The Drosophila transmembrane semaphorin-1a (Sema-1a) belongs to the Sema protein family, which is a major group of axon guidance molecules (Dickson, 2002; Pasterkamp, 2012). During embryonic neural development, Sema-1a functions not only as a ligand for PlexA (“forward signaling”), but also as a receptor for unknown ligands (“reverse signaling”) (Jeong et al., 2012; Yu et al., 1998). In the peripheral nervous system (PNS), bidirectional signaling of Sema-1a plays an essential role in motor axon pathfinding through mediating axon-axon repulsion (Jeong et al., 2012; 2017). In the central nervous system (CNS), however, Sema-1a-mediated forward signaling largely contributes to pathfinding of the outermost FasII-positive axons (Jeong et al., 2012), whereas Sema-1a-mediated reverse signaling through an attractive guidance mechanism is required for midline crossing of commissural neurons (Hernandez-Fleming et al., 2017). Furthermore, the receptor function of Sema-1a also mediates diverse cellular responses in different areas of the Drosophila nervous system, including the giant fiber, visual, and olfactory systems (Cafferty et al., 2006; Godenschwege et al., 2002; Komiyama et al., 2007; Pecot et al., 2013; Sweeney et al., 2011). Defining these cell-type specific and context-dependent functions of Sema-1a will provide a basis for the understanding of mechanisms underlying precise and complex neural circuit formation. To this end, identification of neurons that express native Sema-1a proteins should be an important prerequisite for defining the cell-type specific functions of Sema-1a. Previous in situ hybridization analysis showed that Sema-1a mRNA is robustly detected in a large subset of neurons in the embryonic CNS (Kolodkin et al., 1993). In line with this finding, immunohistochemical staining results showed that Sema-1a expression is mainly found on most commissural and longitudinal axons in the CNS and on five motor nerve branches (ISN, ISNb, ISNd, SNa, and SNc) in the PNS (Yu et al., 1998). However, these two approaches are not suitable for investigating the cellular function of Sema-1a. For the study of cell-type specific neuronal functions of Sema-1a, it should instead be helpful to identify cis-regulatory elements (CREs) that govern the native Sema-1a expression patterns, and also to have GAL4 lines whose expression is driven by the Sema-1a CREs.

Here, we identified three GAL4 lines, R34A03-GAL4, R32H10-GAL4, and R33F06-GAL4 that contain the intronic DNA fragments of Sema-1a that govern reporter expression on motor nerve branches as well as in a large subset of neurons in the CNS. We observed that highly penetrant guidance phenotypes of Sema-1a mutants in the CNS and PNS are significantly rescued by expression of a wild-type Sema-1a transgene driven by each of these GAL4 lines. These complementation results suggest that R34A03, R32H10, and R33F06 DNA elements redundantly serve as CREs generating the native Sema-1a expression patterns during embryonic neural development.

Drosophila stocks

The white mutant allele, w¹¹¹⁸, was used as a wild-type control. The following GAL4 lines were obtained from the Bloomington Stock Center: R90A09-GAL4 (#46907), R32H10-GAL4 (#45212), R33G04-GAL4 (#48116), R34E08-GAL4 (#45221), R34A12-GAL4 (#45217), R34D11-GAL4 (#45220), R33G05-GAL4 (#47346), R34B01-GAL4 (#45605), R33C07-GAL4 (#46194), R33G11-GAL4 (#45964), R33F11-GAL4 (#47545), R34A03-GAL4 (#48120), R33F06-GAL4 (#48115), R33G02-GAL4 (#47546), R32G03-GAL4 (#45591), and R33B07-GAL4 (#45214). All other fly stocks were described previously: Sema-1a^PI (Yu et al., 1998); UAS_B-Sema-1a (Jeong et al., 2012); UAS-taumycEGFP (Callahan et al., 1998); Sca-GAL4 (Klaes et al., 1994).

