the particular neuropil.

To analyze the impact of DA loss on the serotoninergic system, we compared the anatomy of flies lacking dTH in the CNS, compared to dTH-rescue flies and w¹¹¹⁸ control flies. We find that the overall appearance of the serotoninergic system of flies lacking DA is comparable to that of dTH rescue flies or w¹¹¹⁸ controls. However, brains lacking genomic dTH show an increased number of 5-HT immunoreactive somata within the posterior lateral protocerebrum (PLP; Figures 3A–H, PPL). In control w¹¹¹⁸ 5-HT positive neurons in the PLP were described as positioned ventrally to the calyces (Figures 3A–D), near the OL (Figure 3D; Vallés and White, 1988; Monastirioti, 1999; Sitaraman et al., 2008; Cassar et al., 2015; Pooryasin and Fiala, 2015). Giang et al. (2011) further have identified an additional group neurons with faint IR against the 5-HT transporter positioned laterally to the calyces similarly positioned then the PPL1 dopaminergic neurons. In wild-type brains, a small number of TH IR neurons within the PPL1 cluster show very faint 5-HT IR (Figures 3A–C). However, in DA-deficient brains, we find an increased 5-HT IR of neurons juxta-calycal in the PLP (Figure 3H, PPL) at the position of the PPL1 DA producing neurons (Nässel and Elekes, 1992; Monastirioti, 1999; Riemensperger et al., 2013). This increased IR we could verify with two different antibodies against 5-HT (Supplementary Figure S1). However, it is not clear whether the increased 5-HT IR in DA deficient flies derives from the same neuronal populations. Other clusters, like the PMP clusters (Figure 3D), did not reveal any changes in number of visible somata of 5-HT IR. To exclude the possibility that the observed increase in number of these PPL1-like 5-HT neurons observed in DA-deficient brains does not derive from globally increased 5-HT production, we quantified 5-HT IR between DA-deficient flies, rescue flies and w¹¹¹⁸ control flies in the lateral protocerebrum (LP), PLP, PMPd, PMPm and PMPv 5-HT neuron cluster. With the exception of the LP cluster that showed decreased 5-HT IR in dTH-rescue when compared to w¹¹¹⁸ flies, all three strains showed comparable levels of 5-HT IR in the somata of the PMP neurons (Figures 4A–F). It thus appears that increased 5-HT synthesis in the 5-HT/DA PPL1 neurons in the PLP is a selective response occurring in this neuronal cluster that has the capability of increased re-uptake or synthesis of either or both transmitters.

In vertebrate models of PD, degeneration of DA neurons leads to modifications in 5-HT neurons (Zhou et al., 1991; Rylander et al., 2010; Zeng et al., 2010; Niederkofler et al., 2015). To determine whether the same holds true for Drosophila, we analyzed the innervation pattern of 5-HT neurons onto their target region in the vertical MB lobes. The tips of the vertical α- and α’-lobes are densely innervated by both TH- and 5-HT-IR neurons. Whereas TH-IR neurons innervate both regions with similar intensities (Figures 2E1,E2), in wild-type brains, 5-HT neurons mainly send projections towards the α’-lobes, but only faintly to the α-lobe (Figures 2D1,D2, 5A). However, in the DA-deficient brains, there is a strong increase in IR for 5-HT in the α-lobes, but not in the α’-lobes (Figure 5B). This increase in 5-HT-immunoreactive projections onto the α-lobes results in a shift in the proportion of 5-HT positive projections between the two lobe structures (Figure 5C). However, it does not affect the overall size in terms of area surface of the innervated neuropils (Figure 5D).

We next asked whether enhanced DA levels, attained by feeding wild-type flies L-DOPA, would affect the serotoninergic system in the opposite way than lack of CNS DA. To this end we analyzed the IR of 5-HT projections onto their target region in the vertical MB lobes after 10 days L-DOPA treatment. 5-HT IR was significantly decreased in both, α- (Figures 6A,A1), and α’-lobes (Figures 6A,A2) after L-DOPA treatment when compared to control flies. To determine whether the observed reduction in 5-HT IR was caused by altered 5-HT biosynthesis or altered plasticity we used the splitGFP technique (Gordon and Scott, 2009; Pech et al., 2013) to visualize alteration in connectivity between 5-HT producing neurons and the MB vertical lobes. Adult flies expressing one part of the splitGFP in 5-TH neurons and the counterpart in MB Kenyon neurons were fed for 10 days with L-DOPA, and monitored for changes in 5-HT GFP fluorescence. Whereas the general pattern of 5-HT IR appeared largely unchanged and comparable to what has been published before (Pech et al., 2013), we found that certain 5-HT projections were significantly decreased (Figure 6B), as can be seen for the intensity of the signal of reconstituted splitGFP between 5-HT terminals and MB lobes at the tip of the α- (Figures 6B,B1) and α’-lobes (Figures 6B,B2). Thus, 5-HT neurons sharing the same target regions than DA neurons respond with diminished projections with enhanced DA, and enhanced 5-HT IR in the absence of DA. These observations provide evidence of competitive interactions between 5-HT and DA in the Drosophila brain.