Immunohistochemistry

To precisely examine the expression pattern of each GAL4 line in the PNS and CNS, each GAL4 strain was crossed to UAS-TaumycGFP flies. The resulting embryos were immunostained with anti-myc (9E10; Sigma, USA) or anti-GFP (Invitrogen, USA) antibodies and the peroxidase-3,3'-diaminobenzidine reaction (Jeong et al., 2012). To visualize the axonal projection patterns of motor neurons and a small subset of CNS neurons during embryonic neural development, we did immunostaining using anti-FasII antibody, embryo dissection, and imaging as described previously (Jeong, 2017).

Phenotypic characterization

The filleted preparation method was used to better visualize and characterize FasII-positive axon guidance phenotypes observed in the CNS and PNS (Jeong, 2017). The detailed assessment of CNS and motor axon guidance defects were performed as described previously (Van Vactor et al., 1993; Yu et al., 1998). The chi-square (χ²) statistic with a 2 × 2 contingency table was used to assess independence between two categorical variables.

Expression patterns of the GAL4 lines that are driven by flanking and intronic DNA fragments of Sema-1a

To identify CREs that are required to direct Sema-1a expression during embryonic neural development, we took advantage of a collection of GAL4 lines driven by flanking and intronic DNA fragments of Sema-1a (Supplementary Fig. S1). These GAL4 lines were generated by inserting each GAL4 construct into the same genomic location to minimize position effects (Pfeiffer et al., 2008). By crossing to UAS-GFP:NLS, which encodes a nuclear localized GFP protein, the expression patterns of most GAL4 lines were characterized in stage 16 embryos and are available online (Manning et al., 2012). These expression patterns visualized by nuclear localization of GFP reporter protein are clearly valuable for identifying tissue-specific expression patterns, quantification of cells expressing the reporters, and neuronal morphology (Manning et al., 2012). However, in the embryonic CNS, it is hard to identify the cell bodies of motor neurons that extend their axons out of the CNS to form the five motor nerve branches. Therefore, based on their available expression patterns and genomic organization of the Sema-1a gene that contains one ncRNA, one 297{}327 transposable element, and two different tRNA genes in the intronic region, we selected a collection of 16 GAL4 lines for further analysis (Fig. 1). To visualize the axonal expression pattern of each GAL4 line in the PNS as well as in the CNS, we crossed each line to a UAS-TaumycGFP reporter, which is primarily localized to the axonal compartment, and then stained the resulting embryos with anti-myc or anti-GFP antibodies (Jeong et al., 2012). We obtained reporter expression data derived from a collection of 16 GAL4 lines (Fig. 2, Supplementary Fig. S2). Among them, we found that R34A03-GAL4 and R32H10-GAL4 drove reporter expression in ISNb, SNa, and ISN nerve branches, although they also showed expression in lateral bipolar dendrite neurons (LBDs) or sensory neurons in which endogenous Sema-1a proteins were not detected (Figs. 2A and 2B) (Yu et al., 1998). Furthermore, in the R33F06-GAL4 line we observed irregular reporter expression on ISNb or SNa nerve branches in some abdominal hemisegments, consistent expression in LBDs, and also irregular expression in some muscles and unidentified cells (Fig. 2C). In addition, these three GAL4 lines drove robust reporter expression in large subsets of cells in the CNS (Figs. 2A-2C). Given that endogenous Sema-1a is mainly expressed on most commissural and longitudinal axons in the CNS and five motor nerve branches in the PNS (Yu et al., 1998), these findings may indicate that R34A03-GAL4, R32H10-GAL4, and R33F06-GAL4 contain CREs that are responsible for neuronal Sema-1a expression in the PNS and CNS. We also found that the other GAL4 lines barely drove reporter expression in ISNb and SNa nerve branches (Figs. 2D-2F, Supplementary Fig. S2). However, these GAL4 lines showed expression in subsets of cells in the CNS except R32G03-GAL4, in which reporter expression was hardly detected (Figs. 2C-2F, Supplementary Fig. S2). As described previously (Manning et al., 2012), in these GAL4 lines we also observed reporter expression in the body wall, including muscles, sensory neurons, and other cell types (Figs. 2D-2F, Supplementary Fig. S2). Furthermore, we found that R34A12-GAL4 and R33F11-GAL4 showed two different expression patterns, probably because of either multiple insertions of a transgene or line contamination (Supplementary Figs. S2C, S2D, S2H, and S2I).

Rescue of motor axon guidance phenotypes observed in Sema-1a mutants with a selected GAL4 lines

To find out whether R34A03-GAL4, R32H10-GAL4, and R33F06-GAL4 drive the Sema-1a expression required for the axon guidance function of Sema-1a during embryonic neural development, we expressed a Sema-1a transgene using either GAL4 line in Sema-1a^PI loss-of-function mutants. In wild-type embryos, ISNb motor neurons, whose cell bodies reside in the ventral nerve cord, extend their axons in a fascicle toward the ventrolateral muscle region and finally innervate muscles 6, 7, 12, 13, 14, 28, and 30 (Fig. 3A) (Landgraf and Thor, 2006; Van Vactor et al., 1993). Normally, SNa motor axons innervate lateral muscles 5, 8, 21, 22, 23, and 24 (Fig. 4A) (Landgraf and Thor, 2006; Van Vactor et al., 1993). In both ISNb and SNa pathways, selective defasciculation of motor axons at specific choice points is required for correct innervation of their target muscles (Figs. 3A and 4A) (Jeong, 2017). Sema-1a^PI mutants exhibited highly penetrant guidance defects in the ISNb and SNa pathways (98.3% and 76.1% of hemisegments, respectively; Figs. 3B, 3C, 4B, and 4C) (Yu et al., 1998). This can be explained by the finding that bidirectional signaling of Sema-1a is required for selective defasciculation of ISNb and SNa motor axons at specific choice points (Jeong et al., 2012; 2017). Interestingly, neuronal expression of Sema-1a using either R34A03-GAL4 or R32H10-GAL4 robustly restored the ISNb (37.8% and 28.1% of hemisegments, respectively; Fig. 3C) and SNa guidance defects (30.4% and 28.0% of hemisegments, respectively; Fig. 4C) observed in Sema-1a^PI mutants, consistent with their expression that recapitulates the native Sema-1a expression pattern in the PNS (Figs. 2A and 2B) (Yu et al., 1998). A similar rescue effect was also observed in Sema-1a^PI mutant embryos expressing Sema-1a under the control of Sca-GAL4, which drives expression in all neuroblasts and their progeny (34.8% and 23.9% of hemisegments, respectively; Figs. 3C and 4C) (Klaes et al., 1994). In addition, expression of Sema-1a using R33F06-GAL4 partially restored both the ISNb and the SNa guidance defects of Sema-1a^PI mutants (74.3% and 44.3% of hemisegments, respectively; Figs. 3C and 4C), reflecting the fact that irregular and weak reporter expression was detected on ISNb or SNa nerve branches (Fig. 2C). These data strongly suggest that DNA fragments R34A03, R32H10, and R33F06 govern redundantly expression of endogenous Sema-1a responsible for its motor axon guidance function. Furthermore, we also examined the ability of other GAL4 lines, such as R90A09-GAL4, R33C07-GAL4, and R32G03-GAL4, to rescue the PNS guidance defects of Sema-1a mutants. As expected, any of these GAL4 lines did not significantly rescue the ISNb as well as SNa guidance defects observed in Sema-1a mutants (Figs. 3C and 4C). These complementation results are concordant with our observation that reporter expression driven by these GAL4 lines was barely detected in both ISNb and SNa nerve branches (Figs. 2D-2F).

Rescue of CNS axon guidance phenotypes observed in Sema-1a mutants with a selected GAL4 lines

Since R34A03-GAL4, R32H10-GAL4, and R33F06-GAL4 drove robust expression in large subsets of cells in the CNS (Figs. 2A and 2B), we next assessed these GAL4 lines for their ability to rescue CNS axon guidance defects observed in Sema-1a mutants. Immunohistochemistry with anti-FasII antibody visualizes three longitudinal axon fascicles on each side of the midline in the wild-type CNS (Fig. 5A) (Grenningloh et al., 1991). Frequently observed disruptions in the outermost FasII-positive fascicles of Sema-1a mutants (70.7% of hemisegments; Figs. 5B and 5C) (Yu et al., 1998) were robustly rescued by neuronal expression of Sema-1a using either R34A03-GAL4 (2.3% of hemisegments; Fig. 5C), R32H10-GAL4 (7.7% of hemisegments; Fig. 5C), or R33F06-GAL4 (0.0% of hemisegments; Fig. 5C). Along with the fact that neuronal expression of Sema-1a using Sca-GAL4 restored the CNS guidance defects of Sema-1a mutants to wild-type levels (Fig. 5C), these observations demonstrate that DNA fragments R34A03, R32H10, and R33F06 control redundantly expression of Sema-1a required for the axon guidance function of Sema-1a in the embryonic CNS. We also examined the same set of GAL4 lines, R90A09-GAL4, R33C07-GAL4, and R32G03-GAL4, which were used in rescue of the PNS guidance defects (Figs. 2D-2F), for their ability to recover the CNS guidance phenotypes of Sema-1a mutants and found that any of these GAL4 lines did not significantly rescue the CNS guidance defects (Fig. 5C). Taken together, our results suggest that three DNA fragments, R34A03, R32H10, and R33F06, serve as CREs that govern Sema-1a expression required for the axon guidance function of Sema-1a in the CNS as well as the PNS.

Here, we provide evidence that endogenous Sema-1a expression is controlled by multiple CREs R34A03, R32H10, and R33F06 in a redundant manner during embryonic neural development. First, both ISNb and SNa motor axon guidance defects seen in Sema-1a null mutants are robustly rescued by the expression of wild-type Sema-1a under the control of either R34A03-GAL4 or R32H10-GAL4. Consistently, these CREs drove strong reporter expression on both ISNb and SNa nerve branches. We also found that the PNS axon guidance phenotypes of Sema-1a mutants are partially rescued when Sema-1a expression is reintroduced under the control of R33F06-GAL4. This could be explained by the fact that in an irregular manner, we detected low levels of reporter expression on ISNb or SNa nerve branches in transgenic reporter assays using R33F06-GAL4. Moreover, we previously observed mild but significant motor axon guidance defects in embryos transheterozygous for the Sema-1a null allele, which would be expected to have an approximately 50% reduction in the expression of Sema-1a, indicating haploinsufficiency of Sema-1a (Jeong et al., 2012; 2017). Taken together, our findings demonstrate that Sema-1a expression on both ISNb and SNa nerve branches is regulated by multiple redundant CREs such as R34A03, R32H10, and R33F06. Second, Sema-1a null mutation leads to frequent disruptions in the outermost FasII-positive fascicles in the CNS (Figs. 5B and 5C), indicative of a critical role in CNS axon guidance. This CNS guidance phenotype was rescued to wild-type levels in Sema-1a mutant embryos expressing Sema-1a under the control of R33F06-GAL4. We also found that the CNS axon guidance defects of Sema-1a robustly but incompletely rescued by either R34A03-GAL4 or R32H10-GAL4. These results strongly suggest that R34A03, R32H10, and R33F06 serve as CREs to redundantly govern endogenous Sema-1a expression required for the FasII-positive axon guidance function of Sema-1a in the embryonic CNS.

It has been reported that there are usually four developmental enhancer elements per protein-coding gene in the Drosophila genome on average during embryonic development (Kvon et al., 2014). A growing body of evidence suggests that functional redundancy of multiple CREs is a common feature of gene-regulatory networks in vertebrate and invertebrate development (Frankel et al., 2010; Montavon et al., 2011; Osterwalder et al., 2018; Perry et al., 2010) and further demonstrates that functional redundancy contributes to precise spatiotemporal patterns and robustness of gene expression in response to environmental or genetic variation (Barolo, 2012; Cannavò et al., 2016). In addition, changes within multiple CREs with overlapping expression activities have been proposed to provide better opportunities for evolutionary innovations especially for highly pleiotropic genes (Barolo, 2012; Prud'homme et al., 2007). Given that the pleiotropic Sema-1a gene plays a critical role in both motor neuron and CNS axon guidance during neural development (Yu et al., 1998), the multiple redundant CREs of Sema-1a that we have identified here should collaborate to precisely regulate the spatiotemporal patterns and levels of endogenous Sema-1a. Furthermore, R34A03-GAL4 and R33F06-GAL4 lines have been reported to drive reporter expression in some overlapping regions of the adult brain including the antenna lobe (AL), lateral horn (LH), and mushroom body (MB) (Jenett et al., 2012; image data are found at Janelia’s FlyLight site). Interestingly, Sema-1a is required for axon targeting of olfactory receptor neurons and dendritic targeting of olfactory projection neurons (PNs) in the AL as well as axon targeting of PNs to the appropriate areas of the MB and the LH (Komiyama et al., 2007; Sweeney et al., 2007). These suggest that R34A03 and R33F06 also have enhancer redundancy for governing the expression of Sema-1a in the adult brain. In contrast, R32H10-GAL4 line displayed broad reporter expression in the adult brain and ventral nerve cord, but may not be in the brain regions such as AL, LH, MB (Jenett et al., 2012; image data are found at Janelia’s FlyLight site). Taken together, these results demonstrate that the CREs, R34A03, R32H10, and R33F06, show similar, overlapping, or unique enhancer activities for controlling Sema-1a expression in the embryonic and adult nervous systems.

Since transmembrane Sema-1a functions as a ligand (forward signaling) and/or as a receptor (reverse signaling) to mediate axon-axon repulsion at specific choice points in the PNS (Jeong et al., 2012; 2017; Yu et al., 1998), one important question related to Sema-1a expression is which motoneurons express native Sema-1a proteins during embryonic development. Interestingly, we found that the reporter expression patterns of R34A03-GAL4 on ISNb and SNa nerve branches (Fig. 2A) are very similar to the axonal projection patterns of ISNb and SNa stained with anti-FasII antibodies (Fig. 3A). These observations may indicate that Sema-1a is expressed in most ISNb and SNa motor neurons. There is one caveat to this interpretation because the reporter expression of R34A03-GAL4 is also observed in lateral bipolar dendrite neurons where endogenous Sema-1a is barely detected (Fig. 2A) (Yu et al., 1998). Furthermore, R32H10-GAL4 and R33F06-GAL4 drove reporter expression in unanticipated sensory neurons and muscle fibers, respectively (Figs. 2B and 2C). These unanticipated reporter expression patterns of individual transgenic CREs might result from the absence of repressive element(s) located within neighboring CREs, which restrict aberrant CRE activity in the genome (Dunipace et al., 2011; Perry et al., 2011). Another interesting question is how Sema-1a-mediated selective repulsion between motor axons can be achieved if most motor neurons in the ISNb and SNa nerve branches express native Sema-1a proteins. One possible mechanism for the selective activation of Sema-1a repulsive signaling could be through the differential recruitment of co-receptors and/or intracellular signaling mediators (Jongbloets and Pasterkamp, 2014; Pasterkamp, 2012).

The authors have no potential conflicts of interest to disclose.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MIST) (NRF-2018R1A2B6008037 and NRF-2017M3C7A1044816). We thank Alex Kolodkin and the Bloomington Drosophila Stock Center for the fly stocks